IN F O R M A T IO N TO USERS This was produced from a copy of a document sent to us for microfilming. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality o f the material submitted. The following explanation of techniques is provided to help you understand markings or notations which may appear on this reproduction. 1. The sign or "target” for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing page{s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure you of complete continuity. 2. When an image on the film is obliterated with a round black mark it is an indication that the film inspector noticed either blurred copy because of movement during exposure, or duplicate copy. Unless we meant to delete copyrighted materials that should not have been filmed, you will find a good image of the page in the adjacent frame. If copyrighted materials were deleted you will find a target note listing the pages in the adjacent frame. 3. When a map, drawing or chart, etc., is part of the material being photo­ graphed the photographer has followed a definite method in "sectioning” the material. It is customary to begin filming at the upper left hand corner of a large sheet and to continue from left to right in equal sections with small overlaps. If necessary, sectioning is continued again—beginning below the first row and continuing on until complete. 4. For any illustrations that cannot be reproduced satisfactorily by xerography, photographic prints can be purchased at additional cost and tipped into your xerographic copy. Requests can be made to our Dissertations Customer Services Department. 5. Some pages in any document may have indistinct print. In all cases we have filmed the best available copy. Universe Microfilms International 3 00 N ^ E E B R D , A N N A R B O R . Ml 4 81 06 8202462 K e l l e r , W il l ia m Jo h n EFFECTS O F TH E EARLY A N D PERIODIC SCREENING. DIAGNOSIS A N D TREATM ENT (EPSDT) PROGRAM ON TH E HEALTH STATUS OF PARTICIPANTS IN M IC H IG A N Michigan State University University Microfilms International Ph.D. 300 N. Zeeb Road, Ann Arbor, M I 48106 Copyright 1981 by Keiler, William John All Rights Reserved 1981 EFFECTS OF THE EARLY AND PERIODIC SCREENING, DIAGNOSIS ANO TREATMENT (EPSDT) PROGRAM ON THE HEALTH STATUS OF PARTICIPANTS IN MICHIGAN By W illiam J. K elle r A DISSERTATION Submitted to Michigan State U niversity in p a rtia l f u lfillm e n t o f the requirements fo r the degree of DOCTOR OF PHILOSOPHY College o f Social Science 1981 ABSTRACT EFFECTS OF THE EARLY AND PERIODIC SCREENING, DIAGNOSIS AND TREATMENT (EPSDT) PROGRAM ON THE HEALTH STATUS OF PARTICIPANTS IN MICHIGAN By W illiam J. K eller Since 1973 the federal government has required th at each state o ffe rin g a T it le XIX (Medicaid) program w ill also o ffe r the Early and Periodic Screening, Diagnosis and Treatment (EPSDT) program to Medicaid e llg lb le s under the age o f twenty-one years. The purpose of th is study was to determine whether there are indications that EPSDT is b e n e fittin g particip ants 1n Michigan. Two outcome measures were used to assess program e ffe c ts : ral rates and (2) medical costs. (1) r e fe r ­ The primary Independent variab le was the number of life tim e EPSDT screenings received. The general relatio n sh ip tested was whether r e fe rr a l rates and costs vary inversely w ith program p a rtic ip a tio n . A computer-based study was designed to te s t these relationships and two populations of c lie n ts were selected. One consisted o f c lie n ts con­ tinuously e lig ib le fo r EPSDT between January 1, 1974 and December 31, 1979 and numbered 79,754. The other population consisted o f those e lig ib le fo r calendar year 1979 and numbered 245,551. A search o f the EPSDT master f i l e o f 535,753 screening summaries determined the re fe rra l rate a t the la s t (most recent) screening. 56,046 of the former group and 154,187 o f the la t t e r had been screened. 11 W illiam J. K elle r Results showed re fe rra ls decreased ten percent or less between screenings one-five each, given a te s t group size of one hundred or more subjects. Medicaid costs were not found to be Inversely related to life tim e screenings but when Medicaid costs of a ll EPSDT particip ants were compared with the Medicaid costs o f the EPSDT nonpartlcipants, the particip ants showed s t a t is t ic a lly s ig n ific a n t lower costs. The contin­ uously e lig ib le group incurred $26.18 less per person (p _< .0 5 ) .the oneyear e lig ib le s incurred $46.52 less per person (p <_ .0 0 7 ). However, when costs of the screening program i t s e l f were also considered, differences favoring the particip ants were replaced by somewhat greater costs a ttr ib u ­ table to program p a rtic ip a tio n . Other major findings were: (1) R eferrals had decreased annually a t the rate o f approximately eight percent per year. (2 ) Referral rates average nearly f i f t y to one hundred percent higher In D e tro it than in r u r a l, outstate Michigan, w ith race held constant. (3) Blacks have r e fe r ­ ral rates 20-23 percent higher than whites but black EPSDT particip ants show lower medical costs whereas white p articip an ts do not. The study concluded th at the program is achieving modest gains at modest costs. ACKNOWLEDGEMENTS Many people contributed to th is study. most app reciative. committee: To a ll of them, I am I wish to thank s p e c ific a lly the members o f my W illiam Ewens, John H errick, Charles Johnson,and espec­ i a l l y V ic to r Whiteman, chairperson. I am p a rtic u la rly indebted and gratefu l to members of the Survey Design and Analysis D iv is io n , Quality Assurance, Michigan Department of Social Services. Without th e ir co­ operation and in te re s t, th is study would tr u ly not have been possible. They are: Charles Conley, D irecto r; Bruce Burke, Fred C lark, Robert L o v e ll, Joanne H e lfric h and Fred S lid e r. Appreciation 1s also ex­ tended to Steve Bachleda and C arlo tta Devereaux fo r comments and e d itin g . As always, fin a l re s p o n s ib ility fo r th is study and its con­ clusions remains with its author. vi DEDICATION To my fam ily and e s p e c ia lly to my mother who gave so much fo r my education. v TABLE OF CONTENTS Page LIST OF TABLES........................................................................................ 1x LIST OF F IG U R E S ....................................................................................... x x ili Chapter I. INTRODUCTION ............................................................................. 1 II. REVIEW OF LITERATURE............................................................ 11 Background on E P S D T ....................................................... Outcome Studies o f NonEPSDT Screening Programs . Outcome Studies o f the EPSDT Program ..................... R elationship o f Demographic Factors to Health . 11 21 29 38 RESEARCH AND DESIGN METHODLOGY ....................................... 46 O bjective o f S t u d y ........................................................... Hypotheses............................................................................ Q u e s t io n s ............................................................................ Design I R eferral Rate Differences ......................... Table I / I I R e p lic a t io n s .............................................. Design I S ta t is t ic a l Analysis ................................. Design I I Cost D iffe r e n c e s ......................................... Design I I S ta t is t ic a l Analysis ................................. Design I I I Short-Run Cost Differences ................ Design I I I S ta t is t ic a l Analysis ............................ Procedures - Sample ....................................................... C o llectio n o f D a t a ........................................................... 46 46 47 47 50 52 57 60 62 63 64 67 RESULTS..................................................................................... 71 III. IV. Table I R e s u l t s ............................................................... Table I I R e s u lt s ............................................................... Table I I I and IV R e s u lt s .............................................. Analysis of Covariance ................................................... CPA R e s u l t s ....................................................................... Results on Costs, Tests o f Hypothesis 2 . . . . Student's t-T e s t ............................................................... Results on Costs, Tests o f Hypothesis 3 . . . . v ii 71 95 130 133 147 150 151 165 Chapter V. Page SUMMARY AND CONCLUSIONS......................................................................... 167 Suimiary o f EPSDT P r o g ra m ................................................................ 167 Summary o f Screening ProgramOutcomes ...................................... 168 Summary o f Research DesignandMethodology ............................. 170 Summary o f R e s u l t s .............................................................................173 V a lid ity of Referral Rates as Indicators o f Health S t a t u s ..................................................................................................180 Im plications o f Study fo r EPSOTand Social Work ................... 187 Recommendations fo r Future Study ................................................ 191 C onclusions..............................................................................................192 BIBLIOGRAPHY ................................................................................................... 193 APPENDICES.......................................................................................................... 198 Appendix Appendix Appendix Appendix Appendix Appendix Appendix Appendix Appendix Appendix A B C D E F G H I J ..............................................................................................198 ..............................................................................................216 . ..................................................................................... 234 ..............................................................................................251 ..............................................................................................268 ..............................................................................................285 ..............................................................................................302 ............................................................................................. 304 ............................................................................................. 308 ..............................................................................................310 viii LIST OF TABLES Table Reference Table 1. Reference Table I I . Reference Table I I I . I. 1(A ). 1(B ). 1(C ). 1(D ). 1 (E ). 1 (F ). Page Percent o f e lig lb le s and percent of e llg lb le s screened, by a g e ........................................................... 6 Per capita service u t iliz a t io n fo r a l l services, by year o f service and s t a t e .................................. 32 Per cap ita costs fo r a l l services, by year of service and s ta te ....................................................... 32 Average number o f r e fe rr a ls a t la s t screening fo r long-term e lig ib le s by age and number of life tim e screenings ................................................... 74 Average number o f re fe rr a ls a t la s t screening fo r long-term e lig ib le w hites, by age and num­ ber o f life tim e screenings ...................................... 76 Average number of r e fe r r a ls a t la s t screening fo r long-term e lig ib le blacks, by age and num­ ber o f life tim e screenings ...................................... 77 Average number o f r e fe r r a ls a t la s t screening fo r long-term e lig ib le American Indians, by age and number o f life tim e screenings ..................... 78 Average number o f r e fe rr a ls a t la s t screening fo r long-term e lig ib le Spanish-speaking, by age and number o f life tim e screenings ..................... 80 Average number of re fe rr a ls a t la s t screening f o r long-term e lig ib le males, by age and number o f life tim e screenings .............................................. 82 Average number o f r e fe r r a ls a t la s t screening fo r long-term e lig ib le females, by age and number of life tim e screenings .............................. 83 ix Table 1(G ). 1(H ). 1 (1 ). I (J ). Page Average number of r e fe rr a ls a t la s t screening fo r long-term e lig ib le w hite males, by age and number o f life tim e screenings ............................................ 84 Average number o f r e fe rr a ls a t la s t screening fo r long-term e lig ib le w hite females, by age and number o f life tim e screenings .................................... 85 Average number o f r e fe rr a ls a t la s t screening fo r long-term e lig ib le black males, by age and number o f life tim e screenings ............................................. 86 Average number o f re fe rr a ls a t la s t screening fo r long-term e lig ib le black females, by age and number o f life tim e screenings .................................... 87 I( K ) . Average number o f r e fe rr a ls a t la s t screening fo r long-term e lig ib le American Indian males, by age and number o f life tim e s c re e n in g s ...............................88 I(L ). Average number o f r e fe rr a ls a t la s t screening fo r long-term e lig ib le American Indian females, by age and number o f life tim e s c r e e n in g s ...................... 89 I(M ). Average number o f r e fe rr a ls a t la s t screening fo r long-term e lig ib le Spanish-speaking males, by age and number o f life t im e s c r e e n in g s .......................90 I(N )- Average number o f re fe rr a ls a t la s t screening fo r long-term e lig ib le Spanish-speaking females, by age and number o f life tim e s c r e e n in g s .......................SI 1 (0 ). Average number o f re fe rra ls a t la s t screening fo r long-term e lig ib le p a rtic ip a n ts 1n D e tr o it, by age and number o f life tim e s c r e e n in g s .......................93 I(P ). Average number o f r e fe rr a ls a t la s t screening fo r long-term e lig ib le p a rtic ip a n ts 1n f o r ty four Northern Michigan counties, by number o f life tim e screenings ............................................................... 94 Average number o f r e fe rr a ls a t la s t screening fo r long-term e lig ib le p a rtic ip a n ts in D e tro it and Northern Michigan, by race and number of life tim e screenings ............................................................... 96 I(Q ) . I ( R ). Average number o f re fe rr a ls a t la s t screening fo r long-term e lig ib le s , by number and year o f screening (N > 1 0 0 ) ................................................................... 124 x Table I( S ) . II. 11(A). 11(B). 11(C). 11(D). 11(E ). 11 (F ). 11(G). 11(H). I I { I ). II(J ). Page Percent change 1naverage number o f re fe rr a ls a t la s t screening as number o f life tim e screen­ ings Increase by one fo r long-term e lig ib le s , by year o f screening (n _> 10 0 ) ........................................... 126 Average number o f r e fe rr a ls a t la s t screening fo r one-year e lig ib le s by age and number o f life tim e screenings ............................................................................. gg Average number o f re fe rr a ls a t la s t screening fo r one-year e lig ib le w hites, by age and number o f life tim e screenings ............................................................................. gg Average number of re fe rr a ls a t la s t screening fo r one-year e lig ib le blacks, by age and number of life tim e screenings ............................................................... 100 Average number o f re fe rr a ls a t la s t screening fo r one-year e lig ib le American Indians, by age and number o f life tim e screenings .......................................... 102 Average number o f r e fe rr a ls a t la s t screening fo r one-year e lig ib le Spanish-speaking, by age and number o f life tim e screenings .......................................... 103 Average number of r e fe rr a ls a t la s t screening fo r one-year e lig ib le males, by age and number of life tim e screenings ............................................................... 105 Average number o f re fe rr a ls a t la s t screening fo r one-year e lig ib le females, by age and number of life tim e screenings ............................................................... 106 Average number o f r e fe rr a ls a t la s t screening fo r one-year e lig ib le white males, by age and number o f life tim e screenings .......................................... 108 Average number o f r e fe rr a ls a t la s t screening fo r one-year e lig ib le white females, by age and number o f life tim e screenings ........................................................... log Average number of r e fe rr a ls a t la s t screening fo r one-year e lig ib le black males, by age and number o f life tim e s c re e n in g s .................................................. no Average nunfcer o f r e fe rr a ls a t la s t screening fo r one-year e lig ib le black females, by age and number of life tim e screenings ........................................................... m xi Page Average number o f r e fe rra ls a t la s t screening fo r one-year e lig ib le American Indian males, by age and number of life tim e screenings ............................. 112 Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le American Indian females, by age and number of life tim e screenings ............................. 113 Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le Spanish-speaking males, by age and number of life tim e screenings ..................................... 114 Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le Spanish-speaking females, by age and number o f life tim e screenings ..................................... 115 Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le p a rtic ip a n ts 1n D e tro it, by number of life tim e screenings .............................................. 118 Average number o f r e fe rra ls a t la s t screening fo r one-year e lig ib le p a rticip an ts In fo rty -fo u r Northern Michigan counties, by number of life tim e screenings . 119 Average number o f r e fe rra ls a t la s t screening in D e tro it and Northern Michigan, by race and number of life tim e screenings .......................................................... 121 Average number o f re fe rra ls a t la s t screening fo r one-year e lig ib le s , by number and year of screening (n > 100 ) ....................................................................................... 127 Percent change 1n average number of re fe rra ls a t la s t screening as number of life tim e screenings Increase by one, fo r one-year e lig ib le s by year of screening (n _> 100 ) .......................................................... 129 Average number o f r e fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le s , by age and number of life tim e screenings .................................................................. 234 Average number of re fe rra ls a t la s t screening 1n 1978 fo r long-term e lig ib le whites, by age and number of life tim e screenings ............................................. 235! Average number of re fe rra ls a t la s t screening 1n 1978 fo r long-term e lig ib le blacks, by age and number o f life tim e screenings ............................................. 236 x ii Page Average number o f r e fe rr a ls a t la s t screening In 1978 fo r long-term e lig ib le American Indians, by age and number o f life tim e screenings ..................... 237 Average number o f re fe rr a ls a t la s t screening in 1978 fo r long-term e lig ib le Spanish-speaking, by age and number o f life tim e screenings ......................... 238 Average number of re fe rr a ls a t la s t screening In 1978 fo r long-term e lig ib le males, by age and number of life tim e screenings .......................................... 239 Average number o f r e fe rr a ls a t la s t screening in 1978 fo r long-term e lig ib le females, by age and number o f life tim e screenings .......................................... 24:0 Average number o f r e fe r r a ls a t la s t screening 1n 1978 fo r long-term e lig ib le white males, by age and number o f life tim e screenings .................................. 241 Average number o f re fe rr a ls a t la s t screening in 1978 fo r long-term e lig ib le white females, by age and number of life tim e screenings .................................. 242 Average number o f r e fe rr a ls a t la s t screening 1n 1978 fo r long-term e lig ib le black males, by age and number o f life tim e screenings .......................................... 243 Average number o f re fe rr a ls a t la s t screening in 1978 fo r long-term e lig ib le black females, by age and number o f life t im e screenings .................................. 244 Average number o f r e fe rr a ls a t la s t screening in 1978 fo r long-term e lig ib le American Indian males, by age and number of life tim e screenings ..................... 245 Average number o f r e fe rr a ls a t la s t screening In 1978 fo r long-term e lig ib le American Indian females, by age and number o f life tim e screelngs ..................... 246 Average number o f r e fe rr a ls a t la s t screening 1n 1978 fo r long-term e lig ib le Spanish-speaking males, by age and number of life tim e screenings ..................... 247 Average number o f r e fe rr a ls a t la s t screening In 1978 fo r long-term e lig ib le Spanish-speaking females, by age and number o f life tim e screenings ..................... 248 xiii Page Average number o f re fe rr a ls a t la s t screening in 1978 fo r long-term e lig ib le p a rtic ip a n ts in D e tro it, by age and number of life tim e screen­ ings ........................................................................................ 249 Average number of re fe rr a ls a t la s t screening in 1978 fo r long-term e lig ib le p a rtic ip a n ts 1n fo r ty -fo u r Northern Michigan counties, by number of life tim e screenings ................................................... 250 Average number o f r e fe rr a ls a t la s t screening In 1978 fo r one-year e lig b le s , by age and number of life tim e screenings ....................................................... 260 Average number of r e fe r r a ls a t la s t screening in 1978 fo r one-year e lig ib le w hites, by age and number o f life t im e screenings .................................. 269 Average number of r e fe rr a ls a t la s t screening ■ 1978 fo r one-year e lig ib le blacks, by age and number o f life tim e screenings ......................... 270 Average number of r e fe r r a ls a t la s t screening in 1978 fo r one-year e lig ib le American Indians, by age and number o f life t im e screenings ................. 271 Average number of re fe rr a ls a t la s t screening in 1978 fo r one-year e lig ib le Spanish-speaking, by age and number o f life tim e screenings ................. 272 Average number o f r e fe rr a ls a t la s t screening 1n 1978 fo r one-year e lig ib le males, by age and number o f life tim e screenings .................................. 273 Average number o f r e fe r r a ls a t la s t screening In 1978 fo r one-year e lig ib le females, by age and number o f life tim e screenings .................................. 274 Average number o f r e fe rr a ls a t la s t screening 1n 1978 fo r one-year e lig ib le white males, by age and number o f life tim e screenings ......................... 275 Average number o f r e fe rr a ls a t la s t screening 1n 1978 fo r one-year e lig ib le white fem ales, by age and number o f life tim e screenings ......................... 2 76 Average number o f r e fe rr a ls a t la s t screening In 1978 fo r one-year e lig ib le black males, by age and number o f life tim e screenings ......................... 277 xiv Table IV (J ) . Page Average number o f r e fe rr a ls a t la s t screening in 1978 fo r one-year e lig ib le black females, by age and number o f life tim e s c re e n in g s .............................. 278 IV (K ). Average number o f r e fe rr a ls a t la s t screening 1n 1978 fo r one-year e lig ib le American Indian males, by age and number o f life tim e s c r e e n in g s ............................ 279 IV (L ). Average number o f r e fe rr a ls a t la s t screening In 1978 fo r one-year e lig ib le American Indian females, by age and number o f life t im e s c r e e n in g s ............................ 280 IV (M ). Average number o f r e fe r r a ls a t la s t screening in 1978 fo r one-year e lig ib le Spanish-speaking males, by age and number o f life tim e s c r e e n in g s ............................ 281 IV (N ). Average number o f r e fe r r a ls a t la s t screening 1n 1978 fo r one-year e lig ib le Spanish-speaking females, by age and number o f life tim e s c r e e n in g s ............................ 282 IV ( 0 ). Average number o f r e fe rr a ls a t la s t screening in 1978 fo r one-year e lig ib le p a rtic ip a n ts 1n D e tr o it, by age and number of life tim e screenings ...................................... IV (P ). _V. V I. V II. 283 Average number of re fe rr a ls a t la s t screening 1n 1978 fo r one-year e lig ib le p a rtic ip a n ts in f o r ty four Northern Michigan counties, by number o f l i f e ­ time s c re e n in g s ................................................................................. 284 Results o f analysis o f covariance fo r long-term e l i g i b l e s ..............................................................................................138 Results o f analysis o f covariance fo r short-term e l i g i b l e s ..............................................................................................146 Results of comparison o f medical costs fo r long­ term Medicaid e lig ib le EPSDT p a rtic ip a n ts and nonparticipants ............................................................................ 153 V III. Results o f comparison of medical costs fo r short­ term Medicaid e lig ib le EPSDT p a rtic ip a n ts and nonp a rtic ip a n ts .....................................................................................154 IX . Results of comparison o f medical costs fo r long­ term Medicaid e lig ib le white EPSDT p a rtic ip a n ts and n o n p a rtic ip a n ts ........................................................................ 155 xv Table X. X I. X II. X III'. XIV. XV. XVI. Results of comparison o f Medicaid costs fo r short­ term Medicaid e lig ib le white EPSDT p a rtic ip a n ts and n o n p a rtic ip a n ts ....................................................................... 157 Results o f comparison of medical costs fo r long­ term Medicaid e lig ib le black EPSDT p artic ip a n ts and nonpartici p a n t s .............................................................................158 Results o f comparison o f medical costs fo r short­ term Medicaid e lig ib le black EPSDT p a rtic ip a n ts and n o n p a rtlc ip a n ts ....................................................................... 159 Results o f comparison o f medical costs fo r shor ,term Medicaid e lig ib le American Indian EPSDT p a rtic ip a n ts and nonpartlcipants .......................................... 160 Results o f comparison o f medical costs fo r slv r t term Medicaid e lig ib le American Indian EPSDT p a rtic ip a n ts and nonpartlcipants .......................................... 161 Results of comparison o f medical costs fo r long­ term Medicaid e lig ib le Spanish-speaking EPSDT p a rtic ip a n ts and nonpartlcipants ............................................ 163 Results o f comparison o f medical costs fo r short­ term Medicaid e lig ib le Spanish-speaking EPSDT p a rtic ip a n ts and nonpartlcipants .......................................... 164 X V II. Summary o f the analysis o f covariance f o r long­ term e l i g i b l e s .....................................................................................302 X V III. Sumnary o f the analysis of covariance fo r oneyear e l i g i b l e s .....................................................................................308 1 (a ). I(A a ). Number o f long-term e lig ib le screened, by age and number o f life tim e screenings .......................................... 198 Number of long-term e lig ib le whites screened, by age and number o f life tim e screenings .................................. 199 I(B a ). Number of long-term e lig ib le blacks screened, by age and number o f life tim e s c re e n in g s ................................ 200 I(C a ). Number o f long-term e lig ib le American Indians screened, by age and number o f life tim e screenings . . 201 Number o f long-term e lig ib le Spanish-speaking screened, by age and number o f life tim e screenings . . 202 I(D a ). I(E a ). Number o f long-term e lig ib le males screened, by age and number o f life tim e s c re e n in g s ............................ xvi 203 Table I(F a ). I(G a ). I(H a ). I(Ia ). I( J a ) . I(K a ). I (La) I(M a). I(N a ). I(O a ). I(P a ). I(Q a ). Page Number o f long-term e lig ib le females screened, by age and number of life tim e screenings . . . . 204 Number o f long-term e lig ib le white males screened, by age and number of life tim e screenings ...................................................................... 205 Number o f long-term e lig ib le white females screened, by age and number of life tim e screenings .......................................................................... 206 Number o f long-term e lig ib le black males screened, by age and number of life tim e screenings .......................................................................... 207 Number of long-term e lig ib le black females screened, by age and number of life tim e screenings .......................................................................... 208 Number of long-term e lig ib le American Indian males screened, by age and number of life tim e screenings .......................................................................... 209 Number of long-term e lig ib le American Indian females screened, by age and number o f life tim e screenings .......................................................................... 210 Number of long-term e lig ib le Spanish-speaking males screened, by age and number of life tim e screenings .......................................................................... 211 Number of long-term e lig ib le Spanish-speaking females screened, by age and number of life tim e screenings ...................................................... 212 Number of long-term e lig ib le p articip ants screened in D e tro it, by age and number of life tim e screenings ...................................................... 213 Number of long-term e lig ib le p a rticip an ts screened in fo rty -fo u r Northern Michigan counties, by number of life tim e screenings . . . 214 Number o f long-term e lig ib le s screened in D e tro it and Northern Michigan, by race and number o f life tim e screenings ................................. 215 xv i i Table Page 1 1 (a ). Number of one-year e lig ib le s screened, by age and number of life tim e s c re e n in g s .................... 216 II( A a ) . Number o f one-year e lig ib le whites screened, by age and number of life tim e s c re e n in g s .................... 217 II( B a ) . Number o f one-year e lig ib le blacks screened, by age and number of life tim e s c re e n in g s .................... 218 Il(C a ) . Number o f one-year e lig ib le American Indians screened, by age and number of life tim e s c r e e n in g s .......................................................................... 219 II(D a ) . Number o f one-year e lig ib le Spanish-speaking screened, by age and number o f life tim e s c r e e n in g s .......................................................................... 220 I I( E a ) . Number of one-year e lig ib le ihales screened, by age and number of life tim e screenings . . . . 221 Number of one-year e lig ib le females screened, by age and number o f life tim e screenings . . . . 222 II(F a ). II(O a ) . Number of one-year e lig ib le white males screened, by age and number o f life tim e s c r e e n in g s ............................................................................... 223 II(H a ) . Number o f one-year e lig ib le white females screened, by age and number o f life tim e s c r e e n in g s .......................................................................... 224 II(la ). Number of one-year e lig ib le black males screened, by age and number of life tim e s c r e e n in g s .......................................................................... 225 I I (Ja) . Number o f one-year e lig ib le black females screened, by age and number o f life tim e s c r e e n in g s .......................................................................... 226 II( K a ) . Number o f one-year e lig ib le American Indian males screened, by age and number of life tim e s c r e e n in g s .......................................................................... 227 II(L a ). Number o f one-year e lig ib le American Indian females screened, by age and number of life tim e screenings .......................................................... xvi i i 228 Table II(M a ). II( N a ) . II(O a ). II( P a ) . II(Q a ). 111(a). III(A a ). II I ( B a ) . Ill(C a ). III(D a ). III(E a ). III(F a ). Page Number of one-year e lig ib le Spanish-speaking males screened, by age and number of life tim e screenings ...................................................... 229 Number o f one-year e lig ib le Spanish-speaking females screened, by age and number of life tim e screenings ...................................................... 230 Number o f one-year e lig ib le p articip an ts screened in D e tro it, by number of life tim e s c r e e n in g s ...................................................................... 231 Number o f one-year e lig ib le p articip an ts screened 1n fo rty -fo u r Northern Michigan counties, by number of life tim e screenings. . . 232 Number of one-year e lig ib le s screened in D e tro it and Northern Michigan, by race and number o f life tim e screenings .................................... 233 Number o f long-term e lig ib le s screened in 1978, by age and number o f life tim e s c r e e n in g s ...................................................................... 251 Number of long-term e lig ib le whites screened in 1978, by age and number of life tim e s c r e e n in g s ...................................................................... 252 Number of long-term e lig ib le blacks screened in 1978, by age and number of life tim e s c r e e n in g s ...................................................................... 253 Number o f long-term e lig ib le American Indians screened In 1978, by age and number of life tim e screenings ...................................................... 254 Number o f long-term e lig ib le Spanish-speaking screened in 1978, by age and number of life tim e screenings ...................................................... 255 Number o f long-term e lig ib le males screened 1n 1978, by age and number o f life tim e s c r e e n in g s ...................................................................... 256 Number of long-term e lig ib le females screened In 1978, by age and number of life tim e s c r e e n in g s ...................................................................... 257 ixx Page Number of long-term e lig ib le white males screened In 1978, by age and number o f life tim e screenings ................................................... , 2 58 Number o f long-term e lig ib le white females screened in 1978, by age and number of life tim e screenings ................................................... , 299 Number o f long-term e lig ib le black males screened 1n 1978, by age and number o f life tim e screenings ................................................... , 260 Number o f long-term e lig ib le black females screened 1n 1978, by age and number o f life tim e screenings ....................................................... 261 Number o f long-term e lig ib le American Indian males screened 1n 1978, by age and number of life t im e screenings .............................................. 2 62 Number o f long-term e lig ib le American Indian females screened 1n 1978, by age and number o f life tim e screenings .............................................. 2 63 Number of long-term e lig ib le Spanish-speaking males screened In 1978, by age and number o f life tim e screenings .............................................. 264 Number o f long-term e lig ib le Spanish-speaking females screened 1n 1978, by age and number o f life t im e screenings .............................................. 265 Number o f long-term e lig ib le p a rtic ip a n ts screened 1n 1978 In O e tro lt, by age and number o f life tim e screenings .................................. 2 66 Number o f long-term e lig ib le p a rtic ip a n ts screened 1n 1978 In fo r ty -fo u r Northern Michigan counties, by number of life tim e screenings ....................................................................... 267 Number o f one-year e lig ib le s screened In 1978, by age and number o f life tim e screenings. 285 Number of one-year e lig ib le whites screened In 1978, by age and number of life tim e screenings ....................................................................... 286 xx Table IV(B a). IV(C a). IV (D a). IV (E a). IV (F a ). IV(G a). IV(H a). IV ( Ia ) . IV (J a ). IV(K a). IV (L a ). IV(Ma). Page Number of e lig ib le blacks screened In 1978, by age and number of life tim e screenings . . . . 287 Number o f one-year e lig ib le American Indians screened In 1978, by age and number of life tim e screenings ...................................................... 288 Number o f one-year e lig ib le Spanish-speaking screened 1n 1978, by age and number of life tim e screenings ...................................................... 289 Number o f one-year e lig ib le males screened 1n 1978, by age and number o f life tim e screenings . 2 90 Number o f one-year e lig ib le females screened In 1978, by age and number o f life tim e s c re e n in g s .......................................................................... 291 Number of one-year e lig ib le white males screened in 1978, by age and number of life tim e screenings ...................................................... 292 Number o f one-year e lig ib le white females screened In 1978, by age and number of life tim e screenings ...................................................... 293 Number of one-year e lig ib le black males screened in 1978, by age and number of life tim e screenings ...................................................... 294 Number o f one-year e lig ib le black females screened in 1978, by age and number of life tim e screenings ...................................................... 295 Number o f one-year e lig ib le American Indian males screened 1n 1978, by age and number o f life tim e screenings ................................................. 296 Number o f one-year e lig ib le American Indian females screened 1n 1978, by age and number o f life tim e screenings ................................. 297 Number o f one-year e lig ib le Spanish-speaking males screened in 1978, by age and number o f life tim e screenings ................................................. 298 xx i Table IV (N a). IV(O a). IV (P a ). Page Number o f one-year e lig ib le Spanish-speaking females screened 1n 1978* by age and number o f life tim e screening ...................................................... 299 Number o f one-year e lig ib le p a rtic ip a n ts in D e tro it screened in 1978, by age and number o f life tim e screenings ................................................. 300 Number o f one-year e lig ib le p a rtic ip a n ts screened In 1978 in fo r ty -fo u r Northern Michigan counties, by number o f life tim e screenings................................................................................... 301 xxii LIST OF FIGURES Figure Page 1. Regression lines depicting the relatio n sh ip between number o f re fe rra ls and number of l i f e ­ time screenings fo r long-term e lig ib le s * by year o f la s t..s c re e n in g ...........................................................................145 2. Regression lin es depicting the relatio n sh ip between nimiber o f re fe rra ls and number o f l i f e ­ time screenings fo r one-year e lig ib le s , by year o f la s t..s c re e n in g ........................................................................... 148 3. Regression lin es depicting the relatio n sh ip between number o f re fe rra ls and number of l i f e ­ time screenings fo r long-term e lig ib le s , by year o f la s t screening fo r In te ra c tio n m o d e l........................... 307 Regression lin e s depicting the relatio n sh ip between number o f r e fe rra ls and number of l i f e ­ time screenings fo r one-year e lig ib le s , by year o f la s t screening fo r in te ra c tio n m o d e l........................... 312 4. xxiii CHAPTER I INTRODUCTION This Is a study to determine whether the Early and Periodic Screening, Diagnosis and Treatment (EPSDT) program benefits its p a rtic ip a n ts . To th a t end a large volume of ex istin g computerized health data on low income children in Michigan was analyzed using screening re fe rra l rates and treatment costs as outcome variab les. In te re s t was In th e ir v a r ia b ilit y as a function o f program p a r t ic i­ pation with the Influence o f demographic factors also considered. The study's importance lie s In Its contribution to the lim ite d know­ ledge av a ilab le on the effectiveness o f th is la rg e , r e la tiv e ly new and somewhat unconventional program. EPSDT's h is to ry , strategy and c lie n te le a l l make 1t o f p a rtic u la r in te re s t to those in the health and social w elfare fie ld s . Studies have generally shown the poor to have more health prob­ lems and fewer medical resources than higher income groups. In an attempt to address th is problem, EPSDT alms to increase access to medical services but access fo r those with Id e n tifie d , medical needs, not solely low Income. The program's strategy 1s to divide Its pop­ u latio n Into two groups - one seemingly without health problems; the other w ith possible problems and the need fo r services. This division Is accomplished by administering a series of screening tests and pro­ cedures. Medical resources, diagnostic and treatment services, are 1 2 then concentrated on those apparently most in need - those who fa ile d the screening te s t(s ). Screening is a key component in the program and screening is co n tro versial. Although i t has a history in the United States dating from the 1920s, the medical community and public have only moderately accepted i t . Reservations regarding it s usefulness undoubtedly con­ trib u ted to EPSDT's slow pace of implementation. screening makes l i t t l e or no While some believe contribution to maintaining h ealth , others, such as EPSOT advocates, argue th a t r e la tiv e ly small expendi­ tures fo r screening can lessen the need to la te r spend much larg er sums fo r treatm ent . 1 peal. The ratio n a le fo r screening has in t u it iv e ap­ Its basic purpose is to fin d and tre a t problems e a rly , before they advance to a more complicated s ta te . More te c h n ic a lly , screening attempts to shorten the time In te rv a l between problem onset and detec­ tio n in order to consequently shorten the In te rv a l between treatment and recovery. Whether screening accomplishes It s purpose and whether the fa c to r o f time is even important 1n problem detection remain topics o f disagreement. EPSDT was enacted by the United States' Congress in 1967 as an amendment to T it le XIX o f the Social Security Act. Its authorization marked the f i r s t time th a t the United States had Included preventive health services in a larg e, national program. 2 As most programs examples, see Abraham B. Bergman, "The Menace o f Mass Screen­ in g ," American Journal of Public H ealth, LXVII (J u ly , 1977), 601-02 andGunnar B. S tic k le r , "Mow Necessary 1s the 'Routine Checkup'?," C lin ic a l P ra c tic e . VI(August, 1967), 454. M o r ris S. Dixon, J r . , "T itle XIX EPSDT: The Im plications fo r Ped­ i a t r ic P ra c tic e ," B u lle tin of P ed iatric P rac tice , VI (December, 1972), 2. 3 authorized by the Social Security Act, EPSDT is state administered but jo in t ly funded by the federal and state governments. States are required to o ffe r the program to recip ients of the Aid to Fam­ ilie s with Dependent Children (AFDC) program who are under age twentyone although c lie n t p a rtic ip a tio n 1s voluntary. EPSDT has a large e lig ib le population and accordingly a poten­ t i a l impact o f fa r reaching dimensions. N a tio n a lly , some th irte e n m illio n young people are e lig ib le with over h a lf a m illio n o f these liv in g in Michigan. In f a c t , i t is the federal government's larg est health care program fo r poor children and serves more Medicaid c h ild ren than a l l other fe d e ra lly supported health care programs combined. 3 Those In it ia t in g the program were undoubtedly mindful of the mass con­ stituency to be affected and the need to d ire c t resources to th is specific population. However, despite Congressional in te n t and the th re a t to states o f federal fin a n c ia l penalty fo r noncompliance, Implementation pro­ ceeded slowly. The federal government did not issue fin a l program guidelines u n til 1972 and most states did not o ffe r services u n til several years la te r . A ll s ta te s , w ith the exception o f Arizona which has no T it le XIX (Medicaid) program, now have an EPSDT program. Mich­ igan, s ite o f the study, began its program in 1973. When establishing the Michigan program, the T it le XIX agency, the payer o f medical services fo r Department of Social Services (DSS) 3 Department of Health, Education and Welfare: Health Care Financing Adm inistration, EPSD&T: The Possible Dream (Washington, D.C.: Government P rin tin g O ffic e , 1977), c ite d In the Foreward. 4 recip ients and the single state agency responsible fo r EPSDT, chose the Michigan Department o f Public Health (DPH) to adm inister the screening portion of the program. The Department o f Public Health 1n turn contracted with local health departments fo r actual provision of the screening services. The Department o f Social Services was responsible fo r the outreach e f f o r t . These DSS-DPH relation ship s were defined by means o f an interagency agreement and the program structure has remained unchanged to the present, with the exception o f some local health departments assuming the outreach function. The program flow is as follow s: E lig ib le s are system atically contacted and asked whether they wish to p a rtic ip a te 1n the program. Those who request services are scheduled fo r screening a t a c lin ic staffe d by sp ecially trained EPSDT personnel. Those who decline to p a rtic ip a te are simply recontacted a t a la te r tim e, usually in one to two years. The screening is uniformly conducted by a registered nurse and technicians who administer a standard screening package. Those f a llin g a t e s t ( s ) , are referred to an appropriate provider(s) with arrangements made fo r securing the needed re fe rra ls p rio r to the c lie n t leaving the screening s ite . That is , the c lin ic e ith e r obtains a re fe rra l appointment fo r the c lie n t or the c lie n t expresses the preference of making h e r/h is own appointment. A lin k thus exists between screening and the a v a ila b ilit y o f needed treatment services. For each child screened, the resu lts o f the examination are re ­ corded on a special form, the contents o f which are subsequently entered on computer f i l e . For those receiving r e fe rr a l services, as fo r a l l Medicaid e lig ib le s receiving service, enrolled providers b i l l 5 the Medicaid program fo r reimbursement through an automated payments system. Thus, fo r purposes o f conducting th is study, screening re ­ s u lts , medical costs and basic c lie n t demographic inform ation were a ll accessible by computer. As indicated above, there 1s present in the program a fa c to r o f s e lf selection and th is fa c to r complicates e v a lu ative e f fo r t s . gibles have fre e choice over re c e ip t of services. E li­ This means those wishing to p a rtic ip a te in the program can not be denied the opportun­ i t y to do so, even fo r purposes o f research. sidered a r ig h t. C lie n t choice is con­ Thus, random selection and assignment o f program p a rtic ip a n ts is n eith e r experim entally possible nor inherent in the program's operation. Consequently, the question arises as to whether the same fa c to r(s ) which determine program p a rtic ip a tio n might not also be responsible fo r any differences in health status? However, as Philadelphia Health Management Corporation (PHMC) argues in th e ir evaluation o f EPSDT, 4 a counter in te rp re ta tio n o f outcomes is fe a s ib le only i f the a lte rn a tiv e hypothesis is i t s e l f reasonable or has empir­ ic a l support. For example, Improved outcomes in a lon gitu d in al design may be due e ith e r to s t a t is t ic a l regression or experimental e ffe c ts . However, i f f i r s t scores o f the experimental group are lower than f i r s t scores o f a control group, subsequent improvement 1 s more lik e ly a ttrib u ta b le to experimental e ffe c ts . S im ila rly , i f support fo r pro­ gram e ffe c ts is found in a series o f te s ts , each o f which o ffe rs d iffe r e n t a lte rn a tiv e hypotheses, then support fo r the program grows 4 Philadelphia Health Management Corporation, A Study o f The Process, E ffectiveness, and Costs o f the EPSDT Program In South­ eastern Pennsylvania. Part I I I . [P h ila d e lp h ia . Pennsylvania), 1980. 