ACT OF ITY LEVELS. E IMP EVALUATION OF TH HEALTH CARE 0N ACTIV T-HE RURAL'POOR . 0F . . :1. ,....... . . . _ \ .... .. l .. a . . . . . .>..\;¢..c_. 3...? ... . . . . Jr... . .. .z.._\;.....\..fma. 9.4.. .... :2 113;?“ ...........: .... ......Z. 14,....r.;r2./<.:. .11.. .... ......i... ........._4 ...;.L..........«.::;€:... .'.;.. .....:.. .._.,..._........_.:r: ....J ...—v... :.../4 i... . . 1...... .. .. ...... ......,... ' Ilsa... .......,_......:1:, .......;....Z;., ...: .... ...... L}... .11....“ ..¢,:.....;. .42.:: ...... ...: .,._.......i.... ...... 3:327; ... ...... ...... ......i... ......T.... 434...}. ........... ... Z .1 ......f .... ... .. . ......V... 11...... TL ... . . ,l 3.. 5.... . a .4 .... _ a g . . . n I... 31“]. 14244 1., ft . . , .. . , ...}... .... 22.21:...225 . . , ._ .2 ...... ... .. I: f ...... .. t . .:.::;. . t 1.1:..- II... ..............;.V .... .....l... .....r. tiff: . ................_.. 3.31.... .fiwfl...x: “2 ..ME.» .... 2...»... . ...... . x, , .. 2...... , ..t;, ... . J... .. r... ....) , ....(.r£...u5 3...... 4...... .... ....El . . I .3.;..._,..,... ..........;:..:.... ...... .. . . , , .51“)... .............:..,....._.... 1.212.... , e..;.fl..,...f... 5.. . .. . ... .. ..... ......s. : .. ..., ...: 2.......;....:J .....3..:: 1 _. I... . _ ......E. ...............: ... 2.1.... .....Z... ... ... . ........._..... , ...: .... ./...._:.. . ......11 ..1......:.: {Keir l . . .. . 45;... 1 ...... .. .. .... a ..........t y . 2 _. . 4.3.2153 5134354,. . . .....;;.:z .4... . . ... :71 -..... .. .....f: ...}. .. .... . ... _. .... . a 1...... .. 3/54.}..é 5:31.. ...... ,._. ...: ._ : f; 14 «.14. .2 T. . 1...... . :3 /. ... . 3.13.» . 1.; . r a ...;itffi 9.5.5.... .7, 93K 1. ,. ..., .1; » . .. I 2 .... 21.3,? 51...? ... .. r... J... .. rvr; .../i »,...p.r1:£.;. . . I; .,..,.,,,......,_ 3.735.. .... ..r. ...? F4 dIAJ.IJ,r4l. .1. 7f: .. ... 3:7,. 575:»: 5,... 42...? 1,21%? ./ . . . 1!. ... a; .. r... . ... z. _ .... r .... 7.; ,. ...: ...: . ... . 1:: 5.72:1. ... . it. .. «.1. . .3 1 1... :r 7 . :....E;....:;..:..r .. .1. ... ... . r 3.11.2}. ..a,.......v.:( ,/ . ... ., ‘ ... . . ... I 3 r v 11.1%.; i ...,rrv... . 3...}: t .. . :1: 1.... r1313: ...... 51.1.... :1 3:11. 11,. :5»? . .1... . 6...: (I. .12.. :5. ......r Sis '_ é II3IIIIILIIIIIIIIIIIIIIIOIIIIIII9IIIIIIIIILIIIII 5 L um I "' ”I an > ’ yfllVCl’Sli'l .. a l' 3‘" .'}‘.-‘ ”In-"’1, This is to certify that th? - *5 -- a u . , 3 thesis entitled , I. ~ Evaluation of the Impact of Health Care on Activity Levels of the Rural Poor presented by Werner Kiene has been accepted towards fulfillment of the requirements for Ph. D. dPym,inAgricultural Economics MW SM Major professor Date November 10 l 72 0-7639 , . I ~~... ‘~-... g. u,” nu ‘IQ m,‘ v .\. l P‘ ‘ fin . I n .‘I .‘M' c I“ ABSTRACT EVALUATION OF THE IMPACT OF HEALTH CARE ON ACTIVITY LEVELS OF THE RURAL POOR By Werner Kiene Federal and state agencies have been experimenting with amemuudve institutional arrangements in solving the Ameri- can'Wwalth crisis." Yet, so far very little is known mmuttmw impact of programs on the health of the target mefletions. As in other areas of public spending, health mmjmfi;administrators have a fairly good idea of what they mm hnx>their programs but lack information on the socially Nflewnm and desired outputs. Most of the output measures meiindicate how efficient an organization is in providing . "mflts"cfl'care, yet they do not tell how efficient those services are in producing health. (fins thesis attempts to provide some insight into the prmflem.of output identification and measurement of public Pmflecusin general and of health projects in particular. Rm mnmepts of "Derived Demand" and the "New Theory of Mmsmmn'Demand" are applied in developing a framework for idmndfying relevant input-output relationships. This fiammmnmzemphasizes a treatment of attributes rather than cmumntrating solely on the physically observable units. The application of the conceptual guidelines to the emanation of health services results in the definition of Werner Kiene health status in terms of its attributes, i.e., "enhancing role fulfillment" and "reducing deviation from ideal roles." Relevant roles are identified as the ability to play, go to school, go to work, and work at home. (The study suggests ways to group health services inputs according to attributes. However, these concepts are not pursued in further detail.) Questions of the National Health Survey Interview were utilized in organizing a survey instrument on health status outcomes (role fulfillment and deviation from ideal roles). Problems on establishing health related questionnaires, designs, and interviewing are discussed in light of the experience gained in this research. The concepts developed in this study were applied to the evaluation of a rural health project located in Northern Michigan. (The Western Michigan Comprehensive Health Ser— vice Project with its main clinic in Baldwin (Lake County) serves people in an area consisting of the four counties of Lake, Mason, Manistee and Newaygo.) Resource constraints limited the analysis to an investi— gation of Lake County. One part of the empirical analysis consisted of a survey on role fulfillment of all age groups While the other utilized attendance records collected by the school system in the project area. The procedure which utilized a survey instrument was conducted in the format of an "ad hoc comparison" with a comparison county that resembles the treatment county (Lake) . ..b. '-.\. "'i. - ._‘.- I ‘. . ‘ l Werner Kiene h1aleariables except the availability of health services. amiowmonomic data supported the assumption that Montmor- mmy(kmnty, Michigan, met the requirements of a comparison cmmty. The survey results indicated that the project umldreduce days lost from play, school attendance and lwmevmmk but not from work. The examination of school aflmnmnme records did not produce sufficient evidence to emtflfljsh the impact of the health project. Ikmpite the inconclusive results of the empirical analysis, itcouhibe shown that the developed conceptual framework :m atweful guideline in conducting an output~oriented pro— jem:evaluation. Detailed recommendations for additional inwnmigations were reported to facilitate future research onheahxlproject evaluation. EVALUATION OF THE IMPACT OF HEALTH CARE ON ACTIVITY LEVELS or THE RURAL POOR By Werner Kiene A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1972 Copyright by WERNER KIENE 1972 DEDICATION To Heidi ii ACKNOWLEDGMENTS Many individuals were helpful in completing this study. Special appreciation is expressed to the members of the author's Thesis Committee: Drs. Allan Schmid (Thesis Supervisor), Gail Updegraff, and Vernon Sorenson (Major Professor) for their intellectual and personal encourage— ment during this research and to the members of the Guidance Committee: Drs. Dale Hathaway, Mordechai Kreinin, Harold Riley, and Ed Rossmiller. The cooperation of the following individuals and insti-- tutions is acknowledged: The Michigan Department of Public Health (Dr. Kurt Gorwitz, Mr. Glenn Sommerfeldt, Mr. Charles Benda and Mr. David Bell), and the administration of the Health Center in Baldwin, Michigan (Dr. Jose Mejia) for helping in obtaining information about the Western Michigan Comprehensive Health Services Project; Mr. Jack Chase, Director of the Lake County Social Services Department; Mr. Oral McMurphy, Director of the Montmorency County Social Services Department, for their assistance in a survey Of food stamp and commodity recipients; and Mr. Kent Reynolds (Principal, Baldwin High School) who helped with the school attendance survey. Thanks is also expressed to Mrs. Judith Stephenson (Supervisor of the Department's Computer iii .41“: r l ' I ,no" ' , . . ”Add 0 1-- ,.v. a . n ...-a 4‘ Programming Staff) and to Lloyd Teigen for assisting in the computation and analysis of the survey data. Mrs. Barbara Dickhaut and Mrs. Janis Hendrick deserve special thanks for their efficient typing. iv TABLE OF CONTENTS CHAPTER Page I. INTRODUCTION . . . . . . . . . . . . . . . . . l 1” Objectives . . . . . . . . . . . . . . . . A 2. Organization of Study. . . . . . . . . . . 5 II. A THEORETICAL GUIDELINE FOR ANALYZING PUBLIC PROJECTS . . . . . . . . . . . . . . . . . . . 9 l” The Traditional Use of the Economic Model. 9 2. The Expanded Use of the Economic Model . . l5 1 3. Relevance of the Expanded Model. . . . . . 16 ' A. Applications of the Expanded Model to Research in Health Economics. 17 5. The Complementarity of the Traditional I I and the Expanded Model 17 6- Research on Production Functions of Public Projects 18 7. Research Procedure for Program Evaluation—— A Summary 22 III. EVALUATING HEALTH PROJECTS . . . . . . . . . . 26 l. A Summary of Health Services Research. 27 2. Relevant Input-Output (I-O) Relationships- 29 3. Data Research 36 A. Health Data Collection 38 v IL. IV. VI. 5. A Model of the Health Production and Exchange Systems . . . . . . . . . 6. A Summary Model of the Health Status Pro- duction Process. . . . . . . . . . . AN ANALYSIS OF THE WESTERN MICHIGAN COMPREHEN SIVE HEALTH SERVICES PROJECT (WMCHS) . , , l. The Project and its Service Area 2. Rationale for Evaluation 3. Procedure of Evaluation, A. Data . . . . 5. With/Without and Before/After: The Problem of Experimental Design 6. The Problem of Observation Over Time 7. Evaluating the Lake County Health Center IMPACT OF HEALTH CENTER: METHOD I, VARIOUS OUTPUTS. . . . . . . . . . . . 1. Selection of the Comparison County 2. Selection of the Comparison Group—~Scope of Investigation . . . . . . . . . . . 3. The Model. . . . . . . . . . . . . . . . A. Survey . . . 5. Household Questionnaire. 6. Member Questionnaire 7. Analysis 8. Summary. IMPACT OF HEALTH CENTER: METHOD II, SCHOOL ATTENDANCE ONLY. . . . . . . . . . . 1. Conceptual Considerations, 2. Practical Problems~-Data . 3. Practical Problems——Design . A. Procedure. vi A0 A3 A7 A7 53 55 57 6O 63 63 68 69 7O 72 79 82 8A 87 105 107 108 109 109 111 j 5. Analysis 6. Conclusions . . . . . Ln; IEEOMMENDATIONS FOR FURTHER INVESTIGATIONS 1. Summary of Problems 2. The "Ideal" Experiment . 3. Practical Considerations A. Second-Best Designs 5. A General Strategy for Evaluation. 6. Specific Proposals 7. Further Improvements of the Mode——A Dynamic Model. 8. Data . 9. Data Collection. 10. Questionnaires 11. Sampling Frames. 12. Interviewing l3. Long-Run Perspectives. lA. Project Evaluation——A Summary VIII. SUMMARY 1. Rural Health Care 2. Conceptual Framework for Evaluating Social Projects . . . . 3. Application of Concepts to Health. A. Empirical Analysis of Health Status Outputs 5. Concluding Remarks BIBLIOGRAPHY APPENDIX A APPENDIX B APPENDIX C vii 115 123 I29 129 1A0 1A0 lAA IA? 151 153 15A 156 158 I62 16A 165 165 I68 168 I69 170 173 177 179 193 211 219 Table 10 ll LIST OF TABLES Relative Frequency of Individuals in Different Activity Limitation Groups Number of Individuals by Age and Activity Limitations--A Comparison Between the Lake County and Montmorency County Sample Days Missed From Major Activities. Average Values and Regression for Both Lake County and Montmorency County . . . . . Days Missed From Major Activities. Average Values and Regression Results for Lake County Only . . . . . . . . . Indices of Perceived Availability and Quality Of Local Health. . . . . . . . . Indices. Average Number of Halfdays Missed by Students Before and After the Health Center was Estab— lished . . . . . . . . . . . . . . . . . . Average Absence Ratios, Trend Index and Comparative Index of Absence Ratios. Average Absence Ratios and Trend Index of Absence Ratios . . . . . . . . . . . . . Average of Individual Trend Indices of Paired Sample of Students Who Resided in the School District Both in 1966—68 and 1970—72 A Comparison Between Recorded and Reported Number of Half Days Missed viii Page 90 92 95 99 103 10A 11A 116 120 122 125 Appendix Tab le 3 A1 B1 B2 B3 BA 135 B6 B7 C1 Average Utilization of Health Services and Concentration Ratios of Utilization. A Profile of the Four Counties Served by the Health Project and the Corresponding State Averages . . . . . . . . . . Demographic and Economic Profile: Lake County . . Demographic and Economic Profile: Montmorency County Summary of Budget of the Western Michigan Comprehensive Health Services Project for Years B - E Comparison of Socio-Economic Indicators Between Lake County and Montmorency County Eligibility Criteria for OEO, Foodstamps, and Commodity Programs . . Profile of Sample Households. Comparison Between Lake County and Montmorency County Codes of Household Characteristics of Individuals ix Page 207 211 212 213 21A 215 216 217 219 fiv‘m‘_ Figure 9 10 LIST OF FIGURES Market Model for Health Service . . . A Research Procedure for Program Evaluation A Simplified Model of the Health Services Exchange System . . . . . . . Organization of Western Michigan Comprehen— sive Health Services Project. . . Poverty Index of Michigan Counties Graphical Representation of "priority Scheme" Which Underlies the Establishment of "Change Index" . . . . . . . . . . . . . . . . . Calculation of Average Absence Ratios Steps of An Ideal Design. Health Status Production Functions. Dynamics of Health Production Appendix Figures A1 A2 A3 AA Lorenz Curve Crossing Lorenz Curves. Consumption of Gini Coefficient Concentration of Physical Units Received. Page 10 23 A1 52 71 102 112 133 150 15A 198 198 201 201 .. . . .’ e. .... 'u: . . II-I. .l n o ‘ I .‘ ‘ I h . m. CHAPTER I INTRODUCTION Officials at various levels of government feel increasing pressure to report what they actually produce with funds under their administration. Based on economic and organiza- tional theory, the principles of Planning-Programming— Budgeting Systems (PPBS) have been advanced to aid the decision maker in his complicated task. Most of the literature published in the PPBS field con- I centrates on I) the need for reorganized thinking in govern- ment, 2) systems problems and 3) refinement Of Benefit-Cost analysis. The majority of discussions start out with the assumptions that there is an identifiable and measurable set Of inputs and outputs and proceed from there to advance systems and interaction models and discounting procedures. All these models and thoughts are very appealing to those who are aware of the complex systems nature of the public sector and of how this sector is constantly being accused of "overspending," "duplicating and counteracting its efforts" and "underserving the needs of those to be served." Federal agencies were the first to move toward PPBS as a basis for "rational" spending. EXperiments at the Federal level were soon followed by applying PPB thinking to state and local governments. For instance, the state of Michigan 1 o ‘4 'r ...! 0“ up,- ...-..| .. .', o ...u“ -. ~ “'u... .. ‘ '0“ ‘« begun in 1971 to implement a PPB system called Program Budget Evaluation System (PBES). In Michigan, like in many other instances, much organizing and "systemizing" has been completed, and what remains to be done is to "fill in the numbers." It is here where the problems of agency heads, division chiefs, and their middle level staff begin. The problem Of identification: What is their organization's output? The problem of classification: How can this output be grouped into relevant units? The problem of measurement: How can the output be measured? The problem of data collec— tion: How can the information be collected? There is very little organized knowledge upon which those who need information could fall back on. Experience from one area of application is seldomly transferred to another. Exchange is sometimes impossible even within depart— ments due to "language" barriers. Is it possible to form a body Of knowledge which would give guidelines to those who have to "sweat it out" i.e. to those who generate the infor- mation in the system? This dissertation constitutes an effort of providing Such a methodology or guidelines. It grew out of the author's attempts to describe and evaluate the social infrastructure 0f rural communities by means of a computer simulation and by using the PPB framework. The modeling of such a system seemed to be feasible and promised to show results Of heu— ristic value. Yet, the reliance on non—existent output in- dicators made such a model of little use to the actual de— Cision makers. Once it became apparent through a review Of ‘l ..p- ‘u n I O u ...-I . hut! - n.4- V'r‘l \ ‘n... . theliterature and through initial research efforts that thereal bottleneck in evaluating public expenditures is thecflassification and measurement of output, subsequent resauch efforts were focused on these issues. A rural community Health Center in northwestern Michigan wmschosen as a case study for such an investigation. The cmflce of this particular case was based on several consider— ations: 1) Previous contacts between the Health Center and Michigan State University. (Staff members of the College of Human Medicine were involved in planning the Health Center.) ii) Previous research experience of the Department of Agricultural Economics at MSU in the area of rural health care (Sarkarl, Dohertyz). iii) The increasing importance of the variables "health" and “health care" in the welfare function of indivi- duals and communities. (The achievement and main— tenance of a high status of physical and mental well-being has been proclaimed as one of the fore— most goals of our society.)3 iv) The significant increase of monetary inputs in the project area. The Lake County area was served by two aging doctors before the Health Center was es— tablished in 1967. The Health Center's annual bud- get between 1968 and I972 varied between $2.02 million and $3.15 million.II -l ..or.‘_ . -"nJo ... o, u 9" I“ . .~ -.‘ I ... u i. -. . n I n I .. .. \ ..‘I'l . _ , ... . . '~ I ‘v ... \ s \ \ \ Huntllwas probably the most important consideration for concentrating on this particular project. It was felt that shme'mmenmasurement of output was severely limited by a lackcfi‘previous experience, it would be most promising to Imy'u3expose events in a situation of extremely increased inputs. 1. Objectives The Objectives of this study are: i) To develop a theoretical framework for analyzing the input—output relationship of public investment. ii) To apply this framework to the analysis of invest— ment in health care facilities. iii) To demonstrate the procedure of a health project evaluation using a case study. 1y) To produce substantive measures of health care impact on human activity levels. V) To use the collected experience in proposing a plan for a more elaborate and methodologically more ad— vanced evaluation of investment in health care. The objectives indicate that the major emphasis of this Stmhris on developing a methodology for identifying and nwamndng input—output relationships of public projects. TMastudy is divided into a general methodology of public Pmmect evaluation, a specific methodology for health project evaumtion and an empirical treatment of health services evaluation. . '9! I... _-l .. "- . ‘ ' e. u. '\ .~ i 2. organization Of Study Chapter II shows that economic theories can be employed to establish a relevant perspective for the evaluation of a public investment process. The theories are rooted in neo— classical economic thinking commonly referred to as micro- economics. The "New Theory of Consumer Demand" and the con— cept Of "derived demand" are suggested as a guideline for organizing and identifying relevant variables. Dynamic proper— ties are introduced by outlining a systems design as the basic framework for modeling. Chapter III focuses on the problem area of evaluating investment in health care. Here, the general ideas Of pro- ject evaluation as introduced in Chapter II are specifically applied to the health field. New theoretical developments to cope with the peculiarities of the commodities "health" and "health services" are highlighted. The chapter proceeds to summarize past efforts to analyze and evaluate health programs and projects. These efforts consist basically of epidemiological, cost-effectiveness, benefit-cost operations research, health services utilization and peer review studies. The chapter concludes with a model for health and health services. The previously exposed theories are incor— POrated into a framework which can guide an economic investi- Sation Of health programs. A short description of the Western Michigan Comprehen- sive Health Services project introduces the particular case study in Chapter IV. Some of the preliminary explorations Egon ... ‘. ‘0 ll “oo- 5 Nu I. undertaken during the course of this research are briefly reported in order to show what problem areas exist in ana- lyzing health projects of this kind. Since it became apparent that an elaborate input—output model was beyond the time and resource limitations of this dissertation, it was decided to concentrate on the measure— ment of health status as defined in Chapter III. Although most of the time was spent with output measures a parti— cular topic on inputs——the concentration of inputs-—was pursued in depth. The results of this investigation on inputs are reported separately in Appendix A. The empirical research on output measures was divided into two parts: (i) an analysis of health status measures employing a health survey (Chapter V) and (ii) an analysis of school absence utilizing school attendance records (Chap- ter VI). Chapter V describes the procedure of comparison between a sample of residents ofthe project area (Lake COunty, Michigan) and a sample of residents of a "control" area (Montmorency County, Michigan). The procedure and results of a "health outcomes survey" in Lake County and in Montmorency County form the main body of this chapter. Chapter VI gives an account of an investigation of school attendance as a measure of health services output. 30th Chapters V and VI are not intended to produce "hard" results in terms of evaluating the Lake County Health Center. Rather, as previously indicated, they should be regarded as VBhicles for building experience with health outcomes evaluation. This ’cave‘at is especially relevant in light of the small sample size and the limited control of the elements under investigation. Since the overriding objective of this analysis was to open avenues for further work in the area of project evaluation, the experience gained from this research is summarized in Chapter VII. This chapter advances procedures for an expanded investigation of investment in health ser— vices. Although the section concentrates on health, it is hoped that several of the suggested steps are also applicable to other types of projects. Chapter VIII summarizes the total research effort and draws some overall conclusions from the work done. The tables and the Appendix contain information which might be relevant for further research with the data generated for this dissertation. The bibliography was organized by subjects covered in the research in order to serve as a summary of current literature on the various disciplines touched upon in this thesis . FOOTNOTES Shyamelendu Sarkar, "The Copper Country Medical Industry of Michigan as it Serves Rural People" (unpublished Ph. D. dissertation, Michigan State University, 1969). Neville J. G. Doherty, "The Economic Structure and Performance of the Medical Industry in Michigan‘s Grand Traverse Region" (unpublished Ph.D. dissertation, Michigan State University, 1970). U. S., Department of Health, Education, and Welfare, Toward a Social Report (Washington, D.C.: Government Printing Office, 1969) - Bettie L. Nelson, "Profile of Lake County and the Five-Cap Area," Michigan Department of Public Health, 1972, p. 59. (Mimeographed.) CHAPTER II A THEORETICAL GUIDELINE FOR ANALYZING PUBLIC PROJECTS This section attempts to demonstrate the usefulness of the economic production and consumption model in guiding the process of evaluating public eXpenditure programs into rele— vant directions. The discussion is geared to the problem of evaluating health projects, but is general enough to serve as a guideline for the evaluation of other types of programs . l. The Traditional Use of the Economic Model Economic theory indicates that both market—demand and market-supply functions arise from a summation of the indi— vidual Marginal Cost (MC) and Marginal Utility (MU) curves which in connection with a price guides decisions of con- sumers and producers. Prices, in turn, are determined by the interplay of demand and supply as shown in Figure 1. a) Market Failure The model represents the situation of a perfect com— Petitive market without restrictions of entry and where all participants reveal the utility they derive from their action in the market place. Obviously, this ideal situation does not exist in the health area which in economic parlance re— sults in less than optimal resource allocation called "mar- ket failure." Two of the most prominent causes of market 9 - .-.—m... 10 Begum fiance co 3 n Hpcmsa A o< wooEow 53mm no 380 RE mOOfiPAom cufiwom ho nfiv bouncesm .wOOH>pom Spamom now Hoooz posse: 835m flame co BEES I .H onzwflm HIH hem c H h. Huh. niwuw C 1. I 809.com flames n8 posh: moOH>nom npamom mo apaucmsc I 235 ASHES Sam: .8 3333 Hugh wooEom spasm ho AC huh—.380 ‘5. And ...,a u o l -..“.- "-‘l' | ~ g '-. -. V‘.“ ... .‘ ., . ~. \_I i ,_ a, ‘I .‘s‘ \‘ §\ \ ‘~ .~. .\ II 1 . u n 11 failure in the health field are monopolistic conditions and externalities . Monopoly It is a widespread Opinion that medical services are suffering from a monopolistic market structure. This fact changes the determination of market prices and quantities. A removal of the monopolistic situation would cause an in— crease in producers and in the production Of medical ser— vices resulting in a decrease of the price level which, in turn, would result in an increase of the consumption of health services. Competitive pressure could increase pro— ductivity, quantity and quality of services. Some, however, would argue that a completely competitive situation would degrade the quality of service. The following discussion assumes that monopolistic conditions have been abolished and that Figure 1 depicts the actual situation. (A redefinition of the role and functions of medical professionals and in— creased training of medical and paramedical manpower makes the existence of a more competitive market a realistic assumption—~at least for the future.) Externalities Pauly and others show that the simple market model has to be extended to account for externalities.l They argue that individual A derives utility from individual B's de- mand for health services. Since this utility is not repre_ sented in the demand functions of the simple model, the allocative mechanism of the market produces less than .u‘l F , u _. .U s. -l \ ... " v L- 12 desirable results. A's utility from B's consumption of health services can have two different origins depending on the nature Of the externality: (i) communicable diseases and (ii) availability of care. (i) Communicable Diseases Individual A knows that, if B protects himself against communicable diseases, the probability of him (A) contacting these diseases has been significantly reduced. A is a free— rider on B's efforts to prevent diseases. Yet, usually B's success in preventing diseases depends on A's participa- tion in preventive measures. B derives satisfaction from A's consumption of health services and A derives satisfaction from B's consumption of health services. A case of "recip- rocal externality" has been established.2 (ii) Availability of Care Externality arises here because individual A derives satisfaction from knowing that B consumes health services or has at least the option to consume health services. b) Limitations of the Traditional Use of the Model Much of past and current research in health economics uses variations of the health services model as presented in Figure l as the basis for analysis. The most striking fea— ture for non-economists is the assumption of predetermined consumption schedules (MU schedules) which in connection With prices, determine the consumption of health services. This assumes that individuals know what amount of health services is enough for them as long as the price is "right." 'I' r. “'U .- "'o,. o...- I. ‘94 I... ... ‘\ ~.~~ .__ 13 Due to imperfections in the market and because Of the character of the good, some of which were discussed in the section on "market failures", health services are being bought not only by individuals but also by "the public", such as the Office of Economic Opportunity (O.E.O.), which then gives them to individuals. Individuals as well as adminis- trators Of organizations like O.E.O. and the U. S. Department of Health Education and Welfare (H.E.W.) do not know a priori what the marginal utilities of provided services will be. Administrators rely on health services research and evalua- tion to provide them with this information. A sizeable portion of health services research deals with identifying the extent of desirable consumption and utilization of services. (Implicitly at least, this is analogous to estab— lishing marginal utility schedules.) The model as outlined above is, therefore, of little value to public decision makers and to individuals since it does not tell them why different individuals consume dif- ferent amounts of health services at the same prevailing Price. Actually, the model is not even intended to explain this variation, according to its creators. Neoclassical economic theory is based on the assumption that there are different tastes which determine as such the market. In Other words, the difference in marginal utility derived from consuming health services is predetermined and causes, in connection with a given price, different consumption Patterns of health services. 