6 a t the expense o f r iv a l explanations. These are the types o f s itu a tio n s PHMC d e lib e ra te ly constructs in th e ir study and i t is th e ir conclusion th a t s e lf selectio n does not compromise th e ir fin d 5 1ngs. In s h o rt, the lack o f random selectio n does not necessarily in v a lid a te a study and there 1s some em pirical as well as th e o re tic a l evidence to support th is view. Also, of relevance to the issue o f s e lf selection is a compari­ son o f EPSDT p a rtic ip a n ts and no np articipants. Approximately 50% of Michigan EPSDT e lig ib le s are p a rtic ip a n ts , i . e . they have been screen­ ed at le a s t one tim e.*’ This is a reasonably good p a rtic ip a tio n ra te and would seem to suggest th a t p a rtic ip a n ts and nonparticipants are not extremely d if f e r e n t . Age does show some v a ria tio n between the two groups: Reference Table 1. Percent o f e lig ib le s and percent o f e lig ib le s screened by age. Percent o f , E lig ib le s Screened Age 0-5 6-12 13-20 42% 37% 21% Percent o f a EPSDT E lig ib les*3 37% 39.7% 23.3% 5I b i d . . p. 92. - . 103. Michigan Department o f Social Services, Health and Welfare Data Center, " E l i g i b i l i t y S ta tis tic s By County, Report Number EP-293," (Lan­ sing, Michigan). For January, 1981 there were 554,578 EPSDT e lig ib le s , 278,840 (50%) o f whom had been screened a t le a s t one tim e. For Septem­ ber, 1979, there were 485,048 e lig ib le s , 240,455 (49.5%) of whom had been screened a t le a s t once. ^Michigan Department of Public Health and Michigan Department of Social Services, EPSDT Michigan Annual Report, 1978, (Lansing, Michigan), 1979, 10. o Michigan Department of Social Services, Assistance Payments Sta­ t i s t i c s , Publication No. 67, Data Reporting Section, (Lansing, M ichigan), February, 1980, 29. 7 This table shows some tendency fo r younger children to p a rtic ip a te disproportionately in the program. This lik e ly indicates greater parental concern fo r the health of younger children as well as the more in flu e n tia l voice of older children in determining the uses of th e ir time. g Further review o f Michigan program s ta tis tic s suggests d i f f e r ­ ences, although not extreme ones, do e x is t between EPSDT p articip ants and no np articipants:10 P a rtic ip a tio n by sex is comparable fo r a ll age groups, excepting those 13-21 years old. For th is group, 59* of those screened were female; 41* male, a s trik in g d iffe re n c e . Urban- rural differences appear to play some ro le in distinguishing users. During 1978, the r a tio o f screenings to the use o f Medicaid services was 15* higher 1n rural areas. I t was thought th is difference re ­ fle c te d the greater a v a ila b ilit y o f medical services 1n the urban areas. able. S u rp risin g ly, good comparative data on race are not a v a il­ During 1978, 57* o f screenees were w hite, 38* were black, 4* Spanish-Speaking, .3 * American Indian and 1% "other." This d is­ trib u tio n is s im ila r to the ra c ia l composition of those using Medi­ caid services, but since data are not a va ila b le on the ra c ia l comp­ o s itio n o f the e lig ib le population, a s t r ic t comparision o f EPSDT p articip ants and nonparticlpants is not possible. In summary, d i f ­ ferences between particip an ts and nonparticipants e x is t but are not g Although the data displayed are from somewhat d iffe r e n t time periods, since the number of e lig ib le s Involved Is large (106,455 screened and 430,120 e l i g i b le ) , sizable s h ifts in age d is trib u tio n would not be expected to occur 1n a twelve to eighteen month period. l0 M1chigan Department of Public Health and Michigan Department of Social Services, EPSDT Michigan Annual Report, 1978, op. c i t . , p. 11-13. 8 extreme. Those p a rtic ip a tin g in EPSDT tend to be somewhat younger and more ru ra l than EPSDT e lig ib le s in general and, i f a teenager, h a lf again as lik e ly to be female as male. With un lim ited resources, a d iffe r e n t research design would have been p re fe ra b le . Since the program's cen tral purpose is to Improve the health status o f c h ild re n , the Ideal study might per­ form lo n g itu d in a l medical examinations on equated samples o f program p a rtic ip a n ts and nonparticipants. The medical tests and procedures used would be determined by a panel of medical experts. The study would continue fo r many years since e ffe c ts might not be m anifest u n til f a r In the fu tu re . However, a study o f th is magnitude was f a r beyond the w r it e r 's resources. Medical personnel were not a v a ila b le to conduct examinations and re la te d te s ts . long-term study desired. Nor was a However, as noted, res u lts o f screening tests had been retained on computer f il e s fo r v ir t u a lly the program's e n tire h is to ry in Michigan and recent medical cost data were also a v a ila b le . Once arrangements were made to access and analyze these data, a study was possible which used a v a lid and fe a s ib le research design although not an Ideal one. Program evaluation 1s p a rt o f program a d m in istra tio n . This is not to say th a t the products o f evaluation are a pressing, d a ily need or th a t evaluation can m aintain I t s e l f as a p r io r it y In the face o f day-to-day operational and c ris is -c e n te re d demands. How­ ever, as coordinator o f the EPSDT program f o r nearly it s e n tire h is to ry in Michigan, the w rite r is w ell aware o f the need to estab­ lis h an e m p iric a lly based defense o f social programs. During the 9 1970s,the w rite r observed th at an ongoing adm inistrative task was to structure and secure resources fo r the program, a task which neces­ sita te d s e llin g i t a t various adm inistrative levels within the state public w elfare system. states. This situ a tio n lik e ly prevailed in other At the same time the federal government was in the position of s e llin g the program to states so th a t states would implement the program. In the la te 1970s, the federal government attempted to per­ suade Congress to expand the program via new le g is la tio n . In a ll these situ a tio n s * the case fo r EPSOT was u ltim a te ly argued on the very basic level of "does i t do any good?" and "is 1t r e a lly needed?" Obviously, em pirical knowledge concerning the program’ s e ffe c t on health status and medical costs was needed to answer these questions and thereby adm inister the program. The fa c t th a t program Implementation moved slowly and th a t Cong­ ress did not pass new le g is la tio n r e fle c ts , a t le a s t in p a rt, unsatis­ facto ry answers to basic questions o f outcome. In the 1980s,1t ap­ pears these same answers w ill be needed to maintain the existin g pro­ gram o r, a t a minimum, to slow its retrenchment. These circumstances, plus the general public concern over the contributions of social pro­ grams and the increasing need to d is trib u te the poor's diminishing a l ­ location o f resources to areas o f maximum b e n e fit, a l l created impetus fo r undertaking the follow ing study. This d is s e rta tio n presents, in Chapter I I , a review of lite r a tu r e which serves to place the study 1n the context of the program's h is to ry , theory and past findings o f outcomes. Chapter I I I explains the study's research design and methodology and discusses the modes of q u a n tita tiv e 10 analysis used, Including the s ta t is tic a l tests employed and the reasons fo r th e ir selectio n. while Chapter V e n tire study. Chapter IV presents the findings addresses th e ir im plications and summarizes the The obtained data are presented in tables located in e ith e r the body o f the study or in the Appendices. CHAPTER I I REVIEW OF LITERATURE This chapter w ill summarize lit e r a t u r e selected fo r the pur­ pose o f placing th is study in the context o f other thought and re ­ search rele v a n t to th is in v e s tig a tio n . The lite r a tu r e w ill be re ­ viewed under the follow ing four headings resp ec tive ly : {1} Back­ ground on EPSDT, (2 ) Outcome studies on non EPSDT screening pro­ grams, (3) Outcome studies of the EPSDT program and (4) The r e la ­ tionship of demographic facto rs to health status. Background on EPSDT The aim of th is section Is to develop a b e tte r understanding o f the EPSDT program through a survey o f It s e arly h is to ry w ith p a rtic u la r a tte n tio n given to uncovering the program's o rig in a l purpose(s). complish? Why was i t conceived and what was i t intended to ac­ Answers to such questions would help to not only deepen understanding but also to determine whether the program is fun ction­ ing as Intended. U n fo rtun ately, but perhaps not s u rp ris in g ly , ans­ wers to such basic questions are not completely c le a r. F o ltz , through a series o f a r t ic le s ,* has lik e ly established h e rs e lf as EPSDT's p rin c ip a l h is to ria n . The best single source in *Anne-Marie F o ltz , "The Development of Ambiguous Federal Policy Early and Periodic Screening, Diagnosis and Treatment (EPSDT)," M ilbank Memorial Fund Q u arterly/H ealth and S o ciety, L I I I (W inter, 1975) 35-64; Anne-Marie F o ltz , U ncertainties o f Federal Child Health Pol­ ic ie s : Impact In Two States (New Haven, CT: Yale U n iv e rs ity , 12 th is series is "The Development o f Ambiguous Federal Policy: Early and Periodic Screening, Diagnosis and Treatment (EPSDT).'' This a r­ t i c l e is a d e ta ile d h is to ry o f the program's beginnings, it s le g is ­ la tiv e h is to ry and subsequent lengthy development as a regulation by the Department of H ealth, Education and Welfare (HEW, now the Department o f Health and Human Services, or HHS). I t Is r e lie d on heavily in what follow s. F o ltz traces federal support o f health screening, a t le a s t im p lic it support, to 1935 and two sections of T it le V of the Social Security Act. One section established a Crippled C h ild ren 's program, the purpose o f which was to locate and tr e a t crip p led c h ild re n . T it le V also established Maternal and Child Health services which many states used to support w e ll-c h iId conferences emphasizing preventive care and screening. Some states and lo c a lit ie s had established w e ll-c h ild con­ ferences (examinations) fo r lim ite d numbers o f children during the 1920s. T i t l e V strengthened these in it ia t iv e s and the program con­ tinues today. Between 1935 and the 1960s, l i t t l e c h ild health p o lic ie s . innovation occurred in federal During World War 1 1 ,w e ll-c h ild conferences were expanded and the Emergency M aternity and In fa n t Care Program (EMIC) was In it ia te d whereby states received funds to provide pre­ ventive and treatment services to wives and child ren of lower paid m ilita r y personnel. F o ltz says EMIC was successful but was terminated Department of Epidemiology and Public H ealth , 1978); Anne-Mar1e F o ltz and Donna Brown, Health Policy P roject: The Impact o f Federal Child Health Policy under fcftSbt - Tne Case o f Connecticut (New Haven, CT: Yale U n iversity Department of Epidemiology and Public H ealth, 1975; Anne-Marie F o ltz and Donna Brown, "State Response to Federal Policy: C hildren, EPSDT, and the Medicaid Muddle," Medical Care, X I I I (August, 1975), 630-42. 13 a f te r the war. 2 The 1960s saw an unprecedented programs, including child health. number o f In it ia t iv e s in public Foltz notes the major a c tiv itie s : T it le V was expanded through M aternity and In fan t Care Projects (1963) providing comprehensive m aternity and in fan t care and through Children and Youth Projects (1965) which provided comprehensive health services fo r children and youth in selected geographic areas. The Economic Opportunity Act (1964) resulted in the establishment o f neighborhood health centers and the head s ta rt program. Medicaid was authorized (1966) which, although hot a ch ild re n 's program, would finance b i l ­ lions o f d o llars o f medical services fo r children. Federal programs o f th is scope and number had never before been attempted. I t was w ithin th is social clim ate th a t EPSDT was conceived; only one o f many social programs undertaken in th is rare period o f national history when resources and atte n tio n were shifted somewhat to those of the lower class. The ultim ate reasons fo r establishing EPSDT are surely the same reasons fo r th is overall expansion o f public services fo r the poor during the 1960s. Precisely why the United States undertook th is b r ie f period o f social experimentation 1s a matter o f some debate, which although germane, is beyond the purview o f th is study to resolve.^ However, there are a v a ila b le , s p ec ific references to EPSDT's o rig in p Ann-Mar1e F o ltz , "The Development of Ambiguous Federal Policy: Early and Periodic Screening, Diagnosis and Treatment," op c 1 t ., p. 37. 3 For a sample o f the debate see Peter Marris and Martin Rein, D i­ lemmas Of Social Reform (New York: Atherton Press, 1969); Daniel P.~ffoynihan, Maximum Feasible Misunderstanding (New York: The Free Press, 1969) and Frances Fox Piven and Richard A. (Howard, Regulating the Poor: The Functions o f Public Welfare (New York: Pantheon, 19^1)7 14 which are worthy o f mention. Shenkin says* "In common w ith almost a l l le g is la tio n 1n th is era o f social program expansion, the EPSDT proposal arose from the executive branch. . . . 1,4 However, both Shenkin and F o ltz agree EPSDT's th eo re tic ia n s and core o f advocates were w ith in HEW. kin again: Shen­ . . HEW housed the real proponents o f EPSDT. . . . It was s p e c ific perceptions and goals w ith in HEW th a t led to EPSDT."6 More s p e c ific a lly , F o ltz c ite s what is apparently the e a r lie s t w r it ­ ten conceptualization o f EPSDT: "The idea fo r fe d e ra lly sponsored periodic screening fo r low-income ch ild ren f i r s t appeared 1n 1966 in a program analysis prepared 1n the S ecretary's O ffic e o f HEW."6 An HEW pu blication has also credited the Program Analysis as re s u ltin g in the "creation o f E P S D T . T h e analysis was unpublished but because o f the in s ig h t i t gives to the e a rly concept and ra tio n a le fo r the program, references to i t are worth reviewing. F o ltz says the 1966 Program Analysis ou tlin ed three a lte rn a tiv e programs, w ith price tags, which would involve screening and tre a tin g low-income c h ild re n . I t buttressed the case fo r EPSDT by Including 4Budd N. Shenkin, M .D ., " P o litic s and the Health o f C hildren, Medical Care, XIV (October, 1976), 884. 5Ib id . , p. 884 C Ann-Mar1e F o ltz , "The Development o f Ambiguous Federal Policy: Early and Periodic Screening, Diagnosis and Treatment (EPSDT),'' op c i t . , p. 41. Also, F o ltz states elsewhere, "The Idea o f EPSDT seems f i r s t to have germinated w ith in HEW in the 1966 Program Analysis, . . . " i n Anne-Marie F o ltz , "Rebuttal to Dr. Shenkin," Medical Care, XIV (October, 1976), 886. ^Department o f H ealth, Education, and W elfare: Health Care Finan­ cing A dm inistration, EPSD&T: The Possible Dream, op. c i t . , p. 1. 15 a S e lective Service study which indicated a s ig n ific a n t percentage o f draftees were being rejected because o f physical and mental prob­ lems which could have been corrected 1 f id e n tifie d and treated a t an e a r lie r age. “I t was to deal w ith these problems e a rly - and cost e ffe c tiv e ly - th a t EPSDT was es tab lis h ed ."8 I t appears the In te re s t in correcting these problems was prompted by a mix o f considerations. Monetary m otivations were apparently present, or a t le a s t were used as a supportive argument fo r the program. F o ltz says, "The case fin d in g was to l i f t a burden from the population by saving child ren from handicapping con dition s." Q Elsewhere she noted the analysis argued fo r saving society money by preventing d e fe c ts .18 The HEW brochure "The Status o f EPSD&T" says the "immediate reason" fo r EPSDT is to provide poor child ren access to health care because they need i t but 1t also notes th a t another reason is to save the public money by preventing m edically-induced dependency.11 This dual theme is also present 1n another HEW pu b licatio n which prominently stresses the need to address the health problems o f poor children but also notes th a t "Evidence o f the program's cost-effectiven ess is already beginning to come in ." 12 HEW also argues: p C h ild ren 's Defense Fund, EPSDT: Does I t Spell Health Care For Poor C hildren? (Washington, D.C.: Washington Research P ro je c t, In c ., 1977), p. 25 Q Anne-Marie F o ltz , "The Development of Federal Policy: Early and Periodic Screening, Diagnosis and Treatment (EPSDT," op. c i t . , p. 41. 18Anne-Marie F o ltz , "Rebuttal to Dr. Shenkin," op. c i t . , p. 887. ^Department o f H ealth, Education, and W elfare, "The Status o f EPSDT," (SRS, 75-02052) (Washington, D.C.: Government P rin tin g O ffic e , 1975). 12 Department o f Health, Education, and W elfare: Health Care Finan­ cing A dm inistration, EPSD&T: The Possible Dream, op. c i t . , p. 16. 16 By preventing acute Illn e s s and reducing the need fo r expensive In s titu tio n a l care, preventive programs lik e EPSDT represent the long-term advantages o f re ­ moving from the State the fis c a l burden o f caring fo r severely handicapped people, as well as Improving the q u a lity o f l i f e fo r those ind ividu als whose health future is p ro te cted .13 The frequently mentioned purpose of cost reduction is o f par­ tic u la r In te re s t to th is study since costs Is one o f the two out­ come variables which w ill be measured. I t 1s c le a r the poor's medical costs are of in te re s t and Importance whether viewed as re ­ fle c tin g th e ir q u a lity o f l i f e or fin a n c ia l burden to the larg er society. The Program Analysis was c irc u la te d 1n la te 1966, on February 8 , 1967 President Johnson referenced the EPSDT concept 1n an address to Congress and on February 16, 1967 Representative Wilbur H ills Introduced a broad-ranging le g is la tiv e package which included EPSDT. M ills ' proposed le g is la tio n , the Social Security Amendments o f 1967, consumed 112 pages, three paragraphs of which concerned EPSDT. 14 According to Fo ltz the program remained Inconspicuous in sub­ sequent le g is la tiv e hearings, evoking l i t t l e comment. silence was damaging. She says the Her thesis 1s th at Congress was ambiguous on key provisions o f the b i l l - It s costs, scope o f services, e lig ib le population and adm inistration - and these ambiguities hampered la te r program acceptance and Implementation. 15 However, both F o ltz and 13I b i d . , p. 17. 14 Anne-Marie F o ltz , "The Development of Ambiguous Federal Policy: Early and Periodic Screening, Diagnosis and Treatment (EPSDT)," op c 1 t. , P* ^ 1 5 Ib id . , pp. 35-64 and Anne-Marie F o ltz , "Rebuttal to Dr. Shenkin," op. c i t . , pp. 886-87. 17 Shenkin re a liz e these omissions were by design. "Congress f e l t that general directions could be given to the Adm inistration, and the specifics could be worked out in good fa ith ." *® And, (For HEW) " . . . ambiguity was seen as f l e x i b i l i t y ; congressional passage was seen as enabling le g is la tio n to them to get th e ir agencies g o i n g . T h e was to get programs started and work out the d e ta ils la te r . idea Id e a lly , th is is not planning and not the way to le g is la te national health pol­ ic ie s ; pragm atically, i t 1s the quicker appearing and perhaps the only way to get programs established. Of course herein lie s the dilemma: I t may be true th at no p o litic ia n can s e ll an expensive health program to his con stituen ts, but u n re a lis tic costing leads to a public th a t may be­ come increasingly disenchanted w ith federal health programs which cannot liv e up to the expectation placed on them by Congressional and Executive rheto ric .1 8 Assuming Fo ltz is c o rre c t, EPSDT evaluation 1s especially warranted to learn whether the program is meeting it s o rig in a l expectations and, i f so, to thereby e m p iric a lly strengthen the program's reasons fo r e x is t­ ence. The Social Security Amendments o f 1967 (PL 90-248), including EPSDT, passed both houses a fte r an eight-month le g is la tiv e history and were signed into law on January 2, 1968. The law called fo r program Implementation by July 1, 1969 but I t was not u n til June, 1972, four and a h a lf years a fte r le g is la tiv e au th o rizatio n , th a t HEW Issued fin a l 16Budd N. Shenkin, M .D., "P o litic s and the Health of Children," op. c i t . , p. 885. Ib id . , p. 884. 18 Anne-Marie F o ltz , "The Development of Federal Policy: Early and Periodic Screening, Diagnosis and Treatment (EPSDT)," op. c i t . , p. 60. program regulations and guidelines. Even then, states were given u n til July 1, 1973 to implement the program fo r a l l age groups. The long delay which followed le g is la tiv e passage was apparently the re s u lt of HEW attempts to resolve a t le as t some o f the program's le g is la te d ambiguities p rio r to implementation. IQ For example, as mentioned, the law a c tu a lly assigned EPSDT to two existin g programs, T it le XIX - a w elfare program and T it le V - a health program. What was the relatio n sh ip to be between these two programs and th e ir agen­ cies? Which was responsible fo r EPSDT? According to F o ltz , HEW be­ came a "battleground" as various groups lobbied and advocated fo r th e ir version of the program. 20 The controversies, avoided in le g is ­ la tiv e passage, erupted a t the stage o f fashioning reg ulations. Wel­ fare and health agency representatives from both the state and national level were involved as was the National Welfare Rights Organization (NWRO), c h ild health advocates, Congress and HEW's program proponents. Each had th e ir own vision fo r the program. States were p a rtic u la rly in flu e n tia l and, fearing program costs, were the main fa cto r causing the delay in Implementation according to F o ltz. 21 During th is period EPSDT was considerably shaped although many o f the o rig in a l contentions were not resolved and s t i l l remain, p a rtic u la rly the issue o f state versus national program control and concerns regarding program cost and Impact. A key personality in th is formative stage fo r the program was Wilbur Cohen, then HEW Secretary and long-time U niversity o f Michigan 19 ad m in istrato r. As S ecretary, Cohen had ultim ate re s p o n s ib ility fo r promulgating the program's regulations and guidelines and his decisions were c ru c ia l. For example, Cohen resolved the Issue o f a d m in istrative re s p o n s ib ility fo r the program by simply asking T i t l e XIX, but not T i t l e V, to develop program reg u latio n s. Cohen did th is even though the le g is la tio n c a lle d fo r EPSDT regulations in both programs. (An in depth study o f Cohen's ro le in EPSDT would lik e ly be very help fu l fo r understanding the program's e a rly h is to r y .) Even w ith issuance o f fin a l regulations in 1972 and Congressional passage in th a t same year of a penalty provision fo r states with d e fic ­ ie n t programs, implementation s t i l l moved slow ly, or not a t a l l , in most s ta te s . This prompted legal aid attorneys in many states to i n i ­ t ia t e class action su its to get the program s ta rte d . good source fo r recounting these in it ia t iv e s . Peterson is a Generally they were very successful and resulted in many states beginning t h e ir programs in 1973-74 under court order to do so. 22 Michigan implemented it s pro­ gram w ith in three months of a January, 1973 United States d i s t r ic t court order req u irin g implementation. And, once begun, s ta te programs continued to become accepted as a standard, y e t unique, Medicaid bene­ f it. Nonetheless, even though implemented, programs developed rath e r slowly as w ell as d iffe r e n tly across s ta te s . The period since 1973-74 might w ell be considered as a " s ta r t up" phase fo r the program, one in which 1t became In s titu tio n a liz e d . 22 Dramatic changes did not occur; Eric Peterson, "Legal Challenges to Bureaucratic D iscretion : The Influence o f Lawsuits on the Implementation o f EPSDT. Health Policy P roject Working Paper No. 27," (New Haven: Yale U n iv e rs ity , 1975). 20 but the program did operate, children did p a rtic ip a te and th is p a rtic ip a tio n e ith e r d id , or did n o t, have an e ffe c t on the re c ip ­ ients o f service. Several c h a ra c te ris tic s of the Michigan program are worth noting in th is context as they support the choice o f the s tate as a study s ite : Once implemented, the Michigan program quickly began screening large numbers o f ch ild ren and by la te 1975 had screened a quarter m illio n youngsters. 23 C u rrier says th a t by October, 1976 Michigan had done 10% of a l l EPSDT screenings done thus fa r in the United S t a t e s .^ The Michigan program has continued to screen over 100,000 child ren per year and th is substantial p a rtic ip a tio n ra te , plus the existence o f a q u ite heterogeneous population in terms of ra c ia l and urban /ru ral mix­ tu re , make Michigan a good state in which to study the program. In summary, review o f EPSDT's o rig in s and h is to ry reveals a pro­ gram conceived and qu ickly le g is la te d a t the national le ve l but one which has experienced a slow pace o f actual implementation. in a l purposes were apparently several: It s o rig ­ increase access to medical services fo r those in need with the expectation th a t p a rtic ip a n ts ' health status w ill be subsequently improved and medical costs reduced. Before reviewing studies which address how well the program 1s meeting it s expectations, mention should be made th a t a considerable body of "program lite r a tu r e " has been published, much o f 1t by the federal ^Thomas R. K irk , M .D ., e t a l . , "EPSDT - One Quarter M illio n Screenings in Michigan," Public Health B rie fs , LXVI (May, 1976), 492-84. 24 Richard C u rrie r, MA, "Is Early and Periodic Screening, Diag­ nosis, and Treatment (EPSDT) Worthwhile?," Public Health Reports, XCII (November-December, 1977), 527-36. 21 government, which b a s ic a lly provides Information of an operational or "how -to-do-it" nature. An excellen t guide to th is lite r a tu r e is the recently published EPSDT: A Selected Annotated Bibliography, which l is t s over one hundred EPSDT a r tic le s and reports. 25 While th is l i t ­ erature is not d ire c tly relevant to purposes o f th is study, and ac­ cordingly w ill not be reviewed here, i t does provide a deeper and more comprehensive understanding o f the program and o f course i s of In te re s t to program personnel since i t constitutes the program's "technical l i t ­ eratu re. " Outcome Studies o f NonEPSDT Screening Programs Multiphasic screening programs and the physical examination are of in te re s t r e la tiv e to the EPSDT program since they are screening a c t iv it ie s . While they may d if f e r in specifics such as scope of te s t­ ing or type of te s t adm inistrator, th e ir basic purpose 1s iden tical to EPSDT's - shorten the time in te rv a l between onset and detection o f med­ ic a l problems and thereby expedite recovery. In evaluating th e ir e f­ fectiveness a number of studies have used m o rta lity rates as the out­ come v a ria b le . Studies of M o rta lity Rates In the e a rly 1920s, Knight Id e n tifie d the number of deaths occurring to some 6000 holders of ordinary l i f e Insurance who had volunteered to receive fre e periodic examinations between 1914 and 1915. Five and one h a lf years a fte r the examinations, actual deaths to taled 217 among this 25 United States Department o f Health and Human Services, Health Care Financing A dm inistration, EPSDT: A Selected Annotated Bibliography, (Washington, D.C.: Government P rinting O ffic e , 1980). 22 group compared with an expected fig u re of 303 deaths. Knight a t t r i ­ buted the 28% reduction in m o rta lity to the examinations and estimated the res u ltin g monetary value to the company to taled more than $126,000 fo r a cost o f about $40,000. He gave no information about the medical care received by those screened or those not screened. 26 More re c e n tly , Thorner and Crumpacker reported th a t the m o rta lity ra te of executives who p a rtic ip ate d 1n a periodic health examination was less than the rate o f the general population of white males in the United States. The authors f e l t th is differen ce was most lik e ly due to the higher socio-economic level o f those examined and the generally b e tte r level of medical care av a ilab le to them. 27 Roberts, e t a l. studied m o rta lity rates fo r 20,648 men, mostly white executives, who had received employer-sponsored examinations in the northeastern United States between 1950 and 1964. Their m o rta lity rate was compared with the rates o f white males in the general popula­ tio n } w hite, professional males; and two groups of white males receiving c e rta in special classes of l i f e insurance. The study group had a lower m o rta lity rate than three o f the comparison groups and a rate equal to "p re ferre d -ris k males" receiving premium l i f e Insurance. Because the selection process fo r the l a t t e r group excluded those w ith c erta in de­ fects and diseases (not s im ila rly excluded from the study group), Roberts considered i t noteworthy th a t the study group did as well as, not worse 2®A.S. Knight, "Value of Periodic Medical Examination," S ta tis tic a l B u lle tin o f Metropolitan L ife Insurance Company. 2:1 1921 c ite d in t a r b e rt J. Roberts, e t a l . , "M o rta lity Among rales In Per1od1c-Health-Examination Proqrams," The New England Journal of Medicine, CCLXXXI (J u ly , 1969), 20. 27 Robert M. Thorner and E. L. Crumpacker, "M o rtality and Periodic Examinations of Executives," Archives Of Environmental H ealth, I I I (July-December, 1961). 23 than, th is select group. However, he was not w illin g to a ttr ib u te the lower m o rta lity rate to receip t of the examinations, noting th a t a s e lf selection process was operative fo r those receiving examinations and th at th is might have affected outcomes. Roberts did not id e n tify what, i f any, measures were undertaken to insure treatment fo r those determined by the screening to be 1n need o f service. 28 The Commission o f Chronic Illln e s s (CCI) conducted a multiphasic screening c lin ic in Baltimore as part of a 1954 morbidity survey. Five years la te r Wylie found no differen ce in the m o rta lity rates of those screened and those who refused screening and concluded there was no basis fo r believing p a rticip an ts had ben efltted from screening. 29 How­ ever, a twelve year follow -up by K u ller and Tonascia disclosed that those screened, and especially the white females who were screened, had a b e tte r survivorship than those who refused screening. These d if f e r ­ ences were apparently unrelated to variation s in history of chronic disease or d is a b ility a t entry to the study. However, because s e lf selection was operative in the CCI study, K uller and Tomascia concluded the selection bias fo r screening could i t s e l f account fo r the difference in outcome and d e fin itiv e conclusions regarding the value o f screening were not p o s s ib le .^ 28 Norbert J. Roberts, op. c i t . , pp. 20-24. 2®C. M. W ylie, "P a rtic ip a tio n 1n M u ltip le Screening C lin ic With Five-Year Followup," Public Health Report, LXXVI (J u ly , 1961), 596602* 30 Lewis K uller and Susan Tonascia, "Commission o f Chronic I l l ­ ness Follow-Up Study: Comparison o f Screened and Nonscreened In d i­ vid u als," Archives Of Environmental H ealth, XXI (November, 1970), 656-65. 24 CCI apparently did not follow up to insure th at needed treatments were received. Thus, as in other studies where t r e a t ­ ment services are not linked to screening fa ilu re s , i t 1s not known to what extent those screened and found to be in need of treatment received any service other than the screening i t s e l f . In y e t another study concerned with m o rta lity ra te s , Gordon analyzed an epidemiological study o f heart disease which randomly selected 6507 persons fo r examinations and a c tu a lly examined 68.8% of these. The m o rta lity rates fo r two years subsequent to screening were twice as high fo r the unscreened group as fo r the screened group. While the study made no e f f o r t to assure treatment fo r detected problems, examination findings were a va ila b le to the physician. 31 In a l l the c ite d m o rta lity studies, a methodological problem has been the presence of a s e lf selection fa c to r in sorting out those re ­ questing examinations fo r those refusing examinations. Enterline and Kordan had a unique opportunity to get around th is problem and to approximate a controlled study. They compared m o rta lity rates fo r two groups of persons who p articip ated in chest x-ray surveys in Texas and C a lifo rn ia . Screening was done fo r heart disease and tuberculosis. With one group, problems were id e n tifie d Immediately upon the i n i t i a l reading o f the photo-fluorograms and were referred fo r diagnosis. The second group consisted o f those who had a problem a t the time of screening but th e ir film was misread. Consequently, they were not re ­ ferred u n til several years la te r follow ing a second reading of the same 31 Tavla Gordon, e t a l . , "Some Methodological Problems in the Long-Term Study o f Cardiovascular Disease: Observations on the Fram­ ingham Study," Journal o f Chronic Diseases, X (September, 1959), 186206. 25 film . The group id e n tifie d a t the f i r s t reading had a b e tte r sur­ vival rate.**2 In summary, several commonalities are present in the m o rta lity ra te studies. Outcome differences favoring program particip an ts were quite consistently obtained. However, generally loose research meth­ odologies were equally evident and thereby jeopardize the fin d in g s. One pervasive concern Is the v ir t u a lly unavoidable s e lf selection fa c to r. This caused several researchers to be cautious in th e ir con­ clusions although 1t was controlled fo r E nterline and Kordan whose resu lts supported the trend o f fin din gs. Secondly, programs apparently did not ensure th a t treatment services would be a v a ilab le fo r those whose screening resu lts indicated a re fe rra l need. C le a rly , screen­ ing by i t s e l f w ill not contribute to improved health. I f screening were the sole service received In the m o rta lity studies, i t is not reasonable to c re d it the program with differences in outcome. A th ird general q u a lific a tio n regards the use of death rates as the sole de­ terminant of program effectiven ess. This ind icato r is much too spe­ c i f i c , being insen sitive to many changes which are important but are not of a l i f e or death magnitude. Obviously, many meaningful and In te re s tin g changes occur 1n health status which th is in d ica to r can not measure. Studies By Health Maintenance Organizations As prepaid health care plans, HMOs have a structured Incentive to minimize costs. 32 I f unplanned, and thereibre unbudgeted, costs occur P h ilip E. E n terline and Bernard Kordan, "Controlled Evaluation of Mass Surveys fo r Tuberculosis and Hearth Disease," Public Health Reports, LX X III (October, 1958), 867-875. 26 they must be absorbed by the HMO rather than being passed on to the consumer or th ird party payee as is done in the tra d itio n a l health care system. Thus, i t is not surprising some HMOs have shown major in te re s t in using and researching health screening. The Health Insurance Plan o f New York (HIP) was one o f the e a r lie s t prepaid medical plans in the United States. In the la te 1960s, HIP conducted a large scale study involving two random samples, each consisting o f 31,000 women. HIP members. The women were ages 40-64 and a ll The study was concerned w ith mammography and c lin ic a l examination o f the breast. The study group was recruited fo r screen­ ing examinations while the controls followed th e ir usual medical prac­ tic e s . 65% of the study group received the i n i t i a l screening and a large percentage o f these received subsequent rescreenings. A fte r fiv e years o f follow up, the study group had about a 1/3 lower m o rta lity rate from breast cancer than did the control group. However, th is re ­ duction in breast cancer was inexplicably found only fo r those a t ages over 50 and not a t ages 40-49. 33 The Kalser-Permanente Medical Care Program o f C a lifo rn ia is lik e ly the best known prepaid health care plan (HMO) in the United States. In 1964 they undertook a c o n tro lle d , longitudinal study to evaluate the effectiveness o f periodic health examinations. Two samples were randomly selected from th e ir population of enrollees with each sample having approximately 5000 persons, age 35-54. The study group was urged to undergo an annual examination and approximately 65% did so. 33 The Sam Shapiro, "Evaluation o f Two Contrasting Types of Screening Programs," Preventive Medicine, I I (June, 1973), 266-277. 27 comparison group was not so encouraged and sought medical care on th e ir own in it ia t iv e w ithin the Kaiser program. However, approx- imately20-40% o f the comparison group sought examinations each year without urging and by the end o f seven years, 52.8% had received a t le a s t one examination. 34 Several Kaiser researchers have reported study findings. Ramcharan e t a l . analyzed s e lf reports made by study and control groups who completed questionnaires mailed to them b ie n n ia lly . A fter fiv e to seven years of examinations, older study males (age 45-54 a t entry to study) had: (a) a reduction 1n s e lf-ra te d d is a b ility and re­ ported time loss from work, (2) a greater proportion working and (3) a lower s elf-rep o rted u t iliz a t io n of medical services by the sick. How­ ever, no differences were reported on any o f these variables fo r young­ er females and younger males. is not known. Why these age-sex differences occurred The older study males did not report the presence of fewer chronic conditions. not be reduced but i t - Thus, the incidence o f these conditions may may be' b e tte r controlled as evidenced by the older study men reporting less d is a b ility and lower u t iliz a tio n of health services.35 Indicators reported by Dales e t a l . disclosed fewer differences between the study and control groups. Outpatient u t iliz a t io n fo r the physician and laboratory tests was quite s im ila r although the study 34 John L. C u tle r, e t a l . , "Multiphasic Checkup Evaluation Study. 1. Methods and Population," Preventive Medicine, I I (June, 1973), 199206• 35 S a v itri Ramcharan, e t a l . , "Multiphasic Checkup Evaluation Study: 2. D is a b ility and Chronic Disease A fte r Seven Years of Multlphasic Health Checkuups," Preventive Medicine, I I (June, 1973). 28 group had more diagnoses fo r 26 o f 188 s p e c ific diagnoses ( p <_ .0 5 ). The reverse was not found fo r any diagnosis. This d iffere n c e lik e ly occurred because the study group had more m ultiphasic checkups. The number o f h o s p ita liz a tio n s also did not d if f e r appreciably between the two groups w ith the exception th a t old er study women, and to a lesser extent younger women, were h o sp ita lized more. The authors thought these h o s p ita liz a tio n s may have been f o r p re ven tive-th era­ peutic reasons ra th e r than a re s u lt o f advanced disease since most were fo r surgery and gynecology service. In comparing m o rta lity ra te s , no major differences was found in o v era ll ra te but fo r " p o te n tia lly postponable" causes o f death (c e rta in cancers, hypertension, in tr a ­ c ra n ia l hemmorrhages), the control group ra te was twice the study group ra te ( p < .0 5 ). Most o f th is d iffere n c e was due to colon and re c ta l cancer and hypertensive associated causes. Dales suggested th is d iffere n c e might have been due to the study group's re c e ip t o f screening and subsequent followup since s ig n ific a n tly more cases o f hypertension and benign growths o f the colon were diagnosed in out­ p a tie n t c lin ic s . In a d d itio n , p re sc rip tio n dispersal fo r antihyp er36 tensive agents was found to be higher fo r the study group. In a c o s t-b e n e fit analysis o f the screening program, Collen e t a l . concluded th a t over a seven year period of time a net saving of some $800 per man (fo r men age 45-54 a t en try) could be a ttrib u te d to the screening program. The d iffe re n c e p rim a rily re fle c te d the lower d is a b ilit y and m o rta lity rates which enabled the men to work 36 Lorlng G. Dales, e t a l . , "M ultlphasic Checkup Evaluation Study. 3. O utpatient C lin ic U t iliz a t io n , H o s p ita liz a tio n , and M o rta lity Ex­ perience A fte r 7 Years," Preventive M edicine, I I (June, 1973) 221-235. 29 more and longer and consequently to earn more Income. S im ila r d i f - ferences were not demonstrated fo r younger men or fo r women. Findings from the HMO studies are thus mixed. 37 While differences favoring program p a rtic ip a n ts were found, they were not con sisten tly obtained, varying freq u en tly by age and sex fo r unknown reasons. How­ ever, both the Kaiser and HIP studies had complications which would serve to underestimate true d iffere n c es . As noted, the Kaiser study had a sizab le crossover on the screening fa c to r { s lig h tly more than o n e-h alf o f the controls were even tu ally screened) w hile 1n the HIP study only tw o-thirds o f the study group was screened. As a re s u lt, the obtained outcomes, w hile not strongly supportive o f the program v a ria b le , do give p o s itiv e ind icatio ns o f program e ffe c t although o v e ra ll they must be considered Inconclusive. Outcome Studies o f the EPSDT Program Two general approaches have been used 1n an attempt to evaluate the influence o f EPSDT on c h ild health pattern s. One approach has compared cost and u t iliz a t io n rates fo r screened and unscreened e l i g ib le s . The general assumption is th a t these ind icato rs might be i n i t i a l l y higher fo r those screened (because o f re s u ltin g r e fe rr a l needs) but on a longer term basis they should be lower. The second strategy has compared what are e s s e n tia lly r e fe rr a l rates fo r I n i t i a l and repeat screenings. The hypothesis Is th a t r e fe rr a l rates should be lower fo r those rescreened which would be considered In d ic a tiv e 37 Morris F. C ollen, e t a l . , "M ultiphaslc Checkup Evaluation Study: 4. Prelim inary Cost B en efit Analysis fo r Middle-Aged Men," Preventive Medicine. I I (June, 1973), 236-246. 30 o f b e tte r health. Review o f these studies follow s. The Community Health Foundation (CHF) compared cost and service u t iliz a tio n data fo r screened and unscreened e llg lb le s In two North Dakota communities. 38 Diagnosis and treatment data was gathered from the Medicaid claims f i l e fo r a one year period. I t was apparently during th is same in te rv a l th a t the te s t group received th e ir screening. 410 children screened in Minot were compared with 1662 unscreened children 1n Minot and 1920 unscreened children In Bismarck. Results are given below: U tiliz a tio n Differences 1. Those screened used 21 to 30 percent fewer in p a tie n t hospital services.39 2. Those screened used more services in the physician (103%-178%), dental (65%-79%) and outpatient hospital (24%) categories. Cost Differences 1. Total per capita expenditures (Including screening costs) were 36-44% lower fo r the screened group. 2. Per capita expenditures fo r In p a tie n t hospital services were 47-58% lower fo r those screened. 3. Per capita expenditures fo r pharmaceuticals were 18 to 21% lower fo r those screened. 4. Per capita expenditures fo r physician services were 6 to 65% higher fo r the screened group. 38 Comnunlty Health Foundation, "Cost Impact Study Of The North Dakota EPSDT Program," (Evanston, Il lin o is : Community Health Foun­ dation, 1977). (Mimeographed.) 39 In th is example, and fo r those which fo llo w , the f i r s t per­ centage represents the difference between those screened and those not screened 1n the te s t (Minot)community. The second percentage represents the difference between those screened and those not screened in the control (Bismarck) community. 31 5. Per capita expenditures fo r dental services were 17% higher fo r those screened than fo r those not screened in Minot (th e te s t community). However, these expen­ ditures were 2% lower fo r the screened persons than fo r unscreened persons 1n Bismarck (th e control com­ m unity). 6. Per capita o p tic a l expenditures were 71% higher fo r those screened than fo r those not screened in the te s t community but 3% lower fo r the screened persons than fo r the unscreened persons in the control community. This study appears to show the desired re la tio n s h ip between par­ tic ip a tio n in EPSDT, appropriate p a rtic ip a tio n in the health care sys­ tem and improved health. Those screened used fewer in p a tie n t services, more ambulatory services and incurred lower medical costs than those not screened. U tiliz a t io n and expenditure patterns gen erally moved in the same d ire c tio n . However, the CHF cautioned th a t the obtained re­ la tio n sh ip was not necessarily one o f cause and e ffe c t because the s e lf se lectio n process might have resulted in children who were i n i t i a l l y more healthy being the ones who were screened. Also, i t seems u n lik e ly th a t the program is s u ffic ie n tly powerful to reduce in p a tie n t hospital services by 20-30% w ith in only one year. A second study concerned w ith cost u t iliz a t io n was done by Applied Management Sciences (A M S ).^ AMS selected 800 screened and 800 unscreened children from each o f two states and examined the Medicaid claims f i l e fo r the year p rio r to screening, the screening year i t s e l f and the year a f te r screening. Selected findings are displayed 1n the follow ing two ta b le s : Applied Management Sciences, Assessment o f EPSDT Practices and Costs Report on the Cost Impact of the EPSDT Program (S ilv e r Spring, Maryland: Applied Management Sciences, 1976). 32 Reference Table I I . Per capita service u t iliz a t io n fo r a l l services by year o f service and s ta te . State 1 EPSDT NonEPSDT Year 1974 5 .8 6 ^ 1975 10.70 1976 7 .5 8 ** 8 .0 3 ^ 10.17 9 .2 6 ^ State 2 EPSDT NonEPSDT 10.65^ 12.64^ 14.05 14.92 14.04 15.36 ♦D ifference between EPSDT and nonEPSDT sample is s t a t is t ic a lly s ig n if i­ cant a t the .05 le v e l. ♦♦D ifference between EPSDT and nonEPSDT sample is s t a t is t ic a lly s ig n if i­ cant a t the .01 le v e l. In s tate 1 (southern and r u r a l) u t iliz a t io n increased 29 percent fo r the screened group and 15 percent fo r those not screened from 1974 to 1976. AMS argued th a t i f the screened group had experienced the same ra te o f change as the unscreened group, t h e ir u t iliz a t io n would have increased to only 6.74 rath er than 7.58 services. The .84 un its of a d d itio n al u t iliz a t io n (nearly one v i s i t ) is about a 12% improvement a ttr ib u ta b le to the EPSDT program. S im ila r reasoning w ith state 2 (northern and in d u s tria l) data shows u t iliz a t io n increased 8 .5 per­ cent above what I t would have in the absence o f the program. Reference Table I I I . Per capita costs fo r a ll services by year o f service and s ta te . Year State 1 EPSDT NonEPSDT State 2 EPSDT NonEPSDT 1974 $ 8 5 .3 0 ^ $ 1 1 5 .2 5 ^ $ 1 4 6 .9 4 ^ $ 1 9 6 .3 6 ^ 1975 $153.04 $143.53 $198.07^ $254.75^ 1976 $117.27 $129.39 $216.98 $243.01 ♦D ifference cant a t the ♦D ifference cant a t the between EPSDT and nonEPSDT sample is s t a t is t ic a lly s ig n if i­ .05 le v e l. between EPSDT and nonEPSDT sample is s t a t is t ic a lly s ig n if i­ 0.1 le v e l. 33 In state 1 costs Increased 12% between 1974 and 1976 fo r EPSOT non­ p a rtic ip a n ts . Applying th is ra te of change to the screened group suggests th e ir 1976 costs would have totaled $85.53 without program p a rtic ip a tio n . The actual change to $117.27 was equal to a 23% in ­ crease a ttrib u ta b le per person). to the program ($117.27 - $85.53/$85.53 or $21.74 The comparable changes in state 2 were 19 percent and $35.13 per person. In summary, rates of change fo r both service u t iliz a t io n and costs Increased more ra p id ly fo r the screened group in both states although th e ir levels of cost remained lower w ith the exception of the screening year I t s e l f In the rural s ta te . The Increases 1n usage suggested that EPSOT could improve, a t least tem porarily, access to health services fo r poor children and th a t the increased costs resu ltin g from the pro­ gram did not appear to pose a substantial burden to Medicaid. At the same tim e, the AMS study did not demonstrate any short-run cost savings associated with the EPSDT program. The second type o f approach fo r estim ating EPSDT's impact on health is demonstrated by C u rrie r's d iffe r e n tia l analysis of re fe rra l rates. 41 He found th a t during the f i r s t h a lf o f 1976, 62% of those i n i t i a l l y screened were referred as compared with a 49% re fe rra l rate fo r those rescreened. This is a 21% reduction in r e fe rr a ls . A s im ila r Michigan review fo r calendar year 1977 showed these rates to be 62% and 51% respectively (an 18% re d u c tio n ).^ These data suggest th a t Increased 41 Richard C u rrie r, “Is Early and Periodic Screening Diagnosis and Treatment (EPSDT) Worthwhile?," Public Health Reports, XCII (NovemberDecember, 1977), 527-36. 42 Michigan Department o f Public Health and Michigan Department of Social Services, Health Screening: A Call To A B etter L ife , Michigan Annual Report, 1977, (Lansing, Michigan, 1978). 34 contact with the program, as evidenced by rescreening p a r tic ip a tio n , re s u lts 1n fewer health problems. This 1s what the program 1s supposed to accomplish. A te c h n ic a lly sophisticated EPSDT outcome study was rec en tly com­ pleted by Philadelphia Health Management Corporation (PHMC), an EPSDT 43 screening provider in Pennsylvania. They also analyzed health data already stored on computer f i l e but added several procedural techniques fo r the purpose o f protecting the study's "In te rn a l v a lid it y ." The ad­ vantages they note fo r using already obtained data are s ig n ific a n t, namely unobtrusiveness and not adding to service costs through primary data c o lle c tio n . D ire c t service workers and c lie n ts are usually e ith e r unable or un w illin g to a s s is t research projects and paying fo r th e ir assistance becomes expensive. Thus, not only are the PHMC findings important but the methodological adjustments they made are o f In te re s t fo r th e ir contributions in strengthening the "ex post facto" mode o f data analysis. PHMC assessed outcomes in outreach, ris k i d e n t i f i ­ cation and, o f p a rtic u la r relevance to th is study, ris k reduction. Risk reduction was measured by the change In the "health status index" (o r "abnormality ra te " 4^) which equaled: i ic _ ATA TTA-NA where: 43 P hiladelphia Health Management Corporation, A Study o f the Process, E ffe c tlveness, and Costs of the EPSDT Program In Southeastern Pennsylvania, P art I I I I (P h ila d e lp h ia , Pennsylvania), 1980. The "abnormality rate" 1s simply a subset of the health status Index where analysis 1s focused on some, rath er than a l l , of the te s t areas. 35 HS - health status Index; ATA * to ta l number of abnormal te s t areas where treatment is required; TTA * the number o f te s t areas in which a tre a ta b le abnor­ m ality can be found m u ltip lie d by the number of children screened; NA * to ta l number of te s t areas not assessed, an adjustment to elim inate TAs not assessed fo r a given number of child ren . The lower the HS index, the h e a lth ie r the subjects. Their research model, from Campbell and Stanley,^* was: Two-screen sample 01 One screening, occurring when 01 is screened 03 X One screening, occurring when 02 is screened 02 04 The "0" represents an observation a t a given tim e, I . e . a screening. The "X" represents an "experimental treatm ent," i . e . a r e f e r r a l. A l­ though PHMC does not state that a ll those screened were also referred (which would not usually be the case), they note th at the focus is on the outcome of exposure to a screening, those who have had th is exposure and those who have not. The comparisons are between 01 and 02, a lon gi­ tudinal comparison and 02 with 04, a cross-sectional one. Several control procedures were used to v a lid a te findings: 1. Since the 01-02 longitudinal comparison is subject to possible "Instrument" and h isto ry e ffe c t,^ * an 03 to 04 comparison was made fo r 45DonaldT. Campbell and Julian C. Stanley, Experimental and QuasiExperimental Designs For Research, (Chicago: Rand McNally and Company, *®Donald T. Campbell and Julian C. Stanley, op. c i t . , pp. 7-9. 36 the purpose of determining whether time i t s e l f is an in d ire c t fa c to r. PHMC found the 04 HS Index was 21% higher than the 03 HS index and concluded the screening protocol had become more rigorous over time. Since the unadjusted 02 HS was nearly 8% lower than the 01 index, 21% was added to th is 8% differen ce to y ie ld an adjusted reduction o f almost 30% in the 02 index as compared with the 01 HS index. Thus, PHMC determined th a t the incidence of problems decreased nearly 30% fo r the same children over a two year period o f time. 2. Since the cross-sectional comparison of 02 to 04 was subject to selection and regression e ffe c ts , an adjustment was made based on an 01-03 comparison. The HS index fo r the 01s was found to be 26% higher (+26%) than the HS fo r the 03s, indicating th at the longitudinal sample (01s) was I n i t i a l l y a more sick ly group. Since the HS fo r the 02s was 5% lower (-5%) than the HS o f the 04s, the +26% difference was subtracted from the -5% y ie ld in g an adjusted HS fo r the 02s 31% lower than the HS fo r the 04s. Again, those p a rtic ip a tin g in the pro­ gram (the 02s) had about 30% fewer abnorm alities than the n o n p a rtic ipant comparison group. 3. A th ird possible confounding problem was th a t o f maturation in making the longitudinal comparison. S im ila rly , the cross-sectional comparison could be Invalid ated by age differences between the two groups. used. To control both s itu a tio n s , an age-adjustment procedure was B a s ic a lly , a weighted mean was derived which expressed the HS o f the 01s and 04s as i f these groups had the same age d is trib u tio n 37 as the 02s. PHMC observed th a t th e ir model was unable to control fo r a possible in te ra c tio n (sele ctio n -m a tu ratio n ) e ffe c t in the lo n g i­ tudinal comparison or fo r a possible experimental m o rta lity e ffe c t in the 02-04 comparison. However, in th is p a rtic u la r study PHMC argued th a t these uncontrolled facto rs did not confound the re s u lts . In summary of the EPSDT studies, PHMC's central fin d in g was th a t the rescreened group (02) had an approximately 30% lower o verall abnormality ra te compared w ith i t s e l f (01) over time or compared w ith the control group 04 (p < .05 fo r both comparisons). These re s u lts are consistent w ith , and q u ite s im ila r to , C u rrie r's fin d in g th a t r e fe r r a l rates were 20% lower fo r those being rescreened as compared w ith those receiving an I n i t i a l screening. Results from both studies support the view th a t program p a rtic ip a tio n is b e n e fic ia l. Reconciling the CHF and AMS studies is a b it more d i f f i c u l t . CHF's study was a comparison o f EPSDT p a rtic ip a n t and nonparticipant costs and u t iliz a t io n during the screening year only. They found the use o f amublatory services was higher fo r the p a rtic ip a n ts but th a t t h e ir o verall costs remained lower than those Incurred by the nonpartlcipants. 4? k HS* * Z 1 AMS 1n making the same comparison found no (N i) (HS1)/N where HS* * the age adjusted HS; k * the number o f age classes; N1 - number o f te s t areas assessed in the 1th age group o f the standardizing group 02 ( I . e . , TTA1 - NA1); HS1 ■ HS fo r the 1th age group o f the standardized group, in th is case, 01 or 04. 38 s ta tis tic a l difference between p articip an ts and nonparticipants except fo r costs in the northern, in d u s tria l state which were s ig n ific a n tly lower fo r p a rtic ip a n ts . In the AMS three year longitudinal comparison, u t iliz a tio n and costs increased a t a fa s te r rate fo r the particip ants (approximately 10% and 20% fa s te r resp ectively) although to ta l p a r t ic i­ pant costs were lower a t each stage o f the study. Thus, both CHF and AMS studies found EPSDT associated with a higher use of certain medical services although EPSDT users, in spite o f th e ir increased service use, s t i l l incurred lower medical costs than those not p a rtic ip a tin g in the program. Relationship of Demographic Factors to Health An analysis o f the rela tio n sh ip between race and health is faced with several problems.4® F ir s t , since whites are more a fflu e n t than m in o ritie s , income is a v a ria b le . However, studies comparing r a c ia l, health differences seldom control fo r socioeconomic status. Secondly, most data concern m o rta lity ra te s , c e rta in ly an appropriate and impor­ tan t variable but one which is nonetheless not sensitive to any d i f ­ ferences less extreme than l i f e or death. As Reid recen tly wrote: . . . the data on illn e s s and d is a b ility are so new or so Inadequate th at i t is d i f f i c u l t to establish trends to make s t a t is t ic a lly sound conclusions on the subject of m inority health except from data on m o rta lity . ( I . e . , information recorded on death c e r t if ic a t e s .) 48 In what fo llo w s, "race" Is used in a nontechnical sense to re fe r to whites, blacks, Spanish-speaking and American Indians. 40 John D. Reid, Everett S. Lee, Davor Jedlicka and Yongsock Shin, "Trends in Black H ealth." Phylon, XXXVIII(June, 1977), 105-116. 39 Given these q u a lifie r s , discussion o f the ra c ia l variab le follow s. The main point o f the discussion Is th at blacks have poorer health than w hites, a fa c t o f p a rtic u la r in te re s t in Michigan where blacks are a large proportion o f the e lig ib le population. Blacks Succinctly p u t, the s itu a tio n 1s th a t blacks have higher death rates than whites fo r a ll the major causes o f death except suicide. Black-white differences e x is t even before b ir th . 50 Death o f the fetus w ith in the womb is more comnon among blacks than among w hites. Also, newborn blacks are more lik e ly than whites to die during the f i r s t year o f l l f e . ^ 1 Lee notes th a t the chances o f anyone dying from c h ild b ir th in the United States are exceedingly low, less than 1 woman per 1000. How­ e ver, she says there are black-w hite d ifferences and the differences have widened during the tw entieth century a t the same time rates fo r both groups were decreasing g re a tly . In 1973, the maternal m o rta lity rate fo r whites was 3% o f what 1t had been nearly s ix ty years previous. However, the 1973 black ra te was 4 .5 percent o f the much higher rate i t had recorded in 1915 (11/1000 fo r blacks versus 6/1000 fo r whites In 1 9 1 5 ).52 50 Davor J e d llc k a , Yongsock Shin and E verett S. Lee, "Suicide Among Blacks," Phylon, XXXVIII (December, 1977), 448. ®*John D. Reid, e t a l . , "Trends In Black H ealth ," op. c i t . , p. 105. „ Anne S. Lee, "Maternal M o rta lity in the United S ta tes ," Phylon, XXXVIII (September, 1977), 260, 262. 40 S im ila rly , Kovar found the same pattern in studying the trend o f m o rta lity rates between 1950 and 1975. For white in fa n ts , the death rate was 26.8 per 1000 in 1950 and 14.2 per 1000 in 1975, a decrease o f 47 percent. For black in fa n ts , the comparable rates were 43.9 in 1950 and 26.2 in 1974, a decrease o f 40 percent. Thus, although both rates decreased greatly in the 25 year period, i t decreased less fo r blacks, i . e . , the ra c ia l difference widened. Said d if f e r e n t ly , in 1950, the black in fa n t m o rta lity rate was 64 percent higher than the rate fo r white in fa n ts . However, by 1975, the black rate was 85% higher than the white ra te . This means the black in fa n t born in 1975 had a b e tte r chance o f surviving than a black child born In 1950 but a poorer chance o f survival than a white c h ild also born in 1975. These data are p a rtic u la rly in te re s tin g since in fa n t m o rta lity is f r e ­ quently used as a single in d icato r o f national health status. Reid notes th a t among w hites, 106 males are born fo r every 100 females, and the number o f males remains larg er than th at o f females to about age 40. However, among blacks, only 103 males are born a liv e per 100 females, and before adolescence is over, there are more females 54 than males. Wilber says the prevalence o f high blood pressure among blacks 55 1s about twice as high as among whites. Yabura presents data docu­ menting th is claim. He says the death rate fo r high blood pressure ®^Mary Grace Kovar, "M o rta lity o f Black Infants in the United S tates," Phylon, XXXVIII (December, 1977), 370-97. ^John D. Reid, e t a l . , "Trends in Black H ealth," op. c l t . , p. 106. ®5Joseph A. W ilber, M .D., "Hypertension: An E d ito ria l," Phylon, XXXVIII (December, 1977), 353. 41 and related disease ts 58.4 per 100,000 population fo r blacks com­ pared w ith 27.1 per 100,000 fo r w h ite s .5® Moss and Scott c ite data published by the federal government's National Center fo r Health S ta tis tic s on the basis o f a 1974 national health interview survey. This showed proportionately more blacks than whites have hypertension, 22 percent versus 15 percent resp ectively. This pattern was present fo r a ll fiv e age groups except those age 17-24 y e a rs .57 White, in analyzing government s t a t is tic s , finds th a t cancer mor­ t a l i t y increased g re a tly between 1949 and 1967 fo r blacks. In 1949, the cancer m o rta lity rate fo r blacks was 8 percent lower than the white ra te . By 1967, the black rate was 18 percent higher. The to ta l number o f deaths in the black population increased 93 percent between 1947 and i967 while fo r whites the increase was 47 percent. The average annual rate of increase o f cancer m o rta lity was twice as high fo r blacks as fo r whites. Thus, the black death rate from cancer increased both in re la tio n to the e a r lie r black rate and in re la tio n to the rates of w h ite s .5® Kitagawa and Hauser analyzed m o rta lity rates fo r 1959-61. In comparing the more Important causes o f death, they found m o rta lity rates fo r blacks o f both sexes were greater than those fo r whites. For cardiovascular disease, the rate fo r black males was 10 percent eg Lloyd Yabura, "Health Care Outcomes 1n the Black Community," Phylon. XXXVIII (June, 1977), 196 c itin g Edythe Cudlipp, "High Blood Pressure: A Black Epidemic," Essence, IV (October, 1973), 44. 57Abiga11 Moss and Geraldine S cott, "Hypertension: United States, 1974," Phylon. XXXVIII (December, 1977), 357-58. Jack E. White, "Cancer Differences in the Black and Caucasian Population," Phylon. XXXVIII (September, 1977), 297. 42 higher than fo r white males while black females had a rate 50 percent higher than white females. For cancer, the ra te fo r black males was 22 percent higher than fo r white males and black females had a rate 12 percent greater than white females. rate from violence than whites. Blacks also had a higher death 16 percent of the deaths o f black males were from accident, suicide or homicide - 6 percent were from homicide alone. Almost a th ird o f the deaths o f black males aged 15-34 were due to homicide. In comparing the m o rta lity rates o f blacks w ith other m in o ritie s , the black ra te was highest fo r a ll ages fiv e and over. In y e t another way o f comparing black and white health s tatu s, Kitagawa and Hauser noted black-w hite differences in expectation of l i f e a t b ir th . They found blacks have a l i f e expectation s ix years less than whites. The d ifferen ce exists fo r both sexes. Black males liv e 62 years on average versus 68 years fo r white males; black females eg can expect to liv e 70 years versus 76 years fo r white females. Michigan's own program s ta tis tic s r e f le c t a higher incidence o f health problems among blacks. For the f i r s t fiv e years o f the program, the black r e fe rr a l rate averaged 26 percent higher than the white ra te , 67.4 percent versus 53.4 percent. The black rate was also higher than th a t o f the Spanish-speaking and American In d ia n .^ Black Differences By Sex Given the more problematic state o f black h e a lth , i t is also s ig n ific a n t th a t black males have poorer health than black females. ^E velyn M. Kitagawa and P h ilip M. Hauser, D iffe r e n tia l M o rta lity in the United States (Cambridge: Harvard U niversity f»ress, 1973), pp. 106-13. ^M ichigan Department o f Public Health and Michigan Department of Social Services, EPSDT Michigan Annual Report, 1978, op. c i t . , p. 24. Here the "R eferral rate" equals the percent o f screened ind ividuals who are re fe rre d . 43 Reid says, "Age by age, and almost cause by cause, death is more common among black males than among black females, . . . K ita ­ gawa and Hauser s ta te , "There is no doubt that the male 1s the weaker o f the sexes and is more lik e ly to die from almost a ll o f £2 the causes o f death th a t a ffe c t both sexes." They report the excess o f m o rta lity fo r males Is "considerable" a t ages 1-4 and from ages 5 to 45, the death rates fo r black males are two to three times those o f black females. For example, young black males have three times as high a death rate from accidents and suicide and almost six times as high a rate fo r homicide. For older In d iv id u a ls , black males have death rates from cardiovascular disease 40 percent above those fo r black females; fo r cancer, the differen ce is 60 per* 63 cent. The above discussion o f black-white differences is p a rtic u la rly relevant to th is study o f the Michigan program since blacks are by fa r the largest ra c ia l/e th n ic group in the s ta te . They comprise about 40 percent o f the EPSDT e lig ib le population. Spanish-Speaking G rebler, Moore and Gusman in th e ir rath er well known book, The Mexican-American People say, ” . . . there Is no evidence th at Mexlcan- Americans su ffe r from a higher incidence of to ta l Illn e s s or from a ®*John D. Reid, e t a l . , "Trends in Black H ealth," o£. c i t . . p. 105. fiJ> Evelyn M. Kitagawa and P h ilip M. Hauser, D iffe re n tia l M o rta lity in the United S tates, op. c it .,, p. 114. 63lb id . . p. 114. 44 greater prevalence of chronic disease.’ 64 Michigan re fe rra l s ta tis ­ tic s do show a somewhat greater incidence o f problems among Spanish­ speaking as compared w ith whites. However, the difference is not extreme and is much less than the black-white gap. During the pro­ gram's f i r s t fiv e years, the Spanish-speaking re fe rra l rate averaged 7 percent higher than the white r a te , 57 percent versus 53.4 percent.6® American-Indians Kitagawa and Hauser found th a t American-Indians had the second highest m o rta lity rate among American ra c ia l/e th n ic groups.66 Sorkin, a health economist who has done considerable study o f Indian health, compared in fan t m o rta lity rates and deaths from tuberculosis and gas­ tro e n te ritis between 1955 and 1971. He found rates fo r reservation Indians dropped 62%, 87% and 83% resp ectively. As of 1971, Indian in fa n t m o rta lity was 37 percent higher than the white rate but 24 per­ cent lower than the rate fo r blacks. In regard to tuberculosis and g a s tro e n te ritis , the Indian rate was higher than e ith e r white or black r a t e .6^ However, the health indexes of reservation Indians may not be strongly comparable to the Michigan s itu a tio n . Michigan program s ta tis ­ tic s show v ir t u a lly no difference 1n average re fe rra l rates fo r whites 64 Leo G rebler, Joan W. Moore and Ralph C. Gusman, The MexicanAmerican People. (New York: The Free Press, 1970), p. 2% c itin g A. Taher ffoustafa, M.D. and Gertrud Weiss, M .D., Health Status and Practices Of Mexican-Americans. (Mexican-American Study P ro ject, Advance Report I I , Graduate School of Business Adm inistration, U niversity o f C a lifo rn ia a t Los Angeles, February, 1968. 66M1chigan Department o f Public Health and Michigan Department of Social Services, EPSDT Michigan Annual Report, 1978, op. c i t . , p. 24. 66Evelyn M. Kitagawa and P h ilip Hauser, D iffe re n tia l M o rta lity in United States, o d . c i t . . d . 101. 57----------------Alan L. Sorkin, Health Economics, (Lexington, Massachusetts: D.C. Heath and Company, 1975), p. 160. 45 and American Indians fo r the program's f i r s t fiv e years, 53% and 53.6% resp ectively. Based on the above review of the association between race and h ealth , th is study was p a rtic u la rly interested in the program's effe cts on black p a rtic ip a n ts . The size of Michigan's black, e lig ib le population plus evidence showing blacks to have generally poorer health than white and other ra c ia l groups, made the program's e ffe c ts on blacks to be of p a rtic u la r signflcance. Sex In discussing differences 1n black health status based upon sex, Kitagawa and Hauser note th a t the pattern o f female s u p erio rity on v ir ^ tu a lly a l l indices holds also fo r whites. 69 Accordingly, whether sex has an influence on the program's outcomes was also of in te re s t in this study. CD Michigan Department o f Public Health and Michigan Department o f Social Services, EPSDT Michigan Annual Report, 1978, op. c i t . , p. 24 • so Evelyn M. Kitagawa and P h ilip M. Hauser, D iffe re n tia l M o rta lity in the United S tates, op. e f t . , p. 114. CHAPTER I I I RESEARCH DESIGN AND METHODOLOGY As noted above, there are Ind ications in the lit e r a t u r e th a t EPSDT program p a rtic ip a tio n is associated w ith Improved health statu s. The purpose o f th is chapter is to discuss the research design, meth­ odology and s t a t is tic a l techniques used by th is study to determine whether such an association e x is ts fo r EPSDT p a rtic ip a n ts in Michigan. This cen tral concern w ith the program's outcomes res u lts in the study having one o b je c tiv e , the attainm ent o f which was sought be te stin g three hypotheses and by answering two key questions. Objective of Study (1 ). To b e tte r answer the question of whether EPSDT in Michigan 1s improving the health status o f it s p a rtic ip a n ts . Hypotheses ( 1 ). Screenings and r e fe rr a ls are inversely rela te d in number, i . e . the average number o f re fe rr a ls one Incurs is inversely re la te d to the to ta l number of life tim e screenings one has re c e iv e d .1 (2 ). Medicaid costs are inversely re la te d to the to ta l number of life tim e screenings one has received, I . e . costs decline as life tim e screenings increase. *The "average number o f r e fe rra ls " » the to ta l number o f r e fe rra ls divided by the to ta l number of in d ivid u als screened. 46 47 (3 ). Short-run Medicaid costs increase follow ing screening, are greater (follow ing screening) fo r screened than fo r unscreened individuals and (as in #2 above) are Inversely related to the to ta l number of life tim e screenings one has received. Questions (1 ). Do outcomes vary by age, race, sex and,to a lim ite d extent, geographic location? (2 ). Do outcomes vary depending upon whether subjects are con­ tinuously e lig ib le fo r the program? Design I Referral Rate Differences As indicated above, one approach o f th is study fo r assessing In ­ d ire c tly the e ffe c ts of EPSDT used the average number of re fe rra ls a t la s t screening (re fe rr a l rate s) as the dependent v a ria b le . Referral rates were compared with the degree o f program p a rtic ip a tio n on the assumption th a t i f the program is meeting its objective of b e tte r health fo r it s p a rtic ip a n ts , then those being rescreened should have fewer re fe rra ls than those I n i t i a l l y screened and, also , the number o f r e fe r ­ ra ls should decrease as the number of rescreenings increases. Said d iffe r e n tly , th is means r e fe rra ls should be less fo r Individuals who have received more screenings as compared with Individuals who have received fewer screenings. Diagrammatically, th is approach can be represented by Design I , which fo llo w s, per Campbell and Stanley model. 2 Donald T. Campbell and Julian C. Stanley, Experimental and QuasiExperimental Designs, op. c i t . , p. 6. 2 48 Each X represents a screening and each 0 represents the observation o r, in th is case, the determination o f re fe rra ls (o f course, in practice the re fe rra ls o f in te re s t are id e n tifie d a t the la s t, i . e . most recent screening). X 0 X X 0 X X X 0 X X X X 0 X X X X X 0 Since population data were gathered using Design I , one means of data analysis consisted of simply making a d ire c t comparison of the results per the follow ing table display: Table I . Average number of re fe rra ls a t la s t screening by age and number o f life tim e screenings. Age At Last Screening 1 Screen 2 Screens No. Of Lifetim e Screenings 3 Screens 4 Screens 5 Screens or More Under 1 Year 1 Year 2 Years • • • 20 Years Table I allows a comparison to be made, by age, of the average number of re fe rra ls needed a t the la s t screening fo r fiv e or more groups of program p a rtic ip a n ts . These groups are distinguished by the number of life tim e 49 screenings received. (1 ). They are: The Individuals who have received one screening, by age; (2 ). The ind ividuals who have received two screenings, by age; (3 ). The Individuals who have received three screenings, by age; ( 4 ). The Individuals who have received four screenings, by age; (5 ). The individuals who have received fiv e screenings, by age; (6 ). When a v a ila b le , data can be presented fo r those individuals who have received six life tim e screenings, seven life tim e screenings, e tc . When analyzing the ta b le , the question is whether values in each row decrease as one moves l e f t to r ig h t. Age a t la s t screening rs controlled since re fe rra ls are to some extent a function o f age. R eferrals fo r Immunization, when present, were not counted as re fe rra ls since the need fo r Immunizations is a t times a cause fo r re fe rra l and a t other times not a reason to r e fe r . The difference depends solely upon whether the c lin ic gives immunizations as part of the screening process or re fe rs the c h ild elsewhere fo r th is ser­ vice. Thus, because of th is difference in service d e liv e ry , immun­ iza tio n re fe rra ls were t o t a lly excluded from consideration. 50 The Table I format was completed fo r the two groups of program particip an ts selected fo r study in th is research. One group was the population continuously e lig ib le fo r EPSDT since 1/01/74 ( i . e . , they were "welfare recip ien ts" during th is e n tire time period ). Data fo r th is group are presented in the Table I series (explained below). The other group o f subjects was the population e lig ib le fo r EPSDT during calendar year 1979. Data fo r th is group are presented in the Table I format but are labeled as Table I I , simply fo r purposes of d isting uish­ ing the two groups. More discussion of the study groups is presented la te r in th is chapter. Table I / I I Replications (Tables I ( A ) / I l t A ) - I ( S ) / I I ( S ) Tables I and I I were rep lica te d as the re fe rra l rates a t la s t screening varied by sex, race, and to a lim ite d extent geographic lo ­ cation. Tables were t i t l e d as follow s: Tables I / I I - Average number of re fe rra ls a t la s t screening by age and number of life tim e screenings; Tables I ( A ) / I I ( A ) - Average number of re fe rra ls a t la s t screen­ ing fo r w hites, by age and number of life tim e screenings; Tables I ( B ) / I I ( B ) - Average number o f re fe rra ls a t la s t screening fo r blacks, by age and number of life tim e screenings; Tables I ( C ) / I I ( C ) - Average number o f re fe rra ls a t la s t screening fo r American Indians, by age and number of life tim e screenings; Tables I ( D ) / I I ( D ) - Average number of re fe rra ls a t la s t screening fo r Spanish-speaking, by age and number of life tim e screenings; Tables I ( E ) / I I ( E ) - Average number of re fe rra ls a t la s t screening fo r males, by age and number of life tim e screenings; 51 Tables I ( F ) / I I ( F ) - Average number of re fe rr a ls a t la s t screening fo r females, by age and number o f life tim e screenings; Tables 1 (G )/I1 (G ) - Average number o f re fe rr a ls a t la s t screening fo r white males, by age and number o f life tim e screenings; Tables 1 (H )/I1 (H ) - Average number o f re fe rr a ls a t la s t screening fo r white females, by age and number o f l i f e ­ time screenings; Tables I ( I ) / I I ( I ) - Average number o f r e fe rr a ls a t la s t screening fo r black males, by age and number o f life tim e screenings; Tables I ( J ) / 11( J ) - Average number o f re fe rr a ls a t la s t screening fo r black fem ales, by age and number o f l i f e ­ time screenings; Tables I ( K ) /II( K ) - Average number o f re fe rr a ls a t la s t screening fo r American Indian males, by age and number o f life tim e screenings; Tables I ( L ) / I I ( L ) - Average number o f r e fe rr a ls a t la s t screening fo r American Indian fem ales, by age and number of life tim e screenings; Tables I( M ) /II( M ) - Average number of re fe rr a ls a t la s t screening fo r Spanish-speaking males, by age and number o f life tim e screenings; Tables I ( N ) / I I ( N ) - Average number o f r e fe rr a ls a t la s t screening fo r Spanish-speaking females, by age and number o f life tim e screenings; Tables I ( 0 ) / I I (0 ) - Average number of r e fe rr a ls a t la s t screening fo r p a rtic ip a n ts in D e tr o it, by age and number o f life tim e screenings; Tables I ( P ) / I I ( P ) - Average number o f re fe rr a ls a t la s t screening fo r p a rtic ip a n ts in selected outstate counties, by age and number o f life tim e screenings. Tables I ( Q ) / I I ( Q ) - Average number o f r e fe rr a ls a t la s t screening 1n D e tro it and Northern Michigan, by age and number o f life tim e screenings; Tables I ( R ) / I I ( R ) - Average number of r e fe rr a ls a t la s t screening, by number and year o f screening (N > 1 0 0 ); 52 Tables I ( S ) / I I ( S ) - Percent change in average number of re fe rra ls a t la s t screening as number of life tim e screen­ ings Increase by one, by year of screening. For each of Tables I ( A ) / I I ( A ) - I ( Q ) / I I ( Q ) an accompanying table is presented giving the number of screened Individuals represented by Tables I ( A ) / I I { A ) - I ( Q ) / I I ( Q ) . The number of screened in d iv id u als, in t o t a l, and as a function of age, sex, race and to a lim ite d extent geo­ graphic location 1s thus id e n tifie d . These tables are placed in Appen­ dices A and B fo r reference. Design 1 S ta tis tic a l Analysis As discussed, data were obtained per Design I and are presented in the follow ing chapter 1n the Table I format. However, upon analyzing the obtained data an intervening variab le was id e n tifie d which q u a lifie s the Table I and Table I I re s u lts . When year of screening is used as an independent v a ria b le , i t is evident th a t over the years re fe rra ls have been given with less frequency. The overall re fe rra l rate in 1978-79 was approximately h a lf what i t had been in 1973-74. This is a "classic" Campbell and Stanley case of history becoming an independent variable and jeopardizing the study's in tern al v a lid it y .3 For some reason(s) c lin ic personnel made fewer r e fe rra ls 1n the program's la t e r years. This confounds re s u lts . To the extent we fin d those with more screen­ ings having fewer r e fe r r a ls , we are uncertain by using a d ire c t compari­ son of Table I results whether, or to what e xten t, the decreasing r e fe r ­ ra ls are due to program p a rtic ip a tio n or the confounding v aria b le of time. This com plication, which is i t s e l f an in te re stin g fin d in g , led 3 I b i d . , p. 5. 53 to several compensating adjustments In s ta tis tic a l analysis and to a more detailed analysis o f Table I and I I data. One control fo r time o f screening was to analyze re fe rra l rates derived from same-year screenings only. However, since data are a v a il­ able fo r seven years and the Table I series consists o f twenty reports fo r each o f two subject groups, i t was deemed Impractical and unweildy to generate a ll the possible reports th is approach would allow (280 rep o rts). Several approaches were used to avoid such a clumsy method of analysis. Observation o f Table I / I I re fe rra l rates showed th at the pattern o f grand mean change across the number o f life tim e screenings was generally representative o f change across the individual ages. This allowed a g re a tly s im p lifie d Table I / I I analysis since i t meant con­ clusions could be made on the basis o f several grand means rath er than on means fo r each o f twenty-one separate ages. Thus, one means o f co n tro llin g history was to present grand means fo r each year (by num­ ber o f screenings) and determine whether these data change w ith in each year as predicted by Hypothesis 1. This provided a quite s tra ig h t­ forward and d e fin itiv e te s t o f the rela tio n sh ip . Also, Table I format (with age thus con trolled ) was generated fo r screenings which occurred in 1978 only. 1978 was chosen as the year o f study since I t 1s a recent year which contains r e la tiv e ly large numbers o f Individuals with rescreenings. Data so obtained on the continuously e lig ib le group are presented as the Table I I I series; data obtained on those e lig ib le fo r at le a s t a ll o f 1979 are presented as the Table IV series. This approach also controlled fo r h istory. 54 In ad d itio n , formal testin g of Hypothesis 1 was done with history c o n tro lle d , as explained below. In short, the e ffe c ts o f history proved to be manageable and it s presence in the program was an important fin din g o f the study. Also, 1n analyzing the Table I- IV data, recalculations were frequently made to average a ll re fe rra l rates from rescreenings and thereby allow comparison with re fe rra l rates occurring a t I n i t i a l screening. In other words, the comparison was made between rates a t “one screening" and rates fo r "two or more screenings." This sim p li­ fie d numerous comparisons. Because o f the e ffe c t o f "screening year" on the dependent v a r i­ a b le , i t was determined th a t the use o f the m ultip le regression tech­ nique would be most appropriate fo r form ally testing Hypothesis 1, i . e . , whether screenings and re fe rra ls are Inversely related 1n number. Technically, since the Independent variab le "year o f screening" Is a nominal v a ria b le , with each year representing d iffe re n t categories of th a t v a ria b le , and the other Independent variab le "number of life tim e screenings" 1s a m etric, or in te r v a l, v a ria b le , the method o f s ta tis ­ t ic a l analysis used Is termed analysis o f covariance, or more p recisely, the m ultip le regression method o f analysis o f covariance. M u ltip le regression 1s a standard s ta tis tic a l technique used to analyze the rela tio n s h ip between a dependent variab le and set of In ­ dependent variab les. I t analyzes the data from two perspectives: (1) d e s c rip tiv e , determining the lin e a r relation ship o f the dependent variab le on the Independent variables and (2) in fe r e n tia l, evaluating relation ship s in the population by examination o f sample data, fo r 55 example by hypothesis te s tin g . valuable As a d e sc rip tive tool i t has the a b i l i t y to control fo r confounding v a ria b le s , a q u a lity which made i t p a r tic u la rly helpful fo r th is study. The standard assumptions were made 1n using the regression techniques. They are: ( 1 ). The sample 1s drawn a t random. ( 2 ). The c r ite r io n v a ria b le is d is trib u te d norm ally, or a t le a s t can be measured on an In te rn a l scale. (3 ). The regression o f c r ite r io n and p red icto r variab les is 11near. (4 ). A ll the c r ite r io n v a ria b le 's arrays have the same v a rian ce.4 Since the sample sizes were very larg e and the samples were randomly drawn per accepted procedures, the assumptions are reasonable. In a d d itio n , the technique o f Categorical P a rtitio n Analysis (CPA) was used as a t h ir d , and supplementary means of analyzing the data on r e fe rr a l ra te s . CPA is a f a i r l y new and advanced s t a t is t ic a l technique, r e lia n t upon the c a lc u la tin g a b i l i t y of high speed computers, and is not widely known.® I t is designed fo r use w ith categorical data (nom­ in a l and ordinal such as race and m ilita r y rank re s p e c tiv e ly ) which 4Norman H. N ie, e t a l . , S t a tis tic a l Package fo r the Social Sciences. (New York: McGraw H111 Book Company, 1975), pp 341, 399. The te x t notes th a t the assumption o f normal d is trib u tio n may be relaxed when the sample size Is la rg e . ^Richard Andrew B a r t le t t , " P a rtitio n Analysis o f Categorical Data," (Unpublished Doctoral D is s e rta tio n , The U n iversity o f Penn­ sy lv a n ia , 1974). 56 should make i t o f p a rtic u la r in te re s t to social researchers since th e ir data are often ca te g o ric al. CPA 1s a type o f c la s s ific a tio n analysis which seeks associations between variab les. Procedurally, 1t establishes a series of contingency tables pairing In turn each independent variab le with each dependent v a ria b le (s ) of in te re s t and measures the degree o f association between the various "levels" o f each variab le (fo r example, the variab le sex has two levels and in th is study the variab le re fe rra l rate has seven le v e ls ). For the table whose c e lls have the highest s ig n ific a n t re­ duction in prediction e rro r ( I . e . , the difference in e rro r rates ob­ tained by predicting with the predictor variab le rath er than predicting with a simulated predictor variab le which is s t a t is t ic a lly independent o f the c r it e r io n ) , a " s p lit" is made. The s p lit is made on the basis o f those aggregate c e lls responsible fo r the reduction 1n prediction e rro r versus those c e lls not responsible. formed: Two subgroups are thereby one subgroup associated w ith a level o f the dependent variab le a t a frequency o f occurrence which is greater than chance; the other subgroup showing no such re la tio n s h ip . The technique then continues testin g 1n the same manner fo r an association between the two obtained subgroups and each other predictor variab le o f In te re s t. Subsequent s p lits , 1f any, are s im ila rly made on the predictor variab le with the highest s ig n ific a n t reduction in prediction e rro r. The process con­ tinues as long as s ig n ific a n t s p lits can be made and ends when s p lits no longer occur. However, 1t may be the case that no s p lit occurs on any predictor v a ria b les , i . e . there is no relatio n sh ip found between any o f the levels of the variables of in te re s t. 57 For example, CPA could be used to te s t the association between automobile accident f a t a l i t i e s and c e rta in demographic variab les and could f in d , fo r example, th a t 15-17 year old w hite males who own General Motors cars 1n Northeastern Michigan are unusually prone to fa ta l automobile accidents. To achieve th is r e s u lt, s p lits would be made on the basis o f age, race, company-make o f automobile owned and geographic lo c a tio n . CPA 1s o f course somewhat more complicated than presented here but since It s a p p licatio n to th is study's data yield ed no s p lit s , fu r ­ th er e x p lic a tio n o f the technique seems unnecessary. CPA Independent variab les used 1n th is study were: ( 1 ). Sex, (2 ). Race, (3 ). Screening y e a r, ( 4 ). Location, ( 5 ). Age and (6 ). Number o f screenings. The dependent v a ria b le was the average number o f r e fe rr a ls a t la s t screening or the r e fe rr a l ra te s . Design I I Cost Differences The theory o f preventive health and the cost effectiveness ra tio n a le fo r EPSDT both argue th a t e a rly detection and treatment w ill re s u lt 1n reduced, long-run medical costs. Again, th is should occur as deleterio us conditions are not allowed to d e te rio ra te to more advanced, complicated and therefo re costly le v e ls . To assess these long-run changes, the 58 ideal approach would be a tim e-series of the design given below. As before, each 0 would represent an observation, or determination of costs incurred during a given period, and each X would represent a screening: 0 0 0 0 0 0 X 0 0 0 0 0 0 0 X 0 X 0 X 0 0 0X0X0 o x o x o x o x o In th is ideal design, subjects would be randomly selected fo r par­ tic ip a tio n and randomly assigned to the study groups. A control group, having received no screenings, would be included as well as several te s t groups, each to receive a d iffe r e n t number of screenings. number of subjects used would be larg e. The Observations could be done at yearly in te rv a ls and would continue fo r a number of years. Costs could be analyzed in to ta l and by d iffe r e n t provider types, e .g . physician, in p a tie n t h o s p ita l, d e n ta l, e tc . However, as mentioned previously, th is study did not have the resources, or tim e, to under­ take a true longitudinal design. Neither was i t possible to conduct a post hoc, computer-based study which sampled among current e llg lb le s and then analyzed th e ir past cost data. This approach was not fe a s ib le since Medicaid claims' data are re a d ily a v a ila b le fo r only the most recent year to year and a h a lf. Moving back fu rth e r 1n time becomes te ch n ic ally more d i f f i c u l t , such as requiring the use of several d iffe r e n t computer programs. lis tin g the voluntary assistance of technical s ta ff is v ir tu a lly En­ 59 impossible fo r th is level of a c t iv it y . The design which was possible to e ffe c t used data from only the most recent time period but also took In to consideration the re c ip ­ ie n t's past screening h is to ry . This approach is s im ila r to the one above with the exception that data are obtained from only one time period , the most recen t, rath er than from several. S p e c ific a lly , costs were determined fo r 1979 on ly, a period fo r which a l l te s t sub­ je c ts were continuously e lig ib le . The design 1s id e n tic a l to Design I w ith the exception th a t c lie n ts are added to the design who have never been screened. Design I I 0 X X X X X 0 X 0 X0 X X X X0 X X X X0 Some costs obtained by th is design could be considered proxies fo r long-run costs. These would be the costs incurred by rec ip ien ts w ith a greater degree o f program p a r tic ip a tio n , I . e . more screenings. For rec ip ie n ts to have received three or more screenings, program par­ tic ip a tio n must have extended over a number o f years. Th eir most re ­ cent costs, the costs which can be assessed, are therefo re costs In ­ curred a t the end o f a process, i . e . they are in e ffe c t long-run costs. 60 Data obtained by the design could be displayed in a manner s im ila r to the Table I format: Cost (fo r selected v a ria b le ) by age and number o f life t im e screenings. Total Number of L ifetim e Screenings Age At Survey Mid-Point 0 1 2 3 4 5 or More Screenings 5 Years 6 Years 7 Years • • 20 Years Using th is approach, conceivably costs could be presented in to ta l and by provider type and demographic v aria b le fo r each o f the two sub­ je c t groups used in the study. However, th is approach would generate 272 d iffe r e n t tables (8 provider types x 17 demographic variab les x 2 subject groups) each req uiring s t a t is t ic a l te stin g to determine whether the obtained differences were tru e ones. Although data were a v a ila b le to make th is type o f presentation, I t was determined im practical and unnecessary to present and te s t the data by such a procedure. Economy of s t a t is tic a l te s tin g allowed the same hypothesis and questions to be tested in a much more concise manner. Design I I S ta tis tic a l Analysis Athough a h is to ry e ffe c t was determined to be confounding the Design I a n a ly s is , costs, as a dependent v a ria b le , were not s im ila rly 61 subject to such influen ce. While the c lin ic personnel who determined r e fe r r a l needs apparently did adju st th e ir r e fe r r a l c r it e r ia over tim e, thereby making the time (year) o f la s t screening a v a ria b le fo r r e fe rr a l ra te s , th is circumstance had no Influence on costs which were studied In one year only. In essence, while the la s t screening, the '‘treatment" and occasion to c a lc u la te the "average number o f r e fe rra ls a t la s t screening," could occur during any program y e a r, costs were observed only during 1979. Thus, when costs were used as the c rite rio n v a ria b le , time was c o n tro lled as a fa c to r by the study's design, where­ as r e fe r r a l rates were s t a t is t ic a lly c o n tro lle d through using the m u lti­ ple regression technique and other approaches. Consequently, year of screening was not used as a v a ria b le in analyzing the cost data. With number of screenings the o n ly, and necessary, v a ria b le to te s t Hypothe­ sis 2, the regression technique used was te c h n ic a lly b iv a ria te rath er than m u ltip le regression. Also, visual inspection of re s u lts disclosed th a t average, to ta l costs were greater fo r unscreened than fo r screened subjects fo r both the long-term and shorter-term e llg lb le s . To determine whether these sampled mean differences implied a true d iffere n c e in the parent pop­ u la tio n s , Student's t - t e s t was used. The te s t was applied to both study groups and was also repeated as costs varied fo r the fou r "ra c ia l" groups. SPSS includes a standard sub-program fo r conducting Student's t-te s t. The te s t Is the appropriate procedure fo r te stin g differences in sampled mean scores, which was the point o f in te re s t here. The assumptions made in using the t - t e s t are e s s e n tia lly those made when using m u ltip le regression. They are: 62 (1 ). The dependent variab le (costs) is normally d is trib u ted in both populations o f study and ( 2 ). The variance of the dependent variab le has the same value (is homogeneous) fo r both populations. There is no reason to not believe these assumptions are v a lid fo r the study groups, especially since the sample sizes are large and sub­ je c ts were randomly selected from th e ir populations. Design I I I Short-Run Cost Differences Im p lic it in the program's expectation of long-run cost savings is the suggestion th a t short-run costs w ill increase. Implementation of a program which alms to increase access to medical services, and to encourage the use o f those services, should Increase Immediate medical costs i f the program 1s successful. As noted previously, th is aspect o f the program was a cause of concern fo r cost conscious states and did not serve t o Increase it s p o p u larity. Accordingly, i t 1s of in te r ­ est whether, and to what e xten t, short-run costs do increase as expected. This in te re s t is formalized in Hypothesis 3 which states th a t short- run costs do Increase follow ing screening, are greater fo r screened than fo r unscreened individuals and are inversely rela ted to the to ta l number o f life tim e screenings one has received. G rap hically, the hypo­ thesis is depicted well by the follow ing diagram: Y Medical Costs A fter-B efore X No. of Lifetim e Screenings 63 Design I I I allows Hypothesis 3 to be tested: Design I I I 0 X 0 X X X 0 X X X 0 X 0 X X 0 X 0 X 0 X 0 X X 0 For each te s t subject who was screened during the period MayAugust, 1979 (denoted by the X positioned fa rth e s t to the rig h t In the design), medical costs were determined fo r the four month period p rio r to screening, s p e c ific a lly Janu ary-A prll, 1979 and the four month period follow ing screening, s p e c ific a lly September-Deeember, 1979. The Os in the diagram denote the determination of medical costs fo r the given time period, or observations, in a before and a fte r screening pattern . Design I I I S ta tis tic a l Analysis As w ith Design I and I I , the Table I format of data presentation was determined to be th e o re tic a lly possible, but im practical and un­ necessary fo r analyzing the obtained data. Instead, a b iv a rla te re ­ gression was done to te s t Hypothesis #3 per the same ra tio n a le pre­ sented 1n discussing the Design I I S ta tis tic a l Analysis. ber o f life tim e screenings was the independent v a riab le. Again, num­ 64 Procedures-Sample There are d iffe r e n t c lie n t subgroups upon which one might focus fo r th is type of study. Two seem most prominent: (1) those now e lig ib le fo r the program and (2) those continuously e lig ib le since the program's e a rly years (the la t t e r is a c tu a lly a subgroup of the form er). In designing th is study, i t was f e l t basic conse­ quences would flow from studying e ith e r group. S p e c ific a lly , the types of conclusions which could be drawn would be determined by the choice o f which group(s) to include in the study. Some d is ­ cussion o f the ra tio n a le fo r including these d iffe r e n t groups is thus warranted. F ir s t, two reasons w ill be detailed fo r studying those continuously e lig ib le ; next, the advantages w ill be reviewed fo r studying those cu rre n tly e lig ib le . ( 1 ). Approximating A Longitudinal Study - One Argument fo r Studying Those Continuously E lig ib le . I f th is project had been undertaken when the EPSDT program began, a longitudinal study might have been designed. Subjects would have been chosen a t th a t time and followed over the ensuing years. Those subsequently "dropping out" would no longer have been followed; nor would others have been included in the project a f te r i t had begun. course not designed and begun years ago. je c t s e le c tio n , This research was of However, in terms of sub­ an e s s e n tia lly "retrospective longitudinal study" was s t i l l possible by appropriately selecting those who had been continuously e lig ib le since the program began. This approach pro­ duces the same study group which would have been chosen and followed had the study been designed and implemented years ago. S im ila rly , 65 i t excludes, as would a true lon gitud inal study, both those who did not i n i t i a l l y begin the study ( la t e entran ts) and those who might have started w ith the o rig in a l group but la t e r terminated (drop o u ts). ( 2 ). Id e n tify in g Subjects o f Maximum In te re s t - A Second Argument fo r Studying Those Continuously E lig ib le . I t is reasonable to believe th a t c lie n ts who are continuously e lig ib le are more lik e ly to receive m u ltip le screenings. By v irtu e of t h e ir presence and re ­ peated recruitm ent, they have the most consistent opportunity to par­ tic ip a te in the program. Accordingly, th is group is an important inclusion in the study. Since they have possibly received maximum program b e n e fits , i t is c ru c ia l to know what Impact the program has had upon them. s ig n ific a n t. I f they show l i t t l e or no e ffe c ts , th is is highly However, since th is group was presumed to be r e la tiv e ly sm all, i t was believed merely sampling from those c u rre n tly e lig ib le would lik e ly miss or underrepresent them. ( 3 ). G eneralizing Findings to a Larger Population - The Argu­ ment fo r Studying Those Currently E lig ib le . A large proportion of recip ien ts are not continuously e lig ib le and do not p a rtic ip a te regu­ la r ly in the program. However, precisely because these c lie n ts are In the m a jo rity , I t 1s important to know what e f fe c t , i f any, the program is having upon them. Said d if f e r e n t ly , th e ir inclusion is Important fo r assessing the range o f the program's impact even though the program's Influence on them is lik e ly diminished because o f th e ir reduced p a rtic ip a tio n . Studying th is group allows conclusions to be made about a la rg e r and more ty p ic a l c lie n t group. 66 Resolution Regarding the Choice of Subjects. Because o f the p o te n tia l value of studying both those continuously and c u rre n tly e lig ib le , both groups were included 1n the study. Exclusion of e ith e r could have led to shortcomings in the methodology or outcome. Selection procedures (explained below) were thus conducted tw ice, once fo r each group, to thereby include in the study both those con­ tinuously and c u rre n tly e lig ib le . Selection S trateg y. (1 ). To include in the study those con­ tinuously e lig ib le fo r the program, c e rta in s e le c tiv e c r it e r ia were app lied. Subjects must have been: (a) Continuously e lig ib le fo r Aid to Families w ith Dependent Children (AFDC) between 1/01/74 and 1 2 /3 1 /7 9 , (b) Under age 21 (as o f 12/31/79) and (c ) Not members of a Health Maintenance Organization (HMO). (Necessary to exclude since HMO member costs are not entered in the Medicaid claims system.) Per the Table I form at, inform ation regarding r e fe r r a l rates was generated fo r a ll in d ivid u als who met the above c r i t e r i a . tio n numbered 79,754. To fo rm ally te s t hypotheses sample o f 15,951 subjects (20%) was chosen. This popula­ 1- 3 , a systematic A "systematic sample" means every tw entieth subject was chosen who met c r it e r ia a-c above. The f i r s t subject selected, or the " s ta rtin g p o in t," was determined by use of a random numbers ta b le . The high sample ra te is in d ic a tiv e o f In te re s t in th is group. (2 ). To Include in the study those c u rre n tly e lig ib le fo r the program, the follow in g c r it e r ia were used: 67 (a) E lig ib le fo r AFDC during a t le a s t a l l o f calendar year 1979, (b) Under age 21 (as o f 12/31/79) and (c) Non HMO membership. Again, per the Table I form at, inform ation regarding r e fe rr a l rates was generated fo r a l l in d ivid u a ls meeting the above c r i t e r i a . This population numbered 244,551. To fo rm ally te s t hypotheses 1- 3 w ith th is group, a systematic sample o f 16,303 subjects (6.67%) was chosen. Again, a random numbers ta b le was used to determine the f i r s t subject chosen. “C urrently e lig ib le " was determined to mean e llb lb le fo r a t le a s t a ll o f 1979 in order to equ alize the time period of p o te n tia l p a rtic ip a tio n In the Medicaid program. Otherwise, the varying duration o f subject e l i g i b i l i t y would confound the amount of Medicaid costs Incurred. C o llec tio n of Data Computer data needed fo r th is study was stored in two location s: ( 1 ). The C lie n t Information System (CIS) contains inform ation on a ll re c ip ie n ts o f DSS programs. Such inform ation includes demographics and medical cost data in add itio n to basic id e n tify in g Inform ation. (2 ). The “EPSDT Master F ile s " contain each c h ild 's screening re s u lts , by year o f screening. Tapes are a v a ila b le fo r the years 1974 through the present and include the res u lts o f most screenings which have oc­ curred in Michigan. Information regarding r e fe rr a l rates was obtained by conducting a computer count o f specified data stored on the EPSDT Master F ile s fo r those selected subjects who received EPSDT screenlng(s) in the state o f Michigan between January 1, 1974 and December 31, 1979. 68 More s p e c ific a lly , data were gathered in the following manner: C lie n t Selection (1 ). The subjects who had been continuously e lig ib le fo r the program were id e n tifie d by (a) Selecting subjects from the C lie n t Information System per above selection c r it e r ia and (b) Sorting selected subjects In to re c ip ie n t I.D . order. ( 2 ). The subjects curren tly e lig ib le were Id e n tifie d by (a) Selecting the subjects from CIS per the above s e l­ ection c r it e r ia and (b) Sorting selected subjects into re c ip ie n t I.D . order. Referral Rate Data ( 1 ). The desired data were obtained from the EPSDT tapes per the follow ing procedures (a) Hanging each tape, one a t a tim e, and extractin g the follow ing data elements onto a separate tape (A) re c ip ie n t's I.D . number, (B) re c ip ie n t's date of b ir th , (C) re c ip ie n t's sex, (D) re c ip ie n t's descent (race) code, (E) re c ip ie n t's date o f screening, (F) agency which performed the screening and (G) re c ip ie n t's number of re fe rra ls (excluding re fe rra ls fo r Immunizations). 69 (b) The extracted data were sorted into rec ip ie n t 1.0. order and by date o f screening fo r each rec ip ie n t I.D . w ith m ultip le screenings. The re s u lt o f th is process was a "merged f i l e " with the extracted data organized Into rec ip ie n t I.D . order and chronological order "under" each I.D . with, m u ltip le screenings. (2 ). The desired data were secured fo r both the continuously e lig ib le and cu rren tly e lig ib le groups o f subjects by (a) Passing each o f th e ir tapes against the merged f i l e and (b) P u lling o f f Into a separate tape data from the merged f i l e each time there was a match o f re c ip ie n t I.D . num­ bers on the two tapes (while re ta in in g fo r future use a record o f those selected from CIS fo r whom no screening occurred), (c) The s ta tis tic a l Package fo r the Social Sciences (SPSS) was used to calculate the follow ing fo r each of the study groups: (A) the to ta l number o f screenings received, (B) the age a t most recent screening and (C) the number o f re fe rra ls made a t the most recent screening. (3 ). The calculated data were printed 1n the format o f Tables I / I V IP/TVP. ( 4 ). SPSS was used to form ally te s t Hypothesis 1. 70 Long-Run Costs (1 ). Using the subjects and data secured fo r Design I analysis (tapes developed a t completion o f step #4 above) (a ) The FC rep ortin g system was run fo r each group of sub­ je c ts to secure and p rin t the desired cost data in the Table I format and (b) SPSS was used to form ally te s t Hypothesis #2. Short-Run Costs ( 1 ). Using the subjects and extracted inform ation secured fo r Design I analysis (tapes developed a t completion o f step #4 above), (a ) The FC rep ortin g system was run to secure the desired b e fo re -an d -a fte r screening costs fo r each selected sub­ je c t screened between May 1 , 1979 and August 31, 1979 and then to p r in t the re s u lts in the basic Table I f o r ­ mat fo r each o f the two groups o f study subjects; (b) SPSS was used to form ally te s t Hypothesis #3. CHAPTER IV RESULTS This chapter presents the study's findings in the follow ing order: F ir s t, the Table I and I I series are analyzed in re la tio n to Hypothesis I and the impact of demographic factors on re fe rra l rates. Breakdowns then follow showing both h is to ry 's influence on these same results and the results with h isto ry c o n tro lled . To assess fu rth e r the results with c o n tro ls, the regression anal­ ysis of Hypothesis I and the outcomes fo r same-year screenings only are discussed. CPA affords the concluding perspective fo r analyzing re fe rra l rate s . Cost data analysis proceeds by rep ortin g, in tu rn , the Hypothesis 2 regression an alysis, the t - t e s t outcomes and the regression analysis of Hypothesis 3. Table I Results C ontrolling the variable age in the Table I / I I series served a purpose while a t the same time 1t created a need to sim p lify the obtained re s u lts . Any attempt to discuss resu lts 1n d e ta il fo r each o f twenty-one ages 1n some t h ir t y d iffe re n t tables is unwieldy, 1f not unmanageable. Fortunately, review o f the tables discloses th at changes in the grand mean generally r e f le c t s im ila r movement fo r most of the Individual ages. This Is not surprising but th is docu­ mented consistency does give ju s tific a tio n fo r analyzing grand mean changes only. In other words, presenting data by age has established 71 72 th a t i t is not misleading or unwarranted to base conclusions on s o le ly the grand mean. Accordingly, tab le analyses w ill proceed by grand mean analysis. A q u a lific a tio n in in te rp re tin g the ta b le re s u lts arises from the apparent re la tio n s h ip e x is tin g between group size and r e l i a b i l ­ i t y o f data. The obtained res u lts show small groups gen erally have higher r e fe rr a l ra te s . Thus, r e fe rr a l rates are often higher fo r those groups having six-seven life t im e screenings than fo r groups w ith four or less screenings. This 1s exactly opposite the pre­ d ic tio n o f Hypothesis 1, but i t is most lik e ly th a t the explanation is rooted in group s iz e , not treatment e f fe c t . Very few p a rtic ip a n ts have received six or seven screenings and r e fe r r a l rates appear to be u n re lia b le when based on a small number o f subjects. This is another way o f saying re fe rra l rates show considerable variance. Such variance equalizes over a large number o f subjects but can be very d is to rtin g when few subjects are involved. Accordingly, un­ r e l i a b i l i t y of data Is thought to explain the upward turn 1n re fe rra l rates a t the highest le v e ls o f program p a rtic ip a tio n and, because o f t h is , a l l res u lts derived from sm all-sized groups are discounted. R eferral rates fo r those w ith six or seven screenings w ill systemm a tic a lly not be considered or discussed. Also, other grand means based on less than 100 subjects w ill likew ise e ith e r be discounted ro u tin e ly or In te rp re te d with reservatio n . The designation “small group size" appears warranted fo r a t le a s t those groups w ith less than 100 subjects. 73 In the Table I series which follow s, each table o f re fe rra l ra te data has a companion tab le giving the number of subjects from which the re fe rra l rates were derived. This companion series has been placed in Appendix A and has a complementary numbering system fo r easy reference. Each "N" table is numbered Id e n tic a lly to its companion ta b le of re fe rra l rates with the exception o f a lower case l e t t e r 1n parenthesis. For example, to fin d the "Ns" fo r Table 1 (E ), re fe r to Table I(E a ) in Appendix A. The Ns are important to th is study since, as discussed, re fe rra l rates generally show mean­ ingful change over only f a ir l y large groups o f subjects. Table I grand mean resu lts support the Hypothesis I prediction as life tim e screenings increase from one to four screenings. An upturn In re fe rra ls does occur a t screening fiv e but the increase is small and the rate a t that point Is s t i l l a t a lower level than occurs a t screenings 1-3. In short, good, but not p e rfe c t, agreement exists with Hypothesis 1 fo r the long-term e llg lb le s . Furthermore, in comparing i n i t i a l screening ra te s , by age, with rates fo r a ll screenings combined, observation of the table shows repeat screenings obviously have lower re fe rra l rates fo r every age with only a few possible exceptions. Calculations fo r ages one-three show they are also consistent with the p a tte rn , i . e . those w ith rescreenings have lower re fe rra l rates a t every age. For a ll those rescreened, the actual rates are .846 re fe rra ls a t la s t screening as compared with 1.185 re fe rra ls fo r those receiving an i n i t i a l screening. This is a 29 percent decrease; those being rescreened have markedly fewer re fe rra b le conditions than those being screened fo r the f i r s t time. Table I . Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le s by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.063 1.135 1.014 1.156 1.270 1.174 1.218 1.188 1.187 1.161 1.226 1.159 1.181 1.175 1.228 1.202 1.238 1.236 1.452 1.565 1.400 .846 .851 .790 .933 .925 .941 .926 .917 .893 .841 .850 .870 .881 .874 .916 .937 .949 1.040 1.034 .965 1.166 1.25 1.058 .871 .952 .854 .733 .737 .799 .696 .716 .758 .750 .813 .780 .802 .870 .890 .915 1.025 .500 0 1.000 .818 .910 .674 .612 .612 .683 .737 .718 .555 .582 .833 .911 .959 .697 1.000 1.400 0 0 2.000 1.666 .777 .771 .818 .687 .555 .760 .739 .500 .647 .500 .928 .625 .428 .500 1.000 Grand Mean 1.185 .899 .780 .708 .719 -9* +2* % Change As No. Screenings In­ crease By One -24* 4 5 6 • — 7 0 -13% • - — • - 1.333 .200 0 .285 - 0 .333 .500 .500 - 1.500 - 1.000 - 0 .667 - - - - - 1.000 - - - - - .466 -35% .600 +28% 75 Tables I( A ) - I( D ) are concerned w ith the e ffe c t of race on re fe rr a l ra te s . Two questions are foremost: (1 ) Is Hypothesis 1 affirm ed fo r each ra c ia l group and (2) do rates vary across ra c ia l groups? Table 1(A) shows th a t rates fo r whites are inversely rela te d to life tim e screenings with the exception o f moving from four to f iv e screenings. However, again, those w ith fiv e screenings have a lower r e fe rr a l ra te than those w ith one-three screenings. The r e fe rr a l ra te fo r a l l those rescreened is .778, a 25 percent re ­ duction from the 1.039 ra te fo r those i n i t i a l l y screened. Blacks [Table 1 (B )] show a p erfect inverse re la tio n s h ip to program p a rtic ip a tio n over the f i r s t fiv e screenings w ith a meaning­ fu l drop in r e fe rr a ls fo r each add itio nal screening. The re fe rra l ra te fo r a l l those rescreened is .920, a 31 percent reduction from the ra te a t i n i t i a l screening. Black res u lts are also noteworthy in th a t a t each screening, blacks have a higher r e fe r r a l ra te than w hites. At i n i t i a l screening th e ir ra te is 28 percent higher (1.330 - 1 .0 3 9 /1 .0 3 9 ). For a ll repeat screenings, th e ir rate is 18 percent higher (.920 - .7 7 8 /.7 7 8 ). In short, these res u lts show blacks seem to b e n e fit from repeated program exposure and, a t the same tim e, are disp ro po rtion ately in need o f EPSDT as evidenced by th e ir higher level o f problem determ ination. Results fo r American-Indians [Table 1(C )] a re : lik e ly u n re lia b le since no c e ll contains one hundred subjects o r more. Accordingly, these re s u lts w ill be discounted and no conclusions w ill be drawn regarding American-Indians. Table 1(A). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le whites by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 Under 1 1 2 3 4 5 6 7 8 9 10 U 12 13 14 15 16 17 18 19 20 .835 .996 .920 1.018 1.180 1.061 1.143 1.110 1.055 1.034 1.090 .961 .953 .934 1.008 .924 .945 .957 1.111 1.875 2.000 .833 .789 .714 .890 .910 .953 .879 .811 .832 .751 .790 .831 .789 .786 .835 .874 .833 .861 .675 .777 0 1.000 .727 .891 .863 .805 .771 .662 .709 .675 .627 .682 .613 .787 .708 .686 .842 .813 .529 1.200 Grand Mean 1.039 .830 .713 .634 .706 .200 - -20% -14% -11% +11% -72% _ Age % Change As No. Screenings In­ crease By One 3 4 5 _ — _ - - 0 1.000 .666 .926 .738 .605 .558 .537 .631 .581 .584 .455 .634 .755 1.000 .636 1.000 0 - 2.000 1.000 .600 .782 .466 .375 .636 1.100 .692 .666 .555 1.000 .333 0 - 6 7 - - - - - - 0 .250 0 .250 - 0 0 .500 0 - 1.000 - 0 0 - - - - - - - - - Table 1(B). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le blacks by age and number of life tim e screenings. Number of Lifetime Screeninqs 1 2 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.245 1.253 1.104 1.277 1.379 1.287 1.295 1.276 1.327 1.290 1.382 1.366 1.420 1.365 1.420 1.401 1.394 1.374 1.574 1.500 .857 1.000 .884 .857 .985 .944 .928 .972 1.008 .981 .958 .931 .927 .998 .962 .993 1.004 1.024 1.166 1.213 1.042 1.400 1.333 1.666 1.000 1.056 .900 .695 .831 .906 .744 .815 .862 .871 .857 .828 .874 .892 .884 1.033 .846 .428 0 1.000 1.000 .927 .664 .640 .686 .814 .883 .873 .579 .724 1.014 1.031 .957 .727 1.055 1.400 0 2.000 1.000 .642 .909 .800 1.000 .700 .818 .416 .181 .636 .333 .857 .800 .666 .500 1.000 Grand Mean 1.330 .975 .849 .799 .722 1.000 1.000 -27% -13* -6* -10* +38* 0% Age * Change As No. Screenings In­ creased By One 3 4 5 6 7 • _ • _ - - * - - - - - - - - - 2.000 0 - .333 - - 1.000 - 1.000 - - - - - 1.000 - 1.500 - - - - - - - - - - - Table 1(C). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le American-Indians by age and number of life tim e screenings. Number of Lifetime Screeninqs Aqe 1 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 0 .333 1.000 1.400 .333 1.000 .666 2.000 1.307 1.000 .857 .750 .833 1.500 .333 .333 2.000 Grand Mean 1.000 % Change As No. Screenings In­ crease By One 2 3 4 5 _ _ • - - - - - - - - - - - - - .500 .833 1.333 1.142 2.000 .857 1.000 .833 .666 .727 .555 1.166 .428 1.000 2.000 1.000 .500 .500 0 0 .333 0 .250 .142 1.200 .500 0 0 - - 0 - - - - - - - - - - - - .500 .500 3.000 .250 0 1.000 1.500 .500 - .500 2.000 1.500 - 1.000 1.000 - - - - - - - - - - - 1.000 - .896 .375 .750 1.250 -10% -58% +100% +67% 79 Results fo r the Spanish-speaking [Table 1(D )] do support Hypothesis 1 fo r screenings one-four w ith too few subjects having fiv e screenings to place any confidence 1n th is re s u lt. The r e f e r ­ ra l ra te fo r a l l repeat screenings is .720, a 32 percent reduction from the 1.054 recorded fo r i n i t i a l screenings. The r e fe rr a l rate fo r a ll repeat screenings is thus lower than the rate fo r e ith e r blacks or whites. This is re fle c te d in the Spanish-speaking having the lowest r e fe r r a l ra te o f the three groups a t two, three and four screenings. At i n i t i a l screening t h e ir ra te is s lig h tly higher than whites but lower than blacks. In summary, as evidenced by r e fe r r a l ra te s , the Spanish-speaking appear h e a lth ie r than blacks or whites with blacks appearing to be the le a s t healthy o f the three groups. In conclusion, race does not Influence in general the trend o f inverse re la tio n s h ip between r e fe r r a l rates and program exposure; however, i t does appear to ex ert some influence on health status as evidenced by the varying magnitude o f health problems found fo r r a c ia lly d iffe r e n t recip ien ts with equal screenings. P a rtic u la rly fo r blacks, the o v e ra ll Influence o f race on r e fe rr a l rates appears to be meaningful. Over a l l screenings, blacks average 23 percent more r e fe rr a ls than whites and 35 percent more re fe rr a ls than the Spanish-Speaking (1.107 versus .898 versus .8 1 7 ). Tables 1(E) and 1(F) consider sex as a v a ria b le and show i t exerting l i t t l e Influence on the outcome v a ria b le . With the excep­ tio n o f an upward turn 1n re fe rr a ls a t the f i f t h screening fo r males. Hypothesis 1 holds fo r both groups. The downward d r i f t o f r e fe rra ls across screenings is s im ila r fo r both sexes although somewhat sharper Table 1(D). Average number of re fe rra ls at la s t screening fo r long-term e lig ib le Spanish-speaking by age and nunfcer of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.000 1.250 .944 1.040 1.065 .975 1.333 1.022 1.128 1.517 .957 1.026 1.138 .892 .642 .900 .875 .250 0 1.000 .666 .857 .631 1.026 .905 1.093 .696 .456 .648 .761 .698 .871 .886 .871 .692 .588 * 0 _ 1 .0 0 0 - 1 .0 0 0 0 .833 1.125 1.666 2.000 - - - Grand Mean 1.054 Aqe - - - - - - - - - - - - - 0 .722 .888 .838 .555 .551 .379 .647 .392 .875 .454 .909 1 .0 0 0 .778 .679 -26* -13% 0 .667 .166 .181 .666 1.200 .600 .545 .230 .444 .400 .750 .833 - - 1.000 - - 1.000 1.000 - - - - 1 .0 0 0 - .500 0 0 .600 0 - - - - - - - - - - - - - - - .494 CM 1 % Change As No. Screenings In­ crease By One 6 .533 1.000 +8* +88* 81 fo r males. Males drop 30 percent in re fe rr a ls from the i n i t i a l screening to the average of a l l rescreenings (1.201 - .8 37 /1.2 01 ) white females decrease 27 percent fo r the same comparison (1.169 - .8 5 4 /1 .1 6 9 ). Also, w hile males have 3 percent more re fe rr a ls a t i n i t i a l screening (1.201 versus 1 .1 6 9 ), they have 2 percent fewer r e fe rr a ls fo r a l l repeat screenings (.837 versus .8 5 4 ). A ll of these differences between the sexes are sm all. The purpose o f Tables I( G ) - I( N ) is to compare sex differences in outcome fo r each race under study. Tables 1(G) and 1(H) show white males have 4 percent more r e fe r r a ls a t i n i t i a l screening than females (1.059 versus 1.019) and 2 .7 percent more re fe rr a ls fo r a ll rescreenings combined (.789 versus .7 6 8 ). The d ire c tio n o f change across the number of screenings is very s im ila r fo r both groups. The trend of s lig h tly more r e fe rr a ls fo r males does not hold fo r blacks [Tables I ( I ) - I ( J ) ] . Black males do have 2 percent more r e fe r ­ ra ls a t i n i t i a l screening than black females (1.346 versus 1.315) but also have 2 percent fewer r e fe rr a ls fo r a l l rescreenings combined (.8 9 4 versus .9 4 3 ). With Amerlcan-Indians [Tables I ( K ) - I ( L ) ] there is again the problem o f inadequate c e ll s ize . By combining data fo r a l l males and fo r a l l females fo r th is group, somewhat over 100 sub­ je c ts fo r each group is obtained. The o v e ra ll r e fe r r a l ra te thereby obtained is .842 fo r males; .814 fo r females, a 3 percent Increase fo r males. For the Spanish-speaking [Tables I ( M ) - I ( N ) ] , the grand means fo r screenings 1-3 are each derived from over 100 subjects fo r both sexes and Hypothesis 1 1s confirmed fo r both sexes across these three c e lls . The slope o f decrease is d iffe r e n t fo r the two Table 1(E). Average number o f re fe rra ls a t la s t screening fo r long-term e lig ib le males by age and number of life tim e screenings. Number of Lifetime Screenings Age • 0 - - Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.113 1.201 1.023 1.194 1.350 1.203 1.250 1.259 1.182 1.192 1.278 1.192 1.116 1.142 1.141 1.080 1.238 1.153 1.071 2.153 3.000 1.222 .700 .763 .926 .935 .992 .963 .933 .893 .861 .865 .893 .842 .860 .872 .870 .939 .932 .900 1.000 Grand Mean 1.201 .897 .763 .672 .747 .444 -2556 -1556 -12* +1156 -4156 56 Change As No. Screenings In­ crease By one 4 6 2 - 3 5 1 1.000 1.666 .812 .982 .861 .736 .699 .825 .674 .726 .728 .711 .830 .741 .743 .813 .750 .961 .833 1.000 - 0 1.333 .896 .806 .646 .699 .495 .669 .784 .745 .521 .391 .818 .780 1.064 .600 1.000 _ 0 2.000 1.000 1.230 .684 .843 .785 .647 .833 .375 .642 .500 .400 1.000 1.000 .200 .500 - 0 .333 0 .500 .666 - .333 1.000 1.000 - 1.000 - - - - - - - Table 1 (F ). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le females by age and number of life tim e screenings. Number of Lifetime Screenings Aqe 1 2 3 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.012 1.072 1.004 1.116 1.190 1.144 1.186 1.117 1.192 1.130 1.174 1.128 1.246 1.204 1.302 1.291 1.238 1.292 1.542 1.333 1.000 0 .963 .814 .940 .914 .890 .890 .901 .894 .821 .835 .849 .918 .887 .958 .992 .957 1.112 1.081 .959 1.166 _ 1.333 .727 .921 .924 .847 .731 .771 .768 .718 .707 .787 .793 .797 .817 .857 .921 .991 .894 1.060 .428 0 0 .730 .988 .700 .535 .732 .699 .692 .686 .593 .753 .850 1.015 .883 .777 1.000 1.400 0 Grand Mean 1.169 .900 .797 .736 .691 .428 .500 -23% -11% -8% -6% -38% +17% % Change As No. Screenings In­ creased By One 4 5 6 - - - - - - - - - • * 7 - 2.000 .357 .875 .782 .611 .473 .692 .933 .333 1.000 .571 .666 .500 1.000 1.000 1.000 - - 2.000 0 0 0 - 1.000 - 0 - - 0 - - - 0 0 - 2.000 - - - - - - - - - - Table 1(G). Average number of re fe rra ls a t la s t screening for long-term e lig ib le white males by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .871 .948 .995 1.036 1.277 1.096 1.136 1.142 1.047 1.073 1.157 .997 .910 .911 .983 .814 1.022 .950 .500 2.250 3.000 1.666 .625 .733 .893 .885 1.039 .913 .840 .816 .802 .811 .882 .793 .804 .803 .846 .804 .935 1.000 .250 Grand Mean 1.059 .850 .705 .637 .746 .375 0 -20% -17% -10% +17% -50% -100% Age % Change As No. Screenings In­ creased By One - 3 4 5 - - 6 7 - - - • - - - - - - * 1.000 .750 .936 .837 .767 .740 .710 .673 .645 .616 .554 .782 .632 .648 .865 .615 1.000 0 1.000 .800 .812 .741 .732 .546 .483 .770 .509 .585 .454 .555 .736 .750 .800 2.000 - - - - - - - - - - 1.000 1.667 .666 .769 .571 .600 .600 .400 .833 .600 .400 1.000 0 .500 0 1.000 - - 0 - - 0 1.000 - - - - - - - - - - 0 - - - - - Table 1(H). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le white females by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .800 1.043 .841 1.000 1.081 1.022 1.149 1.078 1.063 .994 1.023 .947 .995 .956 1.035 1.024 .893 .960 1.285 1.500 2.000 0 .909 .697 .887 .937 .857 .845 .783 .847 .702 .768 .786 .786 .769 .869 .896 .854 .786 .538 1.200 0 1.000 .625 1.000 .792 .774 .777 .585 .708 .677 .607 .747 .676 .791 .793 .721 .826 .969 .272 1.200 Grand Mean 1.019 .811 .720 .629 .680 -20% -11% -13% 48 % Age % Change As No. Screenings In­ creased By One 4 _ - 5 6 7 _ _ _ - - - - 0 - - - - - - - - - - 0 0 0 - - .500 1.027 .734 .469 .571 .596 .489 .648 .583 .457 .720 .769 1.333 .500 .666 3.000 .333 .571 .800 .375 .272 .666 1.800 .571 1.000 .750 - .333 0 1.000 - 0 - - 0 - - - 0 0 - - - - - - - - - - - - - - - - - - - 0 0 -100% Table 1 (1 ). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le black males by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.301 1.417 1.051 1.332 1.437 1.304 1.360 1.387 1.316 1.323 1.431 1.419 1.331 1.324 1.318 1.301 1.397 1.256 1.333 2.111 1.200 .750 .825 .973 .983 .947 1.003 1.021 1.000 .931 .939 .920 .916 .923 .950 .908 1.050 .948 .823 2.000 Grand Mean 1.346 * Change As No. Screenings In­ creased By One - - 3 _ 1.000 2.333 1.000 1.038 .883 .707 .665 .969 .712 .818 .860 .830 .872 .824 .798 .783 .754 1.055 .833 - 4 5 6 _ _ - 0 1.500 1.076 .821 .574 .645 .479 .854 .800 .945 .489 .323 1.027 .785 1.352 .500 .833 7 - - - - - - 2.000 - - - 1.000 .785 .833 1.400 .666 1.000 0 .333 .428 - 1.000 1.000 0 .500 0 .333 - 1.000 - 1.000 - 1.000 - - - - - - - - - - - - - - - .956 .822 .730 .762 .500 1.000 -29* -14* -11* +4* -34* +100* Table 1(J ) . Average nunfcer of re fe rra ls a t la s t screening fo r long-term e lig ib le black females by age and number of life tim e screenings. Number of Lifetime Screenings Aqe 1 2 3 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.188 1.103 1.161 1.218 1.324 1.270 1.222 1.169 1.339 1.259 1.334 1.316 1.510 1.400 1.492 1.466 1.329 1.463 1.622 1.310 .857 0 1.000 .886 1.000 .906 .911 .944 .994 .966 .986 .924 .933 1.073 1.000 1.029 1.083 1.004 1.278 1.327 .953 1.400 1.500 1.000 1.000 1.075 .918 .685 .975 .841 .777 .812 .865 .919 .844 .831 .944 1.000 .975 1.023 .850 .428 0 0 .928 1.000 .732 .636 .882 .773 .973 .775 .692 1.114 1.000 1.222 .733 .916 1.166 1.400 0 Grand Mean 1.316 .993 .875 .861 .682 2.000 1.000 -25% -12% -2% -21% +193% -50% % Change As No. Screenings In­ creased By One 4 5 6 _ _ - - - - - - - - - - - - 1.000 .375 1.125 .769 .714 .750 .666 .500 0 1.000 .333 .666 .666 2.000 - 1.000 - 2.000 - 7 - 1.000 - - - - - - - - - - - - - - - - 2.000 - - - - - - - - - - - Table I(K ). Average nimfcer of re fe rra ls a t la s t screening fo r long-term e lig ib le American Indian males by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 0 .333 1.000 2.000 1.000 1.250 1.000 2.000 1.833 1.500 .750 1.000 .750 2.000 0 0 _ _ _ - - Grand Mean 1.128 Age % Change As No. Screenings In­ crease By One - - .500 1.000 - 1.166 0 .750 1.000 .500 2.000 1.200 0 1.666 0 1.500 - 1.000 .750 0 0 0 0 0 0 0 1.000 0 0 5 - - - - - - - - .500 - 3.000 .250 0 - 1.500 0 - 0 2.000 - 1.000 1.000 - - - - - - - - - - - - - - - - - 0 - - - - - - - - - - - - .914 .250 .727 -19% -73% +191% 1.200 +65% Table I ( L ) . Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le American Indian females by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Grand Mean * Change As No. Screenings In­ crease By One 1 2 3 4 - - - - - - - - - _ - - - _ 1.000 0 .800 0 2.000 .857 .666 1.000 .666 1.000 1.000 1.000 .500 2.000 - - .800 1.333 1.000 2.500 1.000 1.000 1.500 .500 .333 .714 .666 .600 .666 2.000 5 - - - - - - - - .250 1.000 0 - 1.000 0 .500 .250 1.333 .750 0 0 - .500 - 1.000 - 1.000 - 1.000 - 1.500 - - - - - - - - - 1.000 - - - - - - - - - - - - - .875 - .884 .500 .800 1.333 +1* -43* +6* +67* Table I(M ). Average number of re fe rra ls a t las t screening fo r long-term e lig ib le Spanish-speaking males by age and number of life tim e screening. Number of Lifetime Screenings 1 2 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.500 4.000 1.000 .750 1.230 1.235 1.200 1.230 .947 1.437 .944 1.071 1.142 .875 .600 1.000 .500 .250 1.000 0 Grand Mean 1.098 Age * Change As No. Screenings In­ crease By One - - 0 .777 .700 1.055 1.047 1.095 .821 .583 .758 .600 .593 .757 .740 .714 .363 .750 1.000 0 - 3 - 0 .875 .700 .818 .666 .631 .166 .736 .538 1.000 .714 1.000 1.000 .857 .500 - 4 5 6 0 _ _ - - - - - - - - 0 .333 0 0 1.333 .666 .333 .500 .666 1.000 .500 - 0 0 .600 0 1.000 1.000 - - 1.000 - - - - - - - 1.000 - - - - - - - - - - .752 .719 .515 .538 1.000 -32* -4 * -28* +4* +86* Table 1(H). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le Spanish-speaking females by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .750 .333 .857 1.176 .850 .782 1.500 .722 1.300 1.615 1.000 1.000 1.133 .916 .666 .888 1.000 Grand Mean 1.012 % Change As No. Screenings In­ crease By One - - 2 1.000 1.000 1.000 .555 1.000 .812 1.090 .571 .363 .520 .868 .806 1.000 1.038 1.055 .933 .444 1.000 0 - 3 4 - * - - - - 0 .600 1.125 .888 .444 .400 .529 .533 .266 .769 .263 .750 1.000 .800 1.333 1.666 2.000 - 5 - - • - .666 0 .250 .888 1.000 .600 .400 .200 .428 0 0 1.000 - .500 -» - 0 - 1.000 - - - - - .802 .624 .484 .500 -21% -22% -22% +3% 92 groups but the r e fe rr a l ra te fo r a ll rescreenings is s im ila r , .722 fo r males and .715 fo r females. screenings is 1.098 fo r The r e fe r r a l ra te fo r i n i t i a l males and 1.012 fo r females. These d i f ­ ferences are sm all; sex appears to exert no unique e ffe c t on r e fe r ­ ra l rates fo r the Spanish-speaking. The most notable fin d in g in Tables I( G ) - I( N ) 1s the im p licatio n th a t black males p a r tic u la rly need and b e n e fit from EPSDT. Specif­ i c a l l y , they have higher r e fe r r a l needs a t i n i t i a l screening than any ra c ia l/e th n ic group members o f e ith e r sex and show the la rg e s t decrease in r e fe rr a ls from the i n i t i a l screening to rescreening (con­ sidering a l l rescreenings combined). The decrease is 34 percent (1.346 - .8 9 4 /1 .3 4 6 ), a very meaningful reduction. Tables I ( 0 ) - I( Q ) consider location as a fa c to r. For both D e tro it and outstate p a rtic ip a n ts , the grand means fo r screenings 1-4 are based on s u ffic ie n t subjects to place confidence in the re s u lts . For both groups, Hypothesis 1 1s confirmed over th is range o f program p a rtic ip a ­ tio n . Location does n o t-a lte r the re la tio n s h ip between r e fe r r a l rates and program p a rtic ip a tio n . Also, the slope o f decrease is very s im ila r fo r both groups a t screenings 2 and 3 , w ith some notable, but s t i l l sm all, differences appearing only a t screening 4. O v e ra ll, these res u lts are remarkably s im ila r. What is s trik in g in the location tables Is the higher r e fe rr a l rates fo r D e tro it resid ents. At I n i t i a l screenings, D e tro ite rs average 61 percent more re fe rr a ls than o u ts ta te , ru ra l residents (1.461 versus .9 0 8 ). This is a very large d iffe re n c e , p a r tic u la rly since other var­ iables have yielded intergroup differences o f gen erally only 2-4 percent. Table 1 (0 ). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le participants in D etro it by age and number of life tim e screenings. Number of Lifetime Screenings Aqe 1 2 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.229 1.308 1.322 1.305 1.518 1.355 1.427 1.337 1.436 1.441 1.505 1.601 1.549 1.537 1.588 1.593 1.787 1.578 1.800 1.470 1.000 1.500 .833 1.000 1.323 1.081 1.104 1.106 1.114 1.129 1.158 1.168 1.116 1.281 1.107 1.236 1.217 1.336 1.302 1.600 1.187 2.000 Grand Mean 1.461 % Change As No. Screenings In­ crease By One 3 1.333 - 4 5 6 7 _ _ _ _ - - - - - - - - - - 1.571 1.360 .887 .607 1.126 1.050 .932 1.057 1.074 1.203 .980 1.172 1.046 1.275 1.368 1.555 1.333 1.000 1.000 .600 1.000 .928 .750 .812 1.066 1.000 .875 .666 .642 1.090 1.100 1.285 1.500 2.000 1.000 1.168 1.061 .936 .877 1.000 1.000 -20% -9% -12% -6% +14% 0% - 2.000 2.000 .600 .714 1.111 1.600 2.000 .800 .666 0 1.000 1.000 0 0 2.000 .500 - - 1.000 - - - - - - - - - Table I(P ). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le participants in fo rty -fo u r Northern Michigan counties by number of life tim e screenings. Number of Lifetime Screenings 1 Mean % Change As No. Screenings In­ crease By One .908 2 3 4 5 6 7 .711 .670 .640 .511 .083 0 -22% -6% -4% -10% -84% -100% 95 In considering r e fe rr a l rates fo r a l l rescreenings combined, the gap increases as D e tro it p a rtic ip a n ts have 66^ percent more re fe rra ls (1.133 versus .6 8 4 ). While location does not a ffe c t the program's e ffe c ts , 1t is associated w ith a large d ifferen ce in measured health status. Since re s u lts have already shown blacks to have higher r e fe rr a l rates than w hites, i t might be an ticip ated th at the D e tro lt-o u ts ta te d iffere n c e in r e fe rr a ls Is a c tu a lly a ra c ia l d iffe re n c e . Table I(Q a ) shows th a t blacks are the large m ajo rity o f D e tro it rec ip ie n ts (88 per­ cent) w hile v ir t u a lly a l l o f the selected ou tstate p a rtic ip a n ts are white (98 p ercent). However, fu rth e r analysis suggests th a t location remains an in flu e n tia l fa c to r when race 1s held constant as a va riab le ([Table I( Q ) ] . Urban whites have 33 percent more r e fe rr a ls than ru ral whites a t the i n i t i a l screening (1.186 versus .892) and 45 percent more r e fe rr a ls fo r a l l rescreenings combined (.991 versus .6 8 3 ). The In s u ffic ie n t number o f ou tstate blacks prevents s im ila r analysis o f black r e fe r r a l ra te s , but the wide d is p a rity in white rates argues th at the d iffere n c e obtained on the basis o f location can not be ex­ plained by the ra c ia l composition o f the population. The Table I I series w ill fu rth e r explore th is re la tio n s h ip . Table I I Results As discussed above, data fo r the Table I I series were from re c ip ­ ients who were EPSDT e lig ib le fo r a shorter time period than rec ip ien ts included 1n the Table I series. Accordingly, we might expect fewer Table I I subjects to have received a high number o f m u ltip le screenings, Table I(Q ). Average number of re fe rra ls a t la s t screening fo r long-term e lig ib le participants in D e tro it and Northern Michigan by race and number of life tim e screenings. Number of Lifetime Screenings Location/Race 1 2 3 4 5 6 7 • - Grand Mean Detroit Whites Blacks Grand Mean 1.186 1.506 1.462 1.000 1.193 1.173 .865 1.075 1.060 1.166 .918 .936 1.500 .822 .877 .892 .961 .893 .706 .571 .703 .673 .708 .674 .640 1.000 .652 .487 1.000 .512 1.000 1.000 1.000 1.000 1.135 1.382 1.352 Northern Michigan Whites Blacks Grand Mean .100 0 - - .100 0 .768 .772 .768 97 a development which would r e s t r ic t Hypothesis 1 analysis. Thus, i t was a n tic ip a te d long-term program e ffe c ts might not be as evident with th is group. On the other hand, the population o f th is group Is much la rg e r than th a t o f the long-term e lig ib le s so i t was recognized th is fa c to r might serve to increase the number o f rec ip ien ts w ith rescreen­ ings. Also, Table I I data are more rep resentative o f the to ta l AFDC population where a sizab le proportion o f c lie n ts have frequent changes in e l i g i b i l i t y status. In a c tu a lity , i t made l i t t l e d ifferen c e which population was reviewed. The re s u lts o f both groups were s im ila r a l ­ though trends were somewhat more prominent 1n the Table I I s eries . In addressing Table I I data, the questions are whether Hypothesis 1 is confirmed fo r th is group, whether demographic variab les a ffe c t the res u lts and whether differences occur between Table I and Table I I re s u lts . s e rie s . Discussion w ill proceed as in the presentation o f the Table I The "Ns" fo r the Table I I series are located in Appendix B. For Table I I , adequate c e ll size is present to warrant conclusions fo r screenings 1-5. For these screenings, p e rfec t agreement w ith Hypo­ thesis 1 e x is ts across the grand means. r e fe rr a ls decrease. As life tim e screenings Increase, Hypothesis 1 is confirmed even more strongly than was the case w ith Table I data. Rescreenings combined average .809 r e fe r r a ls , a decrease o f 20 percent from the 1.009 r e fe rr a ls occurring a t the I n i t i a l screening. This is a strong change, consistent w ith the differen ces obtained fo r the long-term e lig ib le s although o f a somewhat lesser magnitude. Tables I I ( A ) - I I ( D ) consider race as a v a ria b le . For both whites and blacks [Tables I I ( A ) - I I { B ) ] , p erfec t agreement ex ists with Table I I . Average nuntoer o f re fe rra ls a t la s t screening fo r one-year e lig ib le s by age and number of life tim e screenings. Number of Lifetime Screenings Age 3 4 5 6 7 8 1 2 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .745 .790 .780 .962 1.034 1.074 1.091 1.088 1.128 1.085 1.139 1.121 1.124 1.140 1.219 1.186 1.243 1.235 1.272 1.208 1.081 .571 .600 .664 .828 .797 .850 .863 .851 .856 .837 .844 .850 .880 .863 .899 .947 .949 1.016 1.152 1.083 1.104 .937 .594 .562 .793 .771 .766 .778 .754 .778 .722 .716 .737 .763 .822 .823 .785 .869 .913 .858 .965 .565 .333 1.285 .581 1.041 .732 .755 .717 .665 .639 .691 .725 .691 .629 .638 .739 .909 .940 .717 .947 .933 .333 .714 .666 .837 .666 .855 .662 .695 .580 .692 .894 .416 .615 .466 .913 .615 .461 .400 1.000 .500 Grand Mean 1.009 .837 .767 .717 .641 .470 1.000 1.000 -17% -8% -7% -10% -27% +112% 0% % Change As No. Screenings In­ crease By One _ - _ - - - 0 - - - - - .800 .857 .166 .200 .285 .666 0 .333 - .333 1.000 .666 - - 1.000 - 1.000 - - - 2.000 - - - - - - - - - - - - - - - - - - - - - - - - - - - 1.000 Table 11(A). Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le whites by age and number of life tim e screenings. Nuntoer of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Grand Mean % Change As No. Screenings In­ crease By One 1 2 3 .673 .678 .712 .920 .965 .985 1.011 1.004 1.046 .990 1.038 .990 .983 .989 1.065 1.031 1.013 1.090 1.200 1.098 1.131 .489 .550 .649 .835 .778 .830 .824 .792 .806 .775 .765 .807 .848 .818 .856 .924 .841 .929 1.046 1.064 .642 .600 .512 .505 .704 .731 .752 .801 .727 .711 .701 .661 .708 .679 .821 .782 .694 .855 .950 .666 1.125 .800 3.000 .388 .903 .666 .715 .764 .668 .625 .598 .616 .598 .666 .557 .571 .774 1.060 .823 .900 .200 .918 .795 .727 .669 .655 .312 0 1.000 -13% -9% -8% -2% -52% -100% +100% 4 5 6 - - 1.000 .200 .200 0 - 1.000 8 _ • 0 .500 .875 .742 .857 .533 .391 .375 .666 1.190 .500 .600 .500 7 - - - 0 - - - - - - - .750 .666 .250 .200 .250 - 0 0 .500 0 - 1.000 - - - - 0 0 - - - - - - - - - - - - - - - - - - - - - 1.000 - - - - - - - - Table 11(B). Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le blacks by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .836 .933 .877 1.036 1.180 1.209 1.226 1.239 1.262 1.224 1.325 1.324 1.342 1.340 1.424 1.375 1.470 1.378 1.377 1.296 .951 .670 .686 .675 .831 .830 .884 .927 .935 .937 .946 .971 .925 .945 .933 .973 .981 1.047 1.127 1.236 1.117 1.289 1.090 .714 .622 .929 .804 .796 .769 .826 .898 .784 .788 .815 .869 .840 .858 .867 .900 .897 .956 .836 .437 .500 .600 .720 1.209 .834 .814 .707 .680 .669 .802 .862 .783 .600 .744 .927 1.027 .900 .730 1.090 1.333 .333 1.000 .750 .764 .633 .900 .800 1.176 .727 .666 .500 .214 .636 .333 .727 .857 .666 .500 1.000 0 Grand Mean 1.146 .905 .832 .791 .725 .846 1.000 -21* -7 * -5 * -8 * +17* +18* Age * Change As No. Screenings In­ crease By One 5 6 _ • - 7 - - - - - - 1.000 1.333 0 - .333 - - 1.000 - 1.000 1.000 - - - - 1.000 1.000 - 1.500 - - - - - - - - - - 101 Hypothesis 1 across screenings 1-5. This is consistent w ith the Table I fin d in g s , with the trend even somewhat stronger. Whites average .769 r e fe rr a ls fo r a ll rescreenings, a drop o f 16 percent from i n i t i a l screenings; blacks average .876 on rescreenings, a decrease o f 23.5 percent from the i n i t i a l level of 1.146. As with the long-term e lig ib le s , blacks show a higher rate o f problems than whites a t i n i t i a l screening (20 percent hig her, 1.146 versus .918) and a greater reduction in problems a t rescreening ( a ll rescreenings considered). For both American-Indians and the Spanish-speaking, the trend o f p erfect agreement w ith Hypothesis 1 continues across those c e lls w ith over 100 subjects. For American-Indians, screenings one-three show the predicted downward trend. Rescreenings average .729 r e fe rr a ls versus .898 r e fe rr a ls fo r I n i t i a l screenings, a 19 percent decrease. screenings 1-4. The Spanish-speaking r e fe r r a l ra te decreases over Here, rescreenings average .736 r e fe r r a ls or 20 per­ cent less than the i n i t i a l screening ra te o f .923. In sho rt, the p e rfec t agreement w ith Hypothesis 1 over a l l sub­ je c ts in th is population (Table I I ) a fa c to r [Tables I I ( A ) - I I ( D ) ] . 1s not disturbed when race becomes That is , race is not a s u ffic ie n tly strong v a ria b le to disturb the general pattern o f re la tio n s h ip , given s u ffic ie n t study group s iz e . This la t t e r q u a lific a tio n is key. If group size 1s la rg e , r e fe rr a l rates are a ffe cted as predicted by the degree o f program p a rtic ip a tio n , I . e . the number o f screenings one has received. Perhaps, since there are more short-term e lig ib le s than long-term ones, the predicted re la tio n s h ip o f Hypothesis 1 is stronger w ith the former group. appear to This is not to say ra c ia l differences do not Table 11(C). Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le American Indians by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Grand Mean % Change As No. Screenings In­ crease By One 1 2 .750 .964 .540 1.087 .739 .851 1.263 1.333 1.166 .941 .681 .857 .588 1.214 .700 .600 1.200 1.000 0 1.000 2.000 1.000 .666 .272 .888 .535 1.176 .941 1.315 1.157 .833 .578 .666 .851 .750 .750 .538 .833 1.000 1.500 1.000 .898 .816 .545 1.000 1.272 -9% -33% +83% +27% - 3 4 5 - - - - - - - - - * - • - 0 6 • 4.000 1.500 1.000 .363 .400 .307 .600 .642 .700 .166 .300 1.000 .333 0 0 0 .333 .500 - 2.500 .400 .666 2.000 .600 .600 1.666 1.500 .500 .500 2.000 1.500 2.000 1.000 1.000 - - - - - - - - - - 2.000 - - * - - 0 - - - - - - - - - .500 0 Table 11(D). Average number o f re fe rra ls a t la s t screening fo r one-year e lig ib le Spanish-speaking by age and number o f life tim e screenings. Number of Lifetim e Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Grand Mean % Change As No. Screenings In­ crease By One 1 2 4 5 .450 .442 .817 .676 .833 .783 .825 .878 .786 .669 .813 .755 .721 .808 .836 .890 .830 .600 1.200 .625 7.000 3 • .400 .384 .652 .914 .687 .704 .563 .584 .467 .716 .483 .702 .611 .926 .920 .714 .875 1.000 1.800 1.000 0 1.000 .500 .875 .444 .440 .625 .777 .833 .600 .571 .500 .500 .888 .714 0 .666 1.000 - 1.000 .250 .500 .600 0 .666 1.000 .500 1.000 1.000 1.000 - 6 _ 1.000 1.000 - .569 .764 .872 .712 .877 1.095 1.096 .948 1.032 1.348 .895 .932 1.106 1.046 1.059 1.109 .892 1.071 1.125 1.461 .500 .923 .776 .680 .622 .571 1.000 -16% -12% -9% -8% +75% 104 e x is t on the outcome v a riab le. When re fe rra l rates are calculated across a l l screenings fo r a ll ra c ia l groups, the res u lts are ordered as follow s: blacks, 1.022 re fe rr a ls ; whites, .850 r e fe rr a ls ; American- Indlans, .821 re fe rra ls and Spanish-speaking .815 r e fe rr a ls . major contrast 1s between blacks and a ll other groups. The Blacks have 20 percent more re fe rra ls than w hites, 24 percent more than AmericanIndians and 25 percent more than the Spanish-speaking. With sex as a variab le [Tables I I ( E ) - I I ( F ) ] , the Hypothesis 1 trend 1s consistent across screenings 1 -5, with a single exception. For males w ith fiv e screenings, there is a small (3%) Increase in re ­ fe rr a ls as compared with males who have had four screenings. This up­ swing does not occur w ith females, where, on the contrary, there is a 24 percent decrease in re fe rra ls from the fourth to f i f t h screening. The reason fo r the male Increase 1s not apparent but closer observation o f the table discloses 1t occurs c h ie fly a t ages 5, 6 , 8 and 10 with additional influence from ages 13 and 15. This same upward movement occurred 1n Table I and these same ages were responsible fo r th a t In ­ crease. A fa c to r peculiar to these ages may be responsible but th is occurrence should not be overemphasized since the rate a t fiv e screen­ ings Is s t i l l lower than occurs a t screenings 1 -3, as again was the case with Table I . While re fe rra l rates are very s im ila r fo r both sexes, females do have 6 percent more re fe rra ls when a ll rescreenings are considered as a group (.813 versus .76 4 ). Referral differences by sex do not appear to be meaningful 1n th is study. Tables I I ( G )-II(N ) present outcomes fo r each sex, by race. with the Table I s e rie s , few strong differences are obtained. As Table 11(E). Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le males by age and number o f life tim e screenings. Number of Lifetime Screenings 1 2 3 4 Under 1 I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .752 .823 .815 1.008 1.078 1.114 1.114 1.132 1.117 1,107 1.181 1.139 1.070 1.098 1.159 1.095 1.182 1.108 1.208 1.636 2.333 .633 .612 .695 .894 .804 .888 .865 .856 .867 .858 .846 .851 .855 .840 .883 .910 .904 .953 .857 1.000 4.333 .916 .517 .543 .823 .791 .786 .798 .752 .781 .681 .724 .732 .705 .807 .741 .703 .827 .739 .750 .625 1.000 Grand Mean 1.018 .774 -24% Age % Change As No. Screenings In­ creased By One 0 0 .636 1.142 .789 .755 .726 .709 .565 .675 .716 .737 .611 .484 .772 .843 .978 .590 .700 .500 - 5 _ 1.000 .750 .760 .888 .871 .681 .727 .565 .875 .461 .611 .500 .400 .933 .800 .500 .500 - 0 .333 .333 .333 .250 .500 .666 0 .333 1.000 .500 1.000 1.000 - .753 .711 .730 .451 1.000 -3% -6% +3% -38% +122% 6 _ - 7 - 1.000 - - - Table 11(F). Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le females by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 4 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .739 .757 .746 .916 .992 1.032 1.069 1.045 1.140 1.064 1.099 1.105 1.178 1.179 1.269 1.250 1.277 1.286 1.277 1.187 1.017 .507 .589 .632 .762 .790 .811 .863 .847 .846 .817 .843 .849 .904 .886 .916 .978 .982 1.048 1.220 1.090 .953 1.000 .675 .585 .764 .753 .745 .761 .757 .774 .761 .708 .743 .822 .837 .901 .864 .905 1.034 .895 1.000 .545 .500 1.500 .523 .902 .677 .754 .708 .627 .721 .711 .736 .634 .647 .765 .705 .962 .907 .833 1.035 1.000 .333 .333 .500 .944 .424 .833 .636 .666 .592 .565 1.120 .222 .750 .500 .875 .500 .400 .333 1.000 .500 Grand Mean 1.002 .836 .782 .725 .550 .500 1.00 1.000 -17% -6% -7% -24% -9% +100% 0% % Change As No. Screenings In­ crease By One 5 - 6 7 _ _ - - - - - - - - - 1.500 1.250 0 0 0 - 1.000 8 - 1.000 - - - 0 - - - - 0 - - - - - - - 0 0 2.000 - - - - - - - - - - - - - - - - - - - 1.000 107 Tables I I ( G ) - I I ( H ) show white males have 2 percent more re fe rra ls than white females a t i n i t i a l screening (.928 versus .908) and 7 percent more a t rescreening (.819 versus .8 6 2 ). Also, white male re fe rra ls decrease less than female re fe rra ls from i n i t i a l screening to rescreening. cent. Males decrease 12 percent; females decrease 16 per­ Among whites, females thus e x h ib it a consistent, but sm all, su p erio rity on outcomes. This d ifference is not present when a ll p a rtic ip a n t responses are considered by sex, Irresp ective o f race. Also, th is trend does not hold fo r blacks [Tables I I ( I ) - I I ( J ) ] . While black males have a 2 percent higher re fe rra l rate a t I n i t i a l screening, th e ir rescreening re fe rra l rate is 4 percent lower than that o f females (1.158 versus 1.136 fo r i n i t i a l screenings; .856 versus .893 on rescreenings - males and females resp ectively. Black males thus drop 26 percent on re fe rra ls from i n i t i a l screening to rescreening (1.158 - .856/1.158) and black females decrease 21 per­ cent (1.136 - . 8 9 3 /1 .1 3 6 ). The comparable decreases fo r whites were 12 percent fo r males (.928 - .8 1 9 /.9 2 8 ) and 16 percent fo r fe ­ males (.908 - .7 6 2 /.9 0 8 ). o f any ra c ia l group. These changes by blacks are the largest Black males, as in Table I , show the most re­ sponse to the program as evidenced by the best Improvement in re fe rra l ra te . Neither American-Indians or the Spanish-speaking follow the black trend or show equally high levels o f re fe rra l need. American-Indian females have 9 percent more re fe rra ls than males a t I n i t i a l screening and 12 percent more re fe rra ls a t rescreening [Tables I I ( K ) - I I ( L ) ] . These higher re fe rra l rates fo r American-Indian females are of in te re s t. Table 11(G). Average Nunfcer of re fe rra ls a t la s t screening fo r one-year e lig ib le white males by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Grand Mean % Change As No. Screenings In­ creased By One 3 4 5 6 7 - - - - 0 - - - 1 2 .664 .698 .748 .968 1.001 1.024 1.019 1.047 1.011 1.017 1.093 1.004 .956 .967 1.042 .961 1.018 1.000 1.041 2.000 2.250 .596 .572 .664 .897 .786 .874 .816 .814 .814 .813 .773 .833 .845 .811 .883 .916 .845 1.036 .760 .625 .928 .811 .710 .695 .666 .333 0 -13% -12% -2% -4% -50% -100% - .500 .359 .461 .696 .745 .788 .825 .793 .663 .674 .664 .665 .612 .788 .652 .631 .916 .617 .600 0 - _ .285 .966 .761 .693 .816 .752 .623 .541 .711 .642 .655 .600 .609 .777 .888 .888 .666 .500 - 0 .500 .538 .950 1.000 .520 .461 .333 .875 .555 .857 .571 .400 .777 0 .333 .333 .500 .500 .250 1.000 - 0 0 1.000 0 - - 0 - - - - - - - - - - - - - - Table 11(H). Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le white females by age and number o f life tim e screenings. Number of Lifetime Screenings Aqe Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Grand Mean % Change As No. Screenings In­ crease By One 1 2 3 .683 .657 .675 .872 .930 .942 1.004 .959 1.082 .962 .982 .976 1.020 1.011 1.086 1.084 1.010 1.121 1.213 1.073 1.069 .365 .526 .631 .773 .769 .783 .832 .771 .799 .738 .756 .784 .851 .825 .829 .931 .838 .861 1.116 1.115 .642 1.000 .684 .552 .711 .717 .713 .778 .658 .762 .727 .658 .752 .745 .852 .909 .751 .810 1.164 .689 1.173 .800 .908 .779 -14% 4 5 6 7 8 - - - - 0 - - - - - - - - _ 3.000 .454 .818 .567 .737 .704 .581 .627 .666 .492 .547 .676 .518 .534 .771 1.266 .750 to 1.272 .466 .642 .550 .300 .400 .538 1.666 .307 .625 .571 2.000 .250 0 0 2.000 1.000 0 0 0 - 1.000 - - - 0 - - - - 0 - - - - - - - - - 0 0 - - - - - - - - - - - - - - - - - - - - 1.000 0 0 1.000 .746 .643 .644 .285 0 1.000 -4% -14% 0% -56% -100% +100% - 1.000 Table I I ( I ). Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le black males by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .864 .975 .911 1.087 1.245 1.256 1.284 1.278 1.284 1.245 1.360 1.354 1.266 1.278 1.343 1.266 1.361 1.224 1.500 1.533 2.000 .676 .692 .721 .898 .836 .918 .932 .917 .951 .932 .970 .896 .882 .888 .903 .898 .987 .920 .909 1.428 3.000 Grand Mean 1.158 Age * Change As No. Screenings In­ creased By One 1.125 .766 .643 .988 .822 .804 .794 .713 .958 .711 .800 .836 .794 .823 .813 .784 .794 .788 .904 .833 - 4 _ 0 .800 1.375 .870 .883 .673 .626 .515 .831 .722 .867 .517 .333 .930 .843 1.120 .500 .714 - .888 .808 -23* -9* 5 6 7 1.333 1.000 1.000 1.000 .812 .833 1.666 .714 .833 0 .333 .428 1.000 1.000 .500 .500 - 0 .333 1.000 1.000 1.000 - 1.000 1.000 - .746 .823 .571 .666 -8* +10* -31* +17* — Table I I ( J ) . Average number of re fe rra ls at la s t screening fo r one-year e lig ib le black females by age and number o f life tim e screenings. Number of Lifetime Screenings Age 7 2 3 4 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .807 .889 .843 .982 1.118 1.162 1.166 1.200 1.238 1.204 1.290 1.297 1.414 1.391 1.482 1.441 1.526 1.450 1.367 1.279 .925 .664 .680 .628 .763 .825 .849 .923 .954 .923 .960 .971 .953 1.005 .977 1.032 1.047 1.090 1.214 1.305 1.099 1.194 1.000 .666 .602 .869 .788 .788 .746 .931 .835 .857 .777 .796 .951 .858 .899 .950 1.000 .974 .971 .836 .437 .500 .750 .600 1.000 .803 .760 .742 .723 .818 .771 1.020 .673 .685 1.076 .925 1.170 .742 .875 1.266 1.333 .333 .500 .500 .428 .388 1.000 .769 .909 .750 .555 .583 0 1.000 .333 .500 .750 1.000 .500 1.000 0 Grand Mean 1.136 .921 .848 .830 .637 1.166 1.000 -19% -8% -2% -23% +83% -14% % Change As No. Screenings In­ creased By one 5 6 1 _ - - - - - - - 1.000 1.333 0 - 1.000 - - - - - - - - - - - - - - - * 2.000 - - - - - - - - - - - Table II( K ) . Average Number of re fe rra ls a t la s t screening fo r one-year e lig ib le American Indian males by age and number of life tim e screenings. Number of Lifetime Screenings Age Grand Mean * Change As No. Screenings In­ crease By One 2 3 _ 4 _ 5 6 _ _ - - - - - - - .533 .769 .569 .933 1.272 .750 1.066 .875 1.615 .900 .750 .833 .461 1.142 .750 .666 1.000 - .800 .166 .777 .615 1.166 .900 1.166 1.100 .750 .545 .857 .875 .600 .875 .500 1.142 .500 - 1.666 1.333 .428 .250 .222 .666 .500 1.000 0 .250 .500 0 0 0 0 .333 0 - - 1.000 .500 1.000 2.000 .250 .333 1.000 0 * 0 2.000 1.000 1.000 2.000 - 0 - - - - - - - .860 .806 .514 .705 1.333 0 -6 * -36* +37* +89* -100* 112 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I I ( L ). Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le American Indian females by age and number o f life tim e screenings. Number of Lifetime Screenings Age Grand Mean 35 Change As No. Screenings In­ crease By One 2 3 4 5 1.000 1.133 .550 1.375 .250 1.000 2.000 1.700 .823 1.000 .600 .875 1.000 1.285 .666 .500 1.200 1.000 0 1.000 2.000 1.000 .571 .400 1.000 .466 1.181 1.000 1.384 1.222 .875 .625 .545 .818 .818 .500 .555 .400 1.500 1.500 1.000 • • _ - - - 4.000 1,000 .600 .250 .500 .500 .500 .750 0 .250 .333 1.333 .400 0 0 0 m - - - - - .939 .832 .578 1.277 1.200 -1135 -3135 +12135 -635 - 1.000 4.000 .333 .500 - - 1.000 - 2.000 1.000 1.666 2.000 1.000 1.500 2.000 - - - - - - - - - - 0 .500 - - - - - 113 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table II(M ). Average number of screenings a t la s t screening fo r one-year e lig ib le Spanish-speaking males by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Grand Mean % Change As No. Screenings In­ crease By One 1 2 .636 .400 .860 .800 .785 .800 .880 .936 .785 .745 .764 .569 .745 .774 .735 .857 .631 .700 1.000 0 7.000 3 • 0 .500 .722 1.030 .571 .666 .702 .750 .518 .724 .695 .846 .869 .916 .636 .625 .600 .500 - .603 .928 .954 .561 .826 1.184 .947 1.081 .969 1.368 .781 .894 1.166 1.093 1.000 1.200 .500 .800 1.000 .929 4 1.000 .500 .750 .250 .400 .300 1.000 .750 .555 1.000 .666 .750 1.200 .333 0 - .766 .724 -18% -5% 5 6 _ 0 - - - - - - - - 1.000 .250 0 .600 0 - 1.000 0 1.000 - - - 1.000 - 1.000 1.000 - 1.000 - .635 .523 1.000 -12% -18% +91% - Table 1I ( N ) . Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le Spanish-speaking females by age and nunber of life tim e screenings. Nunber of Lifetime Screenings Age Grand Mean % Change As No. Screenings In­ crease By One 2 3 4 5 .536 .591 .800 .850 .931 1.000 1.276 .790 1.100 1.326 1.000 .967 1.037 1.000 1.100 1.033 1.000 1.222 1.142 1.461 .500 .222 .500 .767 .553 .882 .766 .789 .838 .786 .600 .863 .918 .696 .844 .941 .935 .941 .560 1.250 .714 - _ _ 1.000 .285 .607 .810 .827 .758 .411 .379 .428 .709 .351 .523 .419 .941 1.142 .833 1.000 1.250 1.800 1.000 - - .918 .785 .637 .611 .714 -1455 -19% -4* +1735 - - - - .500 1.000 .600 .450 .857 .428 .875 .666 .307 .444 0 .500 1.000 - .666 1.000 - - 1.000 - 0 .666 1.000 1.000 - 115 Linder 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 116 Female Indian rates are a c tu a lly second only to blacks and these rates are not consistent with Table I fin din gs. In Table I , American- Indian males had 3 percent more re fe rra ls than females, a ll re fe rra ls considered, while 1n Table I I Amer1can-Ind1an females have 10 percent more re fe rra ls fo r the same comparison (.860 veruss .7 8 2 ). With the Spanish-speaking, the trend is reversed but with only minor differences [Tables I I ( M ) - I I ( N ) ] . Males have 1 percent more re fe rra ls a t f i r s t screening; 2 percent more a t rescreening. Both Spanish-speaking males and females show a 20 percent decrease in re fe rra ls from i n i t i a l screen­ ing to rescreenings combined. In summary, outcomes fo r Tables I I ( G ) - I I ( N ) are generally very s im ila r to those obtained fo r the companion Tables I( G ) - I( N ) . The two population groups under study, shorter and longer-term e lig ib le s , ex­ h ib it s im ila r patterns in re fe rra l rates when both sex and race are variab les. Amerlcan-Indians are an exception to th is g en eralizatio n . The existence o f the inverse relatio n sh ip between re fe rra l rates and extent o f program p a rtic ip a tio n - which holds across a l l p articip ants (Table I I ) - is l i t t l e affected when results are divided by sex fo r each race. I f grand means are based on 100 or more subjects (96 sub­ je c ts fo r Spanish-speaking males with 5 screenings), Hypothesis 1 is confirmed in a l l cases, except one. For whites o f both sexes, the downward trend exists across screenings 1-5 (although rates a t screen­ ings 4 and 5 are v ir t u a lly id en tic al fo r white females). For black males and females, the trend exists across screenings 1 -5, except fo r males a t 5 screenings. For the Spanish-speaking, the trend covers screenings 1-4 fo r both sexes. For Amerlcan-Indians, the trend 117 encompasses screenings 1 -2 . The exception to th is con sisten t, Inverse re la tio n s h ip is black males where there is a 10 percent increase in r e fe rr a ls from the fou rth to f i f t h screening. This is consistent w ith the data a t f i f t h screen­ ing fo r a l l males, and indeed la rg e ly explains th a t upturn since only black males show th is trend. Also, the same ages p rim a rily responsible fo r the outcome fo r a l l males (ages 5, 6 , 8 and 1 0 ), are also responsible fo r the differences w ith blacks. The location fa c to r is considered in Tables I I ( 0 ) - I I ( Q ) w ith re ­ su lts s im ila r to those obtained in the Table 1 se rie s . Screenings 1-4 in Tables I I ) O ) - I I ( P ) are based on a s u ffic ie n t number o f subjects to place confidence in the obtained re s u lts . ou tstate resid en ts, the Hypothesis For both D e tro it and r u r a l, 1 rela tio n s h ip holds across the f i r s t four screenings, i . e . location does not a lt e r the re la tio n s h ip . Furthermore, the slope o f inverse rela tio n s h ip is remarkably s im ila r fo r both groups. I t decreases across screenings 1-4 by 14%, 4% and 7% fo r D e tro ite rs and 11%, 3% and 7% fo r o u ts ta te rs . The second s trik in g re s u lt is th a t the c it y residents have a higher r e fe rr a l r a te . D e tro it­ ers have 61 percent more r e fe r r a ls a t the i n i t i a l screening (1.255 ver­ sus .781) and 56 percent more r e fe rr a ls when a l l rescreening resu lts are averaged (1.066 versus .6 8 4 ). Over a l l screenings, D e tro ite rs average 64 percent more r e fe r r a ls than the ou tstate residents (1.202 versus .7 3 2 ). As w ith the Table I s e rie s , fu rth e r analysis shows th a t location remains an in flu e n tia l fa c to r even when race is co n tro lled as a v ariab le [Table I I ( Q ) ] . Urban whites have 39 percent more r e fe rr a ls than rural Table 11(0). Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le participants in D e tro it by number of life tim e screenings. Number of Lifetime Screenings 1 Mean % Change As No. Screenings In­ crease By One 1.255 2 3 1.081 1.039 -1451! -451! 4 .971 -7% 5 .815 -16% 6 1.000 +23% 7 1.000 0% ♦Since the amount of computer space needed to breakdown group means by age was prohibitive i t was possible to present group means only for this table. Table I I ( P ) . Average number of re fe rra ls a t la s t screening fo r one-year e lig ib le participants in fo rty -fo u r Northern Michigan counties by number of life tim e screenings. Number of Lifetime Screenings 1 Mean % Change As No. Screenings In­ crease By One .781 2 3 4 5 6 7 .698 .676 .630 .525 .176 .333 -11* -3* -7% -17% -66% +89% 120 whites (1.061 versus .7 3 5 ). re s u lts . Comparisons w ith blacks show s im ila r D e tro it blacks average 1.293 r e fe rr a ls a t I n i t i a l screen­ ings; ou tstate* ru ral blacks average .705 r e fe r r a ls . This 1s a d iffe re n c e o f 83 percent more r e fe rr a ls fo r D e tro it blacks. For a l l rescreenings combined, D e tro it blacks average 1.086 r e fe r r a ls ; ou tstate blacks average .569 r e fe r r a ls . blacks have 91 percent more r e fe r r a ls . On th is comparison* D e tro it Over a l l screenings, D e tro it blacks have 96 percent more r e fe rr a ls (1.232 versus .6 2 8 ). These location differences are very large and suggest lo c a tio n , as defined 1n th is study, does Indeed play a large ro le 1n determining the ra te of r e fe r r a ls . I t Is also noteworthy th a t the location d i f ­ fe r e n tia l is greater fo r blacks than fo r w hites. Rural blacks have about h a lf as many r e fe rr a ls as urban blacks (.628 versus 1.232) while ru ral whites have nearly 70 percent as many r e fe rr a ls as urban whites (7 .3 5 versus 1 .0 6 1 ). This Is but another way o f saying th a t rural blacks have a lower r e fe rr a l rate than e ith e r ru ral or urban w hites. This 1s an unexpected fin d in g and c le a rly runs counter to th is study's general fin d in g o f higher r e fe r r a l rates fo r blacks. Looked a t from th is perspective, I t appears race does exe rt an Influence even though I t does not explain the d ifferen ce 1n outcome by lo ca tio n . In summarizing Tables I and I I , the most noteworthy res u lts are as follow s: ( 1 ). Hypothesis 1 1s generally confirmed where re s u lts are based on a s u ffic ie n tly large number o f subjects (100 or more). ( 2 ). Demographic facto rs do not in v a lid a te , or d is t o r t, the Table I I ( p ) . Average number of re fe rra ls at la s t screening in D etro it and Northern Michigan by race and number of life tim e screenings. Number of Lifetime Screenings Location/Race 1 2 3 4 5 6 7 - - Grand Mean Detroit Whites Blacks Grand Mean 1.090 1.293 1.262 .973 1.104 1.087 .873 1.059 1.044 1.136 .952 .966 1.400 .750 .794 .783 .705 .781 .699 .487 .695 .683 .688 .683 .628 .555 .625 .478 1.000 .492 1.000 1.000 1.000 1.000 1.061 1.232 1.208 .125 0 1.000 .333 .735 .628 .732 Northern Michigan Whites Blacks Grand Mean - .125 122 Hypothesis 1 re la tio n s h ip . ( 3 ). Location and (fo r blacks) race do ex ert a meaningful influence on the outcome v a ria b le even though they do not a l t e r the inverse re la tio n s h ip between r e fe rr a l rates and number o f life tim e screenings. Blacks and urban residents gen erally have higher r e fe r r a l rates than whites and r u r a l, ou tstate residents re s p e c tiv e ly . An in te re s tin g , and seemingly meaningful, exception to th is general trend is the lower r e fe rr a l ra te fo r rural blacks. Also, there 1s some in d icatio n th a t black males respond unusually w ell to the program, as evidenced by t h e ir r e la t iv e ly larg e decrease In r e fe rr a ls from I n i t i a l screening to rescreening. ( 4 ). D ifferences between long and shorter-term e lig ib le s seem few and are lik e ly the re s u lt o f u n re lia b le data stemming from study group size rath er than c h a ra c te ris tic s o f the population. Although, as shown above, Tables I and I I do confirm Hypothesis 1, some q u a lific a tio n and reanalysis 1s warranted since the obtained data can also be arranged to show th a t h isto ry has exerted an Influence on r e fe rr a l ra te s . H is to ry 's ro le 1s i t s e l f an In te re s tin g fin d in g but Indicates some overestim ation o f the strength of the Hypothesis 1 re ­ la tio n s h ip has occurred in the Tables analyzed thus f a r . Again, the o rig in a l expectation was th a t overestim ation would most lik e ly occur a t t h ir d , fou rth and f i f t h screenings since these occurred la t e r in the program's h is to ry when re fe rr a ls were given less freq u e n tly. 123 H istory would thereby act to strengthen and confound the Hypothesis 1 re la tio n s h ip . However, data below show the influence did not work q u ite as an tic ip a te d and although h is to ry does ex ert an in flu e n c e , i t does not in v a lid a te Hypothesis 1. The follow in g four tables establish these points: Table I(R ) succinctly shows two important fin din gs of th is study: (1 ) the e ffe c t o f h isto ry and (2) the v a lid it y of Hypothesis 1 despite h is to ry 's confounding influence. H is to ry 's influence can be seen by d ire c t observation o f the columns. For each life tim e screening, r e fe r ­ ra ls rates gen erally decrease as one descends the column, i . e . as the program ages. The descent is not p erfect as in each column there is one year in which rates increase over the previous year. general trend is c le a r. column. However, the This trend Is unequivocal in the grand mean Between 1973 and 1980, the re fe rra l ra te decreased 59 percent (1.483 - .6 0 9 /1 .4 8 3 ). Between 1974 and 1979, the decrease was 44 per­ cent (1.382 - .7 7 4 /1 .3 8 2 ). These are strong changes and occur con sisten tly each y ea r. An in flu e n tia l fa c to r must be operative to so con sisten tly and markedly depress the r a te . In the past f iv e years, the o v e ra ll ra te has de­ creased nearly 50 percent (11180 - .609/11180 * 49%). Despite th is strong e ffe c t o f h is to ry , r e fe rr a l rates and degree o f program p a rtic ip a tio n maintain a c le a r inverse re la tio n s h ip . This can be seen by observation o f the rows (years) where h isto ry 1s con­ t r o lle d . In each row, except one, re fe rr a ls decrease as the number of life tim e screenings increase. from 3 to 4 screenings in 1980. The single exception occurs in moving The upturn is not large and i t is Table I(R ). Average number o f re fe rra ls a t la s t screening fo r long-term e lig ib le s by number and year of screening (n > 100). Number of Lifetime Screenings 2 3 4 5 Year 1 Grand Mef 1973 1.483 1974 1.390 1.142 1.382 1975 1.219 1.025 1.180 1976 1.067 .919 .794 1977 1.008 .947 .877 .715 .957 1978 1.036 .884 .868 .765 .908 1979 .870 .825 .703 .698 1980 .773 .687 .521 .550 Grand Mean 1.185 .898 .777 .708 1.483 1.003 .696 .774 .609 .696 125 noteworthy th a t the rate a t four screenings is based on only 140 subjects, not many over the minimum deemed necessary to obtain v a lid re s u lts . Further analysis of the change in Table I(R ) is given in Table I{ S ). This shows the percent change in r e fe r r a l rates as the number o f screenings increase by one. The comparison to make is the change in each year to the grand mean change of Table I where year of screen­ ing was not held constant. In noting the decrease in r e fe rr a l rates between one and two screenings, fo r the e n tire ta b le , i t can be seen th a t actual change, by y e a r, is less than the grand mean decrease of 24 percent. This most c le a rly shows the e ffe c t of h is to ry in over­ estim ating the decreasing re la tio n s h ip . The average decrease from screening one to screening two, fo r years 1974-1980, is 12 percent, or h a lf o f the grand mean change. The average change fo r movement between screenings 2 and 3 is -12 percent, or only one percent less than the grand mean change. The average change between screenings 3 and 4 is -6 .2 5 percent as compared with the grand mean change of -9 percent. O v e ra ll, with his to ry c o n tro lle d , the reduction in grand mean rates is about 2/3 of the decrease shown in Table I (.1 2 + .12 + .0 6 2 5 /.2 4 + .13 + .0 9 ). Tables I I ( R) and I I ( S ) support the findings of the Table I s erie s. Again, w ith a few exceptions, each column 1n Table II( R ) shows a de­ crease in re fe rr a ls as the years pass. there 1s a consistent decrease. For the grand mean column, Between 1973 and 1980, the decrease in grand means was 59 percent (1.476 - .6 0 3 /1 .4 7 6 ). Between 1975 and 1979, the decrease was 33 percent (1.120 - .7 4 7 /1 .1 2 0 ). In the Table I(S ). Percent change in average number of referrals a t la s t screening as number of lifetim e screenings increase by one for long-term eligib les by year of screening (n > 100). Year 1-2 1974 -18* 1975 -16* 1976 -14* -14* 1977 -6 * -7* • 00 ** Change in Number of Lifetime Screenings 1978 -15* -2 * -12* 1979 -5* -15* -1 * 1980 -11* -24* +6% Grand Mean* -24* -13* -9 * * Grand Mean change is taken from Table 1. 2-3 3-4 4-5 0* +2* Table II( R ) . Average number of re fe rra ls a t la s t screening fo r one-year eligiblesby number and year of screening (n > 100). Number of Lifetime Screenings Year 1 1973 1.476 1974 1.350 1.153 1.344 1975 1.137 1.024 1.120 1976 .995 .885 .824 1977 .994 .904 .890 .787 1978 .933 .869 .859 .800 .923 .895 1979 .794 .743 .695 .695 .646 .747 1980 .691 .612 .538 .557 Grand Meand 1.010 .837 .765 .718 2 3 4 5 Grand Mean 1.476 .961 .923 .603 .709 128 program's past f iv e years* re fe rr a ls decreased 37 percent (.961 .6 0 3 /.9 6 1 ). These trends are a l l consistent w ith Table I data and show r e fe rr a ls have been given w ith less frequency as the program has aged. The e ffe c t o f h is to ry in d is to rtin g Table I I is shown in Table I I ( S) . For each year between 1974 and 1980, the decrease in re fe rr a ls from screening one to screening two is less than the Table I I grand mean change o f -17 percent. fo r these years was The average decrease 9 percent between screenings one and two, again nearly h a lf o f the change represented by the Table I I grand mean. The average decrease from screening two to three was 5.6 percent (versus 8% shown 1n Table I I ) and the decrease between screenings three and four was 3.75 percent (versus 7% 1n Table I I ) . Over a l l screenings, the decrease in r e fe r r a l rates w ith h is to ry c o n tro lle d is 57 percent o f the ra te obtained in Table I I without th is control (.0 9 + .056 + .0 3 7 5 /.1 7 + .08 + .0 7 ). Again, regardless o f the influence of h is to ry , Hypothesis 1 is g en erally confirmed 1n Table I I ( R )- Each row (year) shows a general downward trend 1n re fe rr a ls as program p a rtic ip a tio n Increases, a l ­ though there are several exceptions. In 1978, those receiving a f i f t h screening had more re fe rr a ls than those receiving a fourth screening. However, the f i f t h screening was based on 118 subjects, only a small number above the minimum considered necessary fo r a v a lid base. A second exception is between screening three and four In 1979 where there 1s a le v e lin g e ffe c t ra th er than a continued downward movement. The f in a l exception occurs between screening Table II( S ) . Percent change in average number of referrals at last screening as number of lifetim e screenings increase by one for one-year eligib les by year of screening (n > 100). Changes in Number of Lifetime Screenings Year 1-2 1974 -15* 1975 -10* 1976 -11* -7* 1977 -4* -2* -12* 1978 -7* -1* -7* +15* 1979 -6* -6* 0* -7 * 1980 -11* -12* +4* Grand Mean* -17* -8 * -7* ♦Grand Mean change is taken from Table I I . 2-3 3-4 4-5 -10* 130 three and four in 1980. Here, the fourth screening is based on 262 subjects which would seem to be an adequate number. This change does seem to be a deviation from the general trend but the rate a t the fourth screening remains lower than the ra te a t screenings one and two. In summary, an Inverse re la tio n s h ip between r e fe rr a l rates and number o f screenings is present with year held constant, although i t 1s not a t a l l times a p erfect rela tio n s h ip a t a ll points o f program p a rtic ip a tio n . Table I I I and Table IV Results Although the Table I and Table I I series e stab lish the general v a lid ity o f Hypothesis I , several ad d itio n al approaches were used to explore fu rth e r th is re la tio n s h ip . Tables I I I and IV show r e fe r r a l rates fo r screenings occurring in 1978 only. outcomes, h is to ry 's e ffe c t is c o n tro lle d . By presenting same-year Based upon the data shown in Tables I / I I ( S ) , the expectation Is th a t Tables I I I and IV w ill show Hypothesis 1 holding in a given year but showing less strength than was present in Tables I and I I where h isto ry confounded re s u lts . Accordingly, Tables I I I and IV are o f In te re s t but w ill be discussed In a more summary fashion than Tables I and I I since the la t t e r did confirm Hypothesis I . To sim p lify fu rth e r the follow ing discussion, and In recognltHon o f the volume o f tables used 1n th is study and the re s u ltin g discon­ t in u it y should they a l l be placed w ith in the n a rra tiv e , Tables I I I and IV are placed In Appendices C and E res p e c tiv e ly . The "Ns" fo r Tables I I I and IV are placed In Appendices D and F res p e c tiv e ly . Also, since 131 previous review has generally found l i t t l e d iffere n c e in outcomes between the long and short-term e lig ib le s , Tables I I I and IV w ill be discussed concurrently rath e r than s e q u e n tia lly. As expected, Table I I I and Table IV , giving r e fe r r a l rates fo r a l l subjects in the population without demographic breakdown, confirm Hypothesis 1 across a ll grand means which are based on a s u ffic ie n t number o f subjects, w ith a single exception. In Table IV , re fe rr a ls increase 15 percent between screenings four and f iv e . However, the screening f iv e ra te is based upon only 118 subjects, which is the lik e ly source o f d iffe re n c e . Also, as shown 1n Table 1/ I I ( S) , the ra te o f decrease is lower than obtained without a control on the year of screening. That Is , whereas the decrease between screening one and screening two was 24 percent in Table I and 17 per­ cent in Table I I , i t is 15 percent and 7 percent fo r Tables I I I and IV res p e c tiv e ly . Again, th is is important inform ation since i t says the program's b e n e fits are not as strong as they appear to be without proper a n a ly tic c o n tro l. Race as a fa c to r [Tables I I I / I V ( A ) - ( D ) ] does not disru p t the Hypothesis 1 re la tio n s h ip with the exception o f blacks in the Table IV s e rie s . Here, r e fe rra l rates a t screenings two-four are very s im ila r , although they are 10-11 percent below the I n i t i a l ra te . Black rates remain highest o f the r a d a l/e th n ic groups although the black-w hite d iffere n c e 1s somewhat sm aller than in the Table 1 /I I s e rie s . At i n i t i a l screening, blacks had 28 percent and 20 percent more r e fe r ­ ra ls than whites In the Tables I and I I series res p e c tiv e ly ; the comparable differences are 23 percent and 17 percent fo r Tables I I I 132 and IV resp ectively. The number o f American-Indians was insu f­ f ic ie n t to place confidence in th e ir re s u lts . Differences by sex [Tables I I I / I V ( E ) - ( F ) ] do show a complete reversal from the Table I / I I series with females screened in 1978 showing higher re fe rra l rates than males. The differences were sm all, 6 percent and 2 percent in Tables I I I and IV resp ectively. However, th is is an In te re s tin g development given that males have tr a d itio n a lly exhibited somewhat higher rates of medical problems with females only recently narrowing the d ifferen ce on some indices. Also, rates fo r females ris e in both Table I I I and Table IV between screenings two and three. The increases were small (2%-3%) but th is upturn was not present in Tables 1(F) and 11(F). In both cases, the rates decrease a t screening fo u r. Analyzing the ra c ia l/e th n ic results by sex [Tables I I 1 / IV(G) (N )] does not provide much new inform ation. I t does show th a t females of each group have higher re fe rra l rates than males a t I n i t i a l ing. screen­ Also, i t discloses th a t the increase in female re fe rra ls between screenings two and three is p rim a rily a ttrib u ta b le to white females although the upswing is also evident fo r blacks. This series appears to correct the impression th a t black males show a disproportionately p o s itiv e response to the program as re fle c te d in unusually high de­ creases in r e fe r r a ls . From Table 1 (1 ), the decrease in th e ir re fe rra l rates between i n i t i a l screening and rescreenings was calculated to be 34 percent. From Table 11( I ) the comparable decrease was 26 percent. These decreases were the largest shown by e ith e r sex in any ra c ia l group. However, they do not hold fo r those screened in 1978. Tables 133 I I I / I V ( I ) and I I 1 / IV (J ) show changes by black males and females are qu ite s im ila r. From Tables 111 ( I ) and I I I ( J ) , re fe rr a ls decrease 18 percent from I n i t i a l screenings to rescreening fo r both males and females (1.097 - .895/1.097 fo r males and 1.186 - .974/1.186 fo r fem ales). From Tables IV < I) and IV ( J ), the comparable decreases are 12.5 percent fo r males (1.103 - .8 8 6 /1 .0 1 3 ) and 10.5 percent fo r females (1.045 - .9 3 5 /1 .0 4 5 ). Location differences hold fo r those screened 1n 1978. In the Table I I I s e rie s , D e tro it residents averaged 61 percent more r e fe rr a ls than ou tstate p a rtic ip a n ts a t i n i t i a l screening (1.318 - .8 1 7 /.8 1 7 ), and 51 percent more r e fe rr a ls fo r a l l rescreenings (1.129 - .7 4 6 /.7 4 6 ). In the Table IV s e rie s , the comparable differences were 38 percent a t i n i t i a l screening (1.057 - .7 6 8 /.7 6 8 ) and 40 percent fo r rescreenings (1.077 - .7 7 1 /.7 7 1 ). The l a t t e r differences were less than those ob­ tained 1n the comparable Table I I series but are s t i l l considerable. I t is reasonable to expect these data are consistent w ith the Table 1/ I I series w ith respect to the gen erally higher r e fe r r a l ra te o f blacks not being responsible fo r these urban-rural d iffe re n c e s . Analysis o f Covariance The use o f analysis o f covariance to te s t form ally Hypothesis 1 afforded a d iffe r e n t and supplementary perspective on the predicted re la tio n s h ip between r e fe r r a l rates and number o f life tim e screenings. However, since the population data c o n sisten tly showed a strong, although not always p e rfe c t, 134 Inverse relatio n sh ip between these varia b les , we would expect analysis of covariance to confirm Hypothesis 1. This confirmation did occur. As noted previously, when there 1s in te re s t in analyzing the influence o f two independent va ria b les , one o f which is nonmetric and the other m etric, a useful analysis to employ is the analysis o f covariance. study. Both metric and nonmetric variables were present in th is The o rig in a l independent variab le o f In te re s t, "number of screenings," 1s a m etric v a ria b le. That is , numbers have an ordered relatio n sh ip o f equal difference between them. Tables I / I I ( R ) show th a t th is variab le does a ffe c t re fe rra l rates. However, these tables also show th at tim e, or the "year o f screening," has a s im ila r, depres­ sing influence on r e fe rr a ls . Accordingly, "screening year" is also an independent v a ria b le , but one which is nonmetric. fia b le meaning other than representing a category. "Year" has no quanti­ Moreover, Tables I / I I ( S ) show th a t re fe rra l rates vary somewhat by year as a function o f number o f screenings. S p e c ific a lly , th is is indicated by the d i f ­ fe rin g percentage o f reductions in re fe rra ls as screening year varies. This means the slope o f the inverse relatio n sh ip between re fe rra l rates and number o f screenings varies by year of screening, or said d iffe r e n tly , screening history does not have the same magnitude o f e ffe c t on re fe rra l rates each year. In short, some in te ra c tio n Is occurring between the twoIndependent variab les. Because o f the types o f variables Involved, and th e ir In te ra c tio n , the most appropriate method o f analyzing them 2 fu rth e r Is the m ultip le regression method o f analysis o f covariance. 2 Norman H. N ie, e t a l . , S ta tis tic a l Package fo r the Social Sciences, (New York: McGraw-Hill Book Company, 1975), 381. 135 The nominal variab le "screening year" 1s represented by a set of dummy, in te rv a l variables (0 ,1 ) which serve to tr e a t each year as a separate category and to assign thereby an a rb itra ry m etric value to each year. The standard SPSS program provides several m ultip le regression solutions in the analysis o f covariance context in order to te s t var­ ious hypotheses o f in te re s t. Hypotheses and th e ir solutions are d is ­ cussed next fo r the long-term e lig ib le s . L o g ically, the f i r s t question to answer In th is method of analysis is whether the model which best rela te s dependent and Independent var­ iables is in te ra c tiv e or a d d itiv e . As discussed, the d ifferen ce in these models concerns whether the e ffe c t o f past screenings is constant or variab le across the years. Tables I / I I { R) showed some variab le or in te ra c tiv e influence, however th is strength proved in s u ffic ie n t to be re fle c te d in the analysis o f covariance. The obtained, overall F fo r the in te ra c tio n model, which tests its s t a t is tic a l s ig n ifican ce, was 45.63 w ith the c r it ic a l level of 95F1 11202 * 3.84. This means the in te ra c tio n model was s t a t is t ic a lly s ig n ific a n t a t the .05 level and lik e ly depicts a true relatio n sh ip e x is tin g among the dependent and Independent variab les. In other words, re fe rra l rates are inversely rela te d to past screenings, given control of year of screening and in terac tio n between past screening and year o f screening. 2 However, the obtained R fo r th is model was only .0503 which means the model explains but 5 percent of the variance 1n re fe rra l rate s . This is very low, lik e ly ind icatin g considerable v a r i a b i l i t y In outcome and meaning th a t the pred ictive power of the model is weak, even 136 though the model q u ite lik e ly depicts a true inverse re la tio n s h ip . 2 The low R also means the fo llo w in g , d e ta ile d analyses done o f the regressions are somewhat "academic" since they t e l l much about factors which explain l i t t l e . 2 The overrid ing message o f the obtained R th a t Hypothesis 1 is confirmed but explains l i t t l e is about r e fe rr a l rates. In determining whether the in te ra c tio n or a d d itiv e model was most appropriate fo r the obtained data, the question is whether the Increment in the proportion o f variance in r e fe r r a l rates accounted fo r by the In te ra c tio n is s ig n ific a n t. The comparison is accordingly between the R^ obtained by the in te ra c tio n model (.0503) and the R^ obtained by the a d d itiv e model (.0 4 9 7 ). As can be seen even without s t a t is t ic a l anal2 y s is , there is v ir t u a lly no d iffere n c e in the two R s and formal te s tin g confirmed the d iffere n c e was not s t a t i s t ic a l ly s ig n ific a n t (c alcu latio n s and ANCOVA summary are in Appendix G fo r the long-term e llg ib le s ) . Con­ s is te n t w ith the In te ra c tiv e model's lack o f ad d itio n al explanatory power was the lack o f s t a t is t ic a l sig n ifican ce fo r in te ra c tio n terms in any single year except one. Only in te ra c tio n fo r 1974 was s t a t is t ic a lly s ig n ific a n t fo r the six years studied. Apparently, the in te ra c tio n re fle c te d in Tables I / I I ( R ) was simply not s u ffic ie n tly powerful to be determined s t a t is t ic a lly s ig n ific a n t. Therefore, the a d d itiv e model was considered most appropriate fo r repre­ senting t ie data since th is is procedurally the correct decision when the In te ra c tiv e model 1s not determined superior to the a d d itiv e one. However, the in te ra c tiv e model is presented and discussed b r ie f ly in Appendix H. In any event since the p rediction o f r e fe r r a l rates was not o f d ire c t concern in th is study, the negligibly d iffe r e n t p red ic tiv e 137 a b ilit ie s o f these models was o f l i t t l e consequence. The add itive model is as follows: D1 s + Where: CO Y' « A + B 1° l !°2 + B3D3 + B4D4 + b5d ►B6D6 + B7i 5 ' l i f la s t screened in 1973, 0 otherwise; i f la s t screened in 1974, 0 otherwise; D2 - l °3 m l i f la s t screened 1n 1975, 0 otherwise; °4 - l i f la s t screened in 1976, 0 otherwise; °5 - l i f la s t screened in 1977, 0 otherwise; screened in 1978, 0 otherwise; D6 - l i f la s t NumScren * Number o f life tim e screenings; and Y* ■ Predicted re fe rra l ra te , B * Slope fo r 1975 A ■ "Constant" or YIntercep t fo r 1979 * Slope fo r 1973, Bg * Slope fo r 1974, Bg ■ Slope fo r 1978, B7 * Slope fo r 1979 or Slope fo r NumScren when Dj-Dg * 0 , i . e . when screening year 1s con trolled . The model specifies a lin e a r relatio n sh ip between the Independent and dependent variables and, given knowledge o f the la s t year o f screening and the number o f life tim e screenings, predicts the number o f re fe rra ls . The predictive accuracy o f the model (in comparison to actual outcomes) 1s given by the M u ltip le R -c o rre la tio n . The s ta tis tic a l significance of the M u ltip le R, and therefore the s ta tis tic a l significance of the e n tire model is given by the overall F te s t. e lig ib le s . Results follow fo r the long-term 138 Table V. Results o f m u ltip le regression method o f analysis o f covariance fo r long-term e lig ib le s . M u ltip le R 0.2231 F R Square 0.0497 83.8379 Adjusted R Square 0.0491 Standard Error 1.0548 Variables In The Equation Beta Std. Error B F 0.6596 0.1102 0.0613 115.725 0.5625 0.1749 0.0396 200.948 0.3437 0.0978 0.0407 71.261 0.1604 0.0468 0.0387 17.142 0.1692 0.0565 0.0337 25.111 0.1235 0.0478 0.0296 17.372 NumScren -0.0822 -0.0680 0.0138 35.569 Variable Di °2 °3 D5 °6 Constant B 0.9542 The obtained F o f 83.8379 is the re s u lt of the o v erall F te s t which estimates whether the sample data have been drawn from a population w ith a m u ltip le c o rre la tio n equal to zero, o r e q u iv a le n tly , whether the obtained m u ltip le R c o rre la tio n Is a c tu a lly due to sampling v a r i­ a tio n . T e c h n ic a lly , the n u ll hypothesis being tested is th a t the mul­ t ip l e c o rre la tio n Is zero. The obtained F Is compared w ith the F value given by s t a t is t ic a l ta b le s , a t the desired level o f sig n ific an ce . If the computed F value 1s la rg e r than the s t a t is t ic a l ta b le 's c r it ic a l value, the n u ll hypothesis 1s re je c te d . Otherwise, 1t is concluded th a t 139 the obtained (m u ltip le ) R 1s not a s t a t is t ic a l ly s ig n ific a n t fin d in g . For the Table V data, the c r i t ic a l value fo r u g lO “ 3-84. The obtained o v e ra ll F exceeds th is value so the nu ll hypothesis 1s re je c te d . There is reasonable assurance th at the R r e fle c ts a true re la tio n s h ip . The m u ltip le R, or "R," is the " c o e ffic ie n t o f m u ltip le c o rre la tio n " and gives a measure o f c o rre la tio n between the dependent and independent 2 v a ria b le s . The R (the " c o e ffic ie n t o f m u ltip le determ ination," which is simply the square o f the R) has a more straightforw ard meaning and thus is usually used to in te rp re t the fin d in g s . 2 R is a measure o f the e ffe c t o f a l l the independent variab les combined on the dependent v a r i­ ables. More s p e c ific a lly , i t gives the percent o f variance o f the de­ pendent v a ria b le which is explained by the regression equation. I t is a measure o f the "goodness o f f i t " o f the regression lin e to the actual data. Table V shows th a t the f u ll model explains 4.98 percent o f the variance in r e fe r r a l ra te s . This is not a "good" explanation, meaning the selected independent variab les are poor predictors o f r e fe rr a l rates . 2 C le a rly , a R o f .80 or above would show strong p re d ic tiv e powers. In such a case, knowledge o f p red icto r variables could be used to estimate outcomes w ith l i t t l e margin o f e rro r and would in d ica te a strong assoe la tio n between p red icto r and outcome v a ria b le s . 2 The obtained R in th is study indicates a re la tio n s h ip e x is ts between the variab les o f In te re s t but the re la tio n s h ip 1s subject to much variance and apparent influence from other fa c to r(s ) as evidenced by the considerable s h o rtfa ll from p e rfec t association (1 .0 0 ). 140 The adjusted R Square is a measure which takes into account the number of independent variables in re la tio n to the number o f obser­ vations. Its purpose is to f a c ili t a t e comparisons of the "goodness o f f i t " o f several regression equations th at may vary w ith respect to the 3 number o f independent variables and observations. No such comparisons were made in th is study. Also, because o f the large sample size used in th is study, the adjusted and unadjusted R Squares are v ir t u a lly id e n tic a l. The standard e rro r o f estimate gives the "average" e rro r in pre­ d ic tin g re fe rra l rates upon the basis o f the variables in the regression equation. (Y-Y' ) . T ech nically, i t is the standard deviation o f the residuals The obtained standard e rro r o f 1.0548 means th a t approximately 68 percent o f the actual scores w ill be w ith in one standard deviation o f the predicted score, or Y' ± 1.0548, assuming a normal d is trib u tio n of actual scores about the regression lin e . This 1s a f a i r l y wide mar- 2 gin o f e rro r and was fo re to ld by the low R . The B values are the " p a rtia l regression c o e ffic ie n ts " and indicate the influence o f each independent variab le on the outcome variab le with a l l other independent variables held constant. A primary contribution o f p a r tia l regression c o e ffic ie n ts 1s 1n correcting the overestimation o f a v a ria b le 's e ffe c ts which can re s u lt from using a b iv a rfa te analysis where the influence o f confounding factors 1s not co n tro lled . veying the obtained Bs, several points are o f in te re s t. In sur­ The b a sically descending value o f the slopes of the screening years (B^ - Bg fo r Dg) re fle c ts h is to ry 's e ffe c t in decreasing re fe rra l rates over the Jan Kmenta, Elements o f Econometrics. (New York: The Macmillan Company, 1971), 365. - 141 years. Also, the negative sign fo r Numscren B indicates the existence o f the predicted Inverse re la tio n s h ip between dependent and independent v a ria b le s . The constant .9542 is the predicted number o f r e fe rr a ls per screening in 1979 given no knowledge o f life tim e screenings. The Betas are the standardized Bs, or "standard p a r tia l regression c o e ffic ie n ts ." They are the Bs converted to comparable terms and in ­ dicate the number o f standard deviation units change in the dependent v a ria b le th a t would be predicted when the independent v a ria b le changes by one standard deviation u n it. They are also known as "beta w eig h ts.11 The Betas do not allow fo r estim ating Y values in the o rig in a l raw value units but they are p a r tic u la r ly helpful when the independent variables are measured In d iffe r e n t units since they allow a common means o f comparing the r e la tiv e e ffe c ts of the d iffe r e n t va ria b les . S p e c ific a lly , in Table V the Betas show th a t the f i r s t three years o f screening, and p a rtic u la rly 1974, have the la rg e s t impact on determining the predicted r e fe r r a l rates per the f u l l model. The standard e rro r o f B gives the standard deviation o f B. Thus, 68 percent o f the Bs are considered to be included w ith in plus or minus one standard d e v ia tio n , assuming a normal d is trib u tio n o f Bs. The most common use o f th is s t a t is t ic is in construction o f confidence In te rv a ls fo r the Bs to thereby estim ate the in te rv a l w ith in which the tru e B is lik e ly to be located. The F fo r each B Is the F te s t s t a t is t ic fo r whether the obtained B value is lik e ly a tru e one. population B * zero. Again, the n u ll hypothesis is th a t the I f the computed F value is la rg e r than the s t a t is ­ t ic a l ta b le 's c r i t i c a l value fo r a given confidence le v e l, the null 142 hypothesis is re je c te d . Otherwise, i t is concluded the observed B is not s ig n ific a n t a t the chosen confidence le v e l. Whether the c a l­ culated Fs are s ig n ific a n t is not Indicated by the report and must be determined by f i r s t id e n tify in g from other reports the number o f subjects belonging to each study group and then fin d in g the appropriate c r i t i c a l value fo r the sample in an F ta b le . Fo rtu nately, since the number o f subjects 1s so large in th is study, th is determ ination is q u ite simple. The c r i t i c a l value fo r equals more than 120 subjects. gsFj, (N -2) 1s 3.84 where N-2 The sample o f each screening year is based upon thousands o f subjects so 3.84 is the c r i t ic a l value fo r the F o f each B. Each F is thus s ig n ific a n t a t the .05 confidence level as each exceeds the c r it ic a l le v e l. I t is lik e ly th a t each B r e fle c ts a tru e re la tio n s h ip between the dependent and Independent v ariab le of In te re s t. Testing was also conducted to determine whether the main e ffe c ts o f screening year and number o f life tim e screenings are each s ig n ifica n t when the other is c o n tro lle d . In both s itu a tio n s , the question 1s whether the Increase in the proportion o f variance accounted fo r by the v a ria b le o f in te re s t is s t a t is t ic a lly s ig n ific a n t when the influence o f the other v a ria b le Is c o n tro lle d . In both cases i t was. The e ffe c ts o f screening year are equal to the R d iffere n c e between the a d d itiv e model and the b iv a rla te model fo r number of screenings. The d ifferen ce (.0497 - .0270) y ie ld s an F o f 258.27, s ig n ific a n t a t any le v e l. The 2 e ffe c ts o f life tim e screenings are equal to the R d iffere n c e between the a d d itiv e model and the m u ltip le regression fo r screening year. d iffere n c e (.0497 - .0467) y ie ld s an F o f 5.935, s ig n ific a n t a t the The 143 .05 level of sig n ifican ce (c alc u la tio n s are placed 1n Appendix G). To demonstrate g ra p h ic ally the Hypothesis 1 rela tio n s h ip by y e a r, the follow ing equations would be used based upon the a d d itiv e model and the Table V data: For 1979, NR ■ A + B^NumScren; For 1978, NR a (A + V + B NumScren; For 1977, NR • (A + »# > + B NumScren; For 1976, NR a (A + b4 For 1975, NR = (A + b3 ) + > + B NumScren; B NumScren; For 1974, NR - (A + «*> + B NumScren; For 1973, NR - (A + Bj > + B NumScren; where: NR * Number o f re fe r a ls . By computation, these equations become For 1979, NR■ .9542 - .0822 NumScren; For 1978, NR- (.9542 + .1235) - .0822 NumScren, ■ 1.0777 - .0822 NumScren; For 1977, NR- (.9542 + * 1.1234 - For 1976, NR» (.9542 + .1692) - .0822 NumScren, .0822 NumScren; .1604) - .0822 NumScren, ■ 1.1146 - .0822 NumScren; For 1975, NR« (.9542 + * 1.2979 - For 1974, NR- (.9542 + * 1.5167 - .3437) - .0822 NumScren, .0822 NumScren; .5625) - .0822 NumScren, .0822 NumScren; 144 For 1973, NR - (.9542 + .6596) - .0822 NumScren, ■ 1.6138 - .0822 NumScren. According to N1e, e t a l . , in using dummy variables a ll categories o f the nominal varia b le can not be entered in the regression equation since doing so would render the computational formulas unsolvable. This occurs because the la s t dumny entered would be completely deter­ mined by the preceding variab les. Therefore, one variab le is not entered In the equation but serves as a reference p o in t, or in th is case, the "base y e a r." This treatment does not re s u lt in a loss of information and the selection of the variab le to use in th is fashion 4 is a rb itr a ry . In th is study, 1979 was used as the base year. Accord­ in g ly , each B c o e ffic ie n t fo r Dj-Dg Is equal to the difference in predicted re fe rra l rates between the In tercep t o f the regression equa­ tio n fo r It s year and the Intercep t of the base year or constant. Thus, to obtain the in tercep t fo r a given y e a r, i t is necessary to add the B value and the constant. The above equations are the best predictors of re fe rra l rates by year given knowledge of the independent variab les. lin e is presented on the follow ing page. The graph of each For each year 1t is apparent th a t the slope of the prediction lin e is negative, I . e . the variables are in an inverse relatio n sh ip as predicted. 4 I b i d , , p. 374. I -4 U* % «-*• 5-S * £ o vP 146 Data fo r the short-term e lig ib le s are s im ila r to Table V. Table V I. M u ltip le Results o f m u ltip le regression method o f analysis of covariance fo r short-term e lig ib le s . R R Square 0.1893 F 0.0358 54.2989 Adjusted R Square 0.0351 Standard 1.0212 Error Variables in the Equation .Variable Di °2 °3 D4 °5 °6 NumScren (Constant) B Std. Error B Beta F 0.7449 0.0902 0.0828 80.913 0.6107 0.1570 0.0430 201.779 0.3364 0.0876 0.0420 63.890 0.1830 0.0544 0.0373 24.055 0.1523 0.0533 0.0323 22.225 0.1284 0.0550 0.0271 22.301 -0.0521 -0.0407 0.0136 14.665 - 0.8414 Again, the F s t a t is tic 1s s ig n ific a n t meaning the hypothesis: R * 2 zero is reje c te d . There 1s reason to believe the R and regression lin e do depict a true re la tio n s h ip . The amount of variance explained by the Independent variables is even s lig h tly less than th at obtained fo r the long-term e lig ib le group. However, the regression lines are o f approx­ imate equal v a lid ity fo r both study groups. Again, re fe rra l rates and number o f past screenings are Inversely related when year o f screening is controlled (denoted by the negative B c o e ffic ie n t fo r NumScren) and 147 the F value fo r each c o e ffic ie n t exceeds the c r it ic a l level in d ic a tin g each obtained B value is s t a t i s t ic a l ly s ig n ific a n t a t the .05 confidence le v e l. When the lin e s are computed by y e a r, as previously show, the follow in g is obtained (th e a d d itiv e model is again preferab le to the in te ra c tio n model as shown 1n Appendix I ) . For 1979, NR - .8414 - .0521 NumScren, For 1978, NR ■ .9698 - .0521 NumScren, For 1977, NR - .9927 - .0521 NumScren, For 1976, NR * 1.0244 - .0521 NumScren, For 1975, NR « 1.1778 - .0521 NumScren, For 1974, NR - 1.4521 - .0521 NumScren, For 1973, NR ■ 1.5863 - .0521 NumScren. The graphs o f these lin e s are shown on the fo llo w in g page. The most obvious impression is again th a t o f the depressing influence on re fe rr a l rates of year o f screening and number o f past screenings. The graph on these lin e s per the In te ra c tio n model is presented in Appendix J , recognizing th a t th is model is considered less v a lid than the a d d itiv e one. In summary, the re s u lts o f m u ltip le regression analysis are b a s ic a lly consistent w ith the Table I and Table I I fin d in g s . R eferral rates and number o f screenings are in an inverse re la tio n s h ip but the regression lin e which best explains th is rela tio n s h ip accounts fo r only about 3.5 to 5 percent o f the outcome. CPA Results Categorical P a rtitio n Analysis (CPA) was conducted on the long-term and shorter-term e lig ib le s . As discussed previou sly, CPA tests fo r .8 .7 Nuiriber of Referrals .6 .5 .4 .3 1973 .2 1974 .1 .0 .9 .8 1975 .7 1976 1977 1978 1979 .5 .3 .2 .1 .05 Number of Screenings Figure I I . Regression lines depicting the relationship between number of referrals and number of life tim e screenings for the one-year elig ib les* by year of last screening. 149 associations between variables and "s p lits " on any variables which show a s t a t is t ic a lly s ig n ific a n t degree o f association. For both groups o f e lig ib le s , CPA produced no " s p lits ” on the independent va ria b les , meaning no s ig n ific a n t association (a t the .03 level of confidence) was found between any of the independent va riab les , sin g u larly or in combination, and the dependent va ria b le s , r e fe rr a l rates.® This to ta l lack o f association is lik e ly best re fle c te d in the R squares o f .0 3 -.0 5 obtained by the m u ltip le regression analyses which analyzed the best predictor variab les. Obviously, these cor­ rela tio n s are considerably short o f a perfect (1 .0 0 ) association, a fa c t w ith which the CPA results would seem consistent. Also, review of the population data, by ta b le format, did not indicate th a t age, sex and race were strong predictors o f outcome. While location d i f ­ ferences in the population do seem considerable, apparently they were not o f s u ffic ie n t strength, or featured too much variance, to be recog­ nized by this te s t. C le a rly , other results did ®The independent variables were: of screenings, age, sex and race. screening year, lo c a tio n , number 150 not show location to be a much stronger va ria b le than number and year o f screenings, which, as noted above, were themselves f a r from p e rfe c t p redictors. Results on Costs Tests o f Hypothesis 2 As discussed, a b iv a ria te regression was used to te s t Hypothesis 2 which predicted medical costs would be inversely re la te d to the num­ ber o f life tim e screenings received. Hypothesis 2 was but an extension o f Hypothesis 1 to the v a ria b le costs w ithout the complication o f the Intervening v a ria b le "year o f screening." founding fa c to r was on r e fe r r a l rates only. The e ffe c t of th is con­ Testing a d iffe r e n t de­ pendent v a ria b le , and a d d itio n a lly using only same-year d ata, meant the s t a t is t ic a l analysis of Hypothesis 2 was more s tra ig h t-fo rw a rd . The re s u lts showed th a t fo r the long-term e lig ib le s , an o ve ra ll F o f 13.246 was obtained which exceeds the c r i t i c a l value of 3.84 fo r 95^1 15949 means s t a in e d r , the zero-order c o rre la tio n between r e fe r r a l rates and number of screenings, is s t a t is t ic a l ly sig ­ n if ic a n t . The obtained r was -0 .0 2 8 8 , in the predicted d ire c tio n , but equal to an r 2 o f only 0.0008. This r 2 is so low th a t the independent v a ria b le , although s t a t i s t i c a lly s ig n ific a n t, provides v ir t u a lly no explanation o f outcome on r e fe r r a l ra te s . Knowledge o f screening h is ­ tory accounts fo r only 8/100ths o f one percent o f the variance in costs. In s h o rt, i t 1s q u ite c e rta in a re la tio n s h ip has been determined which y ie ld s v ir t u a lly no inform ation. Thus, the obtained r square means the re s u lt is not meaningful, or u s e fu l, and accordingly Hypothesis 2 151 is rejected . re s u lt. The te s t fo r the one-year e lig ib le s produces the same The obtained r was a -0.0140. However, the computed F was 3.4530 which did not exceed the c r it ic a l level (again, 3 .8 4 0 ). the hypothesis was rejected . Again, For both groups of e lig ib le s , there was in s u ffic ie n t support in this analysis fo r believing costs are related Inversely to program p a rtic ip a tio n . Student's t-T e s t Although b lv a rla te analysis showed v ir t u a lly no relatio n sh ip between costs and program p a rtic ip a tio n , 1t 1 s possible th at expecta­ tions of an Inverse re la tio n s h ip across fo u r-s ix levels o f p a rtic ip a tio n is too demanding a requirement. Also, review of the obtained data sug­ gested a difference in incurred costs did e x is t between those screened and those not screened. More s p e c ific a lly , those screened showed some­ what lower costs than those not screened. i t s e l f be a finding of in te re s t. I f substantiated, th is would Accordingly, tests o f mean difference were conducted fo r both o f the sampled study groups and fo r the four r a c ia l/e th n ic groups comprising each sample. The hypothesis was: Hq: u 1 * Ug or there is no difference in the population means. The a lte rn a tiv e hypothesis was: Hj: Uj f Ug or there is "probably," but not necessarily, a difference in population means. The significance level equals: .05 or the p ro b a b ility is 5 per 100 or less o f obtaining the mean difference a c tu a lly obtained when, in fa c t, there is no mean difference 1n the population. Decision ru le on the use of variance: I f the tw o -ta ile d p ro b a b ility fo r F is greater than the significance le v e l, use the pooled variance estim ate. 152 I f the tw o -ta ile d p ro b a b ility fo r F is less than, or equal to , the significance g le v e l, use the separate variance estim ate. Decision ru le on hypothesis: I f the tw o -ta ile d p ro b a b ility o f obtaining the t s t a t is tic is greater than .0 5 , H is not rejected . 0 I f the tw o -ta ile d p ro b a b ility o f obtaining the t s t a t is t ic is less than, or equal to , .0 5 , re je c t HQ and accept H j. Since I t would be of in te re s t to determine th a t medical costs of EPSDT p articip ants were e ith e r greater or less than the costs of non EPSOT p a rtic ip a n ts , two-ta11ed tests were used. In each table below, the EPSDT p a rtic ip a n ts are designated as "Group 1;" the non particip an ts (number In of screenings ■ 0) are designated "Group 2." Table V II the tw o -taile d p ro b a b ility fo r F Is less than .0 5 , so the separate variance estimate is used. The tw o -taile d p ro b a b ility of obtaining the t s t a t is tic is equal to .0 5 , so HQ is rejected and accepted. There 1s reason to believe the obtained lower costs of $26.18 fo r EPSDT particip an ts is a true d iffe re n c e , i . e . , th at i t exists In the population from which these subjects were sampled. In Table V I I I , per the decision ru le on use o f variance, the sep­ ara te variance estimate is used. Hq 1s rejected and accepted. Per the decision ru le on the hypothesis, The differen ce o f $46.52 which favors the EPSDT p a rticip an ts 1s highly s ig n ific a n t s t a t is t ic a lly . The proba­ b i l i t y is less than one per hundred th at th is obtained d ifference is not a true one. In Table IX , per the decision ru le s , HQ 1s not rejected . There appears to be no true difference in costs fo r whites In th is population. 6Norman H. Nie, e t a l . , S ta tis tic a l Package fo r the Social Sciences, (New York: McGraw-Hill, 1975), 270. Table V II. Results o f comparison of medical costs fo r long-term medicaid e lig ib le EPSDT participants and nonparticipants. Separate Variance Estimate Variable Number of cases 11210 Mean Standard Deviation Standard Error 260.0205 746.7460 7.0530 T Value Degrees of Freedom 2-Tail Prob. 8576.54 0.050 Medcost Group 1 -1.96 Group 2 4741 286.2078 780.535 11.3360 Table V I I I . Results of comparison of medical costs fo r short-term medicaid e lig ib le EPSDT participants and nonparticipants. Separate Variance Estimate Variable Number of cases Mean Standard Deviation Standard Error 312.0948 1143.1900 11.303 T Value Degrees of Freedom -2.70 13996.61 2-Tail Prob. Medcost Group 1 Group 2 10230 6073 358.6192 1101.350 12.978 0.007 Table IX. Results of comparison o f medical costs fo r long-term e lig ib le white EPSDT Participants and non particip ants. Separate Variable Number of cases Mean Standard Deviation 258.1780 673.3970 Standard Error Variance Estimate T Value Degrees of freedom 2-Tail Prob. -0.90 3875.70 0.370 Medcost Group 1 Group 2 5253 2196 274.2891 720.3180 9.291 15.371 156 In Table X, per the decision ru le s , HQ is not reje c te d . The p ro b a b ility is 26.9% th a t the d iffe re n c e favoring the EPSDT p a r t ic i­ pants is not a true d iffe re n c e . This degree o f e rro r 1s too large to accept. In Table X I, per the decision ru le s , Hq is not re je c te d . There appears to be no true d iffere n c e favoring the EPSDT p a rtic ip a n ts , a l ­ though the t value is q u ite close to the acceptable le ve l o f u n certainty. In Table X I I , per the decision ru le s , HQ is rejected and cepted. ac­ This re s u lt is highly s ig n ific a n t s t a t is t ic a lly and is also q u ite meaningful. The p ro b a b ility is very low, 1.7 percent, th a t the reported, average lower cost fo r black EPSDT p a rtic ip a n ts is not a true d iffe re n c e . The amount o f the average d iffe re n c e , $64.29, is a rath er meaningful fig u r e , e s p e c ia lly since 103,939 blacks are in th is popula­ tio n . In Table X I I I , per the decision ru le s , Hq is rejected and Hj accepted. costs. American-Indian EPSDT p a rtic ip a n ts have higher medical The d iffere n c e is highly s ig n ific a n t s t a t is t i c a ll y and highly meaningful in d o lla r terms as i t approaches $200 per in d iv id u a l. Des­ p ite the te s t fin d in g , one in t u it iv e ly fe e ls some reservation about the fin d in g 's v a lId it y given only nine in d iv id u als comprise the nonp a rtic ip a n t group. The number o f Amerlcan-Indians in th is population is 281. In Table XIV, per the decision ru le s , Hq Is not re je c te d . Again, the American Indian EPSDT p a rtic ip a n ts are associated w ith higher costs but the p ro b a b ility o f e rro r is over 50 percent fo r th is fin d in g so the fin d in g is not accepted. Table X. Results of comparison of medicaid costs fo r short-term medicaid e lig ib le white EPSDT Participants and nonparticipants. Separate Variance Estimate Variable Number of cases Mean Standard Deviation Standard Error 314.9151 1263.200 16.749 T Value Degrees of freedom 2-Tail Prob. - 1.10 8420.50 0.269 Medcost Group 1 Group 2 5688 3100 339.9086 847.399 15.220 Table X I. Results of conparison of medical costs fo r long-term e lig ib le black EPSDT Participants and nonparticipants. Pooled Variance Estimate Variable Number of cases Mean 5575 263.5031 Standard Deviation Standard Error 826.236 11.066 T Value Degrees of Freedom 2-Tail Prob. Medcost Group 1 -1.72 Group 2 2455 298.1984 840.824 16.970 8028 0.085 Table X II. Results of comparison of medical costs fo r short-term medicaid e lig ib le black EPSDT participants and nonparticipants. Separate Variance Estimate Variable Number of cases Mean 4082 313.9926 Standard Deviation Standard Error T Value Degrees Freedom 2-Tail Prob. -2.38 5508.36 0.017 Medcost Group 1 Group 2 2826 378.2897 1010.806 1163.285 15.821 21.883 Table X I I I . Results of comparison o f medical costs fo r long-term medicaid e lig ib le American Indian EPSDT Participants and nonparticipants. Separate Variance Estimate Variable Nimtoer of cases Mean 42 255.0812 Standard Deviation Standard Error 562.119 86.737 T Value Degrees of Freedom 2-Tail Prob. 2.17 43.27 0.035 Medcost Group 1 Group 2 9 63.9900 44.803 14.934 Table XIV. Results of comparison o f medical costs fo r short-term medicaid e lg ib le American Indian EPSDT participants and nonparticipants. Pooled Variance Estimate Variable Number of cases Mean 56 254.7877 Standard Deviation Standard Error 384.179 51.338 T Value Degrees of Freedom 2-Tail Prob. Medcost Group 1 0.66 Group 2 9 167.2422 238.030 79.343 63 0.511 162 In Table XV, per the decision ru le s , HQ 1s not re je c te d . The margin o f e rro r is very high on th is obtained d iffe re n c e , as re fle c te d in the s im ila r means and standard d eviation s. In Table XVI, per the decision ru le s , Hq is not re je c te d . The nearly Id e n tic a l mean costs are re fle c te d in a t value which almost guarantees e rro r I f the mean d iffere n c e Is considered a true one. Medical costs fo r the Spanish-speaking appear to bear no re la tio n to EPSDT program. In assessing the t - t e s t outcomes, several coiments are re le v a n t, p a r tic u la rly concerning the res u lts fo r American Indians and the meaning o f cost outcomes in general. In t u it iv e ly , one might suspect th a t the very d iffe r e n t outcomes fo r American Indians are the re s u lt o f the small number o f n o n p a rtic i­ pants sampled. However, a d iffe r e n t explanation 1s possible which 1s consistent w ith the obtained fin d in g s . meaning as an EPSDT outcome measure. Costs can be ambiguous in C le a rly , low costs are considered desirable from a cost-control perspective. I f a program is determined responsible fo r reducing costs, th is Is viewed favorably 1n most quar­ te rs . However, EPSDT was not only Intended to reduce costs, but also to Increase access to medical services. C e rta in ly , the la t t e r re s u lt can increase costs, a t le a s t in the sho rt-ru n. Given th a t American Indians are perhaps the most excluded ra c ia l m inority in the United S tates, 1 t may be th a t EPSDT serves to Increase access, as Intended, fo r th is group. Thus, higher costs o f p a rtic ip a n ts could be In terp rete d as re fle c tin g re c e ip t o f needed medical treatment and as therefore d e s ira b le . Conversely, the lower costs o f nonparticipants may mean care Table XV. Results of comparison of medical costs fo r long-term medicaid e lig ib le Spanishspeaking EPSDT participants and nonparticipants. Pooled Variance Estimate Variables Number of cases Mean 307 233.3746 Standard Deviation Standard Error 395.983 22.600 T Value Degrees of Freedom 2-Tail Prob. Medcost Group 1 h c c -0.14 Group 2 57 241.2861 337.626 44.720 362 0.888 Table XVI. Results of comparison of medical costs fo r short-term medicaid e lig ib le Spanish­ speaking EPSDT Participants and nonparticipants. Separate Variance Estimate Variables Number of cases Mean 350 264.9599 Standard Deviation Standard Error 555.302 29.682 T Value Degrees of Freedom 2-Tail Prob. Medcost Group 1 0.04 Group 2 83 262.9433 359.953 39.510 186.71 0.967 165 is not being received, even i f needed, and may not be an in d icato r of health status. In general, i t is conceivable th a t low costs fo r EPSDT particip ants may be In d ic a tiv e of good health - with EPSDT the explanatory v aria b le while low costs fo r nonpart1cipants may simply indicate lack o f p a r t ic i­ pation in the health care system, regardless o f need. At th is tim e, the meaning of EPSDT cost differences is subject to in te rp re ta tio n . Since there was some expression when in it ia t in g EPSDT th at the program would re s u lt in reduced medical costs, th is study has been interested in deter­ mining whether th is Is occurring. However, any differences are of in te r ­ e s t, whether of increased, or decreased costs and i t is recognized that obtained differences are subject to in te rp re ta tio n o f meaning. Results o f the t-T ests do show some evidence of EPSDT p a rticip an ts incurring lower medical costs, although program costs must also be considered as is discussed in Chapter V. Results on Costs Tests o f Hypothesis' 3 A b iv a ria te regression was also used to te s t Hypothesis 3 which was b a s ic a lly an extension of Hypothesis 2 to the m atter of short-term costs. The basic question was the same as w ith Hypothesis 2, do costs vary in ­ versely with program p a rtic ip a tio n , but vary 1n the short-run? The findings were consistent with those o f Hypothesis 2. For the long-term e lig lb le s , the computed F was 16.392 which ex­ ceeded the c r it ic a l value o f 3.84 fo r g ,^ 15949 * Since the c r it ic a l value was exceeded, we re je c t the null hypothesis th a t r * zero. r of 0.032 is s t a t is t ic a lly s ig n ific a n t but the r^ was only 0.001, The 166 thereby explaining a miniscule amount of v a ria tio n . The fin d in g 1s not meaningful and consequently Hypothesis 3 is not confirmed. In te r ­ e s tin g ly , the obtained slope of the regression lin e Is in the opposite d irectio n from th a t predicted ( I . e . , the sign of the r is p o s itiv e , ind icatin g costs increase with EPSDT p a rtic ip a tio n ). term e lig ib le s , the re s u lt was the same. which exceeded the c r it ic a l level o f 3 .8 4. equal to an r of 0.00031. For the short­ The computed F was 5.049 The obtained r of 0.0176 is Again, the re s u lt was s t a t is t ic a lly sig ­ n ific a n t, but not meaningful and in the opposite d irec tio n from th at predicted. Accordingly, Hypothesis 3 was not confirmed fo r the short­ term e lig ib le s . For both groups o f e lig ib le s , short-term costs were not shown to bear a meaningful re la tio n to EPSDT p a rtic ip a tio n . CHAPTER V SUMMARY AND CONCLUSIONS The purpose o f th is study was to evaluate selected outcomes a ttrib u ta b le to the Early and Periodic Screening, Diagnosis and Treatment (EPSDT) program in order to b e tte r determine whether EPSDT is b e n e fittin g it s p a rtic ip a n ts in Michigan. was done on two outcome v a ria b les : Evaluation ( 1 ) r e fe r r a l rates (number o f referrals/num ber o f In d iv id u als screened) and (2 ) Medicaid costs. The main In te re s t was 1n whether outcomes varied by de­ gree o f program p a r tic ip a tio n , namely the number o f life tim e screenings, w ith a tte n tio n given also to the Influence o f demo­ graphic v a ria b le s . Review was conducted o f the program's h is to ry , purpose and re la te d lite r a tu r e and several suggestions were made fo r fu tu re study. Summary of the EPSDT Program EPSDT is a Great Society program, le g is la te d In the la te 1960s and surviving to the present. The u ltim a te reason(s) fo r In it ia t in g the program was lik e ly the same reason a l l Great Society programs were In it ia t e d , namely p o litic a l ones. The stated reasons were con­ cerns th a t the poor had a higher Incidence o f health problems than upper Income groups w ith less access to medical resources. I t was believed th a t many o f these problems could be corrected, or improved, i f detected and treated an an e a r lie r s ta te . 167 EPSDT was to address 168 these needs by f i r s t divid ing the e lig ib le population into two groups on the basis of screening te s ts ’ re s u lts . Those f a ilin g the tests are considered most in need of medical resources and accordingly are referred fo r diagnosis and/or treatment services and assisted , where necessary, w ith gaining entry to these services. Those passing the tests are encouraged to be rescreened a t a la te r time. Outreach of e lig ib le s to encourage p a rtic ip a tio n is an im­ portant part o f the program. I t is of course expected th a t these procedures w ill b e n e fit p a rticip an ts by Improving th e ir health status. Despite the fa c t th a t Great Society programs, including EPSDT, had Congressional support and were le g is la te d qu ickly, EPSDT was implemented slowly. Enacted Into law in e a rly 1968, 1t was not u n til 1973-74 th a t most states had programs. Often state programs were in itia te d only as a re s u lt o f legal actio n. Michigan's program began in 1973 and, as Is the case w ith most state s, has continued functioning a t a stable level since th at time. Summary of Screening Program Outcomes The lite r a tu r e on screening outcomes can be divided Into two sections: (1 ) studies o f nonEPSDT programs and (2 ) studies of EPSDT. The nonEPSDT lite r a tu r e consists mainly o f m o rta lity rate studies and studies conducted by Health Maintenance Organizations (HMOs). There Is a consistent trend In the m o rta lity lite r a tu r e which shows screening p articip an ts have lower m o rta lity rates than those not screened. These studies span a h a lf century, from the 1920s to the 1970s. However, 169 researchers were generally aware of methodological lim ita tio n s in these studies and were cautious in in te rp re tin g fin d in g s . The lack o f random selection of subjects was near u n iv e rs a l, as 1 s usually the case in a l l screening programs, w hile i t was gen erally not c le a r whether an e x p lic it linkage had been established between screening and re c e ip t of needed treatment services. C le a rly , screening by i t ­ s e lf can not be cred ited w ith improved outcomes. I t 1s also the case th a t m o rta lity ra te s , w hile im portant, are t o t a lly in s e n s itiv e to any change less dramatic than l i f e or death. The HMO studies show mixed re s u lts . Findings were present which favor the screening p a rtic ip a n ts , but these were not consistent across age and sex. Reasons fo r the v a ria tio n s were not known. However, there was considerable mixing of the study and control groups with a sizab le number of controls a c tu a lly receiving "treatm ents." Thus, d ilu tio n o f the treatm ent e ffe c t may not be unexpected. The fou r EPSDT studies reviewed showed consistent res u lts favoring EPSDT p a rtic ip a n ts . The Community Health Foundation, in a North Dakota study, found th a t the to ta l medical costs of EPSDT p a rtic ip a n ts were lower during the screening year than were the costs o f the EPSDT nonp a rtic ip a n ts .* S im ila rly , Applied Management Sciences, in a comparison o f a northern In d u s tria l and a southern ru ra l s ta te , found EPSDT par­ tic ip a n ts incurred lower costs than nonparticipants In the years before, during and a f t e r screening. 2 In both stu d ies, EPSDT was associated ^Community Health Foundation, "Cost Impact Study o f The North Da­ kota EPSDT Program," (Evanston, I l l i n o i s : Community Health Foundation, 1977). (Mimeogrpahed.) 2 Applied Management Sciences, Assessment o f EPSDT Practices and Costs - Report on the Cost Impact o-f the EPSDT Program (S ilv e r Spring, Maryland: Applied Management Sciences, 1976). 170 with increased re c ip ie n t costs although EPSDT p a rtic ip a n ts , in spite o f the increase, continued to have lower costs than those not p a r t ic i­ pating in the program. C u rrie r's Michigan study found th at 21 percent fewer particip an ts were referred from rescreening than were referred from an i n i t i a l screening . 3 This re s u lt is in the same d ire c tio n as the Philadelphia Health Management Corporation fin d in g th a t a rescreened group had a 30 percent lower re fe rra l rate e ith e r compared with i t s e l f over time or compared with a control group receiving an I n i t i a l screen4 ing only. These EPSDT studies, being current and program s p e c ific , gave the best bases upon which to devise a rath er large scale evaluation of the Michigan program. Summary of Research Design and Methodology A computer based study was designed which form ally tested three hypotheses. {1 ). These hypotheses were: Screenings and re fe rra ls are Inversely related in number, i . e . the average number o f re fe rra ls one incurs is inversely related to the to ta l number of life tim e screenings one has received. (2 ). Medicaid costs are inversely relate d to the to ta l number of life tim e screenings one has received, i . e . cost decline as life tim e screenings Increase. (3 ). Short-run Medicaid costs Increase follow ing screening, are 3 Richard C u rrie r, "Is Early and Periodic Screening Diagnosis and Treatment (EPSDT) Worthwhile?," Public Health Reports, XCII (NovemberDeeember, 1977), 527-36. 4 Philadelphia Health Management Corporation, A Study of the Process. Effectiveness, and Costs o f the EPSDT Program in Southeastern Pennsylvania, Part I I I , (P h ilad elp h ia, Pennsylvania), 1980. 171 g reater (fo llo w ing screening) fo r screened than fo r un­ screened in d ivid u als and are inversely rela te d to the to ta l number of life tim e screenings one has received. Regression analysis was used to te s t a l l three hypotheses and Hypo­ thesis 1 was also analyzed through observation of population d ata, in ­ cluding observation o f demographic e ffe c ts . In a d d itio n , Categorical P a rtitio n Analysis (CPA) was done o f r e fe rr a l rate outcomes by selected demographic variab les and Hypothesis 2 was modified to te s t whether the mean costs o f a l l EPSDT p a rtic ip a n ts were d iffe r e n t from the mean costs o f EPSDT nonparticipants. The study's o v e ra ll design was one o f making observations, s p ec if­ ic a lly determinations o f r e fe rr a l rates and costs, and analyzing these data 1n re la tio n to a h is to ry o f screening p a r tic ip a tio n , I . e . " tr e a t­ ment" e ffe c ts , and selected demographic fa c to rs . Actual subject s e le c tio n , outcome ca lcu latio n s and tests were con­ ducted by computer. Two populations o f EPSDT e lig ib le s were determined. One population consisted of In d iv id u a ls e lig ib le continuously fo r EPSDT between January 1 , 1974 and December 31, 1979. 79,754. This population numbered The other population was of In d ivid u als e lig ib le f o r a t le a s t a l l o f 1979, and numbered 244,551. I t was believed th a t both groups m erited Inclusion 1n the study and a subsequent consideration was whe­ th e r t h e ir outcomes varied . For each population, the computerized EPSDT h is to ry f i l e o f 535,753 screenings was f i r s t searched fo r records. For subjects with a screening h is to ry , screenings were ordered by date o f screening with ca lcu latio n s made o f re fe rra l rates a t the la s t , i . e . most recent 172 screening. These re fe rra l ra te s , grouped by number of life tim e screenings, were used to te s t Hypothesis 1 through d ire c t obser­ vation of outcomes. Also, a systematic, random sample was taken from each of the two population groups. The sample of continuously e lig ib le recip ­ ients totaled 15,951 (20% o f the population) while the sample of 1979 e lig ib le s to taled 16,303 (6.67% o f the population). Analysis of covariance and Categorical P a rtitio n Analysis were conducted on those screened In both samples to determine the re la tio n s h ip , 1f any, between re fe rra l rates and various Independent variab les. Medical cost data were obtained fo r both groups (including those subjects never screened) and were analyzed by the b iv a rla te method o f regression analysis to te s t Hypothesis 2 and 3 as well as by Student's t - t e s t fo r mean differences 1n costs between a ll EPSDT p articip an ts and non­ p a rtic ip a n ts . In determining re fe rra l ra te s , immunization re fe rra ls were not considered since reporting was not consistent fo r th is Item. Also, members o f Health Maintenance Organizations (HMOs) were not included In the study since th e ir medical costs are not placed on computer file . The number of subjects studied in th is research 1s noteworthy. To te s t Hypothesis 1, by d ire c t observation, re fe rra l rates were analyzed fo r a combined population of 210,233 subjects. This 1s by f a r the larg e s t group upon which an EPSDT outcome evaluation has been conducted. S im ila rly , the combined sample fo r the two study groups used 1n the cost analysis to taled 32,254 subjects, again the largest 173 sample used in studying the program. The weight o f these numbers contributes to the c r e d ib ilit y of the fin d in g s . Summary o f Results The most important findings of the study are as follow s: ( 1 ). te g ie s . Hypothesis 1 was confirmed by two d iffe r e n t te s tin g s tra ­ D ire c t observation o f population data showed th a t re fe rra l rates were inversely re la te d to the number of life tim e screenings received. When te s t groups consisted o f 100 or more subjects each, group r e fe rr a l rates generally decreased across the f i r s t fo u r -fiv e screenings. More s p e c ific a lly , d ire c t observation o f the population data showed, w ith h isto ry c o n tro lle d , th a t the r e fe rr a l ra te fo r those receiving a second screening was approximately 10 percent less than the ra te received by those i n i t i a l l y screened [12% fo r the long-term e lig ib le s , per Table I (S) and 9 % fo r the short-term e lig ib le s , per Table I I ( S ) ] . S im ila rly , the ra te decreased in moving from the second to th ird screening (a 12% average decrease in Table I(S ) and a 5.6% decrease per Table I I ( S ) and decreased again from the th ird to fo u rth screening [6.25% in Table I(S ) and 3.75% in Table I I ( S ) ] . In sh o rt, the inverse re la tio n s h ip was con­ s is te n tly present but showed some d ilu tio n o f strength as rescreenings Increased. Hypothesis 1 was also confirmed by the m u ltip le regression method o f analysis o f covariance. When year of screening was tre ated as var­ ia b le , in addition to the number o f screenings received, the regression lin e formed to te s t Hypothesis 1 accounted fo r 4.97 percent o f the var­ iance in r e fe rr a l rates fo r the long-term e lig ib le s and 3.58 percent 174 fo r the short-term e lig ib le s . Both results were s t a t is t ic a lly sig­ n ific a n t a t the .05 confidence le v e l. In gen eralizin g , i t could be said th a t the regression lin es accounted very l i t t l e of the variance in outcome. The results of both testin g strateg ies are consistent across both subject groups as well as consistent with each other. The results were also in the same d ire c tio n , although o f a lesser magnitude o f e f­ fe c t , as other c ited EPSDT studies. is open to differences o f view. The "meaningfulness" o f the findings The 10 percent, and less, reductions in r e fe r r a ls , were strong in consistency but "modest" 1n e ffe c t, assuming one considers "treatments" with 10 percent reductive power to be modest. S im ila rly , the regression lin e s , in explaining 3.5 to 5 percent o f r e f fe r r a l rate variance, seem to Indicate th at (re)screenings have some, a l ­ b e it very lim ite d , power to reduce the incidence o f re fe rra b le problems. The difference between .05 and perfect (1 .0 0 ) explanation is considerable, indicating variance In outcomes and the e ffe c ts of other v a ria b le (s ) in determining re fe rra l rates. Number of screenings appears lim ite d in power to predict re fe rra l rates although, over large numbers o f subjects, its consistent association with reduced re fe rra ls is shown. ( 2 ). Hypothesis 2 was not confirmed fo r e ith e r subject group. The obtained re s u lt did not support the notion th a t medical costs are in ­ versely rela te d to EPSDT screenings. This hypothesized relatio n s h ip was found to be s t a t is t ic a lly s ig n ific a n t fo r the long-term e lig ib le group but explained only .0008 of the variance 1n costs. I t was there­ fore concluded th at th is "explanation" of relatio n sh ip provided no mean­ ingful information and the hypothesis was rejected accordingly. The 175 hypothesis was not s t a t is t ic a lly s ig n ific a n t fo r the short-term e lig ib le group. (3 ). Tests of the mean cost differences between EPSDT p a r t ic i­ pants and EPSDT nonparticipants produced the follow ing res u lts: (A ). The mean cost of a ll long-term e lig ib le particip an ts was found to be $26.18 less than the mean costs of long-term e lig ib le nonp a rtic ip a n ts . This re s u lt was s t a t is t ic a lly s ig n ific a n t a t the .05 level o f confidence. (B ). The mean cost of a ll short-term e lig ib le particip an ts was found to be $46.52 less than the mean costs of short-term e lig ib le non­ p a rtic ip a n ts . This re s u lt was s t a t is t ic a lly s ig n ific a n t a t the .007 level o f confidence. (C ). The mean cost o f short-term e lig ib le black p a rticip an ts was $64.29 less than the mean cost of black, nonparticipants e lig ib le fo r the same time period. This re s u lt was s t a t is t ic a lly s ig n ific a n t a t the .017 confidence le v e l. For the long-term e lig ib le s , black p a r t ic i­ pants showed mean costs of $34.69 less than the comparable costs of black nonparticipants. This re s u lt was s t a t is t ic a lly s ig n ific a n t a t the .085 level o f confidence. (D ). The mean cost of long-term e lig ib le American Indian par­ tic ip a n ts was $191.09 greater than the mean cost o f American Indian nonparticipants e lig ib le fo r the same time period. This re s u lt was s t a t is t ic a lly s ig n ific a n t a t the .035 level o f confidence. However, th is re s u lt did not hold fo r the short-term e lig ib le American Indians where the EPSDT p articip an ts had lower costs although a t a confidence level fa r exceeding the .05 le v e l. Since fo r both e lig ib le groups 176 only nine subjects comprised the group o f nonEPSDT p a rtic ip a n ts , confidence was not placed 1n these fin d in g s, despite the s ta t is tic a l significance o f obtained re s u lts . (E ). Mean cost differences between EPSDT particip an ts and non­ p articip an ts were not found to e x is t a t a s t a t is t ic a lly s ig n ific a n t level fo r whites or the Spanish-speaking. There are several considerations in assessing the meaning of the obtained mean differences. a fin a n c ia l cost. For one, screening i t s e l f involves During 1979, the year fo r which the above cost differences were obtained, the determined average cost o f an EPSDT screening In Michigan was $65.56. This cost o f course must be con­ sidered In estim ating any cost savings which might be a ttrib u te d to program p a rtic ip a tio n . Since the rescreening cycle is basically every other year In Michigan, the annual screening cost is reduced by 50% when adding th is cost to the p a rtic ip a n ts ' yearly medical expenses. $32.78. (A). The annual screening cost is thereby determined to be This adjustment yield s the follow ing fo r 1979 data: For the long-term e lig ib le s , there are no savings asso­ ciated w ith EPSDT p a rtic ip a tio n , given consideration o f screening costs ($26.18 • $32.78 * -$ 6 .6 0 ). In fa c t, as shown, medical costs of EPSDT p a rticip an ts are greater than costs fo r nonparticipants. Since 56,046 o f the long-term e lig ib le population had been screened, increased costs associated w ith EPSDT p a rtic ip a tio n equal $369,903.60 fo r th is group ($6.60 x 56,046). 5 This fig u re 1s obtained by dividing determined program expenses o f nearly $6.9 m illio n by the to ta l number o f Individuals screened, 105,239. Data are obtained from the Michigan program s ta tis tic s . 177 (8 ). For the short-term e llg ib le s , the fin a n c ia l savings asso­ ciated w ith EPSDT p a rtic ip a tio n were $13.74 per person ($46.52 $32.78). Since there are 154,187 individuals in th is population who have been screened, group savings equal $2,118,529.38 ($13.74 x 154,187). Since the long-term and short-term e lig ib le populations are mutually exclusive groups, the costs o f the former and savings of the la t t e r could be aggregated y ie ld in g a to ta l cost savings of $1,748,625.78 ($2,118,529.38 - $369,903.60). I t is noted th a t these savings are based upon the determined costs o f conducting the EPSDT program. These costs Include a ll c lin ic a l costs o f screening and outreach costs fo r Wayne County and approximately ten additional outreach workers in the outstate area. The determined costs do not include outreach costs fo r most counties in the s ta te , any outstate transportation costs, or rela te d adm inistrative expenses. These la t t e r cost figures are not a v a ila b le , but, as shown, would need to to ta l approximately $1.75 m illio n to cancel the savings ascribed to the short and long-term e lig ib le groups. In f a c t , i t is lik e ly program costs do exceed savings. Based upon personal knowledge, an estimate o f 75 outreach workers in the outstate area a t $15,000 per worker would not be excessive. would equal approximately $1,125 m illio n . These to ta l costs The Medicaid transportation budget Includes funds fo r EPSDT transportation and fo r medically rela te d transportation fo r recip ients o f other Department of Social Services programs. 1979. This budget equaled approximately $2 m illio n fo r Assuming EPSDT transportation accounts fo r 50% o f th is 178 expenditure* another $1 m illio n in costs is added to the program. This additional cost* and related adm inistrative expenses* would erase any program savings and, in fa c t, would render an overall cost o f o f perhaps $-375—$1 m illio n associated with the program [$1.75 m illio n savings - ($1,125 m illio n outreach costs + $1 m illio n transportation costs + adm inistrative c o s ts )]. There is adm ittedly some estim ation Involved in a rriv in g at th is fig u re . However* assess­ ments of program costs seldom consider a ll costs involved* which are real costs despite d if f ic u lt ie s in determ ination, and the overall impression In incorporating these costs is th a t they lik e ly a t least balance obtained savings a ttrib u ta b le to the program. (4 ). Location differences in re fe rra l rates were found to e x is t between D e tro it and r u r a l, outstate residents. For the long-term e lig ib le s * D etro iters average 74 percent more re fe rra ls than the out­ state residents when a l l screenings are considered (1.352 versus .768). For the short-term e lig ib le s , the D e tro it r e fe rr a l ra te Is 64 percent higher than the-ou tstate rate (1.202 versus .7 3 2 ). These differences are not due to the d iffe re n t ra c ia l compositions of the two geographic areas. For the short-term e lig ib le s , D e tro it whites average 44 percent more re fe rra ls over a l l screenings than rural whites and D e tro it blacks average 83 percent more re fe rra ls over a ll screenings than the rural blacks. Location 1s strongly Influencing re fe rra ls although race also appears to be a fa c to r as evidenced by the above figures where the black, urban-rural difference is nearly twice the w hite, urban-rural comparison (83% versus 44%). 179 Two explanations seem most plau sib le fo r explaining the location d iffere n c e s . residents. For one, rural residents may be more healthy than urban This study has accepted r e fe rr a l rates as re fle c tio n s o f health status and has assumed th a t r e fe r r a l standards are applied equally throughout the s ta te . A ll Michigan EPSDT c lin ic s use the same b a tte ry o f te s ts , w ith the same w ritte n instru ctio ns and r e fe r r a l c r i ­ t e r ia . Consistent w ith th is perspective, 1t would be concluded th a t the urban-rural d ifferences are re fle c tio n s o f d iffe r in g health status. However, a second explanation should be considered. I t is lik e ly th a t treatm ent providers are more a v a ila b le in the urban area, and because o f t h is , 1t is possible th a t more re fe rr a ls are made in th a t s e ttin g . A c lin ic does not wish to Id e n tify a re fe rra b le condition fo r which they can not locate a needed provider. fo r the c lin ic and the fam ily in need o f service. This is fru s tra tin g Thus, i t is possible th a t the supply of providers wields an influence upon r e fe r r a ls . At le a s t in those s itu a tio n s where r e fe r r a l need is m arginal, provider supply may influence whether a r e fe r r a l 1s made. Provider a v a ila b ilit y may also explain the urban-rural d ifferen ce in black r e fe rr a l ra te s . The r e fe r r a l ra te fo r ru ral blacks is about h a lf th a t o f D e tro it blacks. I f providers are generally in shorter supply in ru ra l areas, i t may be they are p a r tic u la r ly lim ite d fo r blacks. Conversely, more services in urban areas, such as more developed public health departments, more special projects and c lin ic s , e t c . , may f a c i l ­ it a t e black r e fe rr a ls in the c it y . I t is not reasonable to believe the obtained urban-rural d iffere n c e in black r e fe rr a ls is an accurate re ­ fle c tio n o f d iffe r in g health status. 180 ( 5 ). Outcomes fo r blacks varied from those o f whites. Blacks had higher re fe rra l rates than whites, in d ic a tiv e o f more problematic health status. This fin d in g is consistent in d irec tio n w ith l i t e r a ­ ture showing blacks to have higher m o rta lity rates than whites. Over a ll screenings, blacks averaged 20 to 23 percent more re fe rra ls than whites (20 percent more fo r shorter-term e lig ib le s per the Table I series and 23 percent more fo r the long-term e lig ib le s per the Table I s e rie s ). In comparing costs, w hite, EPSDT p a rtic ip an ts and no n p artici­ pants did not show s ig n ific a n t differences. However, black particip ants showed lower costs than nonparticlpants a t levels o f s t a tis tic a l s ig n if­ icance 1n which some confidence can be placed. As discussed above, fo r the short-term e lig ib le black p a rtic ip a n ts , the differen ce was $64.29 per in d iv id u a l; fo r the long-term e lig ib le p a rticip an ts the difference was $34.69 per in d iv id u a l. These differences were s t a t is t ic a lly s ig n if­ icant a t the .017 and .085 levels resp ectively. White p a rtic ip a n ts did show lower costs than white nonparticlpants but a t levels o f s ta tis tic a l significance in which l i t t l e confidence can be placed (27-37 percent chance of e r r o r ). The Im plication o f these findings is th a t blacks have r e la tiv e ly more need to p a rtic ip a te 1n the program than other ra c ia l groups, as evidenced by th e ir higher re fe rra l rates and, once p a rtic ip a tin g , Incur lower costs than other blacks who do not p a rtic ip a te . (6 ). matured. Referral rates have declined by year as the program has This statement holds consistently when a ll re fe rra ls are con­ sidered and holds generally when the number o f previous screenings is held constant. The decline has been large. For both the long-term and 181 short-term e lig ib le s , the re fe rra l rate decreased 59 percent in the seven year period 1973-80. year. This is a reduction o f over 8 percent per I t 1s not reasonable to believe health status has improved a t th at ra te . Again, as provider supply may d ic ta te urban-rural differences in re fe rra l ra te s , screening c lin ic s may have accommodated over time to the a v a ila b ilit y o f treatment providers and/or may have refin ed re fe rra l c r it e r ia which, In 1 t9 e lf, may r e f le c t the influence o f the treatment provider. Screening c lin ic s tend to develop a pool of providers who are agreeable to accepting th e ir r e fe rr a ls . In fa c t, c lin ic s are dependent on these providers i f needed treatment services are to be obtained and the program is to function as Intended. In th is s itu a tio n , c lin ic s are undoubtedly amenable to the providers' suggestions regarding appropriateness o f r e fe rr a ls . I f these providers inform c lin ic s th at ce rta in levels o f problems do not need th e ir a tte n tio n , c lin ic s w ill un­ doubtedly heed th e ir counsel. That non-medical reasons do influence re fe rra l rates is indicated by a recent a r t ic le which discussed the a p p lic a b ility to Michigan of national standards 1n growth rates . In Michigan, 25 percent o f those screened under age two are below the national standard and, according to those standards, should be re fe rre d . However, the authors concluded th a t "economics and lim ite d resources force us to assess the wisdom o f re fe rrin g 25% o f those under age 2 to the medical care system, based on only one screening test."** They suggested i t may be necessary to base ^Horner A. Sprague, e t a l . , "Comparison o f EPSDT and NCHS Growth Charts," Preventive Medicine, IX (1980), 406. 182 the growth chart re fe rra l c r it e r ia on a volume o f re fe rra ls which is "economically acceptable." The point is th a t determination of medical need is not based solely on the p a tie n t's condition but also includes consideration of medical resources. ( 7 ). Few meaningful differences in outcomes were found between long-term and short-term e lig ib le s . I t had been anticipated th a t long­ term e lig ib le s would most lik e ly be the ones to show maximum program benefits but th is did not prove to be the case. In general, th is d iv is ­ ion o f the e lig ib le population did not produce new or d iffe r e n t in fo r­ mation and appears to have been an unnecessary d is tin c tio n . An excep­ tion to the s im ila r ity of findings was the difference between long and short-term e lig ib le costs fo r EPSDT p articip ants and nonparticl pants. However, i t does not seem s u ffic ie n tly u se fu l, or necessary, fo r any future study to continue th is dichotomy o f e lig ib le s . V a lid ity of Referral Rates as Indicators of Health Status Given the prominent use of re fe rra l rates 1n th is study, some com­ ment is needed on th e ir performance as v a lid indicators of health status: Since a l l Michigan screenings are conducted by local public health depart­ ments, or th e ir designees, operating with id en tical re fe rra l c r i t e r i a , instructions and tra in in g , there is basis fo r assuming th a t re fe rra b le conditions are determined and processed uniformly throughout the state as well as across d iffe r in g sexes and ra c ia l groups. Also, the purpose of EPSDT, to which a ll screening teams are professionally committed, is to fin d problems and ass is t fa m ilie s in obtaining needed help. Referral rates a re , by d e fin itio n , d ire c t measures of suspected health problems "found" during screening. To believe re fe rra l rates are in v a lid , or 183 even im perfect, indicators of health status is to question the basic in te g r ity o f the program. While th is study does not show re fe rra l rates to be in v a lid In d ic a to rs , and indeed they seem to be generally le g itim a te measures of h e a lth , the study does present findings seemingly in d icatin g th a t r e fe rr a l rates are influenced by non health fa c to rs . For example, r e fe rr a l rates decreased over tim e, even when program par­ tic ip a tio n was held constant. Also, r e fe rr a l rates showed some urban- ru ral differences which seemed too large to be a ttrib u te d to only con­ d itio n s o f h e a lth , and s im ila r ly showed rural blacks to be su rp risin g ly h e a lth ie r than e ith e r rural whites or urban blacks. In s h o rt, re fe rra l rates seem to be imperfect but generally v a lid and useful measures of h e a lth . The study does not provide a basis fo r much more precise asses­ sment of th is in d ic a to r's v a lid it y but i t does not.support an a lte rn a tiv e explanation o f fin d in g s , namely that observed ra c ia l group health d i f ­ ferences are a c tu a lly differences of compliance w ith norms o f personal health care. More s p e c ific a lly , the compliance argument is th a t some groups are more compliant and conscientious than others about keeping r e fe r r a l ap­ pointments and thereby obtaining needed medical care. Therefore, when they return fo r rescreenings, the incidence o f t h e ir problems is decreased, as re fle c te d 1n reduced re fe rra l ra te s . Groups not so compliant w ill return fo r rescreening w ith the o rig in a l problems uncorrected, thereby increasing th e ir subsequent re fe rra l rate s. recounted as rescreenings occur. Th eir problems are simply Thus, the argument is th a t the measure­ ment o f health status through re fe rra l rates Is a c tu a lly a measure of compliance with seeking recommended medical a tte n tio n . 