1A 0) The Traditional Use of the Model and Research in Health Economics The model has been useful in pioneering research and contributed significantly to the development of health economics as an applied field. Most of this research concentrates on health services I and on how to produce these services more efficiently. Very few economic research efforts have been spent on evaluating how to produce health and health status more efficiently. The traditional use of the model does not challenge the researcher to attempt a health status evalu— ation. This statement is not meant to indicate that . there is no economic tool available for evaluating health projects. On the contrary, the principles of economics (i.e. allocation of limited resources to satisfy unlimited wants) form the basic rationale for project evaluation. What I this chapter attempts to expose, however, is the fact that the traditional economic model is an incomplete representa— tion of the actual situation which it tries to simulate. The incompleteness of a model is significant if it forces empirical research to restrict itself to a limited format as prescribed by the model. It is the author's belief that the traditional use of the model unnecessarily restricts the scope of health Economics research. This paper advances the position that a more relevant model will serve not only to broaden the Outlook of research in health economics but will also prOVide 9:: (“I {a -. ... u -‘t 'I out: a -. .n "~¢ up p- u- ‘u‘ .: .:, .3. , \‘\ \ E. 15 I those concerned with evaluating health projects (and other projects) with a relevant procedural guideline. To make economic theory useful in health services evaluation, the . simple model has to be expanded in such a way as to facili— tate explaining the difference in marginal utility schedules which exists among individuals. 2. The Expanded Use of the Economic Model The concept of "derived demand" and the "New Theory of Consumer Demand" are suggested as additions to the tradi— tional model in order to transform it into a useful frame— work for health services research. If put into that frame- work, health services become inputs into the individual's production process of the output health. In other words, the. demand for health services is actually a demand "de- rived" from the demand for health.3 We demand health ser— vices because they have the ability (attribute) to produce health. The "New Theory of Consumer Demand" adds to the concept Of "derived demand" the notion that the same attri— butes can be found in a variety of physical goods.II We de— mand, therefore, different goods because they harbor dif- ferent inherent characteristics which we need in producing desirable outputs. This forces the investigator to realize that utility is not derived from consuming health services but rather from consuming the attributes of health services, i.e. from consuming health. An important fact emerges: The commodity "health" can be produced by consuming various market and non-market goods (health services and others) . 'D'n me... o, "o a.“ I I .... ' o . w. _ .1 'r . u so... . x“. \ n“. (r; u- 4'4 '( which harbor the health producing characteristics. Individuals differ, therefore, in their demand for health services not because of given differing tastes concerning health services but because of a given differing command over various goods containing health producing attributes. 3. Relevance of the Expanded Model The theoretical significance of the expanded model lies in the fact that it explains the variation in the demand for health services. a) Outfit-Orientation The implications for practical research are even more significant. The model is a useful guide for shifting think-— ing away from input analysis (health services) towards input-output analysis (health services-—health). This Shift towards output orientation is especially called for in evaluating public projects, be they in the health area or in other fields. A) Systems Thinking Another advantage of the expanded model is its ability to guide a classification of system components according to relevant attributes. Once the components have been classi— fied it is possible to combine them in such a way as to conceptually maximize desired output, subject to given prices and resources constraints. Research of interacting systems is facilitated by the fact that, regardless of the source commodity, attributes are measured in identical units. For instance, classifying 1? somufl.programs into their characteristics will show which programs have health attributes. If the objective is to unmmnze health, economic theory dictates that the social prqnams under investigation be combined by applying the I eunmrginal principle to both the programs and the inherent cmnacteristics. Those familiar with program planning and budgeting will recognize that this corresponds to the con- ceptual foundation of PPB systems. A. Applications Of the Expanded Model to Research in Health Economics The expanded model suggests an output orientation inkmalth economics. In most instances, this reorientation wmfld.only require a relabeling of the axes of the tradi— thnml model. For instance, in Figure l we would consider Hmalflfl'as the commodity being produced and consumed in— Mmadcfl'health services. Pauly's externality discussion shmfld be recast in terms of individual and collective de- namiibr health.5 Pauly states that individual A derives satnfiaction from individual B's consumption of health ser— vhmm. Using the expanded model we would represent the acmufl.situation, i.e. that A derives utility from B's health. 5. The Complementarity of the Traditional and the Expanded Model The externality example serves to put the expanded mmkfl.into a proper perspective. People do not only get 18 satisfaction from knowing that someone who is sick gets well but also from knowing that this person gets at least some kind of health care, regardless of whether or not this care will result in a desirable health status. The complementarity argument is extremely relevant in applying the discussed concepts in an empirical investi- gation. Information on attributes or characteristics is difficult to obtain. The investigator will often be forced to confine his analysis to proxies of the attri- butes-—usually to the source commodities which harbor the characteristics. For instance, in the case of health projects there is much information available on health services but less on health. Again, it is emphasized that the limited information On actual outputs is no excuse for (i) faulty concep— tualization of the study and (ii) for not trying to get more relevant proxies for the actual outputs. 6. Research on Production Functions of Public Projects An output and systems orientation suggests that the Evaluation problem of public expenditures be put into a Preduction function framework. This framework can be con- ceptualized by the following set of equations: X 00+ x+. ainn Y1 = 8‘11 1 Ym=aml xl+ . . . +arnn xn where Yj = output (3), xi = input (1) and aij = input/output coefficients . F . . 19 a) The Problem of Classifying and Measuring Input and Output Variables. Usually one assumes knowledge of the characteristics ,' of xi and YJ. and of the units with which they are to be I measured, thus leaving the calculation of the input/output coefficients as the major task of the production function analysis. One of the main problems in estimating the input— output relationship of infrastructural investment is to get relevant classification and measurement of inputs (Xi) and outputs (yj) . A related problem is the selection of input and output groups which would permit a comparison of input-output rela— 1 tionships under different forms of organizing the produc- tion process. (This issue could be termed the problem of "additivity" and "scalability"). b) ggnceptuai Solution—-—Classification According to Attributes ‘ ' Again, the awareness of the fact that theoretically all activities can be viewed as consisting of production and Consumption is necessary for an understanding of the con- ceptual solution to the classification problem. In fact, production and consumption are two sides of the same phenom- enon: while I consume a good, I produce the satisfaction which is derived from its consumption. The problem of the analyst is to find out where in the chain of real and conceptual input—output relationships I he should start and call the variable involved the "Desired Output." Under the guidance of the "expanded" model, the .0: A! . - p‘ a... VI . . . --..I | ‘l ‘-.... ‘9‘... 4.. '. U 20 identification of relevant input and output categories has become even more complicated than under the "traditional" model. Suddenly we are faced with a situation in which commodities which were originally perceived as outputs (health services) have to be classified as inputs. What should guide the choice of the dependent variable? I 0) Guidelines for the Choice of Variables I Only vague guidelines can be offered. Experience, intuition and common sense, i.e. the "art" of research . are the major ingredients for a successful investigation. Nevertheless, it can be said that the variables chosen must pass the test of (i) relevance and (ii) feasibility. Scien— . I tific advancements have usually depended on the specifica— tion of relevant variables. To use an example from agri— culture: it is not manure which is the commodity demanded for the production of crops but. its minerals and organic attributes. The Odor of manure is a characteristic which usually is neither desired in itself nor is it considered as a desirable input into the production of crops, although the handling of odorless fertilizer makes a farmer more "respectable" than if he would use manure. Manure has bene— ficial (desired) and costly (undesired) characteristics which enter the decision framework. By defining those de— sirable inputs clearly, science could provide us with dif— ferent methods of organizing our productive efforts. To complete the analogy it should, however, be emphasized that Iihe definition of the characteristics and the subsequent I 21 development of new organizational forms of production evolved rather gradually and Often accidentally. Also, the reor- ganized production brought with it a host Of new problems-— ecological deterioration is just one Of them. The development of crop research has a parallel in feeding research. CrOps are not really the commodities demanded for the production of beef or for the production of meat. Only their inherent attributes are important and therefore demanded. The example could be continued over many stages. What should be inferred from this demonstration is that the some- times accidental recognition of relevant inherent attributes lead to subsequent measurement and a wider set of production possibilities--culminating in such seemingly "impossible" inventions as the development of soybean meat. Are there similar undiscovered Options in the social sciences and how should we discover them? d) Man the Product—Ingredient of Social Research The output of social programs is not to provide a Population with goods and services. The implicit ultimate goal is to produce and maintain individuals in a state which allows them to fulfill their roles as human beings in a dignified and acceptable way. The determination Of those roles evolves out of the complex sociO—political interaction of all individuals. The concept of ultimate goals is not meant to degrade intermediate goals and objectives (or intermediate outputs). l "‘lbl - - suc.‘ P— ~.| . ._ _ __...._«——___——_. 22 However, it is meant to convey the notion that activities (input—output relationships) are organized under the assump- tion of having a direction. It is immaterial whether this direction is conceptualized as the "invisible hand" or as a social value structure. This study does not ask for a judgment on the superiority of objectives and goals but pleads to include pl; relevant input-output relationships into the analysis. The recommendation to those who evaluate social projects is, therefore, to include in the total system of "input-output chains" a production function where "man" is the output.6 7. Research Procedure for Program Evaluation——A Summary Economic theory as interpreted in this chapter suggests the following steps for an evaluation of public projects. (A graphical representation of those steps is shown in the flow diagram in Figure 2.) Disaggregate the program system into relevant input- output relationships by utilizing the principles of Derived Demand and Of the "New Theory of Consumer Demand." Relevancy has to be determined by the original objectives of the investi~ gation. i) Attempt to isolate relevant outputs of the program by checking for correspondence with the social, cultural and political value structure. ii) Attempt to measure the units of input and output by applying findings of social and natural science research. I17 I’ I P I A 2 3 I (bjectives of ’ Research IF i 1 I Concepts of ) Derived Denmd I Disaggregate System I Into Relevant I I-0 Pelatimships I Mutants of the New Theory of Cmsurer Depend Social 1.333% __.I Idmtify Socially Value Relevant Outputs: Os . Stmctme Identify Socially Relevant Social Science Unit f Output (os_1) Which is and Natural “b88139 S 0 NI u n _ Science Output ( ) and ext Igiwnainthe I 0 Research 1119111: ( 8) Social Science Is Methodology Measuring , N0 N0 Possible? I Yes Use Variables in Establishing Production F‘mctims Results of Analysis New STOP }——I Yes ‘ Satisfied N0 Attemts Figure 2. A Research Procedure for Program Evaluation. W 1 ..., u - .... u n u I I ‘ x. 214 iii) If measurement shows no success, settle for relevant proxy variable. iv) If relevant proxy variable is not available or if nwammement of proxy variable is impossible, go to the next stq3"down" in the input—output relationships. v) If measurement is feasible, production functions cm1be established. The degree of satisfaction with the mmlysis and its result will determine where improvement calor should be made. The research documented in this dissertation followed the"philosophy" and procedure as outlined in this chapter am as summarized in Figure 2. The procedure did not Spmfify what should be done at each step, yet it guided the"how" of the research in terms of specifying the outlook theresearcher should take in attacking the problem at hand. N U"! FOOTNOTES Mark V. Pauly, Medical Care at Public Expense (New York: Praeger Publishers, 1971), pp. 21-35. James M. Buchanan and Gordon Tullock, "Public and Private Interaction under Reciprocal Externality," in The Public Economy of Urban Communities, ed. by Julius Margolis (Washington, D.C.: Resources for the Future, Inc., 196A), pp. 52—73. Earl O. Heady, Economics of Agricultural Production and Resource Use (New York: Prentice-Hall, Inc., 1952), p. 113. Kelvin Lancaster, Consumer Demand: A New Approach (New York: Columbia University Press, 19713, pp. 6-12. Pauly, o . cit., pp. 21—35. \ I. CHAPTER III EVALUATING HEALTH PROJECTS Good health is one of the most desirable aspects of a person's life. Despite or just because of its central posi— tion in an individual's utility function, society faces seemingly insurmountable problems in wrestling with the concept of good health and in organizing resources in such a way as to achieve a satisfactory state of good health for its members. At the root of the problems is the lack of a clear defi- nition of good health and how it is produced. Only a satis— factory definition of health will permit us to tackle another set of equally disturbing problems, i.e., problems of distri- bution of health and health producing goods and services. This chapter identifies the position of this particular dissertation in the health research field. The guidelines 0f the previous chapter are applied in identifying relevant input-output categories and existing data are examined in terms of their adaptability to evaluate research. A model 0f the "Health Status exchange system" and a model of the "Health Status production process" are offered as the frame— work for empirical analysis. 26 . . ct- 27 1. A Summary 'of Health Services Research For the following discussion it seems advantageous to divide research efforts into the following broad classes: (a) "Intermediate" Input—Output Evaluation and (b) "Ultimate" Input—Output evaluation . l The terms "intermediate" and "ultimate" are based on the discussion in the previous chapter. They are chosen in order to indicate how well the analysis succeeds in evalua— ting the project's contribution to "man the product." in other words, "ultimate" is a relative term and no value judgment. a) "Intermediate" Input—Output Evaluation Studies falling under this category assume the desira— bility of certain health services and evaluate the process by which those services are produced and delivered. The most prominent representatives of this group are (i) operations research studies which investigate problems such as the allocation of hospital beds, staffing 0f hospitals and scheduling of patients. (ii) Health services utilization studies if they are Only interested in the "proper" or adequate utilization of health services. (iii) Peer review evaluation. This procedure consists of establishing whether patients get that kind and extent of care which is dictated by current medical knowledge and professional standards. ‘l'|nb- I‘l\ .- 'v u ch “'2!- , l "Woo- v Q“. 41. ‘3' M ('11 Q N 28 b) “Ultimate" Input—Output Evaluation Studies following this line of thinking are mainly in— terested in the impact of certain procedures, projects or programs on health related outputs. The various activities in this area are: (i) Epidemiological studies. These studies establish relationships between harmful or beneficial conditions and the incidence of disease. (ii) Project evaluation of the cost-effectiveness variety which determines the impact of health services on certain disease-or health related characteristics and the effect of these on human activity. (iii) Project evaluation of the cost/benefit variety which i attempts to assign a monetary value to the impact of health inputs. The focus of this study is on organizing an "Ultimate" Input-Output evaluation. Both the cost—benefit and the cost— effectiveness approach assume a high degree of information on health outputs and what is even more important, they assume a knowledge about the health services—health (1-0) relationship. This information about health outputs and their relationship to health inputs does often not exist. Epidemiological studies, on the other hand, attempt to Specify an input—output relationship (cause—effect). This thesis is essentially of the epidemiological variety. Traditional epidemiological studies have concentrated on associating the presence of diseases with various causes.2 29 The present study by contrast attempts to associate the level of health status with the presence of various health input combinations. Knowledge created through this kind of a "epidemiological" input—output analysis is a necessary and thus far often unobtainable ingredient for cost- effectiveness and cost-benefit analyses. After having established how this thesis fits into the general picture of health services research it is appropriate to follow up with a discussion of the relevant inputs and outputs embodied in this type of health services evaluation. 2. Relevant Input—Output (I—O) Relationships The health and health services sector consist like other systems of a vast number of 1-0 relationships. We have X-ray machines "producing" diagnostic materials, doctors who "produce" operations, hospitals which "demand" capital for further investment and last but not least, patients who "demand" health services. Chapter II indicates that re— search will only be successful if it is initiated by a care- ful specification of relevant I—O relationships. a) ijectives of the Investigation. Relevancy is largely determined by the objective of the research and the questions asked. If it is of interest to find out M many doctors and nurses are needed to perform (i.e. produce) a particular operation, the relevant inputs will be doctors and nurses. On the other hand, if the ob- J'ective is to determine how to produce a "good" Operation most efficiently the relevant inputs might be skill—level of . . ... or Ia “nod. 0". “.... ... . "a. . I“? ...“ u r... \I 3O performers. Should the objective be to make the patient healthy through an operation, the relevant output will be a healthy patient in both cases. Most peoplerwill claim that the last objective (pro— ducing a healthy patient) would be the only relevant output to them if they were the patients. Unfortunately, very little research has been produced which identifies the healthy patient as the ultimate output. Much of the evalua— tion research assumes that the services as provided have a beneficial impact on the patient's health and that all services are equally needed. This thesis departs from the assumption of a given relationship between health services and health. More so, it attempts to define this relationship. The objective is not to establish a relationship between special types of medication or care and health but to specify the relation- ship between certain institutional forms and arrangements of providing health services and the health status of the popu... lation under consideration. The stated objective suggests that the relevant I—O relationship for this particular kind of investigation is the relationship between health services and health. b) The Output—Health and Its Measurement The World Health Organization defines health as "a state of complete physical, mental and social wellbeing and not merely the absence of disease and infirmity."3 Attempts to utilize this definition in practical research .- 1.1“ i. ; u g 1 . .. f '0 ~- .. ~ -" -' 1“. Hy,- Cr- v“ .1- u ‘. “a s . H. .i Kg ‘5 . ‘— \H ‘M I 7—7 ) 31 have shown that it is impossible to operationally define either complete wellbeing or the absence of disease.” However, the definition is intuitively appealing since it directs our thinking to the achievement and maintenance of health and away from concentrating solely on the treat— ment of disease. This focus is clearly the basis for the current emphasis on providing health through Health Main— ) tenance Organizations (HMO's).5 Mortality Indicators The lack of a clear and operational definition of kmalth and the ease to define death explains the tradi- tional reliance on mortality rates as one of the most accue [ rate health indicators. Mortality rates are indeed useful I long—term indicators of the health status of populations. Hwy have been successfully used in documenting the achieve— nwnt of modern medicine over time and in cross—country comparisons.6 fkalth and Disease Indicators Unfortunately, the total spectrum of health which lies between "perfect" health and death is only roughly captured in an index of mortality. The inadequacy of mortality in— dices becomes even more significant if one considers that the health status of the total living population is some— Where on this continuous line between "perfect" health and death, a line which we apparently can only measure with great difficulty. A definition of the health status of the living population is, therefore, the basic requirement . 'F‘ 1 .1. a "I“ ' . I Vvdtu . . ”u . . “I... -;r. Hui :lP I__——v 32 for rational planning in the health sector. Such a health status index has to be useful both in establishing need levels for further action and in evaluating the accomplish- ment of those programs. But what are valid health status indicators? A multi— tude of professional papers and extensive research have addressed themselves to thisproblem. A review of the literature on this problem suggests a division of the health status indicators into disease-oriented indicators and adjust— ment—oriented indicators . 7 c) Disease—oriented Indicators They are indices of the incidence and prevalence of disease and disabilities. They do not specifically con— sider the impact of the particular disease on the individual's behavior. Their acceptance as an indicator of health depends on our society's implicit understanding of the impact of diseases. Usually, however, such understanding is lacking, eSpecially where individuals have already adjusted to diseases and disability, e.g. have chosen a profession or life style which permits them to be fully productive despite disease or disability. Another critique voiced against disease— -oriented data as a measure of health status is founded in the difficulty of standardizing the collected information, i.e. when is something a disease and when is it just a benign conditions? Disease oriented data are, however, of extreme im— portance in epidemiological investigations where relationships 33 between causes and specific diseases are established. Their value in this kind of analysis explains probably why so many and varied disease—oriented data have been collected. d) Adjustment—oriented Indicators This type of indicator attempts to identify health status by the individual's adjustment to disease and disa- bility. This adjustment manifests itself in two ways: (i) by taking curative action (curative action indicator) and (ii) by changing the kind and level of "usual" activi— ties (role—fulfillment indicator). For instance, a person who has a disease might have to adjust his behavior by taking aspirin (curative action), by doing only light or rm work (activity reduction) or by staying in bed all day (curative action and activity reduction). This example indicates that the two types of indicators do not describe completely different events but are often overlapping. People change their behavior in light of diseases because the new kind of behavior might cure the disease, might prevent the disease from progressing and/or is the only way to live with the disease. Curative action indicator Taking curative action as an indicator of health status produces considerable methodological problems since it uses the level of inputs (the curative action) as a measure of the output (health status). This assumption that curative action depends on the individual's perception of his health Status is the rationale behind using curative adjustment as an indicator for health status. Input levels, however, are +1,» I 1“. s "5‘ I." ..h ._ __._._... . _ 34 in many instances more a function of the knowledge and pur- chasing power of the individual than of his health status. In addition, they depend heavily on the availability of the curative services. A third disadvantage is embedded in the fact that health services are consumed not only to cure a disease but also to prevent further diseases. Extracting that portion of services which reflects the health status at a given point in time creates often insurmountable dif- ficulties. Role fulfillment Indicators Another form of behavioral adjustment to disease and disability is that adjustment which affects the functional role which individuals are expected to play in society. This definition of health is again based on a common under- standing of what constitutes a person's role in society. This dissertation assumes that individual roles can be de— fined for particular segments of the population. The advantage of the "role fulfillment indicator" is its output orientation. This indicator meets most closely um requirements of an "ultimate" output or outcomes— indicator, as was outlined in the previous chapter. In other words, the indicator captures the "characteristics" 0f health, i.e. the degree of ability to fulfill one's role in society. Suggested measures of a "role fulfillment indicator" would be activity restrictions in play, work and recreation, days of school missed, days of play and recreation lost, i. 3'4 In. ‘4. ’u. u“. ’7- . s “u 35 days of work lost. An ideal indicator would also dif- ferentiate between complete activity loss and activity re— duction. (Selma Mushkin refers to this situation as "debility.")8 e) The Inputs - Health Services This thesis will treat health services as inputs al— though it is obvious that they themselves are outputs of a different level production system. For analytical pur— poses it is important to group health services into rele— vant categories. Classification is usually provider oriented (e.g. hospitals, doctors, beds, etc.), disease oriented (T.B. clinics) or patient oriented (e.g. baby clinic, nursing home). Most classification systems rarely seem to satisfy the requirements of evaluation studies as outlined in the preceding chapter. There it was emphasized to classify inputs by attributes. Which characteristics or attributes are relevant will depend largely on the focus of the evalu- ation study. Health services may be investigated in terms Of their attribute of capital intensity or in terms of the Skill level of providers or in terms of their "preventive— ness," or, as suggested in Appendix A of this study, in terms of concentration ratios. The choice of the relevant output will depend on the reasons for using health services. Realizing that we de- “End health services not in themselves but because of some (or all) of their inherent characteristics we have to ask 36 what role those services play in the production process. Again, we rely on the notions of "characteristics" in choosing a relevant classification of health services. 3. Data Research Research on data for health services evaluation has been significantly increased over the past decades partly because of the increasing capacity of electronic computers and partly because of new development in social science thinking. a) Social Indicators and Program Indicators Data used by social scientists are often divided into Social Indicators and Program Indicators. Social Indicators are defined as indicators identifying conditions and prob— lems which are of interest to the collecting agency. (Most of these indicators are collected at the national level although there are some local social indicator efforts)9 Program Indicators are defined as indicators which describe the impact of programs (or projects) on various variables Of interest. b) Confusion in Indicator Terminology The term "program indicator" is often misused and con— sequently misinterpreted. It is a common practice to collect information on outputs of programs and call this informa- tion program indicators. Such a procedure does not really eStablish the output which has been produced by the program but just counts the output which exists in the presence of 34‘ : - III ’M u I‘ I? .. v. ‘ o 37 the program. Yet, this output might have been produced (and usually is produced) by a variety of other inputs p93 supplied through the program. The procedure is a "before and after" analysis and not an evaluation of impact. What characterizes an evaluation of impact is a "with and with— out" analysis, which can only be accomplished by isolating the impact of project influences from non—project influences. Without associating specific outputs with specific input activities Program Indicators and Social Indicators describe the same phenomenon. Although this is a convincing argu— ment much confusion exists in using and interpreting the two concepts. This study, therefore, suggests that the impact of a particular program on a particular output indicator be called the "Program Output Coefficient" (POC) and that the terms ”Social Indicator” and "Program Indicator" be used in describing the ppapg of the output. This distinction can be represented in mathematical terms by the equation I = c + p (Xp) + n(Xn). where I is the state of the output measured by its indicator and p is the "Program Output Coefficient." (C is the con— Stant term in the equation,Xp represents program inputs, alrepresents non—program inputs and n is the output co— efficient of non—project inputs.) In summary, Social Indicatonsand Program Indicators describe states of the output (they are absolute measures) While "Program Output Coefficients" describe their relation u. tfi‘ u.