184 There are several problems with the compliance argument. For one, i t is not in te rn a lly consistent as i t assumes those not given to attending to recommended re fe rra l needs w ill be conscientious about returning fo r rescreenings. I t seems more lik e ly th a t lack of atten tio n to one's own medical needs would be consistent in both s itu a tio n s , meaning those lea st compliant, and thus le a s t healthy, would drop out o f the rescreening process. Secondly, the data are not consistent with the compliance thesis. When reporting r e fe r r a ls , c lin ic s t a f f do not make a d is tin c tio n between i n i t i a l and repeat re fe rra ls fo r the same problem(s). I t is not believed that the incidence of the la t t e r is la rg e , but d ire c t analysis o f I n i t i a l re fe rra ls only was not possible in th is study. However, the thesis would seemingly argue th a t the most compliant groups, the groups with the lowest re fe rra l rates (best h e a lth ), would show the larg es t decrease in re fe rra ls between the i n i t i a l screening and rescreenings. The large decrease would indicate compliance w ith attending to recommended re fe rra ls which would then be re fle c te d in lower re fe rra l rates a t subsequent screenings. However, the data do not show th is pattern . For the long-term e lig ib le s , the re fe rra l rate over a l l screenings was as follow s: (see page 7 9 ). .817, Spanish-Speaking; .898, whites; and 1.107, blacks Consistent with the compliance th e sis , we would expect these groups to be in d e n tic a lly ordered in regard to decreases in re fe rra ls from the i n i t i a l screening to rescreenings. as follow s: The actual arrangement was Spanish-Speaking, -32 percent (1.054 a t I n i t i a l screening; .720 a t rescreenings); blacks, -31 percent (1.330 - .920) and whites, -25 percent (1.039 - .7 7 8 ). The position of the Spanish-Speaking is consistent with the compliance thesis but the position of blacks and whites is r e ­ versed from th a t which the thesis would p red ict. In terms of re fe rra l 185 decreases, whites appear the le a s t compliant group, which is not what the compliance thesis would p re d ic t. For the short-term e lig ib le s , the o verall r e fe rr a l rates were: Spanish-Speaking, .815; American Indians, .821; w hites, .850 and blacks, 1.022 (see page 104). Again, the compliance thesis would expect an id e n tic a l ordering in terms o f r e fe rr a l decreases. Actual decreases in r e fe rr a ls from the i n i t i a l screening to rescreenings were ordered as follow s: blacks, -2 3 .5 percent (1.146 - .8 7 6 ); Spanish-Speaking, -20 percent (.9 2 3 - .7 3 6 ); American Indians, -19 percent (.8 9 8 - .729) and w hites, -16 percent (.918 - .7 6 9 ). Here, blacks are a t the top o f the ranking whereas the compliance thesis would place them la s t. Blacks show the biggest decrease in r e fe rr a ls between i n i t i a l and rescreenings which the compliance thesis would say indicates the most compliance w ith societal expectations of appropriate behavior. However, the compliance thesis would also seemingly p re d ic t blacks to be le a s t com pliant, or a t le a s t less compliant than whites. Also, the black combination o f la rg e s t r e fe rr a l decreases and la rg e s t r e fe rr a l rates is not in te r n a lly consistent with the compliance argument since i t means blacks are the most compliant and le a s t healthy group. In summary, the compliance thesis finds no support In the data fo r explaining black-w hite d iffe re n c e s . A method o f o p e ra tio n a lizin g the compliance thes1s-use o f r e fe rr a l decreases-shows whites to be less com­ p lia n t, but more h ealth y, than blacks, an outcome exactly opposite the compliance thesis p re d ic tio n . The thesis is obviously flaw ed, i f not in v a lid a te d , given It s apparent in a b ilit y to account fo r the outcomes of by f a r the la rg e s t ra c ia l groups. I t is acknowledged th a t, excluding blacks, the compliance thesis 186 does show a consistent and predicted ordering between re fe rra l rates and re fe rra l ra te decreases fo r the other groups. For example, i t is consistent with in te rp re tin g Spanish-Speaking and white outcomes. However, l i t t l e credence should be placed in regarding the Spanish-Speaking as the most healthy group by v irtu e o f th e ir lower re fe rra l rates over a ll screenings. For one, these differences with the white rates are not la rg e . For the long-term e lig ib le s , the Spanish-Speaking showed 9 percent less re fe rra ls than whites (.817 vs .898) while the difference was but 4 percent fo r the short-term e lig ib le s (.815 vs .8 5 0 ). Secondly, these d iffere n c es , which emerge when using overall re fe rra l ra te s , are lik e ly simply due to a la rg e r percentage of rescreenings being represented in the Spanish-Speaking group than in the white group. From Tables I(A a) and I(D a ), i t can be calculated th a t 71 percent of a l l Spanish-Speaking screenings were rescreenings as compared with 54 percent fo r whites. Since fewer problems are found a t rescreenings, the overall Spanish-Speaking rate is lower than the white ra te . With blacks [Table I(B a )] , 54 percent of a ll screenings were also rescreenings so black-white differences can not be explained as a r tifa c ts o f the numbers involved. Black-white comparisons in general appear ju s t ifie d in th is study since both involve such la rg e , and s im ila r, numbers o f subjects. However, since the numbers of Spanish-Speaking and American Indians are so much smaller than blacks and w hites, l i t t l e con­ fidence is placed 1n comparing outcomes fo r these m in o rities with e ith e r whites or blacks. 187 Im p lic a tio n s o f Study fo r EPSDT and Social Work I t is not a n tic ip a te d th a t th is study w ill have far-rea ch in g e ffe c ts on social work or social programming in gen eral. It s focus was not s u ffic ie n tly broad and it s re s u lts were not th a t dram atic, being n e ith e r extremely supportive or detrim ental to the program. should be more s u b s ta n tia l. The impact on EPSDT However, any study, occurring as i t does w ith in a s p e c ific social and h is to ric a l context, w ill have a p a rtic u la r meaning given the concerns o f it s day. Thus, i t may be useful to in te rp re t th is study's findings and p o te n tia l im plications r e la tiv e to the context o f 1981. As noted, EPSDT is a creation of the lib e r a l, w elfare state o f the 1960s. Governmental concern then was to improve social and educational services to the lower classes, lik e ly fo r the purpose o f strengthening the alleg ian ce o f the poor to the la rg e r society o r , more s p e c ific a lly , to the Democratic p a rty. The program was in it ia te d and advocated from the top down-by HEW, an arm of the federal government's w elfare s tate bureaucracy. I t was not s p e c ific a lly demanded from the state or "grass-roots" le v e l, although the 1960s re b e llio n of the blacks was governmentally in terp reted as a demand fo r s h iftin g more o f s o c ie ty 's ben efits in t h e ir d ire c tio n . In 1981, and fo r the foreseeable fu tu re , the concerns o f the federal government, and lik e ly the society a t large a ls o , are q u ite d iffe r e n t. A major goal o f the 1981 federal government is to decrease the costs of p u b lic , social programs. The overriding concern is governmental, short­ term expenditures and th e ir reduction. Receipt o f public services is now viewed negatively since i t r e fle c ts public costs. The Republican govern­ ment's constituency in 1981 is d iffe r e n t from th a t o f the Democratic 138 government in the mid 1960s. To consolidate n a tio n a l, p o litic a l power in 1981 means some dismantling o f the social welfare state which more lib e ra l * governments erected previously to bu ild th e ir own power. Within this context, what then is the meaning of th is study? F ir s t, as noted, the major conclusion was not s ta r tlin g or extrememodest gains fo r modest costs. fin d in g . This would not seem to be a controversial There Is basis in the study fo r d iffe rin g conclusions regarding the program's worth. Some aspects o f the study are supportive o f the program; others are not so p o s itiv e . These pros and cons would seem to have a potential fo r balancing one another. However, in the context of 1981 and to the current national government (perhaps state governments as w e ll) , the "negatives" w ill lik e ly overshadow the supportive findings. S p e c ific a lly , the finding th at the program appears to incur a true cost, over and above savings a ttrib u ta b le to the program, has negative im plications fo r the program. This finding 1s p o te n tia lly balanced by the finding th a t p articip ants appear to be b e n e fittln g from the program. However, i t is lik e ly th at rec ip ie n t benefits w ill not a t a ll be viewed as counter balancing increased costs. I f a program costs in 1981, th is is an important and destructive consideration in judging It s worth. This finding is p a rtic u la rly Important in re la tio n to EPSDT's emphasis on prevention and outreach. The unique, outreach c h a ra c te ris tic of EPSDT is now an Inherent l i a b i l i t y fo r the program. At a time when the focus 1s on reducing program costs, a program which a c tiv e ly encourages the use of Medicaid services w ill not be viewed favorably. Even the fin din g th a t re fe rra l rates are higher fo r blacks than fo r w hites, a fte r continued program p a rtic ip a tio n , can be interpreted negatively. The im p lic a tio n , in 1981, could be th at the program is fa ilin g 189 to elim in ate ra c ia l differences in health status. Social programs which " f a il" are subject to increased s c ru tin y , funding reductions, perhaps even e lim in a tio n . In sh o rt, social program advocates are i n i t i a l l y in a defensive position in 1981. Any studies which do not y ie ld strong ju s t if ic a t io n fo r a program-and ju s t if ic a t io n clo sely tie d to considerations of c o s t-c a ll the program’ s worth in to question. This discussion is not Intended to imply th a t EPSDT, e s p e c ia lly as depicted by th is study, is a f a ilu r e . From the perspective o f program advocates, the program is making gains, a t modest costs, and may even show g reater gains over a longer time period. The point is simply th a t, judged by the governing powers o f 1981, the rath er modest program support shown by th is study may well be a l i a b i l i t y ra th e r than a p o s itiv e or even neutral fin d in g . This is also not a t a l l to say th a t program advocates should place themselves r e fle x iv e ly 1n the position of supporting programs, p a rtic u la rly programs which prove in e ffe c tiv e . Program advocates, Indeed the social work profession should vigorously s c ru tin ize it s own works and make changes where needed. C r e d ib ility and v i t a l i t y w ill only flow from a c r i t i c a l , responsive and innovative handling o f social programming. U ltim a te ly , the most destru ctive position any profession can take is to merely defend it s vested in te re s t. Several concerns o f relevance to social work are raised by th is study. The most prominent are as follow s: 1. Even given adequate evaluations o f social programs, 1t is not evident how to choose among those programs fo r purposes o f d is trib u tin g the dim inishing, and lik e ly in s u ffic ie n t, resources now a llo cated to the social work sector. At the extremes, the decisions may not be d i f f i c u l t . 190 But, most social programs show mixed and ambiguous findings r e la tiv e to e ffic a c y and in such s itu a tio n s choosing among programs is problem atic. For example, given the mixed re s u lts o f th is study, how does one decide whether funds are best spent on EPSDT or on some other s p e c ific social or health program? Such questions are not merely academic. The federal government i n i t i a t i v e of placing health and social programs in to block grants, w ith funds In s u ffic ie n t to finance a l l the programs, w ill force states in to making exactly these types o f decisions. c e rta in ly not new. Such choices are However, they are going to be more freq u en t, and o f greater magnitude, in the near fu tu re than they were in the recent past. The w elfare s ta te of the 1960s was expansionary. The fis c a l crises o f the la te 1970s could gen erally be handled by across-the-board reductions. 1980s w ill see the e lim in a tio n of e n tire programs. The Social work w ill be fo rtu n ate i f I t is involved in such decisions (a t the same time i t has a ro le to play in re s is tin g such changes). Accordingly, i t w ill need a methodology fo r making these choices. 2. The combination o f findings th a t p a rtic ip a n ts not only b e n efitted from EPSDT (as evidenced by reduced r e fe rr a l ra te s ) but also incurred lower medical costs than non-participants (program costs excluded) Is very encouraging. The fa c t th a t these savings were o ffs e t by program costs raises a challenge to social work to reduce program costs-w hile m aintaining q u a lity programming-in order to r e a liz e o ve ra ll Medicaid cost savings a ttr ib u ta b le to the program. A combination o f p a rtic ip a n t ben efits and cost savings would provide the strongest possible ra tio n a le fo r the program. The p o s s ib ilitie s of strong program ju s t if ic a t io n are present in the fin d in g s . The task is s t i l l to r e a liz e the p o te n tia l. 191 3. The fa c t th a t EPSDT was governmentally in it ia te d - w ith strong social work backing but without c le a r grassroots support - challenges the profession to appraise the program o b je c tiv e ly and in consideration o f its support by i t s e lig ib le population. The most basic p rin c ip le o f social work's community organization is to involve rec ip ien ts in the decisions and programs a ffe c tin g t h e ir liv e s . The extent to which Medicaid rec ip ie n ts support EPSDT must be considered 1n combination w ith measures of the program's e ffic a c y . The e a rly h is to ry o f EPSDT is unclear regarding the extent of re c ip ie n t in te re s t in the program. To the extent those in social work posi­ tio ns advocated fo r the program without a base of p o te n tia l re c ip ie n t support, they were pro fessio n ally inconsistent in th e ir lack of a tte n tio n to re c ip ie n t involvement. To the extent th is omission continues, the profession runs the ris k o f in s titu tio n a liz in g the program to it s own purposes and image. Recommendations fo r Future Study Future EPSDT research should focus on: ( 1 ). The re la tio n s h ip between r e fe r r a l rates and rescreening. (2 ). The re la tio n s h ip between program p a rtic ip a tio n and medical costs. ( 3 ). The question o f whether EPSDT Improves access to needed medical services. A central purpose o f EPSDT was to increase the poor's access to "mainstream" medicine. L i t t l e a tte n tio n has been given to whether th is is occurring. (4 ) The d if f e r e n t ia l effectiven ess o f the various screening tests and procedures and the need fo r d ele tin g and/or adding tests to the screening package. 192 Conclusions The ob jective of th is study was to b e tte r answer the question of whether EPSDT in Michigan is improving the health status of its p a rtic ip a n ts . Analysis of r e fe rra l rates indicates the program is having b en eficial e ffe c ts as evidenced by the existence of an Inverse relatio n sh ip between re fe rra l rates and rescreenings. more screenings have fewer r e fe rr a ls . Those with There is variance in th is trend but the trend is consistent and c le a rly evident across the f i r s t fo u rfiv e screenings, given groups o f one hundred or more subjects. S im ilar analyses of cost data do not show the program to be associated with cost reductions but in fa c t to incur fin a n c ia l costs, a t le a s t 1n the period o f time studied. I t is possible th at cost savings would be re a lize d given longer re c ip ie n t exposure to the program or i f more in d ire c t bene­ f i t s o f program p a rtic ip a tio n could be measured. However, present in ­ dications are th a t the program is b e n e fittin g recip ients but Incurring a true cost. In the fin a l analysis of the program's worth, the benefits to recip ients - reductions of 10 percent or less in re fe rra b le conditions must be balanced against true program costs of perhaps o n e-th ird to one m illio n d o lla rs annually or roughly $3-$10 per screening ( $.375 - $1 m il1 io n /1 0 5 ,000 screenings). Based upon t h is , and o th er, studies of EPSDT outcomes, the program m erits continued support. However, considering the study's mixed results and the extraneous factors which appear to Influence re fe rra l ra te s , th is study also suggests th at continued analysis o f program outcomes is warranted. In conclusion, th is study suggests the program is achieving modest gains a t modest costs. SELECTED EPSDT BIBLIOGRAPHY Applied Management Sciences. Assessment of EPSDT Practices and Costs - Report on the Cost Impact of the EPSDT Program. S ilv e r Spring, Maryland: Applied Management S c ie n c e s ,1976. Bergman, Abraham G. "The Menace o f Mass Screening," American Journal o f Public H e a lth . LXVII (J u ly , 1977), 601-02. B u tle r, John A. and Richard K. Scotch. "Medicaid and Children: Some Recent Lessons and Reasonable Next Steps," Public P o lic y , XXVI (W inter, 1978), 3-27. C h ild ren 's Defense Fund. EPSO&T: Does I t Spell Health Care For Poor C hildren? Washington, D. C.: Washington Research P ro je c t, In c ., T57T, Community Health Foundation. "Cost Impact Study o f the North Dakota EPSDT Program," Evanston, I l l i n o i s : Community Health Foundation, 1977. (Mimeographed.) C u rrie r, Richard. "Is Early and Periodic Screening, Diagnosis and Treatment (EPSDT) Worthwhile?," Public Health Reports. XCII (November-December, 1977), 527-3?"! C u rrie r, Richard, MA. "Michigan Leads in Screening M ed ic aid -E lig ib le C h ild ren ," Michigan M edicine, (March, 1977), 136-137. Dixon, Morris S ., J r. " T it le XIX EPSDT: The Im plications fo r Pedi­ a t r ic P ra c tic e ," B u lle tin o f P e d ia tric P ra c tic e , VI (December. 1972), 2 -3 . F o ltz , Anne-Marie. "The Development o f Ambiguous Federal Policy: Early and Periodic Screening, Diagnosis and Treatment (EPSDT)," nd Q u arterly/H ealth and Society, L111 (W inter, 1975), 35-64. . "Rebuttal to Dr. Shenkln," Medical Care, XIV (October, 1976), 886-87. ________ . U n certain ties o f Federal Child Health P o lic ie s : Impact in Two S ta tes . New Haven, CT: Yale U n iv e rs ity , Department of tp^em lology and Public H ealth , 1978. 193 194 F o ltz , Anne-Mar1e and Donna Brown. Health Policy P roject: The Impact of Federal Child Health Poljcy Under EPSDT - The Case of Connecticut. New Haven. CT: Yale U n iversity Department o f Epidemiology and Public H ealth , 1975. _______ . "State Response to Federal Policy: Children, EPSDT, and tHie Medicaid Muddle," Medical Care. X I I I (August, 1975), 63042. K irk , Thomas R ., M .D ., e t a l . , "EPSDT - One Quarter M illio n Screenings in Michigan," Public Health B r ie fs . LXVI (May, 1976), 482-84. Michigan Department o f Public Health and Michigan Department of Social Services. EPSDT Michigan Annual Report, 1978. Lansing, Michigan: 1979. ________ . Health Screening: A C all to a B etter L if e , Michigan Annual Report, 1977. Lansing, Michigan: 1978. Newman, Howard. "The Challenge and P o ten tial o f EPSDT," The Journal o f School H e a lth . XVIV (May, 1974), 243-45. Peterson, E ric . Legal Challenges to Bureaucratic D iscretion : The In ­ fluence o f Lawsuits on tne Implementation o f EPSDT. Health Policy P roject Working Paper No. 27. New Haven, CT: Yale Uni­ v e rs ity , 1975. Philadelphia Health Management Corporation. A Study o f the Process. E ffectiveness, and Costs of the EPSDT Program'in Southeastern Pennsylvania, Part I I I . P h ila d e lp h ia , Pennsylvania: P h ila d e l­ phia Health Management Corporation, 1980. Reis, Janet, Ph.D. and Sharon Herzberger, Ph.D. "Problem and Program Linkage: The Early and Periodic Screening, Diagnosis and T re a t­ ment Program as a Means of Preventing and Detecting Child Abuse," In fa n t Mental Health Journal. IV (W inter, 1980), 262-69. Shenkin, Budd N ., M.D. "P o licies and the Health o f C h ild ren ," Medical Care. XIV (October, 1976), 884-85. Sprague, Homer A ., e t a l . "Comparison of EPSDT and NCHS Growth Charts," Preventive Medicine, IX (1 98 0 ), 398-408. S tic k le r , Gunnar B. "How Necessary is the 'Routine Checkup'?," C lin ­ ic a l P e d ia tric s . VI (August, 1967), 454. United States Department o f Health and Human Services, Health Care Financing A dm inistration. EPSDT: A Selected Annotated B ib lio ­ graphy. Washington, D.C.: Government P rin tin g (JfH ce, i960. 195 United States Department of H ealth , Education, and W elfare: Health Care Financing A dm inistration. EPSD&T: The Possible Dream. Washington, D.C.: Government P rin tin g O ffic e , 1977. United States Department of H ealth , Education and W elfare. "The Status of EPSDT," Washington, D.C.: Government P rin tin g O ffic e , 1975. EPSDT RELATED BIBLIOGRAPHY Breslow, Lester. "An H is to ric a l Review o f M ultiphasic Screening," Preventive M edicine, I I (June, 1973), 177-96. C o llen , Morris F. "Cost Analysis o f a M ultiphasic Screening Program," New England Journal o f Medicine. CCCLXXX (May 8 , 1969), 1043-45. ______ _____ "M ultiphasic Checkup Evaluation Study. 4. Prelim 1nary Cost B e n efit Analysis fo r Middle-Aged Men," Preventive Medicine, I I (June, 1973), 236-46. _______ "D o lla r Cost Per P o sitive Test fo r Automated M ulti phasic Screening," New England Journal of Medicine, CCLXXX (Au­ gust, 1970), 459-63. Cudlipp, Edyth. "High Blood Pressure: A Black Epidemic," Essence, IV (October, 1973), 41-47. C u tle r, John L. "Multiphasic Checkup Evaluation Study. 1. Methods and Population," Preventive Medicine, I I (June, 1973), 197-206. Dales, Loring G ., e t a l . "Multiphasic Checkup Evaluation Study. 3. O utpatient C lin ic U t iliz a t io n , H o s p ita liz a tio n , and M o rta lity Experience a f t e r 7 y ea rs ," Preventive Medicine, I I (June, 1973), 221-235. E n te rlin e , P h ilip E. and Bernard Kordan. "Controlled Evaluation of Mass Surveys fo r Tuberculosis and Heart D isease,” Public Health Report, L X X III (October, 1958), 867-875. Gordon, Tavla. "Some Methodological Problems in the Long-Term Study of Cardiovascular Disease: Observations on the Framingham Study," Journal o f Chronic Disease, X (September, 1959), 186-206. G reb ler, Leo, Joan W. Moore and Ralph C. Gusman. People. New York: The Free Press, 1970. The Mexican-American G rim aldi, J. V. "The Worth of Occupational Health Programs," Journal o f Occupational Medicine, V II (1 965 ), 365-373. 196 JedUcka, Davor, Yongsock Shin and Everett S. Lee. "Suicide Among B lacks,” Phylon. XXXVIII (December, 1977), 448-55. Kitagawa, Evelyn M. and P h ilip M. Hauser. D iffe re n tia l M o rta lity in the United States. Cambridge: Harvard U niversity Press, 1973. Kovar, Mary Grace. "M o rtality of Black Infants in the United S tates," Phylon. XXXVIII (December, 1977), 370-97. K u lle r, Lewis and Susan Tonascla. "Commission on Chronic Illn e s s Follow-Up Study: Comparison o f Screened and Nonscreened In d iv i­ duals," Archives of Environmental Health, XXI (November. 19701. 656-65. Lee, Anne S. "Maternal M o rta lity in the United S tates," Phylon, XXXVIII (September, 1977), 259-66. M arris, Peter and Martin Rein. Atherton Press, 1969. Dilemmas of Social Reform, New York: Moss, Abigail and Geraldine Scott. "Hypertension: United S ta te, 1974," Phylon, XXXVIII (December, 1977), 357-58. Moynlhan, Daniel P. Maximum Feasible Misunderstanding, New York: The Free Press, 1969. N ie, Norman, e t a l . S ta tis tic a l Package fo r the Soda! Sciences. York: McGraw-Hill Book Company, 1975. New Piven, Frances Fox and Richard A. Cloward. Regulating the Poor: The Functions of Public W elfare. New York: Pantheon, 1971. Ramcharan, S a n itr i, e t a l. "Multiphasic Checkup Evaluation Study: 2. D is a b ility and Chronic Disease A fte r Seven Years o f M u lti­ phasic Health Checkups," Preventive Medicine, I I (June, 1973). 207-220. Reid, John D ., Everett S. Kee, Davor JedUcka and Yongsock Shin. "Trends 1n Black H ealth." Phylon. XXXVIII (June, 1977), 105-16. Roberts, Norbert J. "M o rta lity Among Males in Periodic-Health-Exam1nation Programs," The New England Journal of Medicine, CCLXXXI (J u ly , 1969), 20. Shapiro, Sam. "Evaluation o f Two Contrasting Types of Screening Pro­ grams," Preventive H ealth, I I (June, 1973), 266-277. Thomer, Robert M. "Whither Multiphasic Screening?" The New Enqland Journal o f Medicine, CCLXXX (May, 1969), 1037-42. 197 Thorner, Robert M. and E. L. Crumpacker. "M o rta lity and Periodic Examinations o f Executives," Archives o f Environmental H ealth. I l l (July-December, 1961), S Z ttT . White, Jack E. "Cancer Differences 1n the Black and Caucasian Pop­ u la tio n ," Phylon, XXXVIII (September, 1977), 297-314. W ilb e r, Joseph A ., M.D. "Hypertension: An E d ito r ia l," Phylon, XXXVIII (December, 1977), 353. W ylie, Charles M. " P a rtic ip a tio n In M u ltip le Screening C lin ic w ith Five-Year Followup," Public Health Reports, LXXVI (J u ly , 1961), 596-602. Yabura, Lloyd. "Health Care Outcomes In the Black Community," Phylon, XXXVIII (June, 1977), 194-202. APPENDICES APPENDIX A Table 1 (a ). Number of long-term e lig ib le screened by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 Total 331 614 980 1492 1975 2044 1768 1845 2218 2059 2120 2345 1818 1461 1093 622 336 245 73 46 10 13 47 153 463 1000 1257 1284 1505 1607 1563 1565 1829 1493 1539 1255 946 614 369 116 58 6 0 4 17 70 340 606 744 925 791 793 886 774 746 760 597 480 356 209 83 39 8 1 0 3 4 55 145 243 263 235 234 179 213 180 146 126 113 74 33 25 5 1 0 0 1 1 6 27 35 55 32 36 25 23 26 17 12 14 8 7 2 1 0 0 0 0 0 0 3 5 2 7 0 3 3 2 2 0 2 0 0 1 0 0 0 0 0 0 0 0 1 0 1 3 0 0 0 0 0 0 0 0 0 0 0 345 665 1154 2030 3376 4082 4080 4595 4891 4688 4778 5187 4265 3925 3083 2177 1388 863 300 149 25 25,495 18,682 9,228 2,278 328 30 5 56,046 Table I(A a ). Nunfcer of long-term e lig ib le whites screened by age and number o f life tim e screenings. Number of Lifetime Screenings Aqe Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 Total 140 274 464 701 992 974 879 933 1119 1005 1089 1152 893 613 467 237 111 70 18 8 3 6 19 63 237 504 580 598 733 828 773 779 917 709 713 571 391 216 123 37 9 1 0 1 11 37 161 277 342 438 375 379 427 362 339 334 257 188 127 59 17 10 0 0 0 1 1 27 68 107 137 120 119 95 105 77 68 52 45 21 11 4 0 0 0 0 1 0 4 12 10 23 15 16 11 10 13 6 9 7 3 3 0 0 0 0 0 0 0 0 1 4 2 4 0 3 2 2 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 146 294 540 976 1688 1912 1940 2266 2462 2293 2404 2548 2033 1735 1356 868 478 266 77 27 4 12,142 8,807 4,141 1,058 143 20 2 26,313 Table I(B a ). Number of long-term e lig ib le blacks screened by age and number of life tim e screenings. Number of Lifetime Screeninqs Age Total 2 3 4 5 6 7 Total 183 332 497 758 932 1014 854 857 1039 955 964 1140 878 813 604 369 213 171 54 38 7 6 26 84 209 470 630 623 722 714 726 718 837 701 748 620 505 362 228 75 47 5 0 3 6 28 159 301 365 446 386 380 422 379 374 387 309 278 214 139 60 26 7 0 0 2 3 27 69 125 114 99 108 77 95 88 69 69 64 47 22 18 5 1 0 0 0 1 2 14 22 25 12 20 11 12 11 11 3 7 5 3 2 1 0 0 0 0 0 0 2 1 0 3 0 0 0 0 1 0 2 0 0 0 0 0 0 0 0 0 0 0 1 0 0 2 0 0 0 0 0 0 0 0 0 0 0 189 361 589 999 1590 2030 1991 2164 2253 2191 2192 2463 2052 2029 1605 1225 841 563 209 117 20 12,672 9,056 4,669 1,102 162 9 3 27,673 200 Under 1 1 2 3 4 5 6 7 B 9 10 11 12 13 14 15 16 17 18 19 20 1 Number of long-term e lig ib le American Indians screened by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 Total 1 3 1 5 3 9 3 6 13 5 7 8 6 2 3 3 1 0 0 0 0 0 0 0 2 6 3 7 5 7 3 6 9 11 9 6 7 5 1 0 0 0 0 0 0 0 1 8 2 4 1 3 2 4 7 5 6 2 1 0 2 0 0 0 0 0 0 0 2 2 1 4 1 1 2 2 0 0 0 0 0 1 0 0 0 0 0 0 0 0 2 2 2 0 1 1 0 0 0 0 0 0 0 0 0 1 3 1 7 10 22 16 18 27 12 17 24 26 16 15 12 7 1 3 0 0 79 87 48 16 8 238 Number of long-term e lig ib le Spanish-speaking screened by age and number of life tim e screenings. Number of life tim e Screenings 10 11 12 13 14 15 16 17 18 19 20 2 3 4 5 6 Total 6 4 18 25 46 40 27 44 39 29 41 38 36 28 14 10 8 4 1 0 0 1 2 6 14 19 38 53 43 56 57 54 63 63 62 53 39 26 17 4 2 0 0 0 0 5 18 18 31 36 29 29 34 28 24 33 22 10 12 8 3 3 0 1 0 0 0 1 6 6 11 12 5 5 11 13 9 5 4 6 0 2 0 0 0 0 0 0 0 1 1 5 3 0 2 0 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 8 6 24 44 84 103 118 139 139 120 136 141 138 132 94 63 52 30 10 5 0 458 672 343 97 15 1 1,586 202 1 2 3 4 5 6 7 8 9 1 Table I(E a ). Number of long-term e lig ib le males screened by age and number of life tim e screenings. Number of Lifetime Screenings Age Total 2 3 4 5 6 Total 167 298 504 755 990 1048 889 919 1136 1012 1055 1141 916 687 502 262 126 98 14 13 2 9 20 72 246 510 630 624 761 776 772 788 870 722 754 610 423 264 147 30 9 0 0 1 6 32 167 304 387 443 424 402 442 375 392 366 290 234 166 88 26 6 1 1 0 1 3 29 62 116 123 119 121 88 114 94 69 66 50 31 15 7 0 0 0 0 1 1 2 13 19 32 14 17 12 8 14 12 5 11 2 5 2 0 0 0 0 0 0 0 1 3 1 4 3 0 3 1 1 0 1 0 0 0 0 0 177 319 584 1037 1698 2058 2038 2279 2473 2327 2385 2511 2139 1889 1473 981 589 353 79 28 3 12,534 9,037 4,552 1,109 170 18 27,420 203 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I(F a ). Number of long-term e lig ib le females screened by age and number of lifetim e screenings. Number of Lifetime Screenings Age Total 2 3 4 5 6 7 Total 164 316 476 737 985 996 879 926 1082 1047 1065 1204 902 774 591 360 210 147 59 33 8 4 27 81 217 490 627 660 744 831 791 777 959 771 785 645 523 350 222 86 49 6 0 3 11 38 173 302 357 482 367 391 444 399 354 394 307 246 190 121 57 33 7 0 0 2 1 26 83 127 140 116 113 91 99 86 77 60 63 43 18 18 5 1 0 0 0 0 4 14 16 23 18 19 13 15 12 5 7 3 6 2 1 1 0 0 0 0 0 0 2 2 1 3 0 3 0 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 168 346 570 993 1678 2024 2042 2316 2418 2361 2393 2676 2126 2036 1610 1196 799 510 221 121 22 12,961 9,645 4,676 1,169 159 14 2 28,626 204 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I(G a ). Number of long-term e lig ib le white males screened by age and number of life tim e screenings. Number of Lifetime Screenings Age Total 2 3 4 5 6 7 Total 70 136 237 355 501 519 418 464 553 532 545 567 448 293 242 113 45 20 4 4 2 3 8 30 122 263 306 300 369 415 380 408 426 348 353 295 169 92 62 11 4 0 0 0 3 16 79 135 189 216 207 196 223 180 175 161 136 91 52 26 6 0 0 0 0 1 1 15 32 58 71 64 62 48 51 41 33 27 19 12 5 1 0 0 0 0 0 0 2 6 3 13 7 5 5 5 6 5 5 7 0 2 0 0 0 0 0 0 0 0 1 2 1 1 0 0 2 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 73 144 271 494 860 999 970 1134 1247 1176 1229 1231 1019 845 705 399 201 115 22 8 2 6,068 4,364 2,091 541 71 8 1 13,144 205 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Number of long-term e lig ib le white females screened by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 7 Total 70 138 227 346 491 455 461 469 566 523 544 585 445 320 225 124 66 50 14 4 1 3 11 33 115 241 274 298 364 413 393 371 491 361 360 276 222 124 61 26 5 1 0 1 8 21 82 142 153 222 168 183 204 182 164 173 '121 97 75 33 11 10 0 0 0 1 0 12 36 49 66 56 57 47 54 36 35 25 26 9 6 3 0 0 0 0 0 0 2 6 7 10 8 11 6 5 7 1 4 0 3 1 1 0 0 0 0 0 0 0 0 2 1 3 0 3 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 73 150 269 482 828 913 970 1132 1215 1167 1175 1317 1014 890 651 469 277 151 55 19 2 6,124 4,443 2,050 518 72 11 1 13,219 Table l ( I a ) . Number of long-term e lig ib le black males screened by age and number of life tim e screenings. Nunfeer of Lifetime Screenings Age Total 2 3 4 5 6 7 Total 93 158 255 388 460 503 450 421 553 454 480 553 441 376 251 146 78 74 9 9 0 5 12 40 112 235 303 297 369 328 364 345 417 334 364 283 230 158 77 17 4 0 0 1 3 14 79 154 171 206 197 191 198 179 201 188 137 134 106 57 18 6 0 0 0 1 2 13 28 54 48 48 55 40 55 49 34 36 28 17 10 6 0 0 0 0 0 1 0 6 14 12 5 12 5 2 6 7 0 4 2 2 2 0 0 0 0 0 0 0 0 1 0 3 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 98 171 299 517 787 994 987 1056 1134 1078 1068 1206 1031 970 707 543 361 220 52 19 0 6,152 4,294 2,240 524 80 6 2 13,298 207 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I(J a ). Number of long-term e lig ib le black females screened by age and number of life tim e screenings. * Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 Total 90 174 242 370 472 511 404 436 486 501 484 587 437 437 353 223 135 97 45 29 7 1 14 44 97 235 327 326 353 386 362 373 420 367 384 337 275 204 151 58 43 5 0 2 3 14 80 147 194 240 189 189 224 200 173 199 172 144 108 82 42 20 7 0 0 1 1 14 41 71 66 51 53 37 40 39 35 33 36 30 12 12 5 1 0 0 0 0 2 8 8 13 7 8 6 10 5 4 3 3 3 1 0 1 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 91 190 290 482 803 1036 1004 1108 1119 1113 1124 1257 1021 1059 898 682 480 343 157 98 20 6,520 4,762 2,429 578 82 3 1 14,375 Number of long-term e lig ib le American Indian males screened by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 11 12 13 14 15 16 17 18 19 20 2 3 4 5 Total 1 3 1 2 1 4 2 3 6 2 4 2 4 1 2 1 0 0 0 0 0 0 0 0 2 1 0 6 1 4 2 4 1 5 2 3 2 2 0 0 0 0 0 0 0 0 1 4 1 2 1 2 1 2 3 2 2 1 0 0 2 0 0 0 0 0 0 0 2 0 1 4 1 0 2 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2 0 0 1 1 0 0 0 0 0 0 0 0 0 1 3 1 4 3 10 10 9 15 7 10 8 13 5 7 4 2 0 2 0 0 39 35 24 11 5 114 209 7 8 9 10 1 Number of long-term e lig ib le American Indian females screened by age and number of life tim e screenings. Number of lifetim e Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 Total 0 0 0 3 2 5 1 3 7 3 3 6 2 1 1 2 1 0 0 0 0 0 0 0 0 5 3 1 4 3 1 2 8 6 7 3 5 3 1 0 0 0 0 0 0 0 0 4 1 2 0 1 1 2 4 3 4 1 1 0 0 0 0 0 0 0 0 0 0 2 0 0 0 1 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 7 12 6 9 12 5 7 16 13 11 8 8 5 1 1 0 0 40 52 24 5 3 124 Number of long-term e lig ib le Spanish-speaking males screened by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 Total 2 1 11 8 26 17 15 26 19 16 18 14 21 16 5 1 2 4 1 0 0 1 0 2 9 10 18 21 21 28 24 29 25 32 33 27 21 11 8 2 1 0 0 0 0 2 8 10 22 18 19 12 19 13 11 14 14 7 7 2 0 0 0 1 0 0 0 1 0 3 3 3 3 0 6 3 2 3 3 2 0 0 0 0 0 0 0 0 0 1 1 5 2 0 1 0 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 4 1 13 19 45 46 62 73 71 55 67 59 69 65 49 32 22 15 3 1 0 223 323 178 33 13 1 771 Number o f long-term e lig ib le Spanish-speaking females screened by age and number of life tim e screenings. Number of Lifetime Screenings 2 3 4 5 Total 4 3 7 17 20 23 12 18 20 13 23 24 15 12 9 9 6 0 0 0 0 0 2 4 5 9 20 32 22 28 33 25 38 31 29 26 18 15 9 2 1 0 0 0 0 3 10 8 9 18 10 17 15 15 13 19 8 3 5 6 3 3 0 0 0 0 0 0 6 3 8 9 2 5 5 10 7 2 1 4 0 2 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 4 5 11 25 39 57 56 66 68 65 69 82 69 67 45 31 30 15 7 4 0 235 349 165 64 2 815 212 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Number of long-term e lig ib le participants screened In D etro it by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 U 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 7 Total 87 217 326 494 552 639 543 546 692 584 590 712 511 478 352 199 94 95 20 17 4 2 12 22 68 111 229 198 236 231 259 225 334 231 252 207 179 113 76 20 16 2 0 3 0 7 25 62 56 79 60 59 70 67 59 52 58 43 29 19 9 3 1 0 0 0 1 5 10 28 12 16 15 8 16 12 14 11 10 7 2 4 1 0 0 0 0 1 2 5 7 9 5 5 5 0 3 2 1 0 2 1 1 0 0 0 0 0 0 0 1 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 89 232 348 571 695 946 832 882 1006 923 898 1129 816 798 629 431 245 193 54 37 7 7,752 3,023 761 172 49 3 1 11,761 Table I ( Pa) . Number of long-term e lig ib le participants screened in fo rty -fo u r Northern Michigan counties by number of life tim e screenings. Nunfcer of Lifetime Screenings 2 3 4 5 6 7 Total 1,752 1,462 755 250 43 12 2 4,276 214 1 APPENDIX B Table I(Q a). Number of long-term e llg lb le s screened In D e tro it and Northern Michigan by race and number of life tim e screenings. Number of Lifetim e Screenings Location/Race 1 2 3 4 5 6 7 Total 1,045 6,627 7,672 301 2,674 2,975 52 693 745 12 160 172 4 45 49 0 3 3 0 1 1 1,414 10,203 11,617 1,611 26 1,637 1,389 28 1,417 667 24 691 225 8 233 39 2 41 10 0 10 2 0 2 3,943 88 4,031 D etroit Whites Blacks Total Northern Michigan Whites Blacks Total Table 11(a). Nunfcer o f one-year e llg ib le s screened by age and number of life tim e screenings. Total Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 8523 7605 6256 6702 6365 5580 4634 4641 4964 4625 4437 4789 3870 3310 2579 1715 976 799 621 475 122 579 1825 2429 3742 4259 3552 3056 3332 3295 3056 3099 3234 2717 2696 2279 1701 1157 709 340 215 67 16 227 389 917 1535 1560 1484 1707 1437 1385 1458 1280 1178 1192 931 737 552 347 141 86 23 3 7 43 97 299 396 446 434 383 370 277 298 262 210 173 144 100 46 38 15 3 83,588 47,339 18,582 4,044 5 6 7 8 Total 0 0 7 6 43 69 69 77 46 50 39 38 36 26 15 23 13 13 5 1 2 0 0 1 0 5 7 6 5 7 3 4 3 3 3 1 2 0 0 1 0 0 0 0 0 0 0 0 1 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 9121 9664 9125 11464 12506 11165 9696 10196 10134 9489 9314 9642 8066 7437 5978 4322 2798 1914 1146 792 217 51 3 1 154,186 578 Table II(A a ). Number of one-year e lig ib le whites screened by age and number of life tim e screenings. Total Lifetime Screenings Ape Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 8 Total 4281 4056 3574 3881 3902 3268 2793 2737 2881 2735 2621 2742 2213 1757 1327 839 441 363 310 224 76 290 989 1271 2142 2599 2038 1749 1926 1942 1788 1749 1842 1487 1430 1239 848 491 282 128 77 28 5 121 200 507 906 884 828 947 80S 757 791 703 606 599 436 340 228 120 39 24 5 0 2 18 52 165 232 225 238 224 199 159 157 129 104 84 62 33 17 10 5 0 0 0 2 2 24 35 35 45 23 24 21 21 20 15 12 11 5 5 1 0 1 0 0 1 0 4 3 4 5 4 0 4 2 2 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4576 5168 5066 6584 7600 6461 5634 5898 5880 5504 5345 5467 4457 3907 3098 2100 1639 787 489 330 no 47,021 26,335 9,851 2,115 302 32 2 1 85,659 Table ll( B a ). Number of one-year e lig ib le blacks screened by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3938 3289 2473 2612 2220 2132 1690 1717 1895 1731 1640 1888 1508 1420 1151 792 491 396 281 226 41 267 752 1044 1433 1478 1357 1148 1262 1190 1122 1174 1222 1072 1112 913 754 587 383 190 128 38 33,531 18,626 '3 4 5 6 7 Total 11 98 175 356 543 595 569 669 561 548 595 510 512 528 438 363 291 204 92 55 16 2 5 25 43 115 135 195 169 130 147 102 120 no 94 83 73 60 26 22 9 3 0 0 5 4 17 30 30 25 17 22 15 14 14 11 3 11 7 6 4 1 1 0 0 0 0 1 3 2 0 3 0 0 0 1 1 0 2 0 0 0 0 0 0 0 0 0 0 0 1 0 1 2 0 0 0 0 0 0 0 0 0 0 0 4218 4144 3722 4448 4374 4252 3635 3842 3797 3572 3526 3754 3217 3166 2588 1995 1436 1015 589 419 99 7,729 1,668 237 13 4 61,808 Number of one-year e lig ib le American Indians screened by age and number o f life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 Total 28 28 37 23 23 27 19 18 30 17 22 14 17 14 10 5 5 3 1 3 1 1 12 11 18 28 17 17 19 19 12 19 18 27 16 12 13 12 4 2 1 0 0 0 1 4 11 11 10 13 5 14 10 6 10 10 12 5 2 3 3 2 0 0 0 0 0 2 5 3 2 5 5 3 6 2 0 0 0 0 0 2 0 0 0 0 0 0 0 2 2 2 1 1 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 29 40 49 45 64 61 51 54 61 49 55 45 56 40 34 24 19 11 8 6 1 345 278 132 35 11 1 802 - Table II(D a ). Number of one-year e lig ib le Spanish-speaking screened by age and number of life tim e screenings. Number of Lifetime Screenings Age Total 2 3 4 5 6 Total 239 191 141 153 180 126 104 136 125 109 115 118 113 87 67 55 28 28 16 13 2 20 61 93 130 138 120 126 115 131 124 134 139 115 120 104 73 53 35 15 8 1 0 5 13 46 70 64 71 71 65 62 60 60 47 54 41 25 28 16 6 5 1 1 0 0 2 16 24 18 25 24 18 12 15 21 12 6 9 7 3 3 1 0 0 0 0 0 2 4 2 5 4 3 2 2 2 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 260 257 247 331 406 338 321 352 349 316 323 335 298 273 219 162 117 83 40 27 4 2,146 1,855 810 217 28 2 5,058 220 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table II( E a ). Number of one-year e lig ib le males screened by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 Total 4342 3856 3155 3392 3145 2860 2314 2348 2545 2314 2210 2345 1926 1568 1152 709 339 230 48 22 6 297 933 1257 1879 2148 1817 1517 1663 1612 1515 1568 1537 1326 1319 1090 773 490 238 63 17 3 12 116 197 458 775 795 741 856 741 681 725 632 598 576 452 361 255 142 36 8 1 1 1 22 56 147 188 230 203 200 200 148 164 126 95 88 64 46 22 10 2 0 0 0 4 4 25 36 39 44 22 23 16 13 18 14 5 15 5 8 2 0 0 0 0 1 0 3 3 3 4 4 3 1 3 2 2 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 4652 4906 4636 5789 6243 5699 4844 5118 5125 4736 4668 4694 3996 3574 2788 1923 1135 640 159 49 10 40,826 23,062 9,158 2,013 293 31 1 75,384 Table II( F a ). Number of one-year e lig ib le females screened by age and number of life tim e screenings. Total Lifetime Screeninqs Age Total 2 3 4 5 6 7 8 Total 4181 3749 3101 3310 3220 2720 2320 2293 2419 2311 2227 2444 1944 1742 1427 1006 637 569 573 453 116 282 892 1172 1863 2111 1735 1539 1669 1683 1541 1531 1697 1391 1377 1189 928 667 471 277 198 64 4 111 192 459 760 765 743 851 696 704 733 648 580 616 479 376 297 205 105 78 22 2 6 21 41 152 208 216 231 183 170 129 134 136 115 85 80 54 24 28 13 3 0 0 3 2 18 33 30 33 24 27 23 25 18 12 10 8 8 5 3 1 2 0 0 0 0 2 4 3 1 3 0 3 0 1 1 0 1 0 0 1 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4469 4758 4489 5675 6263 5466 4852 5078 5009 4753 4646 4948 4070 3863 3190 2399 1663 1274 987 743 207 42,762 24,277. 9,424 2,031 285 20 2 1 78,802 . 222 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table II(G a ). Number of one-year e lg ib le white males screened by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 Total 2241 2071 1816 1950 1949 1699 1372 1392 1482 1383 1311 1354 1106 859 631 365 160 91 24 6 4 156 505 668 1072 1325 1056 885 973 949 889 899 859 727 696 610 383 213 109 25 8 0 4 64 104 254 471 458 418 484 413 372 402 356 299 288 216 163 96 47 10 1 0 0 0 7 30 84 114 120 121 122 109 90 84 61 50 41 27 18 9 3 2 0 0 0 1 2 13 20 21 25 13 9 8 9 7 7 5 9 1 3 0 0 0 0 0 1 0 3 2 2 4 1 0 1 2 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 2401 2640 2597 3308 3845 3349 2818 2999 2980 2763 2711 2664 2201 1901 1503 947 488 259 62 17 4 23,266 13,007 4,920 1,092 153 18 1 42,457 Table X I(H a). Number of one-year e lig ib le white females screened by age and number of life tim e screenings. Total Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 8 Total 2040 1985 1758 1931 1953 1569 1421 1345 1399 1352 1310 1388 1107 898 696 474 281 272 286 218 72 134 484 603 1070 1274 982 864 953 993 899 850 983 760 734 629 465 278 173 103 69 28 1 57 96 253 435 426 410 463 392 335 389 347 307 311 220 177 132 73 29 23 5 0 2 11 22 81 118 105 117 102 90 69 73 68 54 43 35 15 8 7 3 0 0 0 1 0 11 15 14 20 10 15 13 12 13 8 7 2 4 2 1 0 1 0 0 0 0 1 1 2 1 3 0 3 0 1 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2175 2528 2469 3276 3755 3112 2816 2899 2900 2741 2634 2803 2256 2006 1595 1153 710 528 427 313 106 23,755 13,328 4,931 1,023 149 14 1 1 43,202 Table I I ( l a ) . Number o f one-year e lig ib le black males screened by age and number of life tim e screenings. Number of Lifetime Screeninqs Aqe Total 2 3 4 5 6 7 Total 1951 1656 1242 1337 1075 1064 855 857 969 849 814 915 736 643 478 300 166 125 20 15 1 130 380 527 729 733 684 563 630 574 560 584 598 521 547 415 336 243 113 33 7 2 8 47 87 180 264 292 273 321 288 274 285 250 267 260 209 181 141 85 21 6 0 0 1 15 24 54 60 98 75 64 77 54 68 56 42 43 32 25 10 7 0 0 0 0 3 2 10 12 16 12 6 14 6 2 9 7 0 5 3 4 2 0 0 0 0 0 0 0 0 1 0 3 0 0 0 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2 0 0 0 0 0 0 0 0 0 0 0 2089 2084 1874 2272 2136 2112 1806 1895 1905 1776 1743 1833 1590 1500 1145 855 578 337 83 28 3 16,068 8,909 3,739 805 113 7 3 29,644 225 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I I ( J a ) . Number of one-year e lig ib le black females screened by age and number of life tim e screenings. Number of Lifetime Screenings Age Total 2 3 4 5 6 7 Total 1986 1633 1231 1275 1145 1068 835 860 926 882 826 973 772 777 673 492 325 271 261 211 40 137 372 517 714 745 673 585 632 616 562 590 624 551 565 498 418 344 270 157 121 36 3 51 88 176 279 303 296 348 273 274 310 260 245 268 229 182 150 119 71 49 16 2 4 10 19 61 75 97 94 66 70 48 52 54 52 40 41 35 16 15 9 3 0 0 2 2 7 18 14 13 11 8 9 12 5 4 3 6 4 2 2 1 1 0 0 0 0 1 3 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2128 2060 1848 2186 2238 2140 1829 1947 1892 1796 1783 1921 1627 1666 1443 1140 858 678 506 391 96 17,462 9,727 3,990 863 124 6 1 32,173 226 Under 1 1 2 3 A 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table II(K a ). Number of one-year e lig ib le American Indian males screened by age and number of life tim e screenings. Number of Lifetime Screeninqs Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total - 1 2 3 4 5 6 Total 15 13 17 15 11 16 15 8 13 10 12 6 13 7 4 3 0 1 0 0 0 0 5 6 9 13 6 10 6 10 4 11 7 16 5 8 4 7 2 0 0 0 0 0 0 3 6 7 4 9 3 6 7 2 4 4 2 4 1 2 3 1 0 0 0 0 0 1 2 1 2 4 3 0 3 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2 0 0 1 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 15 18 23 27 31 32 31 27 30 23 31 19 34 16 14 12 8 5 3 1 0 179 129 68 17 6 1 400 Number of one-year e lig ib le American Indian females screened by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 , 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 Total 13 15 20 8 12 11 4 10 17 7 10 8 4 7 6 2 5 2 1 3 1 1 7 5 9 15 11 7 13 9 8 8 11 11 11 4 9 5 2 2 1 0 0 0 1 1 5 4 6 4 2 8 3 4 6 6 10 1 1 1 0 1 0 0 0 0 0 1 3 2 0 1 2 3 3 1 0 0 0 0 0 2 0 0 0 0 0 0 0 0 1 0 2 1 0 0 0 0 0 0 0 1 0 0 0 14 22 26 18 33 29 20 27 31 26 24 26 22 24 20 12 11 6 5 5 1 166 149 64 18 5 402 Number of one-year e lig ib le Spanish-speaking males screened by age and number of life tim e screenings. Nunfcer of Lifetime Screenings 1 2 3 4 5 11 12 13 14 15 16 17 18 19 20 2 3 4 5 6 Total 116 98 66 73 92 65 57 74 65 57 55 57 60 43 27 25 6 10 2 0 0 11 35 50 65 70 60 50 47 70 59 68 65 59 62 53 42 19 10 3 1 1 0 3 6 18 33 35 42 37 36 27 29 23 26 23 24 11 16 5 2 0 0 1 0 0 2 8 12 8 5 10 11 4 9 8 3 4 5 3 3 0 0 0 0 0 0 0 2 4 1 5 3 0 1 1 2 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 128 136 122 158 205 176 158 168 184 154 157 156 155 131 109 83 45 29 7 1 1 1,048 900 396 96 21 2 2,463 229 6 7 8 9 10 1 Number of one-year e lig ib le Spanish-speaking females screened by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 12 13 14 15 16 17 18 19 20 2 3 4 5 Total 123 93 75 80 88 61 47 62 60 52 60 61 53 44 40 30 22 18 14 13 2 9 26 43 65 68 60 76 68 61 65 66 74 56 58 51 31 34 25 12 7 0 0 2 7 28 37 29 29 34 29 35 31 37 21 31 17 14 12 11 4 5 1 0 0 0 0 8 12 10 20 14 7 8 6 13 9 2 4 4 0 3 1 0 0 0 0 0 0 0 1 0 1 3 1 1 0 0 0 0 0 0 0 0 0 132 121 125 173 201 162 163 184 165 162 166 179 143 142 110 79 72 54 33 26 3 1,098 955 414 121 7 2,595 230 7 8 9 10 11 1 Table II(O a ). Number of one-year e lig ib le participants screened in D etro it by number of life tim e screenings. Number of Lifetime Screenings 1 18,473 2 3 4 5 6 7 Total 5,691 1,250 280 76 3 1 25,774 Table I I ( Pa) . Number of one-year e lig ib le participants screened in fo rty -fo u r Northern Michigan counties by number of life tim e screenings. 232 Number of Lifetime Screenings 1 2 3 4 5 6 7 Total 7 *497 4,813 2,031 504 80 17 3 14,945 Table II(Q a ). Number o f one-year e lig ib le s screened in D etro it and Northern Michigan by race and number of life tim e screenings. Number of Lifetime Screenings Location/Race________ 1_________ 2_________ 3_________ 4_________ 5_________ 6_________ 7_________ Total Detroit 2,688 15,312 18,000 709 4,825 5,534 95 1,119 1,214 22 255 277 5 68 73 0 3 3 0 1 1 3,519 21,583 25,102 7,517 112 7,269 4,535 78 4,613 1,897 45 1,942 455 18 473 69 2 71 16 0 16 2 1 3 14,491 256 14,387 Northern Michigan Whi tes Blacks Total 233 Whites Blacks Total APPENDIX C Table I I I . Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le s by age and number of life tim e screenings. 3 4 5 6 7 - - - - - - - - - - - - - - - - - - - - - - - - Age 1 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .500 .250 .583 .800 1.222 .983 1.107 .889 .995 .982 .982 1.051 1.049 1.057 .977 1.196 1.086 1.397 1.357 1.366 .500 .666 .428 .418 .845 .993 .868 .919 .872 .862 .860 .813 .843 .882 .912 1.027 1.054 1.137 1.157 .333 .766 .907 .817 .834 .841 .851 .745 .821 .804 .948 .952 .993 .997 .881 2.058 1.000 1.000 .746 .866 .561 .656 .698 .979 .838 .520 .763 .928 .793 1.285 .833 1.000 2.000 .666 .800 1.111 .428 .600 .777 .667 1.000 1.500 1.750 .500 1.000 1.000 - 1.036 .884 .868 .765 .852 -15* -2% -12% +11% Grand Mean % Change As No. Screenings In creased By One - 1.000 - - - - - - - - - - - - - - - - - - - - - - 1.000 +17% irZZ Number of Lifetime Screenings Table I I I ( A ) . Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le whites by age and number of life tim e screenings. Number of Lifetime Screenings Age Grand Mean % Change As No. Screenings In­ creased By One 4 5 - - - - - - - - 2 3 *p _ 0 .333 0 .666 1.222 1.036 1.129 .905 .900 .766 .836 .912 .816 .944 .886 1.095 1.134 .920 1.416 .666 0 .666 .500 .428 .803 .984 .817 .810 .760 .788 .761 .823 .816 .813 .838 1.017 1.054 .562 1.500 1.000 .866 .826 .928 .790 .797 .774 .687 .691 .715 .876 .866 .942 .974 1.000 2.666 1.000 .932 .817 .814 .682 .815 -12% -0% -16% +20% 0 .815 .852 .526 .666 .516 .695 .735 .444 .681 .666 .600 1.500 1.000 - ,714 1.000 1.000 .250 .666 .833 .333 1.000 1.000 2.000 0 - - - - - 235 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I I I ( B ) . Average number of re fe rra ls a t la s t screening In 1978 fo r long-term e lig ib le blacks by age and number of life tim e scrrenings. i Aqe Grand Mean % Change As No. Screenings In­ creased By One 2 3 4 5 6 7 - 1.000 - _ 0 .636 .909 1.250 .969 1.119 .887 1.097 1.171 1.161 1.205 1.319 1.157 1.094 1.292 1.039 1.636 1.312 1.444 1.000 0 .428 .839 1.012 .915 1.004 .975 .964 .984 .812 .871 .958 1.015 1.036 1.050 1.390 1.117 0 .666 .960 .707 .877 .953 .934 .826 .978 .869 .992 1.047 1.023 .977 .814 1.666 1.000 1.090 .656 .972 .645 .700 .947 1.272 1.000 .518 .916 1.133 1.000 1.142 1.000 1.000 - 2.000 .600 .500 1.000 1.000 .500 0 1.000 1.000 1.000 1.000 1.000 - 1.144 .948 .921 .876 .826 -17% -3% -S t -6% +21% 236 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Number of Lifetime Screenings Table II1 (C ). Average number of re fe rra ls at la s t screening in 1978 fo r long-term e lig ib le American Indians by age and number of life tim e screenings. Nunfcer of Lifetime Screenings Aqe 1 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 _ - - - - - - - - - - - _ _ _ - 0 1.500 .500 0 0 - Grand Mean * Change As No. Screenings Incrased By One - 0 0 0 0 1.000 1.500 - 1.000 - 2.000 - 2.000 1.000 0 1.333 0 1.000 .333 .500 0 3 4 - - - - - 0 - - - .500 2.000 .500 - 1.000 0 - 1.000 - - - - - - - - - - - - - - - - .666 .772 .571 .600 +17* -26* +5* Table I I I ( D ) . Average number of re fe rra ls at la s t screening In 1978 fo r long-term e lig ib le Spanishspeaking by age and number of life tim e screenings. Number of Lifetime Screenings 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.000 Grand Hean % Change As No. Screenings In­ crease By One - - 1.000 .333 .333 .833 1.200 .500 .600 1.000 1.333 .833 - 2.000 1.000 .500 - .811 2 3 4 5 - - - - - * - - - - - - - 0 - - - - 0 1.200 .933 .923 1.416 .800 .285 .571 .800 .708 .842 .611 1.200 .250 0 - .923 1.000 .842 .444 .857 .466 .615 1.000 .941 .600 1.000 1.000 2.000 2.000 1.000 1.000 .250 .333 0 1.000 1.000 1.000 1.000 .500 1.000 .799 .794 .666 1.142 -1 * -0% -16% +72% - - 1.000 0 - 1.500 0 - 1.000 - 2.000 - 1.000 - - - - 238 Age Table I I I ( E ) . Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le males by age and number of life tim e screenings. Number of Lifetime Screenings Age Grand Mean * Change As No. Screenings In­ creased By One 2 3 4 5 6 7 _ .333 .444 1.166 1.727 1.040 1.039 .852 .919 .944 1.030 1.158 .929 1.043 .858 1.113 .895 1.236 1.142 1.833 1.000 .250 .333 .892 1.128 .817 .978 .858 .905 .890 .804 .821 .812 .894 .978 .957 1.076 2.000 .500 .571 .894 .760 .766 .854 .870 .704 .771 .842 .992 .898 .986 .843 .827 3.000 1.000 1.125 .593 1.000 .609 .707 .520 .894 .909 .578 .647 1.200 .571 1.166 .500 1.000 1.005 .886 .835 .754 .906 1.000 6X -10* +20* +10* - -13* - - - - - - - - - - - - - - - - - - - - - - - - 2.000 .500 1.000 1.333 .400 1.000 .750 0 - 2.000 1.500 .500 - 1.000 - - 1.000 - - - - - - - - - - - - - - - - - - - - - 239 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I I I ( F ). Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le females by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .500 0 1.000 .250 .875 .927 1.180 .931 1.062 1.030 .944 .958 1.146 1.074 1.104 1.269 1.250 1.533 1.428 1.250 .500 0 .666 .500 .793 .859 .913 .865 .887 .820 .831 .821 .862 .943 .927 1.071 1.066 1.155 1.111 0 .937 .920 .890 .913 .831 .827 .789 .870 .764 .904 .993 1.000 1.100 .933 1.933 1.000 .750 .871 .791 .500 .576 .857 1.034 .758 .482 .857 .615 1.000 1.375 1.000 1.000 1.066 .882 .901 .776 .805 -17% +2% -14% +4% % Change As No. Screenings In­ crease By One 3 * - * 4 5 _ _ - - - - - - - - .700 0 .666 .500 .500 .800 1.000 1.000 1.000 2.000 - 1.000 - 240 Grand Mean 2 Table I I I ( G ) . Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le white males by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Grand Kean % Change As No. Screenings In­ crease By One 1 2 3 4 5 _ .500 0 1.000 1.571 1.074 1.071 .814 .983 .800 .976 .985 .637 .913 .820 1.026 .928 .857 1.250 1.000 1.000 .333 .166 .878 1.132 .830 .881 .776 .870 .838 .800 .870 .814 .842 .909 1.136 1.000 2.000 1.000 .750 .666 .814 .806 .853 .876 .681 .682 .620 1.000 .833 .968 .722 1.250 2.000 .917 .859 .797 .692 .823 -6% -7% -13% +19% - - - - - - - - - 0 .600 .944 .650 .689 .352 .777 .882 .666 .555 .857 .555 1.333 1.000 - - 1.000 1.000 1.333 .333 - .666 0 - 1.500 0 - - - - - Table II1 (H ). Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le white females by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 0 0 • _ _ Grand Mean % Change As No. Screenings In­ crease By One 0 1.000 1.000 1.192 1.024 .819 .714 .746 .853 .983 .982 .964 1.152 1.375 .944 1.500 0 0 0 .666 .625 .723 .828 .806 .729 .743 .701 .686 .842 .767 .813 .835 1.120 .933 .416 1.000 1.000 .967 1.333 .771 .744 .650 .694 .701 .796 .763 .890 .921 1.190 .714 2.800 1.000 .946 .777 .833 - -18% 5 - - - - - - - - - +Q% - 1.055 .750 .388 .625 .714 .642 .588 .222 .769 .400 .666 1.666 - - .666 - .500 0 .666 1.000 .500 1.000 1.000 3.000 - - - - - .671 .809 -19% +20% 242 Age Table I I I (1 ). Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le black males by age and number of life tim e screenings. Number of Lifetime Screenings Age Grand Mean % Change As No. Screenings In­ creased By One 2 3 4 5 6 7 __ 0 .500 1.333 2.000 1.026 1.065 .875 .862 1.074 1.115 1.333 1.368 1.138 .913 1.195 .842 1.379 1.000 2.250 0 .625 .861 1.123 .800 1.052 .906 .963 .953 .809 .789 .858 1.024 1.050 .931 1.250 - .333 1.061 .698 .721 .882 .865 .746 .912 .971 .955 .925 .974 .829 .666 4.000 1.000 1.285 .583 1.111 .600 .818 .875 1.000 .923 .444 .714 1.500 .600 1.000 1.000 - 1.097 .916 .867 .846 .888 1.000 -16* -5% -2* +5* +13% 2.000 0 1.000 1.000 1.000 . 1.000 0 1.000 1.000 - - _ 1.000 - 243 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I I I ( J ) . Average number of re fe rra ls at la s t screening in 1978 fo r long-term e lig ib le black females by age and number of life tim e screenings. Number of Lifetime Screenings Age Grand Mean 56 Change As No. Screenings In­ crease By One - 1.000 .400 .500 .909 1.173 .898 1.290 1.278 1.207 1.081 1.285 1.177 1.265 1.390 1.156 1.923 1.384 1.304 1.186 2 1.000 - 3 4 5 - - - - - - - - .166 .814 .903 1.022 .986 1.040 .965 1.015 .815 .946 1.053 1.009 1.023 1.145 1.424 1.117 0 .888 .840 .716 1.054 1.000 1.018 .904 1.024 .730 1.033 1.138 1.062 1.106 .954 1.375 1.000 .750 .700 .925 .727 .555 1.000 1.500 1.100 .555 1.200 .714 1.222 1.250 1.000 1.000 .978 .972 .901 .785 -1856 -056 -716 -1256 - - - .750 0 1.000 1.000 0 0 1.500 - 1.000 - 1.000 - 244 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I I I ( K ) . Average lumber of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le American Indian males by age and number of life tim e screenings. Number of Lifetime Screenings Age Grand Mean % Change As No. Screenings In­ crease By One 2 3 4 _ _ - - - - - - - - - - - - - - - - - 1.000 0 - - - - - 0 - - - - - - - 2.000 0 1.000 - - - - - 0 - - - - 0 - 0 - 1.500 - - 1.000 0 0 .500 - - - - - - - - - - - - - - - - - - - .600 .625 .333 0 +4% -47% -100* 245 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table I I I ( L ) . Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le American Indian females by age and number of life tim e screenings. Number of Lifetime Screeninqs 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 - - Grand Mean • - 4 - - - 0 0 0 1.000 1.000 2.000 - 2.000 .500 2.000 1.500 0 1.000 .500 1.000 0 - 0 .500 2.000 - * 0 1.000 1.000 - .714 .857 .750 .750 +20* 0* - - a* 1 % Change As No. Screenings In­ crease By One 3 246 1 CM Age Table I11(H ). Average number of re fe rra ls at la s t screening in 1978 fo r long-term e lig ib le Spanish speaking males by age and number of life tim e screenings. Number of Lifetime Screenings Age ___ Grand Mean % Change As No. Screenings In­ crease By One 2 - 3 - 4 - 5 - - - - - - - - 1.000 .333 1.250 0 1.000 .500 1.000 0 1.000 1.000 .500 - 0 1.142 1.142 1.000 1.833 1.200 .250 .800 .800 .666 .400 .363 0 . .250 0 - • 0 .800 .750 .900 .727 .833 .555 .666 1.200 1.125 1.500 1.000 1.400 _ - .800 .755 .884 .714 1.166 -6% +17% -19% +63% - - 0 0 1.500 1.000 1.000 - 0 - - 1.500 0 - 1.000 2.000 1.000 - 247 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 - Table 1 II(N ). Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le Spanish­ speaking females by age and number of life tim e screenings. Number of Lifetime Screenings 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.000 Grand Mean % Change As No. Screenings In­ crease By One 2 3 4 5 - - - - - - - - - - - - - - _ . _ _ - 1.000 0 - - - - - 0 1.500 0 .666 1.000 1.600 .500 - 2.000 - .821 1.000 1.166 .777 0 1.000 .333 .500 .833 .777 0 1.000 0 2.000 2.000 1.000 1.000 .250 .500 0 1.000 1.000 .500 1.000 0 1.000 .834 .698 .650 1.000 +2* -16% -7% +54% 1.250 .750 .875 1.000 .600 .300 .444 .800 .750 1.333 1.000 1.500 .250 - - 1.000 - - 1.000 - - - - - 248 Age Table 111(0). Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le participants in D etro it by age and number of life tim e screenings. Nunfcer of Lifetime Screenings 2 3 - - - - - - - - - - - - 4 5 Age 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.000 1.000 1.000 1.000 1.277 1.042 1.107 1.220 1.070 1.372 1.480 1.490 1.462 1.375 1.444 1.454 1.760 1.363 1.214 .333 .843 1.285 .945 1.146 1.070 1.315 1.016 1.083 1.072 1.083 1.128 1.442 1.200 1.785 1.333 .500 1.000 .958 1.000 1.434 1.312 1.095 1.333 1.050 1.500 1.333 1.733 1.214 1.166 Grand Mean 1.318 1.116 1.216 .980 1.000 -15% +9% -19% +2% - 1.000 - 2.000 0 - 1.000 - 1.000 - 1.000 1.000 - 1.000 2.000 - - - - - 249 % Change As No. Screenings In­ creased By One - 1.333 .500 .600 1.000 .666 1.500 1.400 1.500 1.000 .333 1.000 Table I1 I(P ). Average number of re fe rra ls a t la s t screening in 1978 fo r long-term e lig ib le participants in fo rty -fo u r Northern Michigan counties by number of life tim e screenings. 250 Number of Lifetime Screeninqs 1 Mean % Change As No. Screenings In­ crease By One .817 2 3 4 5 .769 .789 .515 .882 -6% +3% -35% +71* APPENDIX D t Table 111(a). Number of long-term e lig ib le s screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings Age Total 2 3 4 5 6 7 Total 2 4 12 20 27 229 149 217 240 226 224 312 222 260 218 168 104 83 28 30 0 2 3 7 43 480 313 495 474 495 552 558 515 529 434 379 290 37 58 19 0 0 0 3 30 281 269 398 316 263 287 303 246 252 251 164 134 59 17 16 0 0 0 0 12 71 75 73 67 53 48 62 48 38 28 29 14 6 3 0 0 0 0 0 1 12 5 9 7 5 9 6 2 2 4 2 2 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 2 6 15 30 113 1143 811 1192 1104 1043 1120 1241 1033 1081 935 742 544 187 106 65 2,845 5,683 3,289 627 68 0 1 12,513 251 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table III( A a ) . Number of long-term e lig ib le whites screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings Age Total 2 3 4 5 Total 1 3 1 9 18 138 54 95 121 90 no 160 120 126 123 84 52 25 12 3 0 1 3 6 28 193 132 219 238 234 255 273 232 212 193 167 113 37 16 2 0 0 0 1 15 115 126 181 168 133 128 149 109 105 112 70 39 15 6 2 0 0 0 0 1 38 34 38 45 31 23 34 18 22 12 15 6 1 0 0 0 0 0 0 0 7 3 5 4 3 6 3 2 1 3 1 0 0 0 0 1 4 4 16 62 491 349 538 576 491 522 619 481 466 443 337 210 78 34 7 1*345 2,554 1,474 318 38 5,729 252 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table III( B a ) . Number of long-term e lig ib le blacks.screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 Total 0 1 11 11 8 154 92 115 113 128 105 146 94 127 95 82 51 55 16 27 0 1 0 1 14 268 164 261 221 243 281 260 261 288 218 191 165 99 41 17 0 0 0 1 15 151 130 196 130 122 144 140 123 128 126 87 88 43 9 13 0 0 0 0 11 32 36 31 20 19 22 23 27 12 15 14 7 4 3 0 0 0 0 0 1 5 2 2 2 2 1 3 0 0 1 1 2 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 2 11 13 49 610 424 605 486 515 553 572 505 555 455 375 313 202 69 57 1,431 2,994 1,646 276 23 0 1 6,371 Number of long-term e lig ib le American-Indians screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 2 3 4 Total 0 0 0 0 0 1 0 1 1 1 2 2 2 0 0 1 0 1 0 0 0 0 0 0 0 2 2 0 1 2 1 3 1 3 3 2 2 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 2 1 2 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 1 0 2 0 0 0 0 0 0 0 0 0 0 0 4 3 2 2 3 4 6 5 6 5 3 2 1 0 0 12 22 7 5 46 IN) cn Number o f long-term e lig ib le Spanish-speaking screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings 2 3 4 5 Total 1 0 0 0 1 6 3 6 5 6 5 4 6 6 0 1 1 2 0 0 0 0 0 0 1 15 15 13 12 15 14 14 20 24 19 18 10 8 1 0 0 0 0 1 0 13 10 19 18 7 15 13 11 17 10 6 7 1 2 1 0 0 0 0 0 1 4 3 2 3 2 4 3 2 1 0 1 1 0 0 0 0 0 0 0 0 0 2 1 0 2 0 0 1 0 0 0 1 0 0 1 0 0 1 2 35 32 43 38 31 38 35 40 50 30 25 19 13 3 1 53 199 151 27 7 437 255 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 1 Table I I I( E a ) . Number of long-term e lig ib le males screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 Total 0 3 9 12 11 147 77 115 112 126 99 145 99 139 113 79 48 38 7 6 0 0 2 4 21 252 156 230 228 247 274 273 246 252 203 171 137 70 13 1 0 0 0 2 14 142 150 214 144 147 149 149 127 127 108 73 64 29 2 3 0 0 0 0 8 32 27 41 41 25 19 33 19 17 15 14 6 2 2 0 0 0 0 0 1 2 4 6 5 1 4 2 0 1 2 2 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 3 11 18 55 575 414 606 530 547 545 602 491 536 441 339 255 141 24 10 1,385 2,780 1,644 301 32 0 1 6,143 Table I I I ( F a ). Number of long-term e lig ib le females screened in 1978 by age and number of life tim e screenings. Nunber of Lifetime Screenings Age Total 2 3 4 5 Total 2 1 3 8 16 152 72 102 128 100 125 167 123 121 105 89 56 45 21 24 0 2 1 3 22 228 157 265 246 248 278 285 269 277 231 208 153 75 45 18 0 0 0 1 16 139 119 184 172 116 138 154 119 125 143 91 70 30 15 13 0 0 0 0 4 39 48 32 26 28 29 29 29 21 13 15 8 4 1 0 0 0 0 0 0 10 1 3 2 4 5 4 2 1 2 0 2 0 0 0 2 3 4 12 58 568 397 586 574 496 575 639 542 545 494 403 289 154 e.2 55 1,460 3,011 1,645 326 36 6,478 i 257 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Humber o f long-term e lig ib le white males screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 1 2 3 4 5 Total 0 2 1 6 7 67 28 54 60 55 43 71 58 69 67 38 28 7 4 2 0 0 2 3 12 99 68 100 127 121 131 136 105 100 86 76 55 22 4 1 0 0 0 1 8 54 81 98 82 73 69 82 50 50 48 32 18 8 1 0 0 0 0 0 1 20 18 20 29 17 9 17 9 9 7 9 3 1 0 0 0 0 0 0 0 1 3 3 3 0 3 1 0 0 2 1 0 0 0 0 0 2 3 10 28 241 198 275 301 266 255 307 222 228 210 156 104 38 9 3 667 1,248 755 169 17 2,856 Number of long-term e lig ib le white females screened in 1978 by age and number of life tim e screenings. t 1 2 3 4 5 6 7 10 11 12 13 14 15 16 17 18 19 2 3 4 5 Total 1 1 0 3 11 71 26 41 61 35 67 89 62 57 56 46 24 18 8 1 0 1 1 3 16 94 64 119 111 113 124 137 127 112 107 91 58 15 12 1 0 0 0 0 7 61 45 83 86 60 59 67 59 55 64 38 21 7 5 2 0 0 0 0 0 18 16 18 16 14 14 17 9 13 5 6 3 0 0 0 0 0 0 0 0 6 0 2 1 3 3 2 2 1 1 0 0 0 0 0 1 2 1 6 34 250 151 263 275 225 267 312 259 238 233 181 106 40 25 4 678 1,306 719 149 21 2,873 259 8 9 1 Table 11 1 (la ). Number of long-term e lig ib le black males screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 Total 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 0 1 8 6 4 77 46 56 51 67 52 72 38 65 46 41 19 29 3 4 0 0 0 1 8 144 81 125 95 118 138 129 131 138 106 83 80 44 8 0 0 0 0 0 6 82 63 104 51 67 71 57 71 68 54 39 41 21 1 3 0 0 0 0 7 12 9 20 11 8 10 13 9 7 8 5 3 0 2 0 0 0 0 0 1 1 1 1 1 1 0 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 8 7 26 316 200 306 209 262 271 272 249 278 214 169 143 95 14 7 Total 685 1,429 799 124 9 0 1 3,047 260 Age Number of long-term e lig ib le black females screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 1 2 3 4 5 Total 0 0 3 5 4 77 46 59 62 61 53 74 56 62 49 41 32 26 13 23 0 1 0 0 6 124 83 136 126 125 143 131 130 150 112 108 85 55 33 17 0 0 0 1 9 69 67 92 79 55 73 83 52 60 72 48 47 22 8 10 0 0 0 0 4 20 27 11 9 11 12 10 18 5 7 9 4 4 1 0 0 0 0 0 0 4 1 1 1 1 1 2 0 0 1 0 2 0 0 0 0 1 3 6 23 294 224 299 277 253 282 300 256 277 241 206 170 107 55 50 746 1,565 847 152 14 3,324 Nuidber of long-term e lig ib le American Indian males screened 1n 1978 by age and number of life tim e screenings. Number of life tim e Screenings 1 2 12 13 14 15 16 17 18 19 20 3 4 Total 0 0 0 0 0 1 0 1 0 0 1 0 2 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 2 1 1 0 1 1 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 1 0 2 2 2 2 1 3 1 0 0 0 0 5 8 3 1 17 262 3 4 5 6 7 8 9 10 11 2 Number of long-term e lig ib le American Indian females screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 2 3 4 Total 0 0 0 0 0 0 0 0 1 1 1 2 0 0 0 1 0 1 0 0 0 0 0 0 0 1 2 0 1 0 0 2 1 2 2 1 2 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 2 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 2 0 0 0 0 0 0 0 0 0 0 0 1 3 1 2 1 2 4 3 5 2 2 2 1 0 0 7 14 4 4 29 i\> o CO Number of lonq-term e lig ib le Spanish-speaking males screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 Total 0 0 0 0 0 2 3 4 1 3 2 2 1 4 0 0 1 2 0 0 0 0 0 0 1 7 7 5 6 5 4 5 10 12 10 11 2 4 1 0 0 0 0 1 0 5 4 10 11 6 9 9 5 8 4 1 5 0 0 0 0 0 0 0 0 0 0 1 1 0 0 2 1 1 0 0 0 1 0 0 0 0 0 0 0 0 0 2 1 0 1 0 0 1 0 0 0 1 0 0 0 0 0 1 1 14 14 22 20 14 16 18 17 26 14 12 8 8 1 0 25 90 78 7 6 206 Number of long-term e lig ib le Spanish-speaking females screened fin 1978 by age and number o f life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 1 2 3 4 5 Total 1 0 0 0 1 4 0 2 4 3 3 2 5 2 0 1 0 0 0 0 0 0 0 0 0 8 8 8 6 10 10 9 10 12 9 7 8 4 0 0 0 0 0 0 0 8 6 9 7 1 6 4 6 9 6 5 2 1 2 1 0 0 0 0 0 1 4 2 1 3 2 2 2 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 1 21 18 21 18 17 22 17 23 24 16 13 11 5 2 1 28 109 73 20 1 231 Number of long-term e lig ib le participants screened in D etro it in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 1 2 3 4 5 Total 0 1 1 3 6 83 47 56 59 71 59 77 55 67 48 36 22 25 11 14 0 0 0 0 3 83 56 74 89 71 95 118 96 96 72 78 52 35 14 6 0 0 0 0 2 26 24 28 23 16 21 24 20 22 15 15 14 6 0 3 0 0 0 0 4 5 9 6 4 5 4 6 3 3 0 1 0 1 0 0 0 0 0 0 1 1 0 I 0 0 1 0 0 1 1 0 1 0 0 0 0 1 1 3 16 198 136 165 175 163 180 225 174 189 136 130 89 67 25 23 741 1,038 259 51 7 2,096 Table Ill( P a ) . Number of long-term e lig ib le participants screened in 1978 in fo rty -fo u r Northern Michigan counties by number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 Total 225 403 252 97 17 994 APPENDIX E Table IV. Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le s by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Grand Mean % Change As No. Screenings In­ creased By One 1 2 3 4 5 6 7 _ _ _ 0 .756 .722 .771 .800 .982 .991 1.029 .957 .978 .973 .996 1.020 1.071 1.019 1.098 1.202 1.175 1.347 1.295 1.342 .755 .619 .675 .826 .854 .820 .915 .876 .867 .878 .872 .872 .892 .881 .896 .941 1.013 1.000 1.128 1.274 1.000 .640 .555 .810 .798 .822 .864 .858 .829 .884 .790 .824 .785 .956 .930 1.007 1.010 .943 1.800 1.162 1.285 1.136 .859 .805 .858 .661 .740 .734 .905 .847 .594 .833 .785 .783 1.150 .666 1.000 .500 2.000 1.000 1.000 .947 .866 1.235 .375 .714 .750 .800 .750 1.500 1.750 .666 1.000 .666 .933 .869 .859 .800 .923 1.500 1.000 -7% -1* +15% +63% -33% - - - - - - - - - 7% - 2.000 1.000 - - - 1.000 - - - - - - - - - - - - - - - - - - - Table IV (A ). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le whites by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .716 .657 .655 .789 .988 .955 .974 .941 .946 .883 .942 .918 .925 .948 1.008 1.053 1.142 1.100 1.200 1.296 .695 .573 .628 .813 .875 .816 .921 .879 .805 .838 .793 .807 .905 .895 .870 .925 1.003 1.041 1.000 1.189 .608 .420 .680 .786 .743 .925 .858 .816 .825 .749 .750 .732 .901 .888 .984 1.055 .933 2.583 1.500 1.600 .727 .674 .807 .867 .647 .742 .625 .658 .795 .531 .718 .521 .700 1.250 .333 1.500 .874 .844 .824 ,731 .941 1.500 -3% -2% -11% +29% +60% % Change As No. Screenings In­ crease By One - - 5 6 _ _ - - - - - 1.000 1.200 .888 1.230 .250 .800 .625 .714 .666 1.000 2.000 0 - 2.000 - 1.000 - - - - - - - - - 269 Grand Mean 4 Table IV (8 ). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le blacks by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 7 5 - _ - - - - - - - - - - 0 .809 .809 .940 .829 .991 1.059 1.119 1.034 1.051 1.136 1.100 1.216 1.294 1.147 1.235 1.401 1.231 1.581 1.407 1.386 .791 .698 .741 .853 .825 .833 .912 .871 .939 .945 .997 .987 .879 .886 .944 1.007 1.016 1.017 1.239 1.277 1.000 .666 .743 .955 .792 .910 .792 .888 .903 .972 .873 .955 .842 1.006 1.005 1.042 .990 .928 1.187 1.076 .500 1.700 1.117 .744 .920 .738 .769 .923 1.333 .892 .580 1.055 1.111 .933 1.100 1.000 .750 1.000 2.000 1.000 1.000 .750 .800 1.000 1.000 .500 0 1.000 1.000 Grand Mean 1.029 .912 .913 .923 .871 -11* 056 +156 -656 - 1.000 1.000 1.000 .500 _ - - - - - - - - - - - - - 1.000 - - - - - - - - - - - - - - - - - - - - - - - 1.000 +1556 270 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 56 Change As No. Screenings In­ creased By One 6 4 Table IV (C ). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le American Indians by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1.200 .166 .833 .875 1.000 1.285 .750 0 .888 1.200 .333 1.000 1.500 1.000 .666 1.000 1.000 2.000 1.000 Grand Mean % Change As No. Screenings In­ crease By One - .910 2 3 4 • • _ .857 .200 1.000 .500 .857 1.000 1.400 1.833 .600 .333 1.000 .166 .833 .250 .250 .333 1.000 2.000 - - - - - 2.500 0 1.666 .500 .600 0 2.000 0 .500 .666 1.000 .500 0 - 2.000 - 0 2.000 - 1.000 1.333 3.000 1.000 - - - * - * - 1.000 - .741 .833 1.416 -19% +12% +70% Table IV (D ). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le Spanish­ speaking by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .577 .658 .956 .629 .609 .947 .960 .628 .789 .827 .925 .727 .931 .827 1.384 1.727 1.000 .833 2.000 1.272 1.000 .315 .789 .704 .900 .711 .820 .840 1.000 .723 .714 .760 .800 .733 .842 .588 1.090 .500 .714 2.666 _ _ - - - 0 .571 1.125 .740 .785 .724 .535 .625 .615 .562 .789 1.000 .529 1.000 .909 1.500 2.000 1.000 - - .816 .780 .752 .606 1.000 -456 -19% +65% Grand Mean % Change As No. Screenings In­ crease By One -456 0 0 .888 .500 .428 .428 .800 .500 .800 .600 .500 1.000 .500 1.000 .500 5 - 0 1.000 1.500 0 - 1.333 - 2.000 - 1.000 - - - - Table IV (E ). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le males by age and nunfcer of life tim e screenings. Number of Lifetime Screenings 7 _ 1.000 - .792 1.035 1.000 1.000 -4% +31% -3% 0% 1 2 3 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .729 .722 .780 .821 1.035 1.043 1.040 .975 .967 .968 . .977 1.094 1.000 .969 1.015 1.218 1.095 1.282 1.250 1.75 .966 .596 .688 .891 .892 .895 .970 .826 .912 .901 .887 .890 .863 .836 .835 .957 1.027 .967 .928 1.333 1.000 .708 .428 .818 .799 .807 .829 .842 .856 .876 .727 .803 .791 .950 .827 .916 .926 .809 3.000 1.000 _ .400 1.416 .976 .714 .963 .703 .716 .636 .685 .934 .642 .619 1.142 ,562 1.000 .333 1.000 - .922 .877 .828 -5* -6* Grand Mean % Change As No. Screenings In­ creased By One 5 273 2.000 1.250 1.500 1.000 1.384 .333 1.000 .750 0 1.000 2.000 1.500 .666 .666 - 6 _ 1.000 - Age Table IV (F ). Average number o f re fe rra ls a t la s t screening In 1978 fo r one-year e lig ib le females by age and number of life tim e screenings. Number of Lifetime Screenings Age Grand Mean % Chnage As No. Screenings In­ crease By One 3 4 5 - 6 - - - - - - - - 0 .783 .722 .763 .777 .930 .938 1.019 .938 .990 .979 1.014 .953 1.136 1.073 1.184 1.188 1.244 1.381 1.300 1.327 2 * .421 .642 .662 .764 .814 .741 .862 .925 .826 .857 .855 .855 .917 .927 .948 .927 1.000 1.024 1.171 1.272 .576 .666 .801 .797 .837 .899 .875 .804 .893 .851 .843 .778 .963 1.014 1.082 1.094 1.065 1.714 1.176 3.500 .800 .714 .901 .780 .611 .772 .857 1.102 .743 .561 .969 .428 .952 1.272 .833 1.000 .500 1.000 .800 .692 .714 .750 .500 .600 .750 1.142 .666 1.000 2.000 1.000 - .944 .861 .888 .809 .819 -9% +3% -9% — - - +1% - 2.000 - - - 2.000 +144% 274 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table IV(G). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le white males by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .658 .666 .674 .807 1.004 .990 .996 .975 .986 .922 .949 .992 .836 .905 .955 1.080 1.088 1.028 1.285 1.500 1.000 .556 .601 .876 .898 .916 1.000 .881 .869 .893 .842 .837 .864 .852 .870 .994 .984 1.255 1.111 2.000 .583 .280 .750 .785 .681 .866 .922 .849 .833 .642 .753 .688 .942 .759 .916 .885 .933 2.000 .873 .875 .794 .739 0% -9% -n Grand Mean % Change As No. Screenings In­ crease By One 3 - 4 5 • _ - - - - - - - .333 1.200 .777 .744 .885 .769 .714 .533 .619 .963 .705 .538 .750 .500 1.000 .500 1.000 - 0 2.333 1.000 1.400 .333 1.000 .666 0 6 - 1.000 - - - - - 1.500 0 - - - - - - - - 1.029 1.000 +39% -3 * Table IV(H ). Average number o f re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le white females by age and number of life tim e screenings. Number of Lifetime Screenings Aqe 1 2 3 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .783 .647 .637 .771 .972 .922 .952 .900 .904 .839 .936 .850 1.004 .995 1.069 1.028 1.198 1.133 1.189 1.287 .363 .592 .654 .748 .849 .709 .846 .876 .747 .787 .735 .779 .936 .937 .870 .855 1.023 .870 .977 1.166 .636 .560 .596 .787 .806 .993 .790 .789 .818 .854 .746 .771 .863 .990 1.042 1.216 .933 2.636 1.500 3.500 .333 .600 .875 .848 .482 .785 .777 .700 .590 .333 .842 .272 .900 1.500 0 2.000 .876 .814 .853 .721 .852 2.000 -7% +5% -15% +18% +135% % Change As No. Screenings In­ crease By One - 6 - - - - - - 2.000 .714 .500 .666 0 .750 .600 1.000 .666 1.000 3.000 - - 2.000 - - - - - - - - - 276 Grand Mean - 5 Table I V ( I ) . Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le black males by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .851 .792 .919 .867 1.147 1.140 1.132 1.017 .958 1.064 1.049 1.310 1.260 1.102 1.117 1.406 1.055 1.480 1.166 2.000 .937 .669 .798 .947 .872 .880 .925 .744 .959 .900 .954 .976 .868 .832 .821 1.007 1.043 .849 .875 1.000 1.000 .833 .647 .902 .766 .977 .804 .770 .904 .936 .838 .909 .892 .913 .884 .943 .905 .740 4.000 1.000 1.013 .889 .873 -2% Grand Mean I Change As No. Screenings In­ creased By One -12% 4 5 6 7 _ .500 1.833 1.166 .636 1.250 .652 .800 .846 .833 .933 .500 .714 1.666 .600 1.000 1.000 - 2.000 - 2.500 1.000 1.000 1.000 1.000 1.00 0 1.000 - - - - - - - - - 1.000 .500 - - 1.000 - .9191 1.133 - 1.000 +5% +23% -12% Table IV (J ). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le black females by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 4 5 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 0 .769 .827 .963 .788 .860 .967 1.109 1.050 1.138 1.217 1.145 1.132 1.325 1.197 1.349 1.396 1.350 1.640 1.426 1.363 .500 .726 .681 .768 .778 .783 .898 .991 .992 .990 1.041 1.000 .891 .941 1.052 1.006 .992 1.144 1.315 1.285 .533 .818 1.000 .816 .830 .782 1.014 .902 1.011 .904 .990 .777 1.097 1.106 1.123 1.074 1.103 1.000 1.087 1.500 1.000 .840 .764 .842 .727 1.000 1.733 .846 .619 1.272 .555 1.100 1.200 1.000 .500 1.000 1.000 0 .666 .750 1.000 1.000 0 0 1.500 Grand Mean 1.045 .932 .950 .927 .708 -in +2* -2% -24% - - - - - - - - 1.000 - 1.000 - 278 % Change As No. Screenings In­ crease By One _ _ Table IV(K). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le American Indian males by age and nunfcer of life tim e screenings. Number of Lifetime Screenings 1 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 0 .500 .666 .714 1.500 1.200 .571 0 1.333 1.000 .500 1.000 1.333 0 .500 _ Grand Mean % Change As No. Screenings In­ crease By One - 1.000 0 0 .500 .333 0 1.500 1.666 .666 .400 .666 .333 .750 0 0 1.000 1.000 3 4 - - - - 2.500 0 1.500 .500 .333 0 2.000 0 1.000 1.000 0 .500 0 - 1.000 0 - * - - - - - - - - - - - .775 .619 .708 .500 -20% +14% -29% 279 Age Table IV (L ). Average number o f re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le American Indian females by age and number of life tim e screenings. Nunber of Lifetime Screenings 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 2.000 0 1.333 2.000 .750 1.500 2.000 Grand Mean % Change As No. Screenings In­ crease By One - .666 1.333 0 1.000 2.000 1.333 1.000 1.000 1.000 2.000 1.000 - 1.075 2 3 4 - - _ .800 .250 2.000 .500 1.250 1.000 1.333 2.000 .500 0 1.500 0 1.000 .333 .500 0 - 2.000 - - 2.500 0 2.000 .500 1.000 - 2.000 0 0 .500 1.500 - 2.000 - 0 2.000 - 1.000 2.000 3.000 1.000 - - - - - - - - - - 1.000 - .851 1.000 1.600 -21% +18% +60% 280 Age Table IV(M). Average number of re fe rra ls a t ld s t screening in 1978 fo r one-year e lig ib le Spanishspeaking males by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .333 .700 1.087 .516 .640 .900 .941 .842 .642 .857 .833 .750 .692 .642 3.000 1.500 1.000 1.000 1.000 .416 .875 .550 1.105 .636 .944 .666 1.000 .700 .958 .818 .666 .708 .611 .300 1.250 .400 .500 — _ _ - - - - - Grand Mean % Change As No. Screenings In­ crease By One - .763 - - .600 1.666 .571 .600 .812 .750 .727 .785 .600 .777 1.272 1.000 .750 1.285 0 0 .800 .333 0 .333 1.000 0 1.000 1.000 1.000 - 1.000 - 5 - 0 - 1.500 0 - 1.000 - 2.000 - 1.000 - 0 - - - - - - .741 .829 .541 .875 -3* +12% -35% +62% Table IV(N ). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le Spanish­ speaking females by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 .791 .619 .826 .741 .562 1.000 1.000 .375 .875 .800 1.000 .714 1.125 1.000 1.090 1.857 1.000 .714 2.000 1.272 - .142 .727 .833 .714 .766 .714 1.000 1.000 .740 .388 .714 .952 .761 1.050 1.000 1.000 .555 .800 2.666 0 .500 .800 .923 .888 .615 .250 .400 .416 .500 .800 .727 0 1.090 .250 1.500 2.000 .862 .815 .673 .648 1.333 -5% -17% -4% +106% Grand Mean % Change As No. Screenings In­ crease By One 3 4 5 • . _ - - - - - 1.000 - - - - 1.000 .600 .600 .500 .750 .666 .500 .500 0 1.000 0 1.000 - 1.000 - 1.500 - - 1.000 - - - - - Table IV {0 ). Average nunber of re fe rra ls a t la s t screening in 1978 fo r one-year participants in D etro it by age and number of life tim e screenings. Number of Lifetime Screenings Age 1 2 3 4 • - _ 5 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 0 .689 .821 .828 .784 .912 1.142 1.025 1.103 1.168 1.083 1.216 1.383 1.427 1.323 1.344 1.593 1.469 1.569 1.513 1.397 .750 1.067 .738 .851 .784 .800 1.040 .943 1.072 1.043 1.303 1.090 1.125 1.057 1.083 1.221 1.437 1.093 1.571 1.291 Grand Mean 1.057 1.052 1.186 1.150 1.214 02 +132 -32 +62 2 Change As No. Screenings In­ crease By One - 1.116 .615 1.166 1.000 1.051 1.057 1.219 1.285 1.458 1.071 1.266 1.000 1.320 1.291 1.526 1.294 1.428 - 1.333 - .500 2.250 .666 .875 1.181 .714 1.800 1.833 1.500 1.000 .750 1.250 - 1.000 - 2.000 0 - - 2.000 1.000 1.250 1.000 - 1.000 - 1.500 - 1.000 1.000 - 1.000 - Table IV (P ). Average number of re fe rra ls a t la s t screening in 1978 fo r one-year e lig ib le participants in fo rty -fo u r Northern Michigan counties by number of life tim e screenings. 284 % Change As No. Screenings In­ crease By One 2 3 4 5 .768 .788 .761 .652 .812 +25% +3% 1 Mean 1 1 I*-* ** Number of Lifetime Screenings APPENDIX F Table IV (a ). Number of one-year e lig ib le s screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings Age Total 2 3 4 5 6 7 Total 1 1250 2617 1718 1770 1387 1381 904 1032 991 903 826 940 701 764 642 493 359 265 220 333 0 49 496 588 1108 1197 1291 878 1160 1128 1063 1075 1094 968 939 800 687 532 284 156 113 0 1 50 90 232 426 601 545 672 550 450 473 456 387 371 372 266 190 88 30 37 0 0 0 7 22 78 141 127 118 104 79 74 85 69 54 42 37 20 9 6 2 0 0 0 1 1 9 19 15 17 8 7 12 10 4 2 4 3 3 3 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 1300 3163 2404 3133 3097 3434 2469 2999 2782 2503 2460 2585 2129 2130 1860 1486 1104 649 412 485 19,497 15,606 6,287 1,074 118 2 1 42,585 285 Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table IV(Aa). Number of one-year e lig ib le whites screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings Aqe 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 Tota 1 610 1381 1013 1019 881 859 551 612 584 540 487 563 405 445 373 279 203 no 125 152 23 258 323 574 737 737 512 662 659 601 585 614 529 469 433 360 256 97 54 37 0 23 50 125 239 323 310 347 322 247 247 256 198 182 180 130 72 30 12 8 0 0 5 11 43 83 68 68 70 48 41 49 32 32 23 20 8 3 2 0 0 0 0 0 4 10 9 13 4 5 8 7 3 1 3 1 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 633 1662 1391 1729 1904 2013 1450 1702 1640 1441 1368 1489 1167 1129 1012 790 539 240 193 197 11,192 8,520 3,301 606 68 2 23,689 Table IV(Ba). Number of one-year e lig ib le blacks screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 Total 1 584 1125 633 668 446 468 317 374 351 323 299 332 255 279 242 197 134 141 81 163 0 24 209 240 485 407 491 319 445 414 404 437 411 383 414 323 286 249 170 92 72 0 1 27 39 89 164 245 199 286 197 182 • 198 181 165 163 172 118 107 56 16 26 0 0 0 2 10 34 47 50 42 26 26 27 28 31 18 18 15 10 4 4 1 0 0 0 1 1 5 8 5 2 2 2 1 3 1 0 1 2 3 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 609 1361 915 1253 1056 1259 889 1149 990 938 962 955 835 874 756 618 503 373 193 262 7,413 6,275 2,630 393 39 0 1 16,751 Number of one-year e lig ib le American Indians screened in 1978 by age and number of life tim e screenings. Number of Lifetime Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 2 3 4 Total 5 6 12 8 6 7 8 1 9 5 3 3 4 4 3 1 1 1 2 0 0 7 5 4 10 7 4 5 6 5 6 5 6 6 4 4 3 1 1 0 0 0 0 4 4 3 6 5 2 3 2 2 3 3 2 2 0 0 0 1 0 0 0 0 0 2 0 1 1 0 2 3 1 2 0 0 0 0 0 0 5 13 17 16 20 19 18 12 18 13 13 13 14 15 9 7 4 2 3 1 89 89 42 12 232 ro g Number of one-year e lig ib le Spanish-speaking screened in 1978 by age and number of life tim e screenings. Number of Lifetim e Screenings 1 2 3 4 5 6 7 11 12 13 14 15 16 17 18 19 20 2 3 45 82 46 62 41 38 25 35 38 29 27 33 29 29 13 11 14 12 6 11 2 19 19 44 40 52 39 44 45 47 42 50 45 45 38 34 22 14 7 3 0 0 1 14 16 27 28 29 28 16 26 16 19 22 17 15 11 2 2 2 0 0 0 1 1 9 8 7 7 5 4 5 5 2 1 2 2 2 0 0 0 0 0 0 0 1 1 2 2 0 3 0 0 1 0 0 0 1 0 0 47 101 66 121 98 127 101 117 120 97 102 104 98 99 69 62 49 31 15 16 626 651 291 61 11 1*640 4__ 5 ______ Total 289 8 9 10 1 Table IV(Ea). Number of one-year e lig ib le males screened in 1978 by age and number of life tim e screenings. Nunfcer of Lifetim e Screenings Age Total 2 3 4 5 6 7 Total 644 1361 874 898 678 696 443 530 493 475 396 443 335 397 327 238 167 92 20 12 30 253 292 544 614 660 433 569 535 517 569 541 447 470 371 330 254 123 28 3 1 24 42 121 199 312 276 350 258 235 231 219 197 181 163 120 95 42 2 3 0 0 5 12 43 70 54 64 60 44 35 46 28 21 21 16 9 3 3 0 0 0 1 0 4 6 8 13 6 2 4 3 1 1 2 3 0 3 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 675 1638 1214 1575 1538 1744 1214 1526 1353 1274 1235 1252 1008 1070 889 707 525 263 53 18 9,519 7,583 3,076 534 57 1 1 20,771 290 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table IV (F a). Number of one-year e lig ib le females screened in 1978 by age and number o f life tim e screenings. Number of Lifetim e Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 Total 1 606 1256 844 872 709 685 461 502 498 428 430 497 366 367 315 255 192 173 200 321 0 19 243 296 564 583 631 445 591 593 546 506 553 521 469 429 357 278 161 128 110 0 0 26 48 111 227 289 269 322 292 215 242 237 190 190 204 146 95 46 28 34 0 0 0 2 10 35 71 73 54 44 35 39 39 41 33 21 21 11 6 3 2 0 0 0 0 1 5 13 7 4 2 5 8 7 3 1 2 0 3 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 625 1525 1190 1558 1559 1690 1255 1473 1429 1229 1225 1333 1121 1060 971 779 579 386 359 467 9,978 8,023 3,211 540 61 1 21,814 Table IV(Ga). Number of one-year e lig ib le white males screened in 1978 by age and number of life tim e screenings. Number of Lifetim e Screenings Age 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 Total 328 738 500 503 444 418 275 331 302 285 237 268 190 232 200 136 102 35 14 6 12 133 158 292 385 382 252 329 315 290 317 301 228 231 201 180 127 43 9 1 0 12 25 68 112 163 165 180 146 126 123 130 93 87 79 60 35 15 1 0 0 0 3 5 18 43 35 39 42 30 21 27 17 13 12 10 4 2 1 0 0 0 0 0 2 3 7 10 3 1 3 2 0 0 2 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 340 883 686 868 961 1009 734 889 809 732 701 728 528 563 494 387 268 95 25 7 5,544 4,186 1,620 322 34 1 11,707 Table IV(Ha). Number of one-year e lig ib le white females screened in 1978 by age and number of life tim e screenings. Number of Lifetim e Screenings Age Total 2 3 4 5 6 Total 282 643 513 516 437 441 276 281 282 255 250 295 215 213 173 143 101 75 111 146 11 125 165 282 352 355 260 333 344 311 268 313 301 238 232 180 129 54 45 36 0 11 25 57 127 160 145 167 176 121 124 126 105 95 101 70 37 15 11 8 0 0 2 6 25 40 33 29 28 18 20 22 15 19 11 10 4 1 1 0 0 0 0 0 2 7 2 3 1 4 5 5 3 1 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 293 779 705 861 943 1004 716 813 831 709 667 761 639 566 518 403 271 145 168 190 5,648 4,334 1,681 284 34 1 11,982 293 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Table IV (Ia ). Number of one-year e lig ib le black males screened in 1978 by age and number of life tim e screenings. Aqe 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 6 7 Total 289 568 334 347 203 250 143 174 170 171 141 158 123 147 119 96 54 52 6 6 16 103 124 230 204 251 161 215 196 200 221 211 190 209 151 127 116 73 16 2 1 12 17 41 77 133 97 148 94 94 93 77 93 81 78 53 53 27 1 3 0 0 2 6 24 22 16 23 15 13 12 15 10 7 9 5 5 0 2 0 0 0 1 0 2 1 1 1 1 1 1 1 0 0 2 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 306 683 478 624 510 657 419 561 476 479 467 462 417 444 357 283 228 154 25 11 3 j 551 3,016 1,273 186 14 0 1 8,041 \fSZ Number of Lifetim e Screenings Table IV (J a ). Number o f one-year e lig ib le black females screened in 1978 by age and number of life tim e screenings. Number of Lifetim e Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 Total 1 295 557 299 321 243 218 174 200 181 152 158 174 132 132 123 101 80 89 75 157 0 8 106 116 255 203 240 158 230 218 204 216 200 193 205 172 159 133 97 76 70 0 0 15 22 48 87 112 101 138 103 88 105 104 72 82 94 65 54 29 15 23 0 0 0 0 4 10 25 34 19 11 13 15 13 21 11 9 10 5 4 2 1 0 0 0 0 1 3 6 4 1 1 1 1 2 0 0 1 0 3 0 0 0 1 303 678 437 629 546 601 471 588 514 458 495 493 418 430 399 335 275 219 168 251 3,862 3,259 1,357 207 24 8,709 Number of one-year e lig ib le American Indian males screened in 1978 by age and number of life tim e screenings. Number of Lifetim e Screenings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 2 3 Total 2 2 9 7 2 5 7 1 3 2 2 1 3 1 2 0 0 0 0 0 0 2 1 2 6 3 0 2 3 3 5 3 3 4 1 2 1 1 0 0 0 0 0 2 3 2 2 3 2 2 1 1 1 1 2 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 2 4 10 11 11 10 9 6 8 7 9 6 7 6 5 4 1 1 0 0 49 42 24 2 117 ro U3 cn Table IV (L a). Number of one-year e lig ib le American Indian females screened 1n 1978 by age and number of life tim e screenings. Number of Lifetim e Screenings 1 2 3 4 Total 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 3 4 3 1 4 2 1 0 6 3 1 2 1 3 1 1 1 1 2 0 0 5 4 2 4 4 4 3 3 2 1 2 3 2 3 2 2 0 1 0 0 0 0 2 1 1 4 2 0 1 1 1 2 2 0 0 0 0 0 1 0 0 0 0 0 2 0 1 1 0 1 2 1 2 0 0 0 0 0 0 3 9 7 5 9 9 9 6 10 6 4 7 7 9 4 3 3 1 3 1 Total 40 47 18 10 115 297 Age Number of one-year e lig ib le Spanish-speaking males screened in 1978 by age and number of life tim e screenings. Number of Lifetim e Screenings 2 3 4 5 21 40 23 31 25 20 17 19 14 14 12 12 13 14 2 4 6 5 0 0 2 12 8 20 19 22 18 21 21 20 24 22 24 24 18 20 8 5 2 0 0 0 0 10 6 14 10 16 16 11 14 10 9 11 9 4 7 0 0 0 0 0 0 1 1 5 3 2 3 1 1 3 1 1 0 1 0 1 0 0 0 0 0 0 0 1 0 2 2 0 1 0 0 2 0 0 0 1 0 0 23 52 31 62 51 62 48 60 56 46 52 47 47 52 29 29 21 12 2 0 292 310 147 24 9 782 i Total 298 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 Number of one-year e lig ib le Spanish-speaking females screened in 1978 by age and number of life tim e screenings. Number o f Lifetim e Screenings 1 2 3 4 5 6 7 8 20 2 3 4 5 Total 24 42 23 31 16 18 8 16 24 15 15 21 16 15 11 7 8 7 6 11 0 7 11 24 21 30 21 23 24 27 18 28 21 21 20 14 14 9 5 3 0 0 1 4 10 13 18 13 12 5 12 6 10 11 8 11 4 2 2 2 0 0 0 0 0 4 5 5 4 4 3 4 1 1 1 2 1 0 0 0 0 0 0 0 0 1 0 0 0 2 0 0 0 0 0 0 0 0 0 24 49 35 59 47 65 53 57 64 51 50 57 51 48 40 33 28 19 13 16 334 341 144 37 3 859 299 9 10 11 12 13 14 15 16 17 18 19 1 Table IV(Oa). Number of one-year e lig ib le participants in D etro it screened in 1978 by age and number of life tim e screenings. Number of Lifetim e Screenings Age Under 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total 1 2 3 4 5 Total 1 264 531 262 357 216 266 155 194 190 179 157 185 131 142 125 86 66 65 37 83 0 12 59 84 128 93 150 99 124 151 115 145 166 135 140 108 113 80 64 28 24 0 0 6 13 18 22 39 35 41 35 24 28 30 26 25 24 19 17 7 0 3 0 0 0 2 4 9 8 11 7 5 6 4 6 4 4 0 1 0 1 1 0 0 0 0 1 1 4 2 0 1 0 0 2 0 0 1 1 0 1 0 0 0 1 276 596 362 508 344 465 300 367 381 324 336 387 296 312 258 219 164 137 66 110 3,692 2,018 412 73 14 6,209 Table IV(Pa). Number of one-year e lig ib le participants screened in 1978 in fo rty -fo u r Northern Michigan counties by number of life tim e screenings. Number of Lifetime Screeninqs 2 3 4 5 Total 1,854 1,562 671 164 32 4,283 301 1 APPENDIX G Table X V II. Summary of the analysis of covariance fo r long-term e lig ib le s . Sums of Squares Source 1. 2. (D /O F SS due to interaction model SS due to additive model SS due to screening year adjusted fo r no. of screenings Interaction = .05032 13 1 = 45.63* (2)/DF2 = 83.83* R Additive = ' 04978 302 (a) DF 2 R^ { 2 a )/DF2a = 268.27* (4)/D F. 2 - R No. Screens = .0497 - .0270 (b) SS due to no. screens adjusted fo r screen, y r. ( 2b>/DF2b i)2 _ r2 K Add. K Screen Yr. TaW t = 5' 93* .0497 = .0467 (3)/DF3. SS due to interaction 4. SS Residual W in ter. ' «*M d. * ■ « « ' -04978 ^ In te r . ■ ♦ S ta tis tic a lly s ig n ifican t at the .05 level 6 11,196 7*170r * 1.06 303 Calculations fo r Table XVII 1. tests: For in te ra c tio n (#3 in ta b le ): .00054/.94968 , 6/11,196 .00009/.0000848 1.0613 » not s ig n ific a n t give 2. ggFg m g g * 2.10 For main e ffe c ts of screening year with number of past screenings adjusted (2a 1n ta b le ): .04978 - .02703/1 - .0503 _ - - - - - - 1- - - - - - - ii.lSfi .02275/.9497 _ — I TT7T35‘ .02275/.0000848 * 268.278, s ig n ific a n t given 3. g ^ * 3.84 For main e ffe c ts of life tim e screenings with screening year adjusted (2b in ta b le ): .05978 - .04676 / .94968 _ 5----------T H IS T .00302 / .0000848 „ .0005033/.0000848 ■ 5.935 s ig n ific a n t given 95F6 , 11196 * 2,10 APPENDIX H 304 The f u l l , or satu rated, regression model which includes both independent variables and th e ir in te ra c tio n is as follow s: Y' * Predicted No. of R eferrals * A + + BgDg +■ B^D^ + . . . + BgDg + B^NumScren + BgD^N + BgD2N + B10D3N + * * * B13D6N + Ei * where: * 1 i f la s t screened in 1973, 0 otherwise; D2 ■ 1 i f la s t screened in 1974, 0 otherwise; D3 * 1 i f la s t screened in 1975, 0 otherwise; Dg ■ 1 I f la s t screened in 1978, 0 otherwise. and fo r 1979, Bo ■ In te rc e p t Bl B2 In te rc e p t fo r 1973, In te rc e p t fo r 1974, fo r 1975, B3 * In te rc e p t B7 B8 B9 " Slope fo r 1979 Slope fo r 1974 Slope fo r 1974 h o ' Slope fo r 1975 305 Bg = Intercep t fo r 1978; ■ Slope fo r 1978; and D^N « D, x NumScren NumScren ■ Number of life tim e Screenings. DgN 3 Dg x NumScren 1979 * the "base" year 1n the model, a determination which is a rb itra ry and has no e ffe c t on outcome. The regression lin e s fo r each year are as follows per the in terac tio n model; For 1979, NR ■ A + ByNumScren * .9151 - .0664 NumScren For 1978, NR « (A + Bg) + (By + Bjg) NumScren - (.9151 + .1431) + (-.0 66 4 + -.0 0 5 9 ) NumScren » 1.0582 + (-.0 7 2 3 ) NumScren For 1977, NR ■ (A + Bg) + (By + Bj^) NumScren - (.9151 + .2477) + (-.0 6 6 4 + - .0372) NumScren » 1.1628 + ( - .1036) NumScren 306 For 1976, NR * (A + B^) + (By + Bj j ) NumScren * (.9151 + .2556) +( - .0664 + -.0 5 4 4 ) NumScren - 1.1707 + (-.1 2 0 8 ) NumScren For 1975, NR ■ (A + Bg) + (By + B^g) NumScren « (.9151 + .4313) + (-.0664 + -.0 5 5 8 ) NumScren = (1.3464) + For 1974, NR * (A + Bg) + (-.1 2 2 2 ) NumScren (By + Bg) NumScren - (.9151 + .8960) + (-.0 6 6 4 - 1.8111 + (-.3 6 5 7 ) NumScren + - .2993) NumScren For 1973, NR * (A + B j) + (By + Bg) NumScren - (.9151 + .4744) + (-.06 64 * 1.3895 + (.1385) NumScren + .2049) NumScren The graph of the regression lin e fo r each year is on the follow ing What is most noticeable in these lin e s is the f a ir l y s im ila r page. andf l a t slopes fo r 1975-1979 (which caused the In te ra c tiv e model to be not s ig n ific a n t), the stronger Inverse relatio n sh ip fo r 1974 and the positive slope fo r 1973. The l a t t e r 's deviation from the inverse relatio n sh ip is understandable. 1973 was the f i r s t year of the program and only several dozen children were screened more than once. In the sample, only six rescreenings were selected and these had a higher mean than did the i n i t i a l screenings. po sitive slope. Thus, the regression lin e had no a lte rn a tiv e to a -1973 Number of Referrals 307 1978 1975 1977 1979 1976 Number of Screenings Figure I I I . Regression lines depicting the relationship between number of referrals and number of life tim e screenings for the long-term e lig ib le s , by year of la s t screening for interaction model. APPENDIX I Table X V III. Summary of the analysis of covariance for one-year elig ib les. Sums of Squares Source 1. 2. SS due to interaction model SS due to additive model SS due to screening y r. adjusted for no. of screenings F (ll/D F j W in te r. * -03615 13 (4)/DF4 29.46* (2)/DF2 R2Add. ' -03585 7 2 ^ Add. 1 (4)/DF4 54.29# 308 (a) OF 2 ^ No. Screens ” (2a)/DF.. (4)/DF4 = 267.02 .03585 - .01067 (b) SS due to no. screens adjusted for screen, y r. (2b)/DF2b R2 - R2 Add. Screen Yr. 6 (4)/DF4 = 2.43 .03585 - .03447 (3)/DF3 3. SS due to interaction ^ In t e r . ' ^Add. " 6 .03615 - .03585 4. SS Residual 1 - W in ter. * 96385 ♦S ta tis tic a lly significant at the .05 level 10,216 (4)/DF4 0.53 309 Calculations fo r Table X V III tests: 1. For main e ffe c ts o f screening year with number of past screenings adjusted (2a in ta b le ): .03585 - .01067 / 1 - .03615 1------------------- w ; m - _ .02518 / .96385 _ — — TCTT .02518 / 2. .0000943 * 267.02, s ig n ific a n t given For main e ffe c ts o f (2b in ta b le ): life tim e screenings ggFj 10216 = 3 *8^ w ith screeningyear adjusted .03585 - .03447 / .0000943 6 .00138 / .0000943 _ 5” .00023/.0000943 ■ 2 .4 3 , not s ig n ific a n t given 3. ggFj io216 * ggFg iq 216 For in te ra c tio n (#3 in ta b le ): .03615 - .03585 / .0000943 _ S .00030 / .0000943 , .00005/.0000943 * 0 .5 3 , not s ig n ific a n t given “ 3,84 APPENDIX J 310 The regression lin e s fo r each year are as follow s per the in te ra c tio n model: For 1979, NR ■ A + NumScren .8173 - .0401 NumScren ■ .8173 - s .5767 For 1978, NR - (A + B6 ) - (.8173 + = .9767 + .9767 + (-.3 3 6 ) s For 1977, NR - .6407 (A + B5) 8 (.8173 + 8 1.0379 + M 1.0379 + S For 1976, NR - .5537 (A + B4 ) (.8173 * - 1.0826 + ■ 1.0826 + * .5078 For 1975, NR - (A + B3 ) (.8173 + - 1.1138 + a. 1.1138 + .2206} + (-.0 4 0 1 + - .0406) NumScren 311 “ For 1974, NR * 1.1342 (A + B2) + (By + Bg) NumScren » (.8173 + .4290) + (-.0 4 01 + .1878) NumScren * 1.2463 + (.1477) NumScren = 1.2463 + .8862 - 2.1325 For 1973, NR « (A + B j) + (By + Bg) NumScren - (.8173 + 1.0454) + (-.04 01 + - .2833) NumScren - 1.8627 + (-.3 2 3 4 ) NumScren « 1.8627 - 1.9404 * -.0777 1974 1.8 Number of Referrals 1975 1.0 1979 1977 1976 1973 Number of Screenings Figure IV. Regression lines depicting the relationship between number of referrals and number of lifetim e screenings for the one-year e lig ib le s , by year of last screening for interaction model. 312 1978 313 With the in te ra c tio n model i t can be seen th at the years 1975 and 1974 both have po sitive slopes contrary to the predicted d ire c tio n . 1973 has a negative slope but fo r the same reason its slope was po sitive fo r the long-term e lig ib le s . Only two rescreenings were sampled here fo r 1973, both of which had lower re fe rra l rates and standard deviations than those I n i t i a l l y screened. Thus, th is slope is negative ►'ased upon an inadequate number of subjects. This explanation also appears v alid fo r the 1974 and 1975 outcomes. Again, these were e arly years fo r the program and the sample upon which the regression is done includes a r e la tiv e ly small number of rescreenings. In 1974, two dozen rescreenings were sampled, a l l o f these fo r two screenings only. The mean was higher fo r th is group than fo r those I n i t i a l l y screened (and the standard deviation was lower) so the slope has to be p o s itiv e . In 1975 the mean fo r those with two screenings was a c tu a lly lower than the m®an re fe rra l rate fo r those i n i t i a l l y screened. However, fiv e subjects were sampled with three screenings and th e ir mean rate was highest of the three groups. gave the lin e it s very s lig h t p o sitive slope. This These findings are a c tu a lly consistent with the population data reviewed in Tables I and Tables I I , i . e . , the Inverse relatio n sh ip between re fe rra ls and number of screenings tends to appear consistently only when data are based on large numbers of subjects (100 or more a t minimum). However, i t is noted th a t the obtained F fo r each in te ra c tio n term was less than the c r it ic a l value fo r the .05 confidence le v e l. This means no F was s t a t is t ic a lly s ig n ific a n t fo r any In terac tio n term.