\ a,’ ,1 '( 8 ' 3 to programs—~they are measures'relative to inputs. The division between Social Indicator and Program Indicator should be reserved for differentiating the objective of the collecting agency. Indicators which are collected with the evaluation of a specific program in mind should be called i Program Indicators. Indicators which are collected to record socially relevant events should be referred to as Social Indicators. A. Health Data Collection a) Nationwide Efforts Data collection for national social indicator purposes has been going on for several decades and is largely organized through the National Center for Health Statistics. The basic instrument for data collection is called the National Health Survey which consists actually of several different surveys which are divided into three parts:10 ' (i) The Health Interview Survey-—a continuing nation— wide sampling and interviewing of households; (ii) The Health Examination Survey-—physica1 examination and testing of samples of individuals; and (iii) The Health Records Surveye—sampling and interview— ing of organizations and institutions related to health care. Summaries and analyses of the findings of the National Health Survey are documented in 22 different "vital and Health Statistics Publication Series" which are published by the National Center for Health Statistics.11 Ii 4. :3- .LI 'l'j 'V (J) ( I.) 39 b) StateWide-and'Local Efforts, PrOgram and'PrOJect'Data Much of the experience gathered in establishing the National Health Survey and in subsequent analysis has been useful in research directed towards program evaluation. Due to the variation in programs and projects and due to the lack of universally applicable evaluation procedures, no uniform data system is available at the project or local level. Deshaies and Seidman give an excellent summary of avail— able components of health information systems.12 They divide the available data base into the following groups and discuss how the information has been or could be obtained and what problems are connected with using the particular data: (1) Status of community health consisting of 19 component indicators. (ii) Utilization of health services (18 different measures are reported.) (iii) General population and housing characteristics: (50 component indicators). (iv) Inventory of health facilities and health man— Power (43 component indicators). (v) Status of community environment (38 components). The paper is an outgrowth of the author's involvement in the New Haven Census Use Study which is the first large scale regional—local effort of generating health data for local indicator and project evaluation purposes.13 it cl! no. I I ‘1 § \I.‘ ‘I\I ‘.I Ab . v F 1 Huh A-i . ... «\u :. . ”if u n 3 ~ «\~ ...k n... b0 5. A'MOdel of the Health Status Excugnge‘sysggg From the preceding discussion, it is now possible to establish a model of the health production and exchange system which can set the stage for an empirical investiga— tion of the health production process. The block-flow dia- gram of Figure 3 serves as a simulation of that system. The model differs from traditional presentations by consider- ing "Health" a tradable output instead of treating health services (operations, beds, etc.) as outputs of the system. Health services are treated as inputs in this framework. a) The ”CommodittHealth Status The model employs the concepts which were explored in previous chapters by indicating that health is not demanded as an end in itself but because of its characteristics or attributes. These attributes of health permit the individual to fulfill a socially desirable role and to fulfill his role in a way which is acceptable to him and his environment. What are socially desirable roles? Members of a hetero— geneous society have many roles, but, as previously sug- .geSted, it is possible to isolate some major groups of roles: Infants are "expected" to play, when they grow Older they are "supposed" to go to school and play. Adults are "supposed to" work on the job and around the house. Re— tired people "should" be able to do a little bit around the house. The four major roles (play, go to school, go to work, work at home) are summarized by the term Health Status. Health measured by the summary term of its attributes (i.e. Health Status) is at the focus of the system displayed in Fig.3. Exchange Model P dividual Demand for Health Status Total Demand for Health Status: e.g. Play P School Work Rblic Denand for “work Health Status Q _ Health Status - Pkwsiological Health Related - State of Environment Individuals of Individuals _ | —/_ Health Status Producing Custom Services According to rIheir Remilifiions maractezdstics: Laws e.g. Diagnostic Decisions vent ive Treatment ‘ Producers of Health Services: Doctors Nurses Omanization Dentists of Health Status ___. Dental Nurses Producing Family Aides Services Drivers Administrators Schools Extension Service Individual Himself Fig. 3 — A Simplified Model of‘ the Health Status Exchange System. 4 1 Il‘! Her ‘VUI u ...“. V... I o-A' ‘54. In t.. o'- 7! M1 V .7 ‘u 42 b) Exchange'of Health'Status Individual and public demand together determine the total demand for Health Status (play, school, work and home- work). This Health Status is supplied by the health produc- tion process which acts through improvements and maintenance of the physiological states of individuals and of the health related environment. To make the variety of health status producing inputs manageable they are disaggregated accord— ing to their characteristics. These characteristics can be produced by many different health services producers, such as doctors, nurses and the individual himself. 0) Qggahizing and Reorganizing the Production The present form of organizing the production of a desirable Health Status is governed by a set of customs, rules, regulations and laws. Thus a female patient may or nmst not have an abortion; an abortion "expert" may or must not perform such an abortion. A nurse may or must not nlake an examination. These rules were once established be- cause a majority (representative or not) decided to d0 30 and because the particular organization of the services Seemed to satisfy needs. If we think that needs are not adequately met (given the available resources), we look for Ways to reorganize the present production process. The interplay of demand and supply of Health Status Will determine prices and quantities of Health Status which is the measure of success or failure. Dissatisfaction with prices and quantities of Health Status will put pressure 0“ up: {do 2;," a‘v'u. a,“ It: ‘0- n ”71‘ V.- n! '0' g n.‘ ‘ \ L 143 fixeorganization of the health producing services. The reorganization can occur via technical improvement and/or Based on our ‘Uupugh changes in the rules and laws. previous discussion, the suggested way of successfully re— cmganizing is by concentrating on the actual productive characteristics of Health Services producers. The sug- gested characteristics of inputs are "diagnostic," "pre— ventive," "treatment," yet any other relevant grouping is possible. 6. A Summary Model of the Health Status Production Process Since this study concentrates on establishing a pro- duction function of Health Status as the basis for evalua- ting health projects, aImNbematical summary of these func— : tions is offered herein. Portions of the model will subse— quently be applied to a case study of evaluating a health project. Health is represented by the indicators of Health Eflatus which is expressed by a vector H. hl ha wherehl = play days, h2 = schooldays, h3 = work days and rm = homework days. These are the indicators dealt with in this study. Other research objectives will require con~ Efldering additional indicators of health status. For 1IIIIIIIIIllllllll"""""""' ; an by the attitude of the worker. Similarly, mortality rates (a negative measure) can be included, since they represent complete non-fulfillment of roles. A production function between Health Status H and ’ Health Services HS is established.by H = f(HS) Since Health Services consist of many different (often non— additive) types of services all services are disaggregated into Health Status producing characteristics cl . . .cn (diagnostic, preventive, treatment, etc.). These charact- f eristics are represented by vector HC. An advanced model would, therefore, establish a produc— tion function of H = sLc w ;+_mc: o>_mccchaeoo cca_cu_2 ccw+mwz so co_+mN.cmmLo .v .m_u :. EOL+.MWMWW>OL¢ c_3c_mm c_zu_mm ++m+m w+_mo+cm EOL+ mLmu_>oLa mc_+m+0L EOLw mev_>oLa mcp+m+oL A>+caoo om>mzczv w=_o ++m+m m+__m+cw w:.a +*m+m o+__o+mm use—o w+.zz . A>+czoo cemmzv A>+czoo ow+m_cmzv u_c__o m+_mo+m o___>+oom m>o_c¥ . m omcm_o 0+_.o+mm o_c__o w+__o+mm 04 5 r moo_>me _c+cmo woo_>Low .mu_um2 moom>me >__Emu :o_+mh+m_:_En< 3853 3:55 En: >+czoo cxmn .:_3c_mm 'a 1“ r . . t I“ 53 (county seat of Lake County). Family services are located hla separate building ("Annex") in Baldwin. "Satellite" Clinics are located in each of the adjacent counties (Scott- ville, Kaleva, White Cloud). The Main Clinic and Family Ser— vices have a full time staff, five days a week from 8 a.m. to 5 p.m. (emergency services have been reduced because of cuts in funding). The Satellite facilities are staffed part time. Yet, all types of services, i.e. medical, dental and family services are dispensed to clients residing in Hm Satellite counties by rotating providers from the Bald— whuCenter. Hospital cases are referred to hospitals in Imdington (Mason County) and Manistee (Manistee County). Appendix Table BA summarizes the budget of the past years. 2. Rationale for Evaluation When, in 1971, funding of the project was reduced, it became evident that the administrators (HEW and MDPH) did rmt have sufficient information for identifying the contri- bution of various program parts (medical, dental, family services) to the achievement of the program objectives. Yet, they had to decide which elements of the project were to be reduced and Family Services experienced the largest reduction in funds. Someone in the line of project adminis— trators felt that Family Services output was not worth their nlarginal costs. Was he correct? He was not, according to the Director of Family Services. Neither of the parties lmd a firm informational basis for defending its position. At the same time, it was well known to the project 4'! v. «5' Ci“ "ct 'v '7!- h... "Ii - V... 514 administrators that some area residents including project clients were convinced that the project as a whole "was not worth its money." Yet, the waiting rooms were filled wiUIpatients every morning. Was it used only because the cost to the user was subsidized? What were the benefits of this project? a) Previous Evaluation Attempts OEO monitored all projects and issued quarterly state- nwnts on the progress of each Health Center.7 Yet these I“eports‘gave mainly information on inputs and no informa— tion on the output. In 1971, OEO sponsored a more output- Mfiented evaluation of its health projects. The objectives 0f Hus evaluation were stated as follows:8 "1) Assessment of the Center's success in reaching Hmir target population and determination of the extent to Whidleligible persons are effectively using the health Care services provided. ii) Determination of the degree of patient satis- Isetion with the care provided. iii) Assessment of Center progress in developing a System capable of providing adequate care. iV) Determination of programmatic similarities and dlifferences between Centers and their relationship to performance. Determination of the feasibility of developing V) amethOdology to measure the anti-poverty impacts Of Center Services . H Although the objectives of the GEOMET study emphasizes an01117131113 orientation it fails to show whether the given care helped the target population in improving its health Status. Additionally, it does not show whether different :rnb n-J'J 1 on!!! "I'd - 'an - '1‘ . . . Q ~ ‘. . . ‘- v ‘u u « \ x' 55 institutional arrangements or different input combinations produced different output results. The most promising part of the study is the development of a methodology to measure Hm anti—poverty impact of health center services. b) ijectives of this Study—-Measuring the Output The author's goal was to research the bottleneck of the GEOMET study and many other evaluation efforts, i.e.the isolation and measurement of indicators which could ' be useful in identifying the Health Status output of the Health Center. 3. Procedure of Evaluation The model upon which this evaluation is based was mmmarized in Chapter III by the equation H = f(HS), i.e. Imalth Status is a function of Health Services inputs. I Toestablish this relationship it was necessary to select relevant variables describing input and outputs. a) Choice of Output Variables The concepts of Chapters II and III served to develop alist of possible output indicators. Research in the social and natural sciences was reviewed in order to check ipr previous efforts in measuring the selected variables. Survey methods research and especially the methodologi— cal research of the National Center for Health Statistics kmlped in narrowing down the list of suitable output indicators. Mm following "role—fulfillment" indicators emerged as final candidates: 1) Days missed from work because of health related reasons . i 5. ii) Days when health conditions prevent the individual } from his usual work at home iii) Days missed from school because of health related reasons iv) Days when health conditions prevent children from their usual play. Other indicators, such as an index of activities of daily living (ADL)9, attitudes, community involvement and family finmtioning, were considered as outputs, but were discarded because of resource and time limitations and because it was felt that they were indicators of a different output level. b) Choice of Input Variables Since one of the main functions of an input-output analysis is to facilitate the reorganization of inputs it was decided to group health services according to Characteristics which reflect reorganization possibilities. The following groups emerged: i) Location of service (main clinic, satellite clinic) ii) Type of services (medical, dental, family services) iii) Concentration of utilization of services 1V) Type of provider a) profession (doctor, nurse, aide) b) skill level (skill required to perform function) V) Degree of capital intensity of services . '1 ; I I- ,.|- . n. ."w- 57 A. Data a) Output Data Data on health status output as defined in this study (i.e. role fulfillment) have not been collected by the imalth Center. It was, therefore, decided to collect this information directly. b) Input Data Data on health services inputs were collected by the Health Center. The storage of these data on computer tapes promised to make the handling of input variables a simple task. Yet, an examination of the collection procedure revealed a series of shortcomings. C) Limitations of Input Data (i) Continuity of selection (all data). Wm data collection system has been effective only since 1969 and has been reorganized twice since then (April 1970 mm July 1971). This reporting period is most likely too Short for evaluating the impact of health services. (ii) Accuracy of reporting (Medical and Family Services data). The following discussion refers mainly to the col— lection of data by the family services department and by the nwdical department. Records of the dental department seem tobe more accurate and more complete. The author feels that the reporting system attempts to get too much information from too many people. A short description of the data collection process should clarify fins statement. Every patient—provider contact is recorded . v a sat :9, a u... - -;.I ”H . :7- w“ 1'. . ——— 58 in the following manner: a) personal information, b) eligibility status (payment), 0) reason for visit and (1) services received. Especially information items (c) and (d) are of interest to this research. Yet, an ‘ examination of the collection process reveals that this é information suffers from serious inaccuracies. Most services are performed by teams of doctors, \ nurses and aides but usually only one provider is listed. Also, only one or two types of services rendered are re— ported, even if patients receive several types of services. Similarly, only one or two reasons for visits are listed although many patients are treated for a variety of ail— ments. A different problem arises because of privacy and confidentiality. An investigation on family planning data revealed that several doctors do not report services connected with family planning because they know that this information (in connection with a specific name) will pass through the hands of various nurses and aides before it i becomes an anonymous statistical entry. The problem is aggravated by the fact that most employees preparing the data are members of the projects indigent target population. d) Suggestions for Improvements of Inflt Data Collection (Medical and Family Services Records) A few improvements of the collection process should be recommended here: (i) Simplification of collection forms. The Michigan Department of Public Health (MDPH, Center for Health Statistics) III 1 F!“ u ...] vp. h... u h. I" .‘n ‘~ ‘ v -' ls; db —’— 59 is presently engaged in accomplishing this task. Teaching the utilization of the new forms will contribute to a reduction in inaccuracy. (ii) Sampling; Even with new forms, it will be im— possible and/or costly to fill in all details of the visit as called for by the data form. A sampling procedure would reduce the number but increase the accuracy of reported data. Experiments are needed to guide the sampling method. (Every twentieth patient, or patients with particular charac- teristics, etc .) (iii) Task Analysis. The present data system only sketches what the various providers do and whom they serve (e.g., phone calls about the use of prescribed medications are almost never recorded, but consume a considerable amount of the working hours of some providers). A task analysis is the only reliable procedure to analyze the performed functions. (iv) Consolidation of records. In addition to the information required for the computerized reporting system the medical department collects and stores so—called "patient files." Information for "patient files" is col— lected separately and causes additional work. Since doctors rely on "patient files" for their diagnosis they tend to make their entries complete and accurate. Organizing "patient files" in such a way as to permit their use in eStablishing the statistical records would reduce the bur— den and increase the accuracy of collection- fIIIIIIIIIIIIII_________________________———5 60 (v) Problem oriented charts. One way to reorganize "patient files" is by adopting the "problem oriented chart" which bases the report on the problems as perceived by the patient and not on the doctor's diagnosis.10 Thus, the chart lists "cannot bend over" and not the diagnosis of this problem. It is interesting to point to the similarity between the "problem oriented chart" and the "role fulfillment indicator" which has been advanced in the methodological part of this thesis. 5. With/Without and Before/After: The Problem of Experimental Design Chapter III emphasized that a project evaluation can only be achieved by a "wipg and without" analysis and not just by "before and @3333" analysis. The"with and without" analysis specifies clearly what amount of output was pro— duced by the project inputs. The "before and after" analysis only indicates that outputs have or have not dmnged since the project took effect but it does not identify the specific causes for this change. The "with an without" comparison utilizes the methodology of experi_ Hmntal design by isolating specific input—output relation— ships. The characteristics of an experimental design are Control and random selection.11 The design will usually Consist of (i) stratification of the pOpulation to be treated according to relevant variables which are measured mm known before sampling, (ii) classification of the “In dal- . a“ IIIIIIIlIllllllllI----::——————————————- 61 sample according to relevant variables which are measured after sampling, and (iii) assurance of random occurrence cfi'relevant variables which could not be measured either before or after sampling. Different situations require dif- ferent steps to achieve a controlled experiment. The major obstacle to a "perfect" design is the limited ability to classify or stratify the population according to relevant characteristics. Relevancy is deter— nflned by the influence which these characteristics can Inwe on the effect of the treatment. Put differently, these relevant (but unmeasurable) characteristics act like "inputs" (or independent variables) on "outputs" (or depen- dent variables). In the case of health evaluation research we face the IHpblem of not being able to adequately identify the popula— timm Physiological and mental states are relevant variables and influence the outcome of the treatment. Comparing the health status output of individuals with different treat— ment levels would not be a proper analysis without accounting fin*differences in health status due to different physio— 1C’gical and mental states at the outset of the "experiment.” One way to overcome the difficulty of classifying in— chviduals by their physiological and mental states is to make sure that those states are equally represented in the mnmrol and treatment groups. This can only be achieved indirectly by selecting the control and treatment groups ENJrandom from a parent population. The random selection ‘q ‘5 62 of these groups will assure an equal distribution of physiological and mental states, but it will not always assure a completely controlled design. Only by randomly applying treatment to the individuals within each group we can be sure of such a control. This might seem to be a redundant requirement in the case of most physical experi— ments. Yet in the case of health care it will be often difficult to treat all individuals within a particular treat: ment group exactly alike. If it is impossible to give accurately defined treatment to accurately defined treat— ment groups it is necessary to avoid self selection of treatment caused by unknown criteria (of providers and 7 clients) which could not be included in the analysis. The best way to achieve this is by random application of the treatment. The procedure as described in the preceding paragraph seemed to be infeasible because of time, administrative policies and resource constraints. It was, therefore, de— ; cided to resort to a second-best approach. This approach involved selecting a "comparison" (control) group which resembles the treatment group in a variety of relevant criteria. 12 Criteria were considered relevant when it could be assumed that they would either be associated with M‘have an influence on physiological and mental states. It was assumed that variables such as income, employment and degree of industrialization were relevant ones. Yet umre was no assurance that the chosen variables were an I u'. I... x" _r., ‘~.I ' i" l 4,, 63 sufficiently capable of classifying the two pOpulations into identical groups. Therefore, there was no assurance that health services in the control and treatment counties vmre not "selective" in their treatment of particular in— cfiyiduals. Further experience and better theories might advance the "art" of detecting and isolating this bias. 6. The Problem of Observation Over Time Most of our public projects are productive over a long pmriod of time. In the case of health projects and addi— tional feature emerges: Health projects have a long "gestation period," i.e. it takes a long time to produce mnrbenefits at all. On the other hand if administrators wmnsto know how certain output indicators "behave" iflmy do not want to wait until "time shows the results." hfi;administrators will have to learn to live with this <fimracteristic of the product ”health." The only way how the'Wmiting time" can be somewhat reduced is by develop— hm;sensitive indicators and by employing designs which establish accurate cause-effect relationships. 7. Evaluating the Lake County Health Center Many requirements for an ideal design were not met in ‘Um Lake County case. Baseline information on role ful~ ifillment indicators was in existence only in terms of school eutendance records of the local school. Physiological and nmnufl.states of the pre—Center period was available to some dance in the form of the medical and dental records of ‘ .- f 1" ... . n 1 ’i— 64 clients taken at the time of enrollment in the project. Yet, only elaborate efforts would have permitted an inference of classes of physiological states as called for by the pre- ceding sections. Without a basis for classification of physiological states it was impossible to evaluate the impact of different treatment levels. It was equally impossible to evaluate the impact of similar treatment levels on different classes of physiological states. The amount and quality of information made only second- best procedures feasible. It was decided to experiment with two approaches. One was to apply the format of an "ad—hoc comparison" of the Lake County pOpulation with a pOpula- tion from another county. The other was to analyze the impact of the Health Center on school attendance by examining school attendance records of the Baldwin area school system. a) An Ad Hoc Comparison It became apparent that the only way one could estab- lish a control comparison population for the individuals treated by the Health Center was by identifying a comparable population which lived in conditions similar to those of the Lake County population during the past several years but did not succeed in getting a health project of the Lake County type. It was decided to use economic and demographic criteria for selecting such a population. Data on the quay]- tity and availability of area-wide health services were assumed to assure that the comparison group had received the same amount and quality of health care as the treatment 65 group yggld have had without the availability of the Health Center. Once the two groups were established it was pos— sible to administer questionnaires to a particular section of the two groups. The survey produced indicators of health status which were used to establish the impact of the Health Center. A detailed description of this method ipllows in Chapter V of this study. In SChool Attendance Records as Indicators of Role Fulfill— ment Since these were the only available output data describing both the pre-Center period and the with-Center period it was decided to investigate their usefulness for a.health status evaluation. A description of this research is offered in Chapter VI. 1. ll. FOOTNOTES Manistee COunty: ‘Economic Trends, Suggestions for the Future, Reports of Class Projects of Resource DeveIOpment 816 (Michigan State University, Spring Term, 1965). (Mimeographed.) Mason County Overall Economic Development Committee, Overall Economic Development Program, (Grand Rapids, Michigan: Williams & Works, 1966). Lake County Planning Commission, Lake County: Com— pgehensive Area-Wide Plan for Water & Sewer Service (Grand Rapids, Michigan: Williams & Works, 1971). Newaygo County Community Development, Committee Report, 1967. (Mimeographed.) John Malcus Ellison, A Social and Economic Study of the Negro Problem in Lake County, Michigan‘(Lansing, Michi— gan: Land Use Planning Section, Resettlement Adminis- tration, Region II, 1936). (Mimeographed.) John Malcus Ellison, Op. cit., p. 7. Office of Economic Opportunity, Comprehensive Health Sgrvices Projects: Summary Report (Washington, D.C.: Office of Health Affairs, Planning and Evaluation, Second Quarter, 1971). Office of Economic Opportunity, Study to Evaluate the OEO Neighborhood Health Center Program at Selected Centers, GEOMET Report No. HF—7l (Washington, D.C., 1972) . pp. iii—vi. S. Katz, et a1., "Progress in Development of the Index of ADL,” Gerontology, No. 10 (1970), pp. 20—30. The author was introduced to this concept by Dr. David Nielson of the Western Michigan Comprehensive Health Services Project, in Baldwin, Lake County. F. B. Baker, "Experimental Design Considerations Associated with Large-Scale Project," improving Experi- ‘pgntal Design and StatiStiCal AnalySis, edited by J. C. Stanley (Chicago, Illinois: Rand McNally, 1967), pp. 206"56o 12. 67 Houston recommends the term "ad hoc comparison" for He points out that in such a situation this approach. the term "comparison group" is preferred over the term "control group" because the latter implies that there is sufficient control, a condition which is See Tom R. not guaranteed in an "ad hoc comparison." "The Behavioral Sciences: Impact~ Houston, Jr., Effectiveness Model," Evaluating Social Programs: Theory, Practice, and Politics, edited by Peter H. Rossi and Walter Williams (New York: Seminar Press, 1972), pp- 51-65. CHAPTER V IMPACT OF HEALTH CENTER: METHOD 1, VARIOUS OUTPUTS Chapter IV pointed out the problems of designing a fmalth project evaluation in a way which would facilitate a."with" and "without" analysis. It was pointed out that, because an ideal experimental design seemed infeasible, second—best alternatives had to be explored. This chapter annoys a design called an "ad hoc comparison." Houston describes this procedure as follows: "Here units who were exposed to a program are compared to units who were not, and differences are interpreted as program effects. This procedure may be refined by selecting the comparison group (it is misleading to call this a control group) in such a way that it resembles the program group in various respects." 1 ’Nfis study concurs with Houston that the lack of random Emsignment to the two groups is a basis for misinterpreta- tflon of the results due to bias introduced in selecting the mmmarison group and because of the inability to control the txeatment of the units within each group. But despite its flaws it is believed that this design (or modifications of it) is economically and practically more feasible than designs Imnch meet the requirements of experimental designs more (flosely. Since "better" designs seem to be infeasible, em— Iflrical research has to concentrate on improving "imperfect" CieSigns. The objectives of this chapter are, therefore, 68 69 (ndented more towards gaining experience with an imperfect methodology and not towards producing "hard facts." l-WW The following variables were considered to have an effect on the health related characteristics of individuals: (i)cflimate,(ii) distance from metropolitan areas and in- do app receive any form of public assistance it was decided to differentiate the sample into those two groups. The two groups represent different income levels within the eligible population. This stratification helped also hiisolating the effect of "financial access" to health services from the effect of "physical access." This isola— tfion seemed necessary because it was hypothesized that someone meeting public assistance income guidelines would get at least financial access to health services via Medicare or Medicaid, even without the Health Center, while those just above the Public Assistance income guidelines would not have fins option. Of course, without a Health Center both groups suffer from a lack of "physical access" to health care. 5393 (X2 in regression) Lake county has a large black minority (23 percent of lake county's residents are black) while there are no black residents in Montmorency county. To isolate racial dif- ferences it was decided to separate the Lake county sample into.a "black" and a "non—black" group. 77 Perceived Changes in Income 1968+]2 (X9 in regression) This variable was chosen as a proxy measure for changes in real income. Again, it is emphasized that a change in real income would be one of the "intermediate" mflmuts of the CEO Health Project which should be studied separately as a dependent variable. This analysis treats real income changes as independent variables. Eflucation of Head of Household (Xlu in regression) The level of education is of considerable importance ih.determining a person's health.2 This study stratified 13m population into one group which consisted of families \ 1 vfiwre the head of the household had eight or fewer years cfl‘schooling and into another group where the head of the rwusehold had nine or more years of education. &ufitary facilities and heating (X5, X6 in regression) Epidemiological studies have established a relation- Nup between the health status of a population and the 3 Inesence or absence of certain environmental conditions. IUthough both counties did not report any water borne (fiseases it was decided to stratify the sample according Mathe availability of sanitary facilities. It was believed lflmt sanitary conditions might serve to capture socio— ecmumfic conditions not represented in the other income related measures. Similar considerations led to the in— mhmion of home-heating as a variable for stratification. CA m y .. “JO-AV I“... n...‘ n "‘ n nh‘ 1.: I- ~ \ V" I. 1.1o . is ‘\L IIIIIIIIIIIIIIIIIIIll-llll---——________ 78 It should be emphasized that the improvement of both kmme-heating and sanitary facilities are explicit targets of CEO Health Centers. They are "intermediate" outputs of the Health Center, if one uses the language of Chapter II cfi‘this paper. It would be advisable to make an additional evaluation of the Health Center's impact on heating and sanitary facilities. The present investigation limits it— self to treating home—heating and sanitary facilities as input (or explanatory) variables. lhmrition (X7 in regression) As in the case of heating and sanitary facilities, changes in nutrition were considered as explanatory varia— 1 bles not affected by the Health Center. EH Methodological Problems A look at the variables indicates that they describe a.situation which could be summarized as the socio—economic i mnflition of the families interviewed. The growing aware— rmss of the importance of these variables in determining imalth status was the main reason for their inclusion as Separate variables. There are, however, methodological problems connected vuth a procedure which treats these variables as exogenous: (km of the prime objectives guiding the establishment amiadministration of CEO Health Projects is directed towards 13m explicit change of these "environmental" variables. A (flmnge of the variables in the desired directions is there~ fine an output of the CEO project. A comprehensive evaluation . ....Ic J..- f»: .u- . VG n\u ‘L ‘\.L sh ulna olv ado “B n q o .1 Van ‘vav 3V F . "IU- LIU CV .A- .flun "A ..N P u ... u ..\ : . "I INHI- Q‘U I J ' '| In U P II b nu ‘ I I \ h I I 1“ v N A~h 1\U Alli I- c n. \ v 11 r k ~ I .. a t! —-—’" 79 would have to include an analysis of these "intermediate" outputs and relate.them to "ultimate" outputs of health status. Resource and time limitations forced this study to concentrate mainly an identifying and measuring the im— pact of the Health Center on "ultimate" outputs (i.e. changes in health status) and to treat "intermediate" out- puts as exogenous variables. 4. Survey The lack of applicable data necessitated the develop— ment of a survey instrument which would facilitate collect— ingiflmadesired data. The current survey literature guided iflm procedure for administering the total survey (Kish, ‘ lensing and Morgan,5 Moser6). The methodological series cfl‘the National Center for Health Statistics served as a gaude to surveying problems in the health and health ser- 7 vices area. a) §grvey Instrument The survey instrument consisted of a questionnaire vduch evolved out of an iterative process consisting of Inmmrous feedbacks from staff members of the Michigan State Ikdversity, of the Michigan Department of Public Health and (fi‘the Western Michigan Comprehensive Services Project. Two Ifllot surveys helped to make the instrument applicable to actual use. Since part of the information pertained to the total Ilouse'hold and was the samefor every household member, it VHS decided to divide the survey instrument into two parts: A» VT? .. . .. f Out 'Iflf‘ Na s“ v..“‘ 'J a 1" AL ‘A I) [______ 80 (i).Household questionnaire and (ii) Member questionnaire. b) 'Sampling Eflocedural'AlternatiVeiconSidered , One alternative is a simple random sample from a sampling frame and an interview at the family's residence. The Social Services Departments of both counties offered s cooperation in establishing a list of families receiving food stamps and Surplus Commodities, respectively. During the month of the interview (September 1972) the Lake County Social Services Department distributed food stamps to 134 families (403 individuals) who did not receive any other form of Public Assistance (referred to as NA——Not Assistance clients) and to 179 families (H86 individuals) who received cmher forms of Public Assistance (PA clients). The Mont— nmrency County Social Services Department administered dur— ing the same month the distribution of Surplus Commodities to 162 NA families (#23 individuals) and to 76 PA families (201 individuals). A random sample of 50 clients could have been drawn in each county but a pilot survey with another sample in— dicated that most of the families were extremely difficult and costly to locate and lived far apart. When it turned out that the refusal rate was rather high (i.e. twenty percent refused to answer any questions), it was decided to use a different procedure. Erased‘ure Applied in this 81:qu Food stamp and Surplus Commodity recipients were asked ———7' 81 to volunteer the desired information at the time they picked up their stamps or commodities, respectively. Since most clients tend.to pick up their food stamps during the first few days of the month and since commodities are delivered at three specific days, it was decided to interview both NA gng PA clients and to interview willing clients who were next in the waiting line when the interviewer had an opening. 0) Interviewing Interviews in Lake county were conducted in offices of the Social Services Department in Baldwin on September 5, 6 and 7, 1972. The interviews were handled by the author and two Graduate Assistants of the Department of Agri- cultural Economics at M.S.U. on September 5. On September 6 interviewing was done by the two Graduate Assistants and on Ekptember 7 by the author alone. One interview lasted for approximately 5 to 20 minutes. Eighty £923 household inter— Views were completed. The refusal rate was zero. In Montmorency county it was necessary to interview the heads of the households (or their spouses) while they were sitting in their cars waiting to receive the Surplus Commodities. Interviews were conducted at three different locations: Atlanta (on September 1), Hillman (on September 6) and Lewiston (on September 8). All interviews were con- ducted by the author and lasted between 5 and 15 minutes. Emre too, 84 househOld interviews were completed. Two individuals (one male, one female) refused to answer the .‘,, H' duh oil ,- u. '7‘ cu in u' 82 the questionnaires. The interviews consisted of two parts. First the respondent was asked to answer a Household Questionnaire which was applicable to all members of the household. Then he was asked to answer health related questions for himself and all other members of his household. The interviewers read the questions to the respondents and entered the re— sponses into the appropriate places on the questionnaire. Respondents were the heads of the households or their spouses. No provision was made to differentiate between kinds of respondents. 5. Household Questionnaire The household questionnaire was designed to take stock of the socio—economic conditions of the surveyed families. A.copy of the questionnaire is presented in Appendix C. Question (0) established the duration of residency in the county. Questions (d-g) were included to discover whether there was a difference in mortality and institutional confine- nwnt between the two populations. It was impossible to de— tect any differences in the analysis-~mainly because of the low frequency of responses. Yet, an inclusion of this question is important in a large scale investigation: By excluding this information it would be possible for an area Which has a low health status but sends all its sick people into institutions to show a higher aggregate health status than an area which has not such a low health status but .L u,“ "C a... ':.r J“ ; . —7’ 83 treats all its sick.pe0ple at home. Question (d) served also to test claims that families attract distant relatives (grandchildren, nephews, nieces, etc.) because of the availability of free health care. The small sample, however, did not warrant an actual test of this hypothesis. Questions (h-w) helped to obtain information on inde— pendent variables called for by the model. In spite of the pulot surveys, it became clear during the interview that rum all questions were equally well suited for obtaining tum desired information. For instance, questions (h—l) resulted in unambiguous answers by all respondents. Ques— tion (m), on the other hand, which asked respondents for improvements in heating was perceived differently by dif— ferent people. Many respondents volunteered that their home twating is "better" because they got newer equipment. Only a.few respondents explained that their house is actually warmer during the winter. Question (O) proved t0 be insuf- ficient in describing changes in nutrition. Older people tended to eat less, younger people ate more. Using the vmmd nutrition directly, question (p) seemed to be more amiable for expressing changes in nutrition. Questions (r and s) asked to assign a rank to the quality and availability Cf health care in the area. It was interesting to observe iflmt respondents who were older than 35 Years had considerable gyeater difficulty in grasping the concept of a continuing Scale for the availability of health services than those ’n 'i z, —f—' 84 below 35 years of age. A difference in education is suggested as the reason for this difference. Additionally, very few people could specifically answer question (s) which asked to identify reasons for changes in health care. (The ques- tion was asked to indicate what the respondents thought had influenced the availability of care in the area.) 6. Member Questionnaire The respondent (head of the household or spouse) answered first the Member Questionnaire for himself and subsequently for all the members of his family. The number— ing system and specific statements guided the interviewer to the appropriate position in the questionnaire. A copy cu'the questionnaire can be found in Appendix C. Subjective Evaluation of Health Questions (1—3) asked for a subjective evaluation of the respondent's health between 1971-72 and 1968, where the respondent's health status could be marked on a scale con— sisting of six positions. Older respondents here, too, fwd greater difficulties in assigning a numerical value to Eiconcept than did younger respondents. Egeventive Care All respondents were certain in answering questions (4*?) for themselves, yet, male respondents had consider- able problems in remembering doctor and dentist visits and other health related events of their children and wives. @mwtions (8—11) were inserted to introduce the concept of disability into the respondents' thinking. No further use + "-_ __‘.._________..+, ...—h—n— 5 {a .1 .I- r: .61 .I y p N I u Ia p . . . 0 .«l. a. . fwd -.I. .r .. w m .11» hi :5 at a» «\u «\V Q.» 31 {L Vi a C «nu a .1 « ROI. L.v W . if. C» ..C a t Cb nu ....d. ..1& 51v Ah. :75 CF- .fiti - it w- A P b. ..Au RU. x)& u alt ‘1 flu ANN n-v .-1 Nuav 11:: . J I.|. ch N\L s b4 «\V Q5 «HIV h. . QM he 7n.) (an Ials -.v ...J r n F; ._v f. h ARI I. H ..a. An. 3.. an. an: . V 3‘ ct \ \ u v. s U r u 1. n P.“ .n i ... s v r . _.|ul a s .1 n v 5. . u h. x > tr 1 I‘! H’C —i— 85 was made of these.questions in the analysis because of the lack of complete and consistent answers. Insability and days missed becauSe cf health. Question (12) checked for total disability and the rwed for constant attention. Questions (lA—l7), (19—22), (3A~36,A2) and (38-41) followed a common pattern of identify- ing first the "chronic disability status" of an individual and then the days he/she could not perform his "role" as determined by his chronic disability status. Chronic dis— ability was identified in terms of the individual's ability to perform major activities (roles) which were relevant to Ins age. The relevant role for children between ages 0-5 was tijlay "as usual." The relevant role for children between 6 and 17 years of age was to go to school and play "as usual." Adults between 18 and 65 are supposed to work "as usual" and to perform home work "as usual." Retired Emople past 66 have their role specified as being able to work around the house "as usual." The first question of each group (1A, 19, 3A, 38) identified whether the individual was totally limited in performing his/her major role. The second question (15, 20, 35, 39) established whether the individual was limited in the kind of major activities. The third question (16, 21, 36, A0) checked whether the individual was limited in the amount of role fulfilling activities. An individual could be assigned to only one limitation class within a particular activity group. (Individuals who needed total :zrc: nay: . t u.‘ «41:3310n 2a m “We {N \. IIIlllIllllllIlIIIII-II::———————————————' 86 care were already identified in question (12).) The fourth question of each group (17, 2A, 42, Al) determined the num— ber of days an individual missed from his "Egggl" activities as identified by the chronic disability categories. The previous procedure of establishing chronic limita— tions first and then identifying the days missed for each limitation class followed the procedure used in one section cfi‘the Health Interview Survey.8 The Census Use Study Report No.12 indicates the application of this procedure to the New Haven (Connecticut) Census Use Study.9 In interviewing the respondents it could be observed that the respondents who were clearly in one particular limitation category had ng_problem in identifying themselves Math the role limitation as described. Yet, individuals who had only recently been forced into a more severe level cfi‘limitation had difficulties in classifying themselves. Emamples are injured veterans, peOple in retirement, indivi- duals who survived severe diseases. Despite these limitations it was felt that the frame— work as presented offers a viable methodology for classifying long term limitations in terms of a "role—fulfillment indicator." Employment status Questions 22-33 served to separate unemployment due to health conditions from unemployment due to conditions in the labor market. Furthermore, these questions attempted to put thernmmer of days missed from work because of health reasons into a relationship to the number of days on which the ' Fr. 4'- ‘rn .:.-..m.re Ac ’ L “,n ‘nr -4:4env.aUe "jun, Y“ L‘V‘O *:lfice‘i [“1" " nun... I ~.. “ r "' ‘1‘»5LaOJ .zzrnrngr,u Q ..urv..u.“b U ”h". .fi‘ ' "a (3.18 Ga", ‘5‘. . ‘ ‘1": :rr‘ n; "“- me at . .r Wfi‘h‘: A .. «cum/Y. {’n‘ fian‘ 0 v 'JV“~ U —i— 87 individual was actually employed. The month by month gnocedure was time consuming, yet forced the individual to cmmcentrate on his employment for the last year. A check between the answers to questions (27-31) and the answers ‘miquestion (32) proved to be completely inconsistent. Ikmpondents tended to associate full time with a "full vmmking day" (8 hours) and not with a full working year. 'Hfis suggests that the time consuming breakdown by months, weeks and days was necessary to define the work opportunity hirelation to the days missed. The apparent confusion on whet constitutes "full time" employment resulted in incon— sflstent responses to question (33) WhiCh attempted to identify acflenge in the amount of employment. Therefore, question (33) wasrmm treated in the analysis. 7. Analysis The collected data were coded and stored on computer taPes. Calculations were handled on computers of the Michi— gfllState University Computer Center. The analysis of the (Eta was divided into the following steps: a) profile of smmfle households, b) distribution of activity limitations, (U average numbers of days lost from role fulfillment, CU work opportunity unrelated to health vs. work days massed, e) regression analysis of days lost from role ful— Ifillment, f) perceived availability and quality of local imalth care, g) perceived health. mummy 55"! Wr'n Y“ . I ..:.ue:ws 1n ,1C :13; n‘mfii’ I‘- ... all)“ ' L as: longer If 3“ .I ‘\ I seal-en 3 of f n ' ' ’ CECElOu I ‘r‘fv :f‘ h ...“. ,U pEI‘Cf 55in ° .1 {up T.' . Mn. ..Ontm0r( . Eefi‘imq vLiUu 7 O \S“‘ NC A « .3 .octor 3 ..1‘1» ' ‘ . cb‘S With :3?" ' “w 18 SUI‘p 88 a) Profile of Sample .Hous'ehoildg Appendix Table B7 offers a profile of the sample households and only a few remarks are needed here. It appears that most characteristics are evenly distributed emwng the two counties. The Lake county sample, however, consists of more Not Assistance (NA) clients (70%) than the antmorency sample (58%). Furthermore, there are no black residents in Montmorency while 44% of the Lake county sample are black. It is interesting to note that NA clients have lived longer in the area than public assistance (PA) clients (Section 3 of Appendix Table B7). Section 6 of Appendix Table B7 indicates that in Lake County 50 percent of the sample reported an improvement in iflEir families' nutrition which contrasts with 32 percent 0f the Montmorency sample. An unexpected result is shown in section 7 of the table. Fifty four percent of the Lake SaMfle would see a doctor more often if they had more income ‘H‘if doctor services were more readily available. This Cmfinests with 44 percent of the Montmorency sample. This result is surprising in the light of the greater supply Ofmedical services in Lake county. Yet, it pOintS OUt ifimt Montmorency residents feel that they get medical atten— tion even without the presence of a health center of the Size of the Lake county project. On the other hand, it is Possible to interpret this result as reflecting the greater awarehess Of need for health care of the Lake county sample. the: So?“ - arm "wt; (99). ‘z-} years com; !a .. l ”I. “an." :w-Vu ‘l ( \llll‘ 1" < ‘ S S at nu. l .l .l e 1 3 .fl Co it axe .nnu add vsnu .u are n.l n ..l. U ml .1; . Q A Q 0 AU «D s. O . I; e c h . 4». w w h ..T s 1Q ..r . V e P 0 e \ll . Isl‘ , “1. AV t .htu .fiu ‘ .-. Ck. 1?; L§V n5» E. r . an..." 5V nun flu 3-4 ..IJ u t I. . AU n Q ... ..K «\W a .v u, u .H A. U V0 c A N n: c an. .. .r. . a. .. h u . u... xx ... ... x ... .. ‘\ ‘V‘Ll V —i— 89 The heads of households had a lower level of education (9.1 years completed) in Montmorency than their Lake counter— parts (9.9). Interestingly, the heads of families who were on public assistance had in all cases more years of educa- tion than the heads of families who were 323 on public assistance (Section 10, Appendix Table B7). b) Distribution of Activity Limitations Respondents were asked to classify themselves and the members of their families into five limitation groups. Group 1 consisted of individuals who needed help to move around inside or outside the house. Group 2 was composed of individuals who could n93 fulfill their activity (play, school, work, homework) because of health reasons. Those who were limited in the kind of activities they could per- fbrm formed group 3. Individuals who were limited in the amount of activities were assigned to group 4. Group 5 consisted of individuals with n9 limitations at all. Thus grmn31.represented the cases with most severe activity limitations while group 5 included those who had no (or no reported) limitations. Question 11 of the Member Questionnaire (see Appendix (D helped to assure that limitations classification referred to periods which have been in existence for more than six nmnths. It is suggested to improve the accuracy of this information in further studies by asking for the duration (fi'the limitation'afggy a limitation state has been estab— lished. Table 1 presents the relative frequency distribution fable 1. Relative Different diltatim We around 2‘9811'103 fulfill activity is . . “143.33 in kind 0" Activity 0? activity "‘0 limitations 90 Table l . Relative Frequency Distribution of Individuals in Different Activity Limitation Groupsa County Activity Lake Montmorency Limitation Percent l=needs help to move around 1.2 0.6 2=cannot fulfill activity 6.9 6.0 3=limited in kind of activity 18.5 11.7 4=limited in amount of activity 4.5 7.0 5=no limitations 68.5 74.5 aThe percentages were obtained by adding up corresponding groups of activity limitations of all roles (play, school, work—job, work—no job, homework) and establishing percentages. The absolute nunbers upon which this calculation is based are shown in Table Sf individuals in The table 1 are severely lin T'ionmorency sampl lizited) individl her 0:“ individual ‘u‘h'w ‘ ‘ ....nv'lfl‘uals oetwe IIIIIIIIIIIIIIIII-""":7_____________7 . 91 of individuals in different limitation groups. ' The table indicates that the Lake County sample had more severely limited individuals (group 1, 2, 3) than the Montmorency sample and fewer unlimited (or less severely limited) individuals. Table 2 presents the absolute num- ber of individuals in the various activity limitation groups. Individuals between zero and five years of age and above the age of sixty five report only on a single activity (play and homework respectively) while school children and indi— viduals of age eighteen to sixty five report on two activi— ties (play—school and work—home work respectively). The discussion, up to this point, treated limitation classes as descriptive variables of the population. Yet, it is obvious that one of the long-run outputs of health care is the reduction of limitations. This study does not treat differences in limitations as outputs because of the short duration of the evaluated project and because of the small sample size. A treatment of the conceptual problems introduced by changing roles is offered in Chapter VII of this study. 0) Average Numbers of Days Lost From Role Fulfillment To isolate influences of agg and major activity on the number of days lost from role fulfillment it was neces— sary to base further steps of the analysis on age and activity groups. Additionally, it was decided to investi- gate how the number of days lost varied between limitation EPOUES . mouse lbw metre sale mlo howls +he melee sale mlo hedgehog hoes secs Add. HH< Abfibflnodx g W0 Zn. WHO-Zhuzg ES 10Hgn >.nw§0hv Xhocagnwcg 62.4 “4% ”HHHOIHw38UvN—HHuhI-HVIH \r..\~nn.u\rd\lmd.\94\\ U~a Munfwn\ \ANN MQNIwafl-Unhs gain” [‘0 LUQ§2 usuu‘ M‘UNIQMNNU \flwvfhdi-AUHV 3911K... And—ha .nnUhuvquanI— abomnfiuhl-QQEOHU < om HHH m eeHanece me MOH e mmHuHceoeAmv He eH He me e mm m mH e eH eH e NH e mm mH em me mm mm m emosmzom HH m e m e m H m m m H m m m m H H m we em Hesse m as . mmancce om om mm m em m m m e m m m m m Hooeem H H m 2 9 H m mm em OMHuHcece m em em emHuHceceHHv mmH H mw mm mmH m mm em m e H m m m e e e H H m HaHd H H H acme +ee meumH sHue mue meme +ee meimH HHse muo mecHeceHeaH seHsHece HHe HHe seHsHece mm< sezmmezezgz mmeq m @3ng 3:300 5:808»:on one one peace excH the ecczemm ccmHssssco suuchHeceHsaH seH>Hpc< sec ems as mHszeH>HccH no scessz .m cHeme I)! 1 :z t| ( H some. +oo hOIwH sale mic some. +ph melee x. Hip mic asbestosis: horseshoe H .....c. HH< \nufibflboiu. g Foaming—9.2% “EH 93 .gsosw ago: oeo: one CH maodoH>HUCH on ddOhMlsoaa one CH maosefi>flosH wsHooo so ooooaonoo ohm odozofi>aecw mo gonads HopOD mean .sccHeeeHEHH on can one son on menus .seH>Hecc mo oesoso CH oopHsHH me new new on woenw .mpH>Hooo mo coax CH wouHEHH me new ooh on moons poccmo one new OS moewm “meowuopfisfla ocnm .moH>Hpoo HHHMHSQ poceoonm .ocsoso o>os 0o oHoe meooena ”meoaaom mo poems .mpfi>flpoo Hafimazm .seHsHecc so ceases cH ccpHEHHue .seHsHecc so ccHx cH cceHEHHum snoecH we cHsoem occHeceHEHH scHsHecec wwm QANV+AHV cam chseHchcH so cohesz Hence Ir. HHH HHHuHceoe meH NOHnHmmce mm mm mm mm a see c e m m m 02 m m s s e o>om mH ma 0 mm mm mmoz mm mm em em m s e H H e as m a mH mH m asst m H I. wow< mmlwa palm mlo wow< +mm mwnwa Halo mic macaeopfieaq mpfi>aoo< HH< HH< seH>Hec< MozmmOZBZQz LIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII--------filnnllnl mw< Umfifiucoo . N OH 9mm. Table 3 dis amber of days 1c andftontmorency c 23, and 214 show t . a: .‘vu (I) of limitatit fewer days 10513 . (age 18-65: me, too Lawe cc Eisntmorency coun‘ :ssed from work ~ Seaith. A c0] :iKS county indi‘ Isnsiderably few h l" 5 “RUE Orency Coun ~ \ un'fih I a ..qu sized that Aimed "unemplob’ "M‘ . 1 mmplogment was The fact t tr L . k Opportunlty -."".-----' i 9” Table 3 displays in columns H and 5 the average number of days lost from role fulfillment for Lake county ; and Montmorency county, respectively. Rows 4, 8, 12, 16, 20, and 24 show the values for the age—activity groups regard— less of limitation. Lake county individuals experience , fewer days lost from role fulfillment in all but the work group (age 18-65, row 20). The other rows display days lost from role fulfillment for the various limitation groups. Here, too Lake county individuals miss fewer days than their Montmorency counterparts. (Exceptions are rows 9, 1M, 17 and 21.) d) Work Opportunity Unrelated to Health vs. Workdays Missed As previously stated, it is necessary to analyze days missed from work in light of work opportunities unrelated to health. A computation of days employed indicated that Lake county individuals of age 18—65 were employed for considerably fewer days during the year (137 days) than Montmorency county individuals (193 days). It should be emphasized that the concept of work opportunity days ex— cluded "unemployment because of health." (Health related unemployment was counted as days lost from work activities.) The fact that Lake county individuals have fewer work opportunity days and additionally miss more work days because of health (17.1 days) than Montmorency county indi— Viduals (8.6 days, see row 20 of Table 3) should be recog— nized by relating work opportunity to "days missed from work because of health." One way to accomplish this is by 63mm Anon; 3.3; 3.3; an»; 33; Anna; CLO. ~ Amos; «no...» 3:; Sam; Rad; s OL. m1»! ....m 04 a; H.0I fin: m4! :4: ....Q m.¢l 3.x.l 061- and 04H a.: < : Anon; Aim; $23; 1.2.; “an“... 39... 2.3.; Roz.» 3:; a .c 3:; q 0.x. a.mvl 3.3 c." .... 1?: ....«I :J: a6 a.mwl a.m.! 04.2! QM~.O 04" 6.: n h u L x N M. H . .5 as «53 . . . t> .9~ . ll.- 021.. . L— “Ends-gal: .unvz . 3:- I3 I53... . OCR—I. Jay-2 . J 7-9321» 45.2 0‘13! n ~< J! » IL). s I n I.» . a: .a‘fnlc . 0:1tlc uIv-sl n-lnIu ...-1.; Ila. . lu- v)!- IF ‘u a nu: . rib-I; In! .Onlul- .5 . ..uLd22ln L: Jis'fi :31... - «(I. o P.- .vl~ . II< as H ~I~ «15:5 :>< » \ISNI) nan-driv- .I.u.< . I Id.- .i f- Isl-n Elv»3- 15.4354. . —ul~.-.Z 4“: lug.» .FvHIIJ~ .. .u‘u l... u. "5 ‘5‘ Nay-\EEJE! ..i)( y ‘5V~ ~‘vni3 -.-VN 5‘0 u) Nng‘ ~§QNUS~N~ 3.x a Q 1,1! 3“ .9. ..‘X 5! {X Nil. 3‘ 5.! NY 5 F ifwl‘lut l>ll~ wild! IKE—Q 1i .BH‘IED‘K :4 F. ...n- ... ... , u . . . ... q- . . Q. ~. 1;: x. um (I . ,xm . . (I III¢(.>| {\(t1 y .1. . . 1.. RI— 95 358 0x5 o5 .3 gapgomno 5 has .508 33:28 spa-mo: Em 895 moaomE o5 85» 9839.50 o5 5C 63350 who a «to NH .3253 Baroda? page.» 3.23% 5 Pagan .5333 «33 arm 3 £88 mmn m 8 #5335 8 um £3328 no 2593 5 2.33.2.2. EVE... co BE 5 835$ 29.58 3 2.329235 2 352» 8323252 3: <2 x 2... 2.2..- 22...... .2... 2...... 22.. 2.2.. 22.. 2.2.... 2.22 22...... .22....- 2 My. a... 2... .22 ...... 2.2 .2. 22.2 . 2 2.3 2.2m 2 m N. 2.2.2 .22.... 2.2.2 .222. 2W... 22.. 22.... 22.. 2...... 2...... 2.... .... . 2. 2 2.2... 2.... .2. .2.- 22... 2...... 2.2... 2.... 2...... 22.2 .2... .... 2 . 2 8%... fix 3me amp-u 8mm. swam. swam awn. amt :2me Swan ...-22. as... 2.2 2.2 m M a man 2 ... a 22... .2... 2.2.... 22.. 2...... .2... 2...... .2... 22.....- 2.... 2.22 . 2 2 2...... 2.. 22.. a... 2... 2...... 22.. 22... 2...... 22.2 2.2. 2... .2 M. 2 222. .2. .2... 2... 22...... 222 22... 22.. 2...... 2.2... .22 . : .. 22.... 22.. 22... 22.. ......- 222. 22.... 2... 222- 2222 2...... .2- . m 2 22.. .2... .2... .22- 2....- 222 22... 22.. 222 .2... 22.. .2 2... . .2. 2 2w... 2... 2.2. .2. 2... .22 2.2... 2.... 222 22.... 22.- 2 W 2 .22. 2... 22.. 22.... 22.... 222 2.22 22. .22 22,... 2...... . m o.m m.m M St» a $2.2m. AH; ammo ammo swam ammo E 5w“. :ch 323 8.22% Emu m... 3.0 .3. mm 2.2 w a 22.. 22.. 2...... 2.... 2.2.... 22.. 222 2.2... 22.. 222 222. 22... 2... . M . man 92 2 o 92 m :10 m ”mm.” mm.“ “mm” 2.2%“. aw... 3&2...” TEHJ 22mm...” 2er 3?... 82mm 2”me 2.22- 22.2 2.: a... .3 2 2 2.. as. 2.2 £222 awn. 9.22.2...” EM.” aw...” awn 3w... swam 22mm 2.....- 22.2 ....n a... m M M. M ... . 45:3 . . .5 . .5 25 .8” .5 Ware-cw ...qu .uxmmufi-Mm ...-”WWW” oflhummlm ore-2M...m..s.m.vm_m_~uM awn .wwugfifi pun: “Mix-w :flm mfi WE. Wm .muufiofinm ”on“ “aflflhufi .gawflsfl 8339.3 322.33 8393 ”MN mag WW owwn m 0W8 own: Mm: “a UN”;— uquurd gigs 2.!”th mm 083M «3 guns“ as am gmg Badge: ( .2 2 a on a no .mofigfig .89.: EEC Bum-E was $5038. U |||Il .3500 honor—95:0: v.52 3:300 mi 299.. Lam magnum ...oanmmnmy. uc< moSHd> 9Wokb>< zultiplb’ing the I ,. -[1 2,; workdays i W01 For instanc xcrk days during each= 2&0) the 3 and the index f0} 52121213222 theS< will change the 1 17;: to (17.1 x . ini'idntmorency c: rhe preced. ":rfikn «au‘ "‘ JQayS. To .- 2cszed to calcul. ...dlvidual (of ti umy adjusted EDP n ‘ .«028QUI‘6 will 2? we -.Ierent se ' Recression A A linear r in ~'. WDact of va from h- ‘ qCtlvities II II II I—J‘ I—’- I—" '4- H- H. II II I—’ H .4 LA) ’f— 96 nmltiplying the number of days missed by an index of (all workdays:work opportunity days). For instance if it is assumed that there are 2M0 work days during a year (12 months x u weeks each x 5 days each = 2N0) the index for Lake county will be 2M0/l37 = 1.75 and the index for Montmorency county will be EMU/193 = 1.2M. Multiplying these indices with the number of reported days will change the number of days missed in Lake county from 17.1 to (17.1 x 1.75) = 29.9 and the number of days missed a in Montmorency county from 8.6 to (8.6 x 1.2M) = 10.7.10 The preceding example was based on average work Oppor— tunity days. To make the measure more specific it is sug- gested to calculate in further studies the index for every individual (of the working group) directly and use indivi- dually adjusted values as the basis for computation. Such a procedure will facilitate a comparison of adjusted values for different sections of the sample. e. Regression Analysis of Days Lost from Role Fulfillment A linear regression routine was employed to establish the impact of various variables on the number of days lost from activities. Dependent variables were: Y1 = Days missed from activity, where i = 1 u: age 0—5, play activities i = 5 8: age 6-17, play activities 1 = 9 ... 12: age 6—17, school activities i = 13 ... 16: age 18—65 homework activities 1 = 17 ... 20: age 18-65, work activities i = 21 ... 2“: age 66, homework activities. Al.kflependent V XI= County input that I "treat son g1 count§ X2 = Race: X3 = Assist 0 = 01 f Xh = s: C1 :75 "S '1 0'? (I >- J' N :1 c—r m H, o-oJ r . 'm' H- m o o a) :3 CT (D roar—'- r-ov h 9 not i: X7= Nutri' did n- id = Need , doctd X9 = Incom go u. X10: Sexzp 11= Enrol 1 = e Physi the 1 Denta last Educa years Of ed .‘ .t+ \- “lmation for B IIIIIIIIIIIIIIlllllI----::—————______, 97 £11 independent variables were specified as dummy variables. XI = County of residence. (This was the major health input variable. Dummy "I" specified the group that lived in Lake county (i.e. received the "treatment"). Dummy "O" specified the "compari— son group,"i.e. those living in Montmorency county. X2 = Race: 1 = white, 0 = black X3 = Assistance status: 1 = not on assistance, 0 = on assistance. X4 = Length of residency: 1 = lived longer than four years in area, 0 = lived between one and four years in area. X5 = Sanitary facilities: 1 = had all sanitary facili— ties, O = did not have all sanitary facilities. X6 = Heating: 1 = heating improved, 0 = heating did not improve. X7 = Nutrition: 1 = nutrition improved, 0 = nutrition did not improve. X8 = Need for doctor services: 1 = would need more doctor services, 0 = do not need more. X9 = Income: 1 = income went up, 0 = income did not go up. XlO= Sex: 1 = Male, 0 = Female. X11= Enrollment in Health Center in Baldwin: 1 = enrolled, O = not enrolled. X12= Physical examination: 1 = had exam during the last five years, 0 = did not have exam. X13= Dental Examination: 1 had exam during the last year, 0 = did not have exam. X14= Education: 1 = head of household had 0—8 years of education, 0 = head had 9 or more years of education Estimation for Both Lake County and Montmorency The first run of the 24 equations involved the total Sample population. Since X2 (race) and X11 (enrollment in Health Center in Baldwin) were only applicable to the Lake county population, they were omitted from the regression equation. Table 3 presents the coefficients of the esti— mation. Significance levels are listed in parentheses. Coefficients with a significance level less than 0.10 are in Squares. rEhe result. having access to reducing the num significant coef as the county co ILES reduces "da ‘Anu 2;: that part of 2:02 during the I‘1 ihe regre: 201ml» My ("treatme m 10% from 1 enables Was n t —i— 98 The results suggest that being in Lake county (i.e. having access to the treatment) has the greatest impact on reducing the number of days missed. The other statistically significant coefficients are not so complete and unambiguous as the county coefficients. The presence of sanitary facili— ties reduces "days missed" in two equations but increases them in the third. The fact that sicker people go more often to the doctor is represented by more "days missed" by that part of the population which had a physical examina- tion during the past five years. Estimation of Lake County Population Only Since race (X2) and enrollment (X11) are variables 9 which are relevant to Lake county exclusively it was de- cided to run a separate set of regressions for the Lake county population only. Table M presents the results of. the analysis. Length of residency, nutrition, income, sex, physical exam and level of education had some influence on the number of days missed. However, there were too few significant variables to allow a particular interpretation. Discussion of Regression Results The regression results indicated that being in Lake county ("treatment") has the strongest effect on reducing days lost from role fulfillment. The impact of other Variables was not uniform and/or significant enough to Warrant an interpretation. Yet, several variables seemed to be more important than others. The distribution of squares both in Table 3 and Table 9 suggests to concentrate in future studies on P 5.. .\ 3.3; 3.8.; “riots Con; 275$ 33; 3mm; 3mm; “Nos; Chm.» 50m.» ammo.» 38.» x 9.0] .....zl 5.: m5 sél Md n o. :4. a.m.: .....m QJI a.m.. «fin «(Ll sm~.o Q: mum < a :CNJ 1.1!; $3.“; :9”; 2:3; 34o; Sac; 2.5.. 2.55.» :13; «men.» «9.6.» «new.» on? wolnl FI=H M-nw 5'”! .....u M.“ 3.9 :.~..I. «we. asfil mnnhl “rem H.5H-l NQN.° an: n \H M é n\ N I . I. . lrtly M, “'0 W l t ||I0|«l Inl.l.l . . PM: &~ Illllouglul. Ilt. I ll -.r.> 95le .l...— i .5: OZIO .Lph; ...-2|: ‘.¢I..fl(.|¢ .CZ .0.:|: Sn .QHHID .202 .LJJZIC ..CZ ......u:lC _ a: nan—(l: Ex». :lalb .I€.~3~|fl XQEQZIC . I! :IAVIu . 5.. “ii-Ta flu; “gala. . — _r.V-. F—I— nuaélq 1: . 0.: In [All . nvq5I~ .5: .LJZZI~ .K. 997...:I... ..hnwnn u ~. I > . I . . .-.~.\. ....x .. . . ..l.l~ .... .....I....:~ ... 2 ...l.~!) ..I\\ P..\\..\ ...}. ; _> . ....\ .~ {It :IVII . 3. e :L >,!.~ . t l. ..3; n ””8830.“ $8293 5 woman ”EU 3 fix manmflumxrn 33 0a $.3th mmo p35 3332m .333 ooQNngm as man 00% mgfivmavm 9.3 g 8» .cflgumm 9:3 Ba 52: 98388 "alone 8 Bug on cg... 83 ..fido hpgoo 0.3 0a. mhmmmh 0H Quests .8 £595 5 835»: .3338 no HE 5 E333” .8033; ocnm VA 1! . . . . Na; 88; . ..m m m am .. . . . NJ Ami. v 38 V A08 v Ammm v 8. I . m.mo~ Nmmd z z m . E; 23 V Ame a Amma V 8A.: v A3. . s . may a i- a m a m. mmmmb Aqéu 197 m6 m on a o N w. s m: m o m m M MM . 9% AA z . . . . . m .V ASS ANS; A95; AME; 3 . . m +mm. Hm Ame»; Awake 8“me mom V Amwm c Ammo ewes “five Ammo as as. Z- a a- .. aw i I? 0 ll .. . . . . «Mm. Ame: 3?; 88¢ . . . ow “mm Ammww 8“me mm; flown Amwaw Awflmw awn“: Amswaw Amdmv mam 9: oswr m R: mm". o H S mm x .. O. .l a I . . . m . . . . . SE; 893 Amma v Amom v Am: V . . .m m S . am: as; Ammm 6 Age v 88 o A2; v E . . . NA... 3- s m. as o m Awfiqnv ca 99” TN. ad mg: .. T .A m a H m. m w a. AA 0 A: 3 . . . A 3 ANN? A85 :8; AR: 893 A93 A8: ea; 3%; . . 2m m miA : Amwmw. Ammwaw Ammmmw mmwmmn ST .23- mam? ma: 9mm? men 0.5 Tmm m 3mm- m mm 03 o - . . . . . . . Aem:.o Amma.u Aamk.c . Aamfiw Aomaww Ammmwv tum V emamuv. Ammm... Amfluv Awwwmw Ammm a Amlumnv m6: =.~ m.~ ~.mm god : om Tm VA m 2 c o e u u . a o a o mgvv m E 21a; A J 38 v Ammm v 33 v Ram V 32. V SS V 39 v 3% v 2mm V Amomuv Am . . . . m.H A m.~ «Mow? m4 Tm: m.mu A: «.7 am a.m- a m m m: o m h m H s Sm 0 NW N o o A: A. . = Re. Re. A H . 38; 8mm; :3: 8mm; A8: A33 A8: 33; 88; Sam; . . . 9 a A 3w mummy 31 EN. 3? new- Na. new as- 921 a.m.. m ma m E a: o a om m Sum S q 9 - . . . . . a. HR. A. AA a 8; A93 A25 Ammmmv swam 8va Amway Ammm o Asmmow Amméw A flow AMwTa WHAT m . m4 .1? do 36 3 mum M NH was; as; A8: A83 :8; Ass; Am: Ammo; A83 A3: A23 A33 Amt; . . . A Km Tm m4 TH NS 3.0.. m4: To: p.m. m4: m.m m4 :6. m 0 SN 0 m a m o 2 AA m S m.m m Stm m Ame... £an ABM; Sammy AmmAAw; Ammmv Ammmmv Amman Amomww Amman 8m: Ame”; a.mmmv m m: Nfl 0 N m mum w m 0° 0 NI I u I I ' o a. o I :03 a | o - n o - ANS; 2.8; 3mm; 8%; Es; :34 8.5; 2.2; 88; 2.8; ANS; 8R; 3th a Ana m4 dm a.m. m6: a.m- m6 9m 24.. HA ma m.~.. ad. m4: 306 a.m. m a a 0.3 A. a w m ism m A53 ASS A83 A83 A82 83; 3&3 3mm; A85 :8; as; Ame; A83 a a one mm: 3: m6 p.m- m.~ m.~ Aim o.~- Wm 97 a.m.. m.~ 3.7 535 as mum < : Chm. Am J ammo flow; 89; 83; 88; 8mm; Amos; A. . . . . a.m.. e.g. S: m.» p.m- Md m.~ Aim o.~.. A Mam 8“me Ammmuv Ammwmv 1:- and as m A m AA a N m mzo H .....Lw +muo .4. of. .um of. is uozno «Home 82 .058 8am 68.6 82 ..hmszuo 82 Humane gnaw-0 ma _ To 62.2 Ermxfia .xmwuflrfia HM must“: uMMMoMum Summon” 8 “was“ 5 685 as. fizz-H a: one...” dam 33 + 31m; .92 82M cause-aw 393, a man a a 2x Q5 tomxooo Hafiz awn: .Wm .38: .0334 88 5398 «m .25 83 5335 Evans. 8335 .. .. a t. .. .. .. a a ... a. 0% ... u. .. ...a .... a ... 5...”... 0.3.5 .3550 mid.“ ho.“ modsmmx Sfimwmtmmm 9.8 mum—Jada, gun—02¢ 639.933 .332 BEE 9832 £85 . u wanna variables X1 (U (sanitary facili 3:13 (education) naire and from t 1‘} Perceived Ave This sectj ability and qua] Lake county 8am; I've respondents quality of healt :ndicating "p001 *t— 100 variables Xl (treatment), X4 (length of residency), X5 (Sanitary facilities), X7 (nutrition), X12 (physical exam), X13 (education) and to omit the others from the question— naire and from the analysis. f) Perceived Availability and Quality of Local Health Care This section summarizes efforts to measure the avail- ability and quality of health services as perceived by the Lake county sample and by the Montmorency county sample. The respondents were asked to "rate the availability and quality of health care in the area" on a six level scale indicating "poor” to "good." To aid the respondents in identifying a position it was decided to label the poorest position (~3) and the best position (+3). To prevent respondents from immediately settling for the "in between," no average position was marked. Individuals who insisted on a middle position were asked to reconsider their answers and choose either (—1) or (+1). It was felt that this procedure would not introduce any bias since truly indif- ferent individuals would make their choice at random and those who were not truly indifferent would choose the side which comes closer to their previously "hidden" position. The rating was done both for 1968 and l971/72. The scores were later transformed into an ascending order between 1 = poor and 6 = good. To get a measure for the change between 1968 and 1971/72 an arbitrary "rule" was developed which would assign the highest value to a change from very low to very high and the lowest score to a change from very high to very low. Positions between the highest and ‘= . ._ n - mg to a pr. change to a ] alower posi‘ following ya; Priority I: Priority ii: Eriority ill Priority IV: Priority y; Priority VI: *t— 101 highest and the lowest "change index" were assigned accord— ing to a "priority scheme" which generally assumes that a change to a higher position is preferred to a change to a lower position. Further priorities are assigned in the following way: Priority I: Moving from below average to a position above average. Priority II: Above average and moving up. Priority III: Above average and staying at same level. Priority IV: Below average but moving up. Priority V: Above average but falling. Priority VI: Below average and staying at same level. Priority VII: Below average and falling. Priority VIII:Moving from above average to a position below average. This "rule" was established arbitrarily and reflects the author's intuitive evaluation of the importance of various changes. A different weighting would, of course, result in a different priority scheme. In spite of its arbitrary character, it was believed that this measure of change is superior to a procedure which would just record up— ward or downward change regardless of the starting and final position. The graphical representation of the applied "priority scheme" is shown in Figure 6. Once the scores for I968, l97l/72 and the "change index" for 1968-l971/72 were obtained it was possible to compare the two counties. Table 5 shows that the Lake county sample ranked availability and quality of Vin: 102 moaahoocs Sofinz :oEoQom apHLOflhQ: mo QOflQMpComom v ‘0 U) :'<“.QL_H \omzmmH y/ as we g . o :oEQmHHnmpmo onp : xowcfl mmcmgo: m p mop Hmoflnmmaw .m mLSMHm :xooca owcm£o_ as we as we as we. as we a a t a (Ur-1 /s,r \omzmmr—y kom—tmmt—o a m m .2 m m wmzm as we as mm \0 in :- ww nova \O Lh-‘ZI‘MNH :1; \OmrmNH \0 In 3* an Olrfi local health can iontmorenc y coun‘ 33018 5. Indice Qualit; T index IIIIIIIIIIIIIIIIIIIIIIIII-llIl-::r——————_____ 103 local health care significantly higher than the Mbntmorency county sample for 1968 and for l97l/72. Table 5. Indices of Perceived Availability and Quality of Local Health Care. Index ' Lake Montmorency Significance Sample Sample Level of Difference 1968 3.39 2.63 0.01 1971/72 4.50 3.31 0.01 1968-71/72 25.21 18.61 0.01 Table 5 shows that Lake county also had a higher "change index”, thus, indicating greater improvement between 1968-1971/72 than Montmorency. The advantage of the "change index" results from the fact that it gives a relative measure of change for each individual. The disadvantage is its dependence on an arbitrary "priority scheme" as outlined in the preceding paragraphs. Further research is required to determine the sensitivity of the "Change index" to various "priority schemes." This section presented a measure which utilizes the perception of clients as an indicator for evaluation. It is Obvious that this indicator measures more the in— put than the output of health services. To evaluate the Perception of health services ggtpgtg it is necessary to examine perceived health. s) W The analy: pattern as the < .h: indicators i :he Member Quesi :‘ois procedure :' indicators whic group and the < ————i 104 g) Perceived Health The analysis of perceived health followed the same pattern as the one described in the preceding section. The indicators were obtained from questions I and 2 of the Member Questionnaire (see Appendix C). The merit of this procedure is that it establishes time related output indicators which can be compared between the treatment group and the comparison group. Table 6. Indices of Perceived Health Significance Index Lake Co. Montmorency of Sample Sample Difference 1968 4.28 4.04 0.06 l97l/72 4.18 3.94 0.06 1968—71/72 21.53 20.27 0.06 Utilizing the procedure of the previous section it was posSible to establish Table 6. The table indicates that the Lake county sample rated its health significantly higher than the Montmorency county sample although the differences were less pronounced than the differences observed for health services. This can be interpreted as indicating that health services have improved considerably in Lake county,yet have not yet produced too noticable improvements in health status. This chapt county Health PI as "ad hoc comps ration of role I :issed from majc “(€710 bud V eived indice p. .Nu waitive impact fulfillment and concept if 105 8. Summary This chapter presented an evaluation of the Lake county Health Project by employing an analysis referred to as "ad hoc comparison." Emphasis was placed on the deri- vation of role fulfillment indicators measured by days missed from major activities. A discussion of the use of perceived indicators of health concluded the chapter. Based on the assumption that the Montmorency county sample constituted a valid comparison group it could be established that the Lake County Health Project had a positive impact on reducing the number of days missed from role fulfillment. A detailed discussion of the methodolog— ical and conceptual problems which surfaced through this analysis is given in Chapter VII. 1. Tom R. Houst Impact-Erie 31115, edit New York: N Aaron Anton and Overall Quarterly , Michigan Dep Surve: Re n.» ....- Leslie Kish, & sons) Inc . m JOhn B. Lans Methods (Ann Research, Un CI A. Moser W U. S. Depart Public Healt Publication WW I 4:. U: 3- Depart Public Healt Illness and mm Printing Off U' 3- Depart Census Use S Sstemsll w 6Ike‘e’I‘lgn l “waver. the are Orlly rel missed. Com days would b not SUbjecte I 106 FOOTNOTES 1. Tom B. Houston, Jr.,"The Behavioral Sciences: 10. Impact-Effectiveness Model," Evaluatinngocial Pro— rams, edited by Peter H. Rossi and Walter Williams New York: Seminar Press, 1972), pp. 60—61. Aaron Antonovsky, "Social Class, Life Expectancy, and Overall Morality," The Millbank Memorial Fund Quarterly, Vol. 45, No. 2 (April, 1967), pp. 31-73. Public Health, Michigan Health Michigan Department of Procedures Manual (Lansing, Survey: Reference and Michigan, 1970). Leslie Kish, Survey Sampling (New York: John Wiley & Sons, Inc., 1965). John B. Lansing and James N. Morgan, Economic Survey Methods (Ann Arbor, Michigan: Institute for Social Research, University of Michigan, 1971). C. A. Moser and G. Kalton, Survey Methods in Social Investigation, 2nd ed. (New York: Basic Books, Inc., 1972). U. S. Department of Health, Education and Welfare, Public Health Service, Vital and Health Statistics Publication Series (Washington, D. C.: Government Printing Office, 1963-)- U. S. Department of Health, Education and Welfare, Public Health Service, Age Patterns in Medical Care, Illness, and Disability: United States. 1968-1969, Series 10, No. 70 (Washington, D.C.: Government Printing Office, 1972), pp. 77, 79. U. S. Department of Commerce, Bureau of the Census Qensus Use Study, Report No. 12: Health Information §ystem~II (Washington, D.C.: Government Printing Office, 1971), pp. 129, 131, 132. However, the adjusted values of "days missed from work" are only relevant for a comparison among work days missed. Comparing them with play days or home work days would be inappropriate because these values are not subjected to the same transformation. HPACT 0F HEALT One of the hdlvlduals betwe to prepare themse an informal lear ieam and study h lissed school day Whities. If sch reasons they are heconomic terms his loss is a lo; :onsumption shall Placed on the fac‘ slrable. Once th« gut indicator has Problem of devisi: ihenomenon "schoo iialyze attendancr =s‘lduin area scho< =lll'rent liter-stun PDlicability of : "filtration . Desh. CHAPTER VI IMPACT OF HEALTH CENTER: METHOD 11, SCHOOL ATTENDANCE ONLY One of the principal roles our society expects individuals between ages 6 and 17 to fulfill is the ability to prepare themselves for the future life by means of formal and informal learning. A loss or reduction of this ability to learn and study has severe consequences in future years. Missed school days are, therefore, foregone investment oppor— tunities. If school days are missed because of health reasons they are also relevant indicators of discomfort or, in economic terms, indicators of disutility. How much of this loss is a loss in investment and how much is a loss in consumption shall not be discussed herein. The emphasis is placed on the fact that good health has utility and is de- Sirable. Once the relevance of school attendance as an out— put indicator has been established, the analyst faces the problem of devising a procedure to measure and analyze the Phenomenon "school attendance." This chapter attempts to analyze attendance data which have been collected by the Baldwin area school district (Lake County, Michigan). The current literature gives only a few references as to the applicability of school attendance records in health program evaluation. Deshaies and Seidmanl recommend this approach 107 or hthill et. evaluating scho (a) of attended a= b/c. Atten due to changing (e.g_ teacher 5 indiVidual studs. and averaged for schools (aas) tc 01‘ students . T0 accomp] necessary to isc cedm‘es as disct design Would reg ‘hich did not he 3.11 other Pelevg P°Du1ation . IIIIIIIIIIIIIIIIIllll--:::————————————* 108 and Tuthill et. a1.2 propose to utilize attendance rates in evaluating school health programs. 1. COnceptual'ConSiderations The use of attendance records as program evaluation tools necessitates the transformation of the raw data into units which are comparable over time and across classes, schools and school districts. The most desirable way of measuring variations in school attendance is the ratio (a) of attended days (b) over attendable days (c), that is a = b/c. Attendance can vary between full (a = l) and zero (a = 0). Such a ratio standardizes the measure of 1 school attendance for fluctuations in attendable school days due to changing weather conditions and/or political events (e.g. teacher strikes). The ratio (a), if defined for an individual student, is called (aI) and can be aggregated and averaged for classes (aaC), parts of classes (aaP) and schools (aaS) to facilitate comparisons among various groups of students. To accomplish a "with and without" analysis it is necessary to isolate non—project effects using design pro- cedures as discussed in the preceding chapter. An ideal design would require the identification of a control group Which did not have access to the Health Center but shared all other relevant socioeconomic criteria of the treatment pOpulation. despite the use (iii) Reco 01‘ attendable d (M Acc Considerably am (V) Comp the end or the (V1) Many Sitating extens Nation can be Despite 1; that if attends 1‘ °°mplete for It was in the school (1131 1°81ca1 States "‘0 differed ’2! Wived. It hau 109 2. Practical Problems--Data An examination of the available attendance records of the Baldwin area school system (Elementary, Junior High, Senior High) revealed the following: (i) Records for the years 1966—1972 are available for only 70 percent of all classes. (ii) There is a lack of uniformity in reporting, despite the use of standardized reporting forms. (iii) Records are inconsistent in showing the number of attendable days. (iv) Accuracy and completeness of reporting varies considerably among teachers. (v) Completeness of reporting drops generally toward the end of the school year. (vi) Many records are not yet summarized, thus neces- sitating extensive efforts of compilation before any com— putation can be undertaken. Despite these dificiencies, it was generally observed that if attendance is recorded for a specific day the record is complete for all members of the class. 3. Practical Problems--Design It was impossible to identify a control group within the school district. Without a classification of physio— logical states it was impossible to identify individuals Who differed 9n1y_in the amount and/or quality of care re~ ceived. It had to be assumed that those desiring the Project's servi and that others only comparison rolled in the H Health Center. Meet influen Previous health levertheless, i heIllaintained b en1'Olled and no 01‘ this ratio 0 attl"ibutoble to th"huh the Can of the asgumpti dents were subj e fluences and tha inPOI-tones of no the pillar upOn “implies, for had twice the an the Health Cente Mount after the that a ten Perce contPibute more 5”" 0f enrolled Enrolled familie ——1 110 Project's services would get services to "satisfy their needs" and that others would get them from other providers. The only comparison which could be made was between those en— rolled in the Health Center and those not enrolled in the Health Center. However, this required disregarding non— project influences such as differences in income, race, previous health history, and access to other types of care. Nevertheless, it was hypothesized that "some control" could be maintained by comparing the ratio of absences between enrolled and not enrolled. It was believed that a decrease cfi‘this ratio over the life of the Health Center would be attributable to the health services which were provided through the Center. This belief seemed justified in light of the assumption that both enrolled and not enrolled stu— dents were subjected to similar changes in non-project in— fluences and that these changes did not change the relative importance of non-project influences. This assumption is the pillar upon which the validity of this procedure rests. It implies, for instance, that if the not enrolled group Pmd twice the amount of income as the enrolled group before the Health Center was established it would have twice the amount after the Center was established. It also implies that a ten percent increase in disposable income did not contribute more (or less) to the health status of the chil- dren of enrolled families than to the children of not enrolled families. To compen and design the assessing the H The general app index of (absen divided by (abs It was establis random phenomen between classes the health 0 .Cc-rted absences PSported them 0 If A1 is lit average abs \ wl -. . .n)~ dividing the Sealth Center ( ~Soup (MN), 18 hexanple of 8 Procedure provi IIIIlIlllllllllIIIII-I-E:7——————————————“* 111 4. Procedure To compensate for the discussed limitations of data and design the following procedures were employed for assessing the Health Center's impact on school attendance. The general approach was to base the investigation on the index of (absences of the enrolled group of the class) divided by (absences of the not enrolled group of the class). It was established that the non-reporting of absences was a random phenomenon and did not introduce bias in a comparison between classes. (This could have had serious effects on the health outcomes if the before-project teacher re— ported absences only in winter and the after-project teacher reported them only during the months with the mild weather.) If A1 is the number of absent days of individual (i), the average absence (AA) is defined by AA = % SUMAi where @ = l . . .n). The average absence ratio (AAR) is obtained by dividing the average absence of the group enrolled in the Health Center (AAE) by the average absence of the not—enrolled group (AAN), leading to AAR = (AAE): (AAN). Figure 7 gives an example of such a calculation. It was assumed that this procedure provides sufficient information for testing the difference in school attendance "without" and "with" the Health Center. However, the data do not lend themselves for calculating total days missed. Since it was impossible to get year by year regis- tration records of the Health Center, a directory dated "Fall 1971" was used to divide school classes into HC-enrollees 3 c my oh F4 .0. 1% AV \rish. -. k a % UTHHOCHQM\ Ram. UtHHOcHqu \ . . a I -u-.-. -.-...\»\I . mnuvakflshleurvnl \ 112 mospmm oocomn< owmao>< so coapmazoamo .N ofismflm Z®‘l~ QqunLuQ P-krrwklllhlrl Ll (\ o - AQDOLM OOHHOLCO E02 AD OOMWHE WKNU @HWEV muEmO3pW \ d An pounnE wasp HHpEV Mo Shhnm WOHQME OOEOMQ< quSOhm UOHHOLC s0 KQUCH UZOMB m5<dDML0QEOU mOflPNE OOCQMQ< % 1'1 UUCMthQaV Quadsxfi0>< -Mw ToNtQerfi ha A a... a .thA. n). fluv— .....nJa_...sl.n.~< ....1. vflHVHu-Lma. aha>ruininra,«unwraandfib qu~4lw XLVHV-4 Usaflvfuflfi A QDNAQHUVN 116 .a canoe mo whohsss QSIUoUGSOL one Eons poo end when Hmcawaao can 509% ponawpho ohms wanna wasp pom mohsmfls one an .mao>oH opmaw op Loses mononpsohmm ca maonasz Aw ”opoz mo.a sm.a om.a one mmma mm.o mm.a one mmma Am coapoom Song pocampnov moapmm :mMBm<: one :mmommm: so owmno>< mo coapoom sm.a mm.a Amauaav ao.a Amuse emma am.o mm.a Aaauoav mm.a Asiev mmma mm.a sm.a Awlsv mmma ss.o :m.a Aslmv omma msma : osma mama . mama Aw Cowpoom Sous pocamppov mpmow Hoosom o>apdoowcoo 03% so moapmm mo owmho>< Am coapomm ma.a Amav sm.a Aaav mm.o Amv oa.a asv amma mm.a Aaav am.a Aoav sm.a Asv ae.a aev mmma mm.a Amv am.a Asa wmma mm.a Ase aa.a Ame mmma Ir, mssasma as\osma mm\sema se\mmma ammommmuammemav no cause AmmommmvnAmmem¢v It. so oapmm mmamd mmommm .lr nmao>oa opmaw am3©a>H©cH mo moapmm A< coapomm mowpmm oocomn< Amsosm poaaopco 802 an commae.mzmp mammv mesonspm o> so xoUCH moapmm mocomn< “Amsoam ooaaoaso mm commas ammo wammv so guaam it; apmneasoo mo sopca cache Am<¢v moaumm cocomn< owmho>< mo amps .WO h.m apmm postage so ameca osapssmdsoo one Nessa scone .moacmm postage omssosa .w tapes tIIIIIIIIIIIIIIIhHuIIIIIIllllllllI-I------ hfitiduals wi‘ assumes that 01 ever time ("38” Sate Classes-- unified into e ..w tne same 5 “ age and being Table 8. 1: students b( 117 Individuals with Similar Characteristics.” The other part assumes that only the same individuals should be observed over time ("Same Classes—~Same Individuals"). Same Classes——Individuals with Similar Characteristics This procedure examines a paired sample of classes over time. One class made up of students born in 1954 and one consisting of students born in 1955 is traced through grades 6, 7, 8, 10, ll and 12 respectively. Students are divided into enrollees and non—enrollees. While they are not the same students, they share characteristics such as age and being in the same class. Table 8, Section A shows that in 1966/67 in the class of students born in 1955 the enrolled group missed 1.61 times as many days as the NOT enrolled group. During the following years the enrolled group continued to miss more days than the NOT enrolled group, yet in 1971/72 they missed only 1.35 times as many days as the NOT enrolled group. A reverse trend appears to hold for the students born in lQSH who started out with a ratio of 1.10 in 1966/67 and ended up with a ratio of 1.42 in 1971/72. Section B of Table 8 averages the absence ratio over both the BEFORE and the AFTER two-year periods. The absence ratio between enrolled and not enrolled of those born in 1955 decreases from 1.59 to l.35. The group born in 1954, however, shows an increase from l.Ol in 1966—68 to 1.39 in 1970—72. The average of overall grades indicates an increase in the absence ratio from 1.30 in l966—68 (BEFORE) to 1.37 1&1970-72 (A1" Section C of l" The conf “AWp‘A‘ ‘ U‘Jhtrlcbe perlc ~3.,‘F L.‘ . ‘ Wcmo resm the ideal , _ 3» $5343: The I‘eQ u “100‘ Distric‘ 43K l at the tré 13% L. . ‘e the (We Oil Q . Studer may let h —f— 118 in 1970—72 (AFTER) for students born in l95U—55 (See Section C of Table 8). The conflicting results do not permit accepting the hypothesis that school attendance of the enrolled group has improved. In order to test whether this conflict is caused by students who were not living in the area throughout the complete period under investigation, a time series of those students residing in the School District both BEFORE and AFTER the Health Center's establishment is examined next. WWW The requirement of residing in the School District at the two points in time plus being enrolled in the grades for which attendance records were available eliminated 60 percent of the students. The attendance of the remainder was again divided into four cells: enrolled, NOT—enrolled, AFTER AND BEFORE. Two possibilities are open in the case where only those students are examined who have been residing in the School District both in 1966—68 and in 1970—72. One is to look at the trend of average absence ratios, the other is to compare the averages of individual trends. To make this difference clearer, consider EAl, EA2, ... EAn to be indivi— dual absences of enrolled students AFTER the Health Center was established and NAl, NA2,... NAn to be absences of NOT enrolled students AFTER the Health Center was established. Similarly let EBl, EB2,... EBn be individual absences of enrolled students BEFORE the Health Center was established Iantial, IiBZ,.. students BEFORE absences for ea .:.::;.:. = -‘- (sums H IlllllllllllIIII--::————————————————”F’ 119 and NBl, NB2,....NBn be individual absences of NOT enrolled students BEFORE the Center was established. The average absences for each of the four cases are: AAEA # % (SUMEAi), AANA = % (SUMNAi), AAEB = % (SUMEBi), AANB = l (SUMNBi). 1’1 The average absence ratios of enrolled over NOT enrolled are AARA = (AAEA): (AANA) in the AFTER-period and AARB = (AAEB):(AANB) in the BEFORE period. The trend of average absence ratios is calculated as the ratio of AFTER over BEFORE or TA = (AARA):(AARB). Individual absence ratios (IAR) are established for the four cells, AFTER, BEFORE, enrolled, NOT—enrolled by dividing each absence by the total average absence of its period (TAAA for AFTER and TAAB for BEFORE), resulting in IARAEi = (EAi):TAAA IARANi = (NAi):TAAA IARBEi = (EBi):TAAB IARBNi = (NBi):TAAB Individual trends (IT) are calculated by dividing for each individual the AFTER absence ratio by the BEFORE absence ratio. Thus ITEi = (IARAEi):(IARBEi) for students who are enrolled in the Center and ITNi = (IARANi):(IARBNi) for NOT enrolled students. The averages of the individual trends (AIT) are AITE = % (SUM ITEi) for enrolled and AITN = % (SUM ITNi) for NOT enrolled students. Trend of Average Absence Raios (TAf—An Analysis of Group Progress Table 9 shows that even by limiting the analysis to those who resided in the school district during both periods -« .: ui'v ~ v-dh 120 Table 9. Average Absence Ratios and Trend Index of Absence Ratios of Students Who Resided in the School District Both in 1966—68 and in 1970-72 a) Year of Average absence ratios (AAR) Trend index of birth 0 Ratio of (halfdays missed by the enrolled group): Absence Ratios students (Halfdays missed by NOT-enrolled group) (AFTER):(BEFORE) BEFORE AFTER 1966/67 l967/68 l970/7l 1971/72 A) RATIOS OF INDIVIDUAL GRADE LEVELS 1955 (6) 1.71 (7) 1.73 (10) 1.60 (11) 1.27 1 195u (7) 1.12 (8) 1.10 (11) 1.u2 (12) 1.u3 B) AVERAGE OF RATIOS OF TWO CONSECUTIVE SCHOOL YEARS (obtained from Section A 1955 (6-7) 1.72 (10-11) 1.U3 0.83 195A (7—8) 1.11 (ll-l2) 1.43 1.28 F) AVERAGE OF "AFTER" AND "BEFORE" RATIOS (obtained from Section B) 1 _ _ _ - Lash/55 (6 7 8) 1.u1 (10 11 12 1.u3 1.01 Note: a) Numbers in parentheses to grade levels. similar resul‘ chapter. Th0 those born in it 1.01. The 53111105: I’ES iiith th halprogress .-'.s summarized :‘nelr absence higher for ti. ,roup . Again f 121 similar results are obtained as in Section B of this chapter. Those born in l955 show a trend index of 0.83, those born in 1954 have one of 1.28 and the combined index is 1.01. The comparable figures in Table 8 are 0.8M, 1.37 and 1.05, respectively. Average of Individual Trends (AIT)——An Analysis of Individual Progress With this procedure it is possible to observe indivi— dual progress of students over the course of the project. As summarized in Table 10, individual students increased their absences in each group. The increase was, however, higher for the enrolled group than for the NOT—enrolled group. Again, enrolled students born in 1955 showed a lower increase (1.14) than those born in 1954 (1.37). 0) Cross Section Analysis with Reference to Grade Levels The problem with the previous comparison is that one does not compare the students at the same year in their own lives. One could hypothesize that those enrolled in the Health Center tend to increase their absences with growing age because of other, thus far unexplainable reasons. To isolate these effects, one has to examine students at com— Parable points in their lives, i.e. compare 6-7-8 graders BEFORE the Health Center was established with 6-7—8 graders AFTER the Health Center was established. The available records limit the investigation to a comparison between 6—7-8 graders in 1966-58 and in 1970—72. From Table 8, Section A, one learns that those born in table 10.Aver Sam; Dist tear of Er :lrth of St students (i C \— l955 Enl NOE \ , e95“ hm NOF. \ “3391 a) The div dua. rat: mis: 122 Table 10.Average of Individual Trend Indices of Paired Sample of Students Who Resided in the School District Both in 1966—68 and in 1970-72.a -— - Year of Enrollment Average of Birth of Status Individual Students (Health Trend Be— (Enrolled):NOT—Enrolled) Center) tween 1966-68 and 1970-72 1955 Enrolled 2.41 1.14 NOT Enrolled 2.10 1954 Enrolled 2.07 1.37 NOT Enrolled 1.50 Note: a) The individual trend indices were calculated by dividing individual AFTER-absence ratios by indivi— dual BEFORE—absence ratios. Individual absence ratios were established by dividing the half days missed by each student by the average number of half days missed by the student's total class. 1955 (BEFORE) 1959 (AFTER) 1 absence ratio pears in Sectz‘ averaged, P851 in grades (6-7 completely a: a reductior Eea‘lth Center ;erlod (0011113811 fl 12? comparing finisher this 1' Silure or due TflL V” “(W Ue determl‘ The aha] Jpn: 0f the Ch), 1 ‘ “Wee in ti ifctl 0!) C: W01; ‘ 5rL “WEI" I Obser at: 4h Cehtel.‘ Cz‘r uUH'e ‘ Hts (Sectj 123 1955 (BEFORE) have a higher absence ratio than those born in 1959 (AFTER) while those born in 1954 (BEFORE) have a';ggg; absence ratio than those born in 1958 (AFTER). The same ap— pears in Section B of Table 8 where the results are partially averaged, resulting in a comparative index of 1.25 for those in grades (6—7) and of 0.83 for those in grades (7-8). Only by completely summarizing (Section C of Table 8) one arrives at a reduction in the absence rate from 1.30 in the BEFORE Health Center period to 1.25 in the AFTER Health Center period (Comparative Index: 0.96). It is worthwhile to note that grades (7—8) in 1966—68 have consistently a lower absence ratio than the grades with which they are compared. This observation holds both for comparing similar grades at different points in time and for comparing similar classes at different points in time. Whether this is a peculiarity due to the data collection pro— cedure or due to actual differences in half days missed can- not be determined from the small sample of 12 classes. 6. Conclusions The analysis could not establish clear evidence of the impact of the project on the school attendance of the students enrolled in the Health Center. Comparing grades at the same level (comparative index of absence ratios: 0.96 in Table 8, Section C, would suggest a positive influence of the Health Center. Observing the same classes over time (trend index of absence ratios: 1.05) suggests a negative influence of the Health Center. If this evidence is split up into its com— ponents (Section B of Table 8) one sees that in both instances, different 013: test whether ' in either sit in the analys system is the study feasibl 2) Recorded Let us difference in “WE and th ficient evide Lishpent of t question the and against 116 survey d5 cussed in Chg her of days i irevious sch< fiance record: discussed pm Table 1 THE average 1 re port Only . —>— 124 different classes produce indices above or below 1.000. To test whether the differences are statistically significant in either situation one would have to include more classes in the analysis. An organized attendance record—keeping system is the basic requirement to make such an expanded study feasible. a) Recorded vs. Reported Absence Let us assume that more data were available and a difference in school attendance could be shown between the BEFORE and the AFTER period. Would the analysis be suf— ficient evidence to attribute this difference to the estab- 1ishment of the Health Center? To shed some light on this ‘ 1 question the attendance records of seven students were com— pared against survey data reported for the same seven students. The survey data were obtained as part of the survey dis— cussed in Chapter V. Parents were asked to report the num- ber of days their children could not go to school during the previous school year because of disability or health. Atten- dance records were obtained from the school records as discussed previously in this chapter. Table.11shows a comparison among the two sets of data. The average number of recorded absences is 61 whereas parents report only an average of six absences per year. The differ— ence could be even bigger considering the fact that school attendance records are usually incomplete. The downward bias Of recorded absence due to incomplete school records might, however, be comparable to under—reporting of absence by Student's parents, thus resulting in the same Table 11A C of . Individual . Student \_v_ Sb’li: 7 Average 125 Table 11.A Comparison Between Recorded and Reported Number of Half Days Missed. A Sample of Seven Students. .School.Records: .Survey Data: Individual F—Recorded Number of Reported Number Student Sex Half Days Missed of Half Days Missed 1971/72 A F 106 4 B F 52 16 C M 94 8 D F 14 2 E F 83 8 F M 15 O G M 60 14 SUM: 7 424 L 52 Average 61 I 7 relationship 1 aratio of 61 Table 1] for a variety some of them. then at all 1 Health Center ‘3) Overall 5 In tryi attendance is the school at lescribe thef 126 relationship between recorded and reported absence, i.e. a ratio of 61 to 7. Tablelj.clear1y indicates that students miss classes for a variety of reasons, health related reasons being only some of them. Should one use school attendance records then at all in evaluating the health status output of the Health Center? b) Overall School Attendance In trying to interpret TablelJ.one realizes that school attendance is an indicator of general social functioning of the school age population. However, since OEO Health Centers describe their overall objective as enhancing the state of Egtal social functioning of a target population, it can be concluded that OVERALL school attendance isga relevant out— put indicator of OEO Health Centers. Any investigation using OVERALL reported attendance as an indicator necessitates, however, a careful design for controlling other influences. The ability of the teaching staff to attract students to classes and the activation of disciplinary codes to enforce school attendance are probably Very strong factors of school attendance and have to be ex- plicitly considered in future research designs. Since such a design will most likely involve an analysis across school districts, the availability of and the access to complete attendance records is a basic requirement for future research utilizing school attendance. n tealth Re A neas Althou 3t Health C iEWES revea absences as ”mine of t IIIIIIIIIIIIIIIIlI---::—————————————* 127 0) Health Related School Attendance A measure of OVERALL school attendance is, however, a weak proxy measure for evaluating the impact of investment in health care facilities on the physical health of the target population. Information on absence caused by illness and disability has to be collected specifically as reported in Chapter V. d) Reflections on School Attendance Although no specific cause-effect relationship between the Health Center and school attendance could be advanced, it was revealing to get an estimate of the magnitude of absences as well as to detect the lack of sufficient re— porting of this rather important loss of economic resources. To appreciate this loss, the average number of half days (38) missed by the eight grades in 1970—72 was expressed in percent of the approximate number of half days of attend- able half days (270) during the reported period. It turns out that the average student missed about 14 percent of his school time. Put differently, the school district which finances the schools get only 86 percent of its spent re- sources to their intended use, i.e. provide formal educa— tion for the school age population. Whether this is a high or low figure can only be determined by comparative studies of different school systems, which requires extensive data and lies beyond the SCOpe of this study. nC.D uh I ‘4 H 1H»- m \l - AG nu“ NU hi so .u - hub :1. 0V ~\U c n. . 5M “U1. 3 .s a U .H U.- 128 FOOTNOTES John C. Deshaies and David R. Seidman, "Health Informa- tion Systems," Socio—Economic Planning Science, Vol. 5 (1971). pp. 515-535. Robert W. Tuthill, et al., "Evaluating a School Health Program Focused on High Absence Pupils: A Research Design," American Journal of Public Health (January 1972), pp. 40-42. RE: The pre problems fa health care be useful f ihapteI. pre T0 but advantagec in this 37 CHAPTER VII RECOMMENDATIONS FOR FURTHER INVESTIGATIONS The preceding chapters gave some indications of the problems faced in attempting an evaluation of a rural health care facility. Although the investigation was con— fined to rather limited subsections of the target popula- tion it was possible to gain some experience which will be useful in further studies on health care impact. This chapter presents the author's ideas on how he would organ- ize future research efforts in evaluating health projects. 1. Summary of Problems To put specific proposals into a perspective it is advantageous to reiterate the major problems encountered in this study and to propose some solutions to these problems. a) Control of Variables Role fulfillment of individuals is produced by a variety of individual and collective activities. Experience with this study indicated that there are three types of events which influence the production of health status. They are (i) differences in the services applied to individuals, (ii) differences in environmental conditions affecting health status and (iii) differences in the physiological and mental states of individuals. If we are interested in 129 analyzing the events, we hi constant. Ti neasure them both cont In order burr effect Sudltechnic (l) stratiff variables wl deation of are measure Sicurrence '1‘. ”measured ““188 and . flammable Usuall iHlOLmt and mental cor able Whig, Classificg diViGUals 0“ health is - “(1148 0 130 analyzing the health status impact of one or more of these events, we have to be able to keep the remaining events constant. To keep them constant we have to identify and measure them or assure that they are randomly distributed in both control and treatment groups. In order to isolate the effects of the project from other effects is is necessary to employ prOper techniques. Such techniques will have to involve three basic elements: (1) stratification of the population according to relevant variables which are measured before sampling, (ii) classi- fication of the sample according to relevant variables which are measured after sampling and (iii) assurance of random occurrence of relevant variables which could not be measured either before or after sampling. The less able the re— searcher is to measure relevant variables the more efforts will have to be spent on assuring a random distribution of unmeasurable variables within each selected stratum or class and on assuring identical distribution of the un— measurable variables between the selected classes. Usually there is some information available on the amount and kind of health services consumed and on environ- mental conditions, yet there is almost no information avail— able which would lend itself to a useful and operational classification of physiological and mental states of in— dividuals. Yet since the effect of similar health services On health status will differ from patient to patient de— pending on his physiological and psychological state it is important to control for these states. If the population cannot be SL1. characterist sary to desi an equal pro known in hot by random as however, forces the I identify whe viduals wit? 131 cannot be sufficiently measured according to those relevant characteristics of physiological—mental states it is neces- sary to design the "experiment" in a way which would assure an equal proportion of all states even if content is not known in both the control and the treatment group, i.e. by random assignment of peOple to each. However, the disadvantage of such a design is that it forces the researcher to dilute his objectives. He cannot identify what effect health services have on various indi— viduals with similar physiological states. He has to be satisfied with identifying the impact of particular health services on the total class or stratum of people. The only way he can be sure that he has a true control group is if there are identical distributions of physiological states both in the control group and in the treatment group. In order to get this assurance, it is necessary to randomly select both control and treatment groups from a common parent distribution or to attempt a measurement of the states which could prove that the two samples have the same known distribution of physiological states. Yet once the physiological states are kpppp it is considerably more efficient to classify the sample according to these States and to attempt separate evaluations for each state. Once it has been established that the distribution of unmeasurable characteristics is the same for the treatment groups it has to be assured that treatment is administered "evenly" within each treatment level. This implies that every unit within a particular treatment group gets the den of trea erai identiC graphical 8‘1 The dilE the conpari the treatme tions of a rthe onl tepulation bl classify mental stat 331d close < in ' 1‘ 1h achf 132 same amount and kind of treatment or has the ppm; chance of getting the same treatment. To control the randomiza- tion of treatment it will be necessary to administer sev— eral identical treatments for each treatment group. A graphical summary of this procedure is offered in Figure 8. The dilemma which the project evaluator faces is that he often deals with single projects and populations so small that the assumption of equal distribution of physiological states cannot be established even for the tota1_population of controls and treatment groups. Random sampling of single control and treatment groups does not improve on the incom— parability of distributions, since it would just result in the comparison of two samples which differ by more than the treatment, a situation which contradicts the assump— tions of a controlled experiment. The only way a health outcomes evaluation of a gmall pppulation can be achieved under controlled conditions is by classifying the samples according to physiological and mental states prior to treatment. Considerable experimenting and close cooperation with medical professionals is called for in achieving such a classification of physiological and mental states. b) Individual Role Adjustment Specifying role fulfillment as a relevant output con— cept raises the questions of: "What are relevant roles?" The study suggests four major roles in relation to the age Of the individual, i.e. play, school attendance, work, home work. Yet, experimenting with these groups shows 1-. ..-! Q. ~|l .. c -_~.._ ‘1"- us. ‘2’. . . o .,, \ N... | ,e. , .-.. t i . . :E‘au, "Mi-3: . .‘1 n ‘ ‘ . : “ETE‘a‘c l D \ i v. . 133 1. Parent population with particular distribution of physiological characteristics. i Distribution of physiological characteristic L 7 II. Division in centrol and treatment groups (across) III. Repetitions of treatments (down) Control Treatment Treatment (Treatment Level 1 Level n Repetitio Level 0) 1 (Observati l) 0 o O . Repetitio 2 1; (Observatio 2) - . g o g _ Repetition 3 (Observatio 3) F‘\\\\\~ . O O o. . . O Q Q . . O . . O . . Repetition n- (Observatlo n-l) O Q 0 Repetition n :Observatio n) ’ (\ K O Q . \‘\ Figure 8. Steps of an Ideal Design. that there each of the The pro terized by instance, t condition w iays from ‘n several yea sally to hi EEO "usual" ——_—,— 134 that there exists considerable variation of roles within each of the four groups. The process of growing up and getting older is charac— terized by selecting and adapting to different roles. For instance, the parents of a preschooler who has a heart condition will report that their child is missing many days from his "usual" major role, i.e. play. Yet after several years of learning to adjust emotionally and physi- cally to his weak heart the child has redefined his role and "usual" play is not any more his relevant role. The inability to play football on days when all his classmates play football "as usual" will not be reported by his parents as days missed from a major activity. Similarly, an individual who is injured might report a substantial number of days missed from his "usual" work during the first weeks or months after his injury. As time progresses he will have received treatment and his injury will have been cured to some degree and he will report fewer days missed from his usual activities. Yet, he also might have redefined his role and report only days missed from his newly adapted role—-however perceived. Many activity components which were vital in fulfilling his pre—injury role are irrelevant in fulfilling his new role. If he is able and willing to choose a role that does not require activity components which became unfunc— tional because of the injury, he will report fewer days missed from his role than if he would have refused to rede— fine his role. An injured football player might have to all his fut "play" them an individt field with easily ing Compone were no EXT not be a n6 1:) the m mid not l C? alternat alternatiw he will 016 E‘PEViOuS n his (a P02 SEQ— is man; f 135 all his future games, yet, once he becomes coach he can "play" them even if his injury never heals. Similarly, an individual who lost a leg might not be able to plow a field with a pair of oxen but he probably could do the job easily and in a fraction of the time if he had command over a tractor. The fact that he lost his leg might turn out to be no limitation at all if he could choose as his new role the task of doing research on the utilization of tractors. The preceding examples indicate that role adjustment is a product of a complex process determined by the follow~ ing components: (1) the need for adjustment (if there ‘ were no external pressure requiring action there would not be a need for any action—-neither cure nor adjustment), (ii) the availability of cure (if there were cure there would not be a need for adjustment), (iii) the availability of alternative roles (there must be a socially acceptable alternative to which the individual can turn, otherwise he will classify himself as being unable to fulfill his previous role), (iv) the feasibility of assuming another role (a role which is beyond the reach of the individual because of cultural, geographical or educational reasons iS not an alternative). The problems involved in evaluating health projects by means of role—fulfillment days are demonstrated by two related situations: on one side it is possible that two individuals with identical physical conditions and identical c report diff in the othe with identi ii- 136 identical care, yet differently perceived roles, will report differing amounts of role fulfillment days missed. On the other side, it is conceivable that two individuals with identical health conditions and different levels of care will report identical amounts of role fulfillment days missed just because they perceive their role dif- ferently. The evaluator does not know whether more role fulfillment days are due to health services or due to better role adjustment. He does not want to give any "credits" to the health project which are actually "earned" by role adjustment which is unrelated to the project under investigation, or miss giving credit for role improvement (shift) caused by the project. Health services seem to perform two tasks. Their primary task, as perceived by our society, is to aid the individual in striving for "ideal" roles. Their secondary task is to aid the individual in performing whatever role he has assumed well and comfortably. In other words, health services tend to affect role fulfillment more from the curative and preventive side leaving the task of role djustment to other social institutions or to the individual imself. However, health services do have some impact on Ole adjustment through mental and medical treatment which llows individuals to adjust to new roles. Examples are rtificial organs or psychiatric help in adjusting to ew roles. In the context of analyzing health projects, it is portant to realize that both better role adjustment and ——— better cure status if m let, the pr adjustment when it is restore the which enabl There exist 311d prevent Eat: role . IIIIIIIIIIIIIIII-I-I-E:___________—Ffi 137 better cure and prevention will result in better health status if measured in terms of role fulfillment days. Yet, the preceding discussion on determinants for role adjustment shows that role adjustment is called for only when it is (because of technology and costs) impossible to restore the physiological and mental state of the individual which enabled him to fulfill his previously specified role. There exists a trade-off between role adjustment and cure and prevention in terms of producing role fulfillment. Yet, role fulfillment produced through cure and prevention is often more highly valued than role fulfillment through role adjustment, because of the social preference for "ideal" roles. 0) Deviations from Ideal Roles——A Separate Output The role fulfillment indicator does not measure this second objective, i.e. movement toward "ideal" roles. In order to get an indicator for this output it is necessary to get a measure of the kind and degree of deviations from the ”ideal" roles. An analysis employing such an indicator of deviations from "ideal" roles will have to be standardi— zed for age and occupational differences, otherwise it would be possible that a control group which is exposed to high occupational hazards might be compared with a treatment group which works under much safer working conditions. Resulting differences would be mistakenly attributed to the project. Related problem of T his role. I‘ inherent r< .he case of have his an stttach hur situation? at fulfill Lily a qual capture thi (r) ) uestatl 138 d) Quality of Role Fulfillment Related to the problem of role adjustment is the problem of how painless or easy an individual can perform his role. Moving one's arm without discomfort is quite a different role fulfillment than moving it with pain. In the case of a bad arm the individual will probably attempt to adjust his role in such a way that he does not have to move his arm too often. But what can he adjust if his stomach hurts and nothing is available to correct this situation? He might still perform his major activities and fulfill his role, yet he fulfills it under pain. Only a qualitative indicator of role fulfillment will capture this potential output of health services. I e) Gestation Periods of Health Production ‘ Most project evaluation is done on a short term basis I and is often done mainly to facilitate "mid—course cor— rections;"Administrators often perceive that they cannot afford to wait a whole generation before they decide on the merits of a project. Yet by its character, health :are in one period affects morbidity and health status in later periods. Care during childhood may mean fewer >roblems as an adult. In order to have a basis for impact .nference, it is necessary to have long run information in the reaction of health status under different institution- ,1 and production arrangements of providing health services. The experience of this thesis suggests the need for ime series and longitudinal data in understanding the ’ ' tion Ong term effects of health inputs. Only this informa will enable Such inf ormz period whicl to want. H< . .. .31 (1) gestati< by utilizin, next study. LAVCEVEQ we analysis. rIIIIIIIIIIIll----"E:__________F 139 will enable a confident evaluation of health projects. Such information is expensive and requires an evaluation period which is longer than most administrators are willing to want. However, there seems to be no shortcut to solve the gestation problem directly. The only way the waiting period might be shortened is by utilizing experience from one study in designing the next study. Once we know more about the relationships involved we can refine the design and methodology of the lanalysis. Once we have a better methodology it can be hoped that the chosen indicators will become more sensi- tive and more reliable even in a shorter observation ‘ period. 1“) Sensitivity of Indicators Health production in deveIOped nations is probably taking place already at decreasing returns to scale. dajor gains are to be expected only in attacking the dis- ;ributional imbalance of health status. But, despite the lossibility of major returns by distributing health care mre equally, it is very difficult to point towards any hort term changes in health impact indicators which re attributable to particular projects. Especially, ultimate output indicators change only ?ter a considerable period of time. (Infant mortality 1d pain levels for some disorders are perhaps the only ceptions.) Intermediate output indicators such as rticular indicators of health attitude and preventive practices I are not the cess. Rese ship hetwee we how thi cutputs as and we can Pfsduction A ‘I ‘ } lbw P7 ihug 1‘ a1 h ere exists — 140 practices react faster. Yet, as their name suggests, they are not the complete target of the health production pro— cess. Research is needed which establishes the relation- ship between intermediate outputs and health status. Once we know this relationship, we can rely on intermediate outputs as relevant proxy measures of the ultimate output and we can be confident that the search for alternative production methods is headed in the right direction. 2. The ”Ideal" Experiment The preceding discussion suggests that individuals be grouped according to the following criteria: (i) age (ii) roles (iii) physiological and meantal state at the outset of experiment. Each of the resulting cells has to be divided into a :ontrol group and treatment groups. The treatment (dif— erent levels and kinds of health services) will be evalu— ted by measuring the change in pwp output indicators: (1) deviations from "ideal" roles (ii) role fulfillment days a. number b. quality 3. Practical Considerations Physiological and Mental States Thus far only age and broad roles are identifiable. are exists pp_easily obtainable classification which would permit :ion accordir it seems that scape-ration v The ma reassure the iiffer only 1‘ Although such ferent physic fiesirable ste 111 p .3 7.,- w ‘ v.1» dnCiaE cannot t lMl ould permit the investigator to group the client popula— Yet, ion according to physiological and mental states. t seems that such a classification could be achieved in ooperation with medical researchers. The main target of such a classification will be 0 assure the comparison of two similar individuals who iffer only in the kind and amount of treatment received. lthough such a grouping will still result in lumping dif— rent physiological and mental states together it is a sirable step since it reduces the variation within cells d enlarges the variation between cells——a feature which laracterizes the controlled experiment. The remaining :uses of variation within cells has to be compensated for random selection so the cells have identical distributions the unclassified physiological and meantal states. Where is cannot be done, causality is in doubt. Deviations from "Ideal" Roles As previously explained, this output indicator of riations from "ideal" roles is important because it allows to evaluate how well the health project performs its smary task, i.e. to aid individuals in assuming "ideal" es. The problem which the investigator faces is to ntify ”ideal" roles. "Ideal" roles are dependent on age culture. It seems that the four basic roles (play, 301, work, home work) have to be further refined in order warrant the establishment of deviations. The refinements . have to consist of specifying various functions within each role. Activity of suggested f This study daily livin ”ideal" pol 951 and men if? ‘ J'iw "‘J the fa I‘i VA will .mn 1H2 each role. Such refined indicators will be similar to Activity of Daily Living (ADL) indicators which have been suggested for charting progress of chronically ill people. This study proposes to develop age specific activities of daily living indices for all individuals. To arrive at this information, it will be necessary to ask questions such as: "Can you play football?" or "Can you move your arm above your head?" or "Can you see this dot?" It might appear that defining the deviations from "ideal" roles is the same thing as identifying physiologi— cal and mental states of individuals. However, it should be emphasized here that although the tasks to identify the two measures might be similar or even the same, they are aimed it two different things. Measuring the absolute physiologi— :al and mental state is an input measure, while measuring hanged deviations from ”ideal” roles is an output measure. nly the fact that we use some role measures and infer from hem the physiological and mental states causes us to confuse me two concepts. Inferring from the role on the underlying lysiological and mental state simplifies the definition ' states but it introduces the need for more elaborate signs if one wants to establish causality between physio— gical states and deviations from "ideal" roles. Qlassification of Health Services Inputs Although health services can be more simply identified measured than physiological states it has to be emphasized t their classification still poses serious problems. These problems ar inputs and order to es be able to sill have 0 be scaleabl ficult to h tide of inp Most Problems by PPOgrams an here Suffer SHE kidney Yet 3 , o x. 3‘ Elizes th site inpUts bill be har. Logical Sta End magnitu. Sibdy. The 01 lbiposing he bites and 01 Gui:Dll'tsl MI in be deve' ——7———wr “ * nth 1A3 problems arise mainly from the multitude and variety of inputs and input combinations applied to individuals. In order to establish production functions it is necessary to be able to show what impact a change in an input category will have on the outputs. In other words, inputs have to be scaleable or additive. Additionally, it is very dif— ficult to handle all the different combinations of a multi- tude of inputs. Most evaluation research has tried to overcome these problems by examining the impact of a few clearly defined programs and inputs on the conditions of individuals who were suffering from clearly defined categorical diseases, e.g. kidney diseases. Yet, in evaluating comprehensive health programs one realizes that the magnitude and variation of health ser- vice inputs is too great to warrant such an analysis. It will be hard to find individuals who have identical physio— logical states and have received all inputs in the same kind and magnitude with the exceptipn of the one input under study. The only way this problem can be solved is by de- composing health services inputs into their relevant attri— butes and observe how changes in the attributes affect outputs. Much future research and experimenting is needed 0 isolate relevant attributes so that useful input scales an be developed. The t the measure role categc insures the ;Tom) only 7' 144 4. Second Best Designs The basic requirement of the "ideal experiment" is he measurability of physiological and mental states, refined ole categories and input scale. Such refined measurement nsures that the treated population differs from the control roup pply in terms of identified treatment, i.e. we examine ole fulfillment of individuals in a specific physiological— ental state under specific treatment conditions and under ontrol conditions. In reality we will not be able to completely identify nd measure the state of individuals at the outset of the xperiments. In order to make an analysis at all we will ave to group the population in such a way that we can ssume that both the control group and the experimental roup have reasonably identical distributions of variables hich have an impact on role fulfillment. For instance, we have reason to believe that age, sex, cc and socio—economic conditions are correlated with ysiological and mental development. If we cannot identify ese physiological and mental states directly and use m as the basis for classification, we will have to be isfied with measuring the variables with which they are ociated. Epidemiological and socio—medical research been concentrating on establishing these relationships should, therefore, be tapped to facilitate classifi- ion of individuals in further studies. a) Social . It is randomly dl treatment g: ahealth Ce: in the Heal izportant t Cut the cou; SZudy will < \‘q in the ( Vulp‘le t0 0t —-———’—*——t 145 a) Social Experiments It is possible to organize an experiment which would randomly divide a population into a control group and a treatment group. The treatment group would be enrolled in a Health Center and the control group would not be enrolled in the Health Center. In order to assure control it is important to ascertain that control is maintained through- out the course of the experiment. The objectives of the study will determine how the experiment has to be controlled. If we are interested in identifying whether the enrolled group produces more role fulfillment days than the not enrolled group, we will be satisfied by comparing the control and the treatment group before and after the experiment. However, such an evaluation does not control for the pos— sibility that the control group might have received similar \ treatment from other sources. Therefore, if our objective is to evaluate treatment we have to control for variations in treatment. The only way such an evaluation can be organized is by classifying gpg_measuring health services inputs in the treatment group gag in the experimental group. It will be comparatively Simple to obtain this information on the treatment group, because they can be observed through the project. Yet, it is rather difficult to obtain this information from the control group unless it is observed on purpose. 3) Ethical Dilemma of Health Experiments This raises an ethical question on health services experiments, Why should one person get treatment because he is in ti person be c control grc bility of s society obj impossible El..Derit'lents denied care that the de. a\. "' Inferen O :3 f—J CD Cutput relationship is offered in Figure 9. Figure 9a hows the health status production function of a particular roup of individuals who have the same amount and kind of ervices available to them. Figure 9b displays the health tatus production function of individuals with particular BmOgraphiC and physiological characteristics under increas— ig availability of health services (regardless of individual >nsumption). Figure 9c combines the two factor-product [notions into one single production function which could estimated by regressing health status on the two dif— rent forms of health services (available and consumed). a) Healtl b 4) Health °) Health “Blue 9 150 a) Health Status ?’ //’"7fl Health status production function of a group of indi- viduals with particularly demographic gpd physio- logical characteristics who have identical health ser- vices available to them. Quantity of health ser- vices consumed (grouped _l’ according to attributes). b) Health Status Health status production function of a group of indi- viduals with particular demographic gag physio- logical characteristics who have different health services available to them. Quantity of available services (grouped *according to attributes). Health status production surface of a group of in- divuals with particular demo— graphic and physio= logical charac- 0) Health Status ,/ Consumed health “*4____,_-_—.. a.. 9-- . services / L teristics. /. f1 “/ ,w / ’ / . / l l .4 ’- Available Health services re 9. Health Status Production Functions Similar 1“ different Fro; that the 1 individua available less we k1 they get, services ; comes a p; PPOXy Will Size of t] The specific ] briefly 01 are eXhz tric met; cOm ling Whi: h0u< mail 0f 4| uIIIIIIIIIIIIIIl----'-'_——_——'fijflhwh*fififi 151 Similar functions can be established for all groups of different demographic and physiological characteristics. From the preceding discussions it becomes apparent that the more we know about health services and about the individuals to whom they go the less will the variable of available health services influence the health status. The less we know about the individuals and the health services they get, the more will the variable of available health services influence the health status mainly because it be- comes a proxy for care consumed. The acceptability of this proxy will increase with increasing pOpulation and sample size of the examined groups. 6. Specific Proposals The generalized strategy will have to be adapted to specific research situations. Two such situations are Driefly outlined here. ;) Utilization of Available Census Information The National Health Survey instruments collect health elated information throughout the United States. The pro- dure applied to arrive at the information is that of a "highly stratified multisage probability design. In the first stage, primary sampling units (PSU's are selected from a universe of 1,900 such units which are geographically defined and which collectively exhaust the territory of the 50 states and the Dis— trict of Columbia. Each PSU consists of a standard metrOpolitan statistical area (SMSA) or one or two contiguous counties. In a series of successive samp— ling steps, there is selected a final sampling unit which consists typically of a cluster of 6-9 neighboring households, called a "segment." Data are secured, mainly through personal interview, for each member of these sample households. The design makes each week Unit sens or m Appl require tk ofhealth economic c Fitneasux Le. chror Survey iru C0uld be StElte 0P 152 week's interviewing a probability sample of the entire United States, and weekly samples are additive in the sense that they can be combined for 13, 52, 10“, or more weeks." Applying the generalized strategy to these data would require the PSU's to be grouped according to availability of health services attributes and according to socio- economic characteristics of the sampled population. Out- put measures (days missed from roles, deviation from roles, i.e. chronic disability) can be obtained from the Health Survey interview. The results of such analysis would indicate how suc- cessful different levels of health service attributes were in affecting the output indicators; role fulfillment and role deviation. From this information it will be possible ~to draw conclusions on the impact of certain public pro— grams and projects which were in effect during the years the Health Survey was taken (approximately during the past 15 years). It should be emphasized that the presented sugges— ions are only speculative at this state since a detailed tvestigation of the survey procedure was beyond the scope this study. Further research and close c00peration with 9 National Center for Health Statistics is required be— e such a project could be launched. Utilization of Newly Collected Information The information desired in the generalized strategy 3 be collected in several communities throughout the or the country. Since it will be difficult to get data on ph; sumed, it 1' tetcarefu] ion on the thepopulat Since everal yea labile it w 3Xperience ESin the N 1‘ CbSerVin \i—"rr—WWHW 153 data on physiological conditions and health services con- sumed, it is suggested to obtain this information on small but carefully selected samples and infer from this informa- tion on the distribution of these characteristics throughout the pOpulations of interest. Since such a study will have to be undertaken over several years and since the American population is extremely mobile it will be necessary to control for population shifts. Experience gathered in local data collection efforts, such as in the New Haven Census Use Study, will be of great help 7 in observing communities over time. 7. Further Improvements of the Model--A Dynamic Model Since the researcher will usually enter the health Droduction process in the middle of the individual's life nd because of the previously discussed gestation period, t is advantageous to relate inputs and outputs by dynamic >del which will handle feedback and other time related Operties of the health production process. Such a model 1 be conceptualized by the following equation: HSC HSA H = f(H t—l’ t—l) t t-1’ re: (H) = health status of a particular population group 8 fulfillment deviation) (HSC) = health services which are consumed by a cular population group (HSA) = health services which were available in the ity. .s “...-4 Hm sub501 (t) is a 1 an of hez fining pre bythe pre ing block Available Health Services 154 The subscripts indicate that the health status of period (t) is a function of the health status in previous periods and of health services which were available or consumed during previous periods. The feedback mechanism described by the preceding equation can be represented by the follow— [Health Status (H)! R t Consumed Health ing block diagram. Available Health Services (HSA) Services (HSC) Figure 10. Dynamics of Health Production The flow diagram shows that health status is both a measure Of need in one period and a measure of success in another. AS a measure of need it determines (in connection with the availability of health services) kind and quantity of ser— Vices which will be consumed. As a measure of success it indicates the output of the consumed health services which was either produced directly or through intermediate out— Duts. Delay functions represent the delay (gestation period) Detween the administration of inputs and the appearance of >Utputs. 8-aat_a The empirical work undertaken in this study showed that ata SyStems are not complete at best and usually not avail— - d. ble at all. Data collection is expens1ve and complicate ugh—.92.... for succe a) Attri Res health se to combin :eptually state. F1 attributes search on 3011151 be ( should be Other pri‘ Organizat; in such :11 ”Mei 7’— 155 Cooperation with community organizations and simplifica— tions of the collection procedure are necessary requirements for successful research. a) Attributes of Health Services Research is needed to identify relevant attributes of health services. Only such a classification will enable us to combine and add health services analytically and con- ceptually even if they are non—additive in their physical Furthermore, grouping health services according to (Re- state. attributes will make systems analyses manageable. search on combinable attributes of certain health services could be done in the format of several M.S. theses) It should be emphasized that the health services attributes of other private or public activities (nutrition, community organizations, family structure, etc.) should be included in such an analysis. b) Physiological and Mental States Cooperation with medical research is called for in establishing a manageable and relevant classification System. Much of the experience of public health research and epidemiological research has to be incorporated in this kind of investigation. 0) Health Services Outputs Role Deviation. Cooperation with other social sciences and with medical sciences is necessary to establish a more refined classification of roles. As previously indicated, research is needed to identify degrees of deviation f1 156 from ideal roles. Chronic disability states as used in this study are only imperfect measures of role deviation. Role fulfillment. This indicator will follow the procedure described in this study by identifying days missed from the individual's major roles. Mortality. The usual procedures of calculating and analyzing mortality rates give satisfactory results. d) Intermediate Output Indicators Since many health projects attempt to affect changes of behavior and environmental conditions which might have im- pacts on health status, it is recommended to further improve the collection and analysis of this information. However, the relevancy of these data as intermediate output indicators can be determined only by examining their impact on health status. 9. Data Collection a) §ghool Data School boards in the selected communities can be asked for cooperation in organizing and maintaining attendance records. Once the school boards and the school administra— tors realize that these records will be used they will probably exert more care in documenting the information. The usefulness of school attendance records for evaluating education and schooling should make the additional efforts worthwhile even to the school administrators. In order to be able to utilize school attendance data as output indicators of health projects, it is necessary to establish the relationship between health related and health unrelat of the tion on known 1 health or sc‘noc o) m T? problems recall ; year). establis Haven Ce T‘r Every t} 157 unrelated reasons for absence. A carefully selected sample of the school age population will give sufficient informa— tion on the distribution. Once this sample distribution is known it can be inferred on the total school population and health related absence can be calculated for the total class or school or for any specific group of school children. b) Survey The survey reported in this study suffered from recall Droblems. Further investigation should employ a much shorter eecall period (four weeks) than the ad hoc comparison (one 7ear). The procedure could be a modification of the one established by the Health Survey Interview and the New aven Census Use Study. The sample could be divided into four sub—samples. very three months a sub—sample could be interviewed and sked to report health conditions over the preceding four eeks. Such a procedure is the least expensive way of l) interviewing the whole sample only page every year and .i) still capturing seasonal trends in sickness and activity ,mitations. Resource constraints will determine the size the sample and into how many sub-samples it would be vided. Experiments will have to be made to determine ether the procedure would not be equally reliable if sub— nples could be asked to report their health status by >ne. The trade-off between (i) shorter recall periods i phone interviews and (ii) longer recall periods and sonal interviews has to be established. Research by the ional Center for Health Statistics might be directly applic Th' gather of heal lens of search reader for Sea 19-22, h‘hfiher SEVeral It : 3111018 5( liestiOI TOUP arr —-7 158 applicable to this problem. 10. Questionnaires This section offers a brief discussion on how experience gathered in this study could be utilized in future studies of health impact evaluation. (A full treatment of the prob- lems of constructing questionnaires for heath evaluation re— search is, however, beyond the scope of this study. The reader is referred to the publications of the National Center for Health Statistics for detailed treatment of these problems). 1) Deviations from "Ideal" Roles This study used chronic limitations as a proxy for de- riations from "ideal" roles (see questionnaire items lM-l6, .9—22, 34—36, 38-40). To give the interviewer a check on hether the respondent has understood the question correctly, everal questions (items 8—12) were asked as an introduction. It is recommended to expand this introduction and to in- Lude some objective questions on health conditions. These lestions might be function—oriented, such as: "Can you move >ur arms over your head?" or disease—oriented, such as: "Have u had malaria?" or treatment—oriented such as: "Have you en treated for malaria?" Research on the Health Survey In— rview which includes many similar questions will have to be viewed to come up with acceptable questions. Once suitable estions are selected it will be necessary to combine answers these questions with answers from the questions on limita— ns in arriving at valid indicators of deviations from eal" roles. —i—_‘ 1' 159 Confusion Between Role Fulfillment and Role Deviation The questionnaire of the ad hoc comparison attempted classify the respondent into one of the five chronic nitation groups: (1) need help to move around, (2) unable do major activity, (3) limited in kind of major activity, ) limited in amount of major activity, (5) no limitations all. The objectives of the questions were to isolate Jiation from ideal roles from role fulfillment. These dif— ?€HC€S were not communicated with sufficient precision in a questionnaire used here. Therefore, after limitations I kind" or "in amount" have been established the follow- ; question should be asked: "On how many days did illness injury keep you from that kind of activity (play, work, .) which you should be able to perform even with limitation?" Fulfillment of Roles Questions l7, 23, Al and H2 asked for the number of s missed from roles. If these questions are asked over lOPt period of time (and include the change suggested :he preceding paragraph) they will be sufficient for Lining an indicator of role fulfillment. Fulfillment of Two Roles This study specified Egg roles from the school age lation (play agg_school) and the working age population ework agd work). it seemed that respondents did not always separate the events and gave the same or conflicting answers, for days and home work days. For instance, many respondents 'ted that they missed two weeks (14 days) from work, IIIIIIIIIIIIIIIIIIIII--""—'——_————————_*fi— 160 when further probing indicated that they actually missed only 10 days from wggk and were limited in their home work during the four additional weekend days. The interview administered in this study showed that the kind of occupation determines whether an individual will miss more work days (on the job) than home work days or whether he will miss fewer days than home work days. Although no organized analysis of this phenomenon could be obtained, it seemed that individuals who have jobs demanding hard physical efforts would miss more work days on the job than at home. In other words, they were too sick to work on the job, but not too sick to do the usual home work. On the other hand, individuals who had less strenuous jobs would miss more home work days than work days on their jobs—-i.e. :hey could do their job but needed rest time at home and ere, therefore, unable to perform their home work. Children eemed to be often too sick to go to school yet not suffi- iently sick to reduce their play activities significantly. It was evident that the two roles were not the same for st individuals. Yet, by requiring the respondents to re- 111 events over a long period of time (one year) much of the .fferences were not reported at all or in an incorrect manner. - appears that in the case of a long recall period, there very little gain in asking the working pOpulation and the hool age population about their fulfillment of twp roles. rther experimenting is necessary to determine a recall riod Which is short enough to enable the respondent to :all how he or his family members fulfill two specific roles. IIIIIIIIIIllllll""""‘:T_________—_*_ 161 e) Work Loss Not Related to Health Further experiments are necessary to make questions (24-31) of the member questionnaire more suitable for an investigation. It seems that combining self—employment and outside employment in one question causes unnecessary confue sion and reduces the reliability of the answers. Many respon— dents said that if they are not outside employed they are fully self—employed at home, thus resulting always in a full time total employment even when some of the time was non— productive. It appears that dividing the questions into questions on work and questions on home work serves to treat self-employment sufficiently. Dividing self-employment into further categories ioes not produce additional benefits. It is, therefore, sug— gested to apply questions (27—31) to outside employment only ind to adapt the questions to the selected recall period. ) Environmental Information The household questionnaire of this study offers some sight into what questions should be asked to classify the dividual's environment. The regression results indicate at length of residency in an area, nutrition, availability health care and education of the head of the household fluence health status most significantly. However, further udies must establish whether these environmental variables uld not be captured by relying on income differences exclu- vely. This study eliminated income differences to a large ent by concentrating on the food stamp and Surplus Commod— recipient group. 162 11. SamplingFrames The following records should be examined for their suitability as sampling frames: Census sampling frame. Attempts should be made to obtain access to the sampling frame used by the National Census. Although most of the Census information is inacces- ible to non-governmental users it might be possible to get cease to information which is useful in establishing a ampling frame. (This thesis did not investigate the easibility of such a procedure.) Michigan Health Survey (Project ECHO). The Michigan epartment of Public Health investigates the incidence of nvironmental conditions in its ECHO Survey (Evidence of ommunity Health Organization).8 One of the phases of the irvey consists of locating all residences of the survey “ea. The proposed study could be based on a sampling frame Irived from the Michigan Health Survey, wherever such a rvey has been taken. Should the proposed study concen- ate on communities which are not included in the Michigan alth Survey it would be possible to attempt to utilize a procedure devised by the Michigan Department of Public thh. The major disadvantage of this approach is that does not differentiate between permanent and seasonal idences. However, a differentiation between the two is remely crucial in surveying an area which has many sonal residences—~a prevailing phenomenon throughOut :hern Michigan. Postal Addresses. The fact that most rural addresses IIIIIIIIIIIIIIIIIIIIIIlllllll--::——————____l 163 are listed by rural route or postal box makes this a less desirable tool than it might appear on the surface. Yet, postal boxes and rural routes pose no problem if the survey is handled by mail. Phone Directory. The most significant disadvantage of this approach is rooted in the correlation between income and the access to private telephone services. Utility Bills. Power companies keep files on indus— trial and non-industrial users. Furthermore, they divide their records into those of seasonal and year—around cus- tomers. This division constitutes a considerable advantage in tourism areas. 1 Plat Book. Each county has usually an organization which sponsors the publication of a plat book. This plat book lists names of house and property owners and references them according to their position on a map. The plat book constitutes a sampling frame of the population that owns property yet it leaves out non—owners. Yet, non—owners may be heavily represented in that part of the population that is to be interviewed. Court List. Judges keep lists of area citizens upon whom they call for participation in juries. Cooperation with the courts is necessary in order to get access to this usually unavailable information. Lists of Population Groups. If the survey addresses a particular population segment it is sometimes possible to find an existing list of this group. An example is the use of the population receiving food stamps as a proxy for IIIIIIIIIIIIIIIIIIIIIIllllllll---——______l 164 the low income group as demonstrated in this dissertation. Similarly, one can use school records as the basis for investigating the population with school age children. The matching of various lists could result in the establishment of a complete frame, yet it usually creates more problems than it solves. l2. Interviewing Since the questionnaires are sufficiently simple it would be possible to obtain answers first by mail and inter- view non-respondents personally at a later time. To facili- tate such a "mixed" interviewing procedure it is necessary to establish kind and magnitude of reporting bias arising under the different procedures. Interviewers could be local residents who have received sufficient training or full time "itinerant" interviewers. Reward for Participation. Experience in the ad hoc comparison indicates that most respondents feel bothered by the questions. The knowledge of doing something "for the society" constituted the only incentive to participate. It is suggested to offer the respondents of future surveys some kind of a personal reward. One way of rewarding the participants would be to offer them useful publications of the Extension Service as an immediate reward for answering the questions. Such a procedure would fulfill at the same time an Extension function-~if the publications are selected appropriately. Distributed publications should be attrac- tive and of immediate relevance to the respondents, such as brochures on gardening, nutrition, sewing, etc. 165 13. Long Run Perspectives Future studies will have to be reformulated and reorganized several times. Experiments with questionnaires will finally produce an acceptable survey instrument. Once such a stage has been achieved, it is possible to attempt a formal cooperation between the selected communities and the University which would result in a division of labor. The communities would be responsible for collecting the informa- tion and the University would be reSponsible for organizing, upplying and interpreting the information as required by he communities. It can be expected that other indicators ill be developed simultaneously with the health status ndicators. The availability of numerous indicators will rarrant an elaborate computer and storage system for these ata. Experience gathered through efforts such as the TelFarm ystem at Michigan State University will be a valuable guide a developing such a system. The clients of the proposed rstem would be communities instead of individual enter— °eneurs. 14. Project Evaluation—~A Summagy The preceding discussions indicate the problems en— untered in project evaluations. The long gestation period health projects and the need for control are probably the > areas which will pose the most difficult problem to :ure research. As indicated in the discussion on ethical vblems of human experiments, our social values object giving one person care and denying it to the other. 166 This ethical foundation is also reflected in our political process. We do not want to admit that we give special health services to underprivileged families in one community and deny it to underprivileged families in another community. Yet, our "ideal" experiment would suggest just this. In reality we see that social demonstration projects are political pawns and if one community gets a special OEO health project, other communities will demand "their" rojects on grounds of equality. How can social production function research be accom— lished with constantly changing inputs, lacking controls nd long gestation periods of production? No certain answer an be given. Yet, this chapter on recommendations for urther studies should be concluded by emphasizing that the ong run examination of physically observable health services iputs will not be feasible because of the changing physical iture of these inputs. Only a disaggregation of services :to attributes and an examination of these attributes 11 make a long term production function analysis possible. d health status production can only be analyzed over a 1g time—~longer than most of our past projects have been existence. FOOTNOTES A grouping of this kind may involve an implicit choice of a common denominator in order to warrant some kind of additivity of different diseases and conditions. U. S. Department of Health, Education and Welfare, Public Health Service, Benefit—Cost Analysis of Kidney Disease Programs (Washington, D.C.: Government Printing Office, 1968). This assumes that there is no systematic self—selection by patients and/or no systematic selection by program administrators. The term "attribute" refers to the discussions in Chapters II and III. U. S. Department of Health, Education and Welfare, Public Health Service, Health Survey Procedure: Con— Questionnaire Development, and Definitions in cepts, the Health Interview Survey, Vital and Health Statistics, 2 (Washington, D.C.: Government Printing Series I, No. Office, 1964). U. S. Department of Health, Education and Welfare, Public Health Service, Age Patterns in Medical Care, illness, and Disability, Vital and Health Statistics, 70 (Washington, D.C.: Government Print— Series 10, No. 76—88. ing Office, 1972), pp. U. S. Department of Commerce, Bureau of the Census, Qensus Use Study, Report No. 12: Health Information System—II (Washington, D.C.: Government Printing Office, 1971). Michigan Department of Public Health, Michigan Health Reference and Procedures Manual (Lansing, Survey: Mich., 1970). CHAPTER VIII SUMMARY 1. Rural Health Care Recent years have witnessed a dramatic decrease in the vailability and quality of health care in rural areas. ifferent groups have advanced various explanations for 118 decay of what once seemed to be an adequate system. is most important causes are believed to be the concentra— ion of medical professionals in densely populated and high icome areas, restriction of entry into the medical profes- ion through alledgedly outdated licensing procedures and he medical profession's interest in highly advanced surgi- Ll procedures necessitating large and costly hospitals. e resulting abundance of specialists and the lack of neral practitioners has upset health services consumers d has received considerable political attention.l Federal and state agencies have been experimenting with ernative institutional and service arrangements in solving American "health crisis." Yet, so far very little is wn about the impact of programs on the health of the get populations. Like in other areas of public spend— , health project administrators have a fairly good idea hat they put into their programs, but lack information he socially relevant and desired outputs. Most of the 168 fill. 169 output measures used indicate how efficient an organization is in providing "units" of care, yet they do not tell how efficient those services are in producing health and changes in activity levels of people. Experience and intuition cause administrators to assume a positive correlation between the inputs (units of care) and the outputs (health). Very little factual knowledge is available to support or dispute their assumptions. However, if administrators try to establish a formal relationship they realize that they face a two—fold problem. One is to identify what the socially relevant outputs are and the other is to find a way how to measure the output. This 1 thesis attempts to provide some insight into the problem of optput identification and measurement of public projects in general and of health projects in particular. 2. Conceptual Framework for Evaluating Social Projects Economic theory provides the basic framework for evalua— ting input and output relationships of public projects. Yet, what is output of one production process is input into another production process. Focusing on an irrelevant process will produce irrelevant conclusions. For instance, it may be important to find out whether organization A produces hospital beds more efficiently than organization B and whether organi— zation A produces doctors less efficiently than organization . But, as long as we do not know whether hospital beds roduce more health than doctors or which combination of the W0 produces more health, we are concerned with irrelevant IIIlIIIIIIIIIIIIlI-l-I-IE:T__________" 170 information. The notion of "Derived Demand" indicates that the demand for many goods is derived from the goods they actually produce. The "New Theory of Consumer Demand" expands on this and shows that the same derived demand situation is applicable to our consumption pattern. We demand goods not because of themselves but because of their several attributes which we desire. The thesis suggests that these two theoretical concepts be used as a guideline for arriving at relevant input—output relationships. The concepts urge the evaluator to break down the input-output chain into its relevant components T) by asking questions such as: "Why is this product (service) 1 1 demanded?" and "What are the inherent attributes of the product (service) which make this product valuable to the consumers?" It should be emphasized that the thrust of this dis- sertation is to jiggg identify the relevant relationships and Ep2p_cope with the measurement problem—~and not the other way around.2 Once relevant relationships have been» defined, it is the task of the investigator to find out which of the variables cannot be measured at all given the present state of the art. The remaining variables will be measured directly or by relevant proxies. 3. Application of Concepts to Health ) Qutputs (Health Status) The proposed conceptual guidelines resulted at first in dentifying "health" as the relevant output. In attempting IIIIIIllllllll----::———————————"* 171 to measure health it turned out that by applying the "New Theory of Consumer Demand" one realizes that "health" too, is not demanded as an end in itself. Rather, the individual's desire for being able to fulfill his {913 in society causes him to demand health. It was, therefore, decided to measure health status in terms of its attributes, i.e. as role fulfillment. Several role fulfillment indicators could be conceptu- alized. For practical purposes, only a few could be selected. The chosen indicators were measured in terms of their absence and were expressed in the following way: (1) Days lost from wogk because of health reasons. 1 1 (ii) Days lost from pomp work because of health reasons. (iii) Days lost from school because of health reasons. (iv) Days lost from play because of health reasons. The use of role fulfillment indicators poses several conceptual problems. The process of growing older is char- acterized by selecting and adapting to different roles. Once people have adjusted to a certain role (e.g. working slowly or not performing manual labor), they will report OHly days lost from fulfilling their new roles. Yet, there Xists a social preference for "ideal" roles. Since the ttainment? of "ideal" roles constitutes another objective f the health production process it has to be included in n evaluation of health services. Because of the difficulty of defining "ideal" roles it as decided to use the following proxy indicators: f , 172 (1) Needs help to move around inside and outside the house. (ii) Cannot fulfill his role (play, school, work, home work), at all. (iii) Limited in the kipg_of his role fulfillment (i.e. cannot work in a factory). (iv) Limited in the amount of his role fulfillment (i.e. 3annot work as many hours as before) 3) Inputs (Health Services) One of the major problems of any evaluation is the classi— ‘ication and grouping of inputs. In evaluating health pro- ects one faces a multitude of different health services l nputs. How should one treat them in an analysis? This study suggested to apply the concepts of the "New heory of Consumer Demand" also to the problem of identify- ag input classes. Such a procedure would disaggregate ealth services or activities and goods producing health 1to relevant attributes, (e.g. preventive, diagnostic, *eatment, skilllevel of providers, etc.) and base the .alysis on these attributes. Once the input-output rela— onship between attributes and health status has been tablished it is possible to use prices as a guideline for termining which health services and goods should be used the production process. Because of resource and time litations it was decided to concentrate in this study on ,1th status outputs, and to pursue the identification measurement of inputs only marginally. 173 Resource constraints precluded an indepth analysis of the input side of the attempted input-output analysis. Yet, some experimenting with existing data was undertaken. Particularly the question of number of services consumed was of interest to this investigation because such information constitutes the cornerstone for a complete production function analysis. Since most input data are reported in terms of health services per unit (patient) it was of interest to get a measure which would indicate how evenly the inputs were distributed among the units. Appendix A describes the analysis of concentration of health services inputs supplied by the Health Center. Lorenz curves and Gini coefficients are suggested as considerable improvements over the existing methods of reporting averages only. .4. Empirical Analysis of Health Status Outputs Once "role fulfillment" and "role deviation" were es- :ablished as the major "ultimate" output indicators of 1ealth services it was decided to apply these concepts in rhe case of a rural health project located in Northern Michi- an. (The Western Michigan Comprehensive Health Services roject with its main clinic in Baldwin (Lake County) serves 30ple in an area consisting of the four counties of Lake, ison, Manistee and Newaygo.) Since the major objective of this part of the study was . examine the feasibility of measuring health status output was decided to limit the evaluation to segments of the 174 population. The empirical study explored two approaches for obtaining data on health status output: a survey and existing records of school attendance. Attempts to establish an experimental design as the >asis for the ”health status output" survey showed that :reatment units have to be classified according to physio— .ogical and mental states in order to accomplish a "with nd without" analysis. Lacking feasible procedures to easure the physiological and mental states it was decided o resort to an "ad hoc comparison" which compares the reatment group with a comparison group that resembles the reatment group in all characteristics with exception of 1 1e treatment, i.e. health services. Basing the matching of treatment and comparison group 1 regional, local and personal socio—economic characteristics, was decided to compare a sample of Lake county food stamp cipients with a sample of Montmorency county (Michigan) rplus Commodity recipients. In developing the questionnaires for the survey it was ided to utilize questions of the National Health Survey the greatest possible extent. The main reasons for this ategy were lack of experience in establishing health ated questionnaires and the desire to have comparable a. Although no comparison between national data and se of this study was attempted it was felt that such orts would be desirable for future (and preferably more lete) surveys for which this study would be the initial ‘ all 175 The heads (or the spouses of the heads) of 84 families were interviewed in each county when they came to pick up their commodities or food stamps. The respondents would nswer questions about their household, their health and bout the health of each member of the household. Although here is a considerable recall problem with health inter— iews about periods which extend longer than two weeks in he past, it was necessary to ask the question with refer— nce to the whole past year in order to facilitate a statis- ical analysis of the collected information and to avoid asonal variations. An analysis of the data indicated that the Lake county ample missed ggwgg days from play, school, and home work 1d more days from work than the Montmorency county sample. [22 all differences were significant at the 10 percent >Vel however.) The fact that Lake county individuals have ss work opportunity (unrelated to health) tend to make e difference in work days missed even more severe. A regression analysis established that other variables trition and level of education of the head of the house— d) had some, yet not conclusive, impact on role fulfill- t. Confining the analyses to a specific income group mmodity and food stamp recipients) controls probably necessary for a variety of other health related variables. investigation over the total income range is necessary ind out whether income would be a sufficient proxy for tifying the population according to health related ables other than health services consumed. 176 A comparison of "role deviation" indicators showed that the Lake county sample (treatment) had more individuals in categories of severe limitation than the Montmorency sample (comparison). Since these indicators can only gauge longer term changes it was decided not to use them for an evaluation of the Lake county project which had existed ot more than five years. The "role deviation" indicator 's, therefore, a social indicator and not a "program output oefficient" as defined in this study. However, the "gesta' ion period" of health production is not only a problem in tilizing ”role deviation" indicators but also (though to a esser degree) is dealing with "role fulfillment" indicators. The other major empirical investigation consisted of xamining school attendance records of the Baldwin School istrict. Several different procedures were utilized to (pose differences between students who were enrolled in is Health Center and those not enrolled. The major advantage of this investigation as compared the survey was availability of recorded data both for e "before" and the "after" project period. However, con— ntrating on classes within the project posed several ign problems. Control could be considerably improved utilizing the procedure employed in this study in an hoc comparison" or in more elaborate designs suggested this research. The investigation of the limited number Lake county school classes did not give any basis for irming or denying the production of health status output the Health Clinic. 177 5. Concluding Remarks The emphasis of this study was on methodology and conceptualization. It was possible to develop a theoreti— cal framework for evaluating public projects, particularly health projects. The research succeeded in isolating relevant health output concepts and identifying suitable measures to collect the desired information. An applica- tion of the developed concepts and measures to the evaluation of a particular rural health project did not provide enough evidence for clearly assessing the impact of the project on activity levels (role fulfillment) of the examined poverty group. However, a major result of this thesis, as presented in Chapter VII, is the experience gained in survey concepts and design which can be used in future investigations. 178 FOOTNOTES John Walsh, "Medicine at Michigan State (1): Education and Legislation," Science, Vol. 77 (1972), pp 1085-7. Relevancy is ultimately a political decision but analysis can provide inputs for that process. 3.... .21: . BIBLIOGRAPHY BIBLIOGRAPHY A) Planning and Evaluation—General Asimov, Morris. Introduction to Design. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1962. Blalock, Hubert M. Jr. Theory Construction: From Verbal to Mathemati- cal Formulations. Edited by Herbert L. Costner and Neil Smelser. Englewood Cliffs, N .J .: Prentice-Hall, Inc., 1969. Caro, Francis G. , ed. Readings in Evaluation Research. New York: Russel Sage Foundation, 1971. Easton, David. A Systems Analysis of Political Life. New York: John Wiley and Sons, Inc. , 1965. Lancaster, Kelvin. Consumer Demand: A New Approach. New York: Columbia University Press, 1971. Rossi, Peter H. , and Williams, Walter,eds. Evaluating Social Programs: Theory, Practice and Politics. New York: Seminar Press, 1972. Schmid, A. Allan. Analytical Institutional Economics: Challenging Problems in the Economics of Resources for a New Environment. Paper presented at American Agricultural Economics Association Annual Meeting, University of Florida, August, 1972. Stolper, Wolfgang F. Plannirg Without Facts: Lessons in Resource Allocation from Nigeria's Development. Cambridge, Mass.: Harvard University Press, 1966. Suchman, E. A. Evaluative Research. New York: Russell Sage Founda— tion, 1967. JNESCO. Main Trends of Research in the Social and Human Sciences. Part One: Social Sciences. Paris: UNESCO, 1970. B) Planning and Evaluation—-Health uster, Richard, Ieveson, Irving, and Sarachek, Deborah. "The Produc— tion of Health: An Exploratory Study." Journal of Human R_e_sources, Vol. 4 (Fall 1969), pp. 411—433- 179 180 Baum, Martin A. Primary: Health Care Delivery Simulator. Washington, D. C. . The Mitre—1: Corporation, 1971. Bellin, Seymour 8. , and Geiger, H. Jack. "The Impact of a Neighbor- hood Health Center on Patients' Behavior and Attitudes Relating to Health Care: A Study of a Low Income Housing Project." Medical Care, Vol. 10, No. 3 (May-June 1972),p pp 2211-239. Borgatta, Edgar F. "Research Problems in Evaluation of Health Service Demonstrations . " The Milbank Memorial Fund Quarterly. Health Services Research II, Vol. 44, No. 4, Part 2 (October 1966), pp. 182-201. Bridgman, R. F. "Some Methodologic Problems in Health Practice Research and Health Planning." International Journal of Health Services, Vol. 2, No. 1 (February, 19727, pp. 51-62. DeGeyndt, Willy. "Five Approaches for Assessing the Quality of Care." Hospital Administration, Vol. 15 (Winter 1970), pp. 21—142. Densen, Paul M. "Some Practical and Conceptual Problems in Appraising the Outcome of Health Care Services and Programs. " Outcomes Conference I-II. Edited by Carl E. Hopkins. Department of Health, Education, and Welfare (1970), pp. 22-23. )onabedian, A. "Evaluating the Quality of Medical Care." The Milbank Memorial Fund Quarterly. Health Services Research II, Vol. 4F, Part 2 (July 1966), pp. 166—206. lunn, C. L. "The Economic Value of Prevention of Disease." Indian Journal of Economics, Vol. 4 (1924), pp. 127-141. etter, R. B., and Thompson, J. D. "A Planning Model for the Delivery of Health Services." Department of Administrative Sciences and Department of Epidemiology and Public Health, Yale University , n . d. (Mimeographed) . Drst, Brian E. The Grisly Analytics of Death, Disability, and Disbursements. Professional Paper No. 81. Arlington, Va. : Center for Naval Analyses, 1971 nzberg, E. , and Rogatz, P. Planning for Better Hospital Care. New York: King's Crown, 1961. PO, A. 8. "Planning an Information System for Health Services." Journal of Methodoloa in Medical Research Information and Documentation, Vol. 11, No. I (1972), pp. 1-8. 983', B. J. "Evaluating National Hospital Policy: A Systems Approach." Health Services Research (Spring 1972), pp. 11-22. ”— 181 Kalimo, E., Sievers,K ., Purola, T., and Nyman, K. The Utilization of the Medical Services and Its Relationship to Morbidity, Health Resources; and Social Factors. Publications oTthe National Pensions Institute of Finland, Series A3. Helsinki, Finland: Research Institute for Social Security, 1968. Klarman, H. E. "Syphilis Control Program. " Measuring Benefits of Government JInvestments. Edited by R. Dorfman. Washington, D. C.. The Brookings:- Institute, 1965. Klarman, H. E., Francis, J ., and Rosenthal, G. O. "Cost-Effectiveness Analysis Applied to the Treatment of Chronic Renal Disease." Medical Care, Vol. 6 (July-August 1968), p. 118. Kuhner, A. "The Impact of Public Health Programs on Economic Develop- ment: Report of a Study of Malaria in Thailand." International Journal of Health Services, vol. I, No. 3 (1971), pp. 285-92. Leveson, Irving, Ullman, Doris, and Wassall, Gregory. "Effects of Health on Education and Productivity." Inguiry, Vol. 6 (December, 1969), pp. 3-11. Lipworth, L. "Cost Effectiveness of a Preventive Program in Renal Disease." The Milbank Memorial Fund Quarterly, Vol. 47, Part 2 (July 1969), p. 70. Mishan, E. J. "Evaluation of Life and Limb: A Theoretical Approach." Journal of Political Economy, Vol. 79 (July—August 1971), 5.8.7-705 VIushkin, Selma J. "Health as an Investment." Journal of Political Economy (October, 1962), pp. 129-157. Javarro, Vicente. "Systems Analysis in the Health Field." Socio— { Economic Planning Science, Vol. 3 (1969), pp. 179-189.__ - >ffice of Economic Opportunity. An Evaluation of the Neigqborhood Health Center Program: Summary of Results and Methodoloa. Washington, D. C., 1972. lffice of Economic Opportunity. Study to Evaluate the OEO Neighbor- hood Health Center Program at Selected Centers . Volumes I, II and III. Final Report, GEOMET Report, No. HF-71. Washing- ton, D. C. ,1972. rtiz, Jorge, and Parker, Rodger. "A Birth-Life—Death Model for Planning and Evaluation of Health Services Programs." Health Services Research (Summer 1971), pp. 120-143. chulberg, Herbert C. , Sheldon, Alan and Baker, Frank, eds. m Evaluation in the Health Fields . New York: Behavioral Publi- cations, 1969. ——'— 182 Shapiro, S. "Bind Result Measurement of Quality Medical Care. " The Milbank Memorial Fund Quarterly. Health Services Research II, vol. 35, Part I (April, 19 7, pp. 7—30. Sheldon, Alan, Baker, Frank, and McLaughlin, Curtis P. S osium on Systems and Medical Care ,_ Harvard University, 19 Cambridge, Mass.: MIT 83, 1970. Stewart, Charles T., Jr. "Allocation of Resources to Health." The Journal of Htmen Resources, Vol. 6, No. 1 (1971). Texas Hospital Association. Health Resources Planning Unit. Toward A Health Services Simulator. An Interim Report. San Antonio, Texas: Texas Hospital Association, 1971. Torrance, George W. , Thomas, Warren H. , and Sackett, David L. "A Utility Measure of Effectiveness for Health Care Programs." Paper presented at the 39th National Meeting of the Operations Research Society of America, Dallas, Texas, May 5-7, 1971. Dorrance, George W., Thomas, Warren H., and Sackett, David L. "A Utility Maximdzation Model for Evaluation of Health Care Programs." Health Services Research (Summer 1972), pp. 118-133. Tuthill, Robert W. , et. a1. "Evaluating a School Health Program Focused on High Absence Pupils: A Research Design." American Journal of Public Health. (January, 1972), pp. 40-142. llmenn, John E. , ed. The Application of Mana ment Science to the Evaluation and Design of Regional Health Services. Hempstead, N. Y.. Hofstra University, 1968. .S. Department of Health, Education, and Welfare. Public Health Service. Estimating The Cost of Illness. Health Economics Series No. 6 Washington, D. C. . Government Printing Office, 1966. S. Department of Health, Education, and Welfare. Public Health Service. Benefit/Cost Anaiysis of Kidney Disease Programs. Washington, D. C. . Governmental Printing Office, 1968. 3. Department of Health, Education, and Welfare. Public Health Service. Proceedings of a Conference on Conceptual Issues in the Analysis of Medical Care Utilization Behavior. Rockville, Md.: Health Services and Mental Health Administration, 1969. L Department of Health, Education, and Welfare. Public Health Service. Outcomes Conference I-II: Methodology of Identitjying, Measuring and Evaluating Outcomes of Health Service Programs, Systems and Subsystems. Rockville, Md.: Health Services and Mental Health Administration, 1969. — 183 Weisbrod, Burton A. Economics of Public Health: Nicasuring the Economic Impact of Diseases . Wladelphia: University of Pennsylvania Press, 1961. C) Other Social Science—Health Research Anderson, James G. "Causal Model of a Health Services Systems." Health Services Research (Spring 1972), pp. 23-143. Boulding, Kenneth E. "The Concept of Need for Health Services." _T_'he Milbank Memorial Fund Quarterly. Health Services - Research II, Vol.“lflf, No. ll, Part 2 (October 1966), Burton, Lloyd Edward and Smith, Hugh Hollingsworth. Pubiic Health and Community Medicine for the Allied Medical Professions. Baltimore: Williams and Wilkins Co., 1976. Economics of Health and Public Polic . ‘ lic ' Campbell, Rita Ricardo. Washington, D. C.: American Enterprise Institute for Policy Research, 1971 . Clark, D. , and MacMahon, B. , eds. Textbook of Preventive Medicine, Boston: Little Brown and Company, 1967. Doherty, Neville J. "The Economic Structure and Performance of the Medical Industry in Michigan's Grand Traverse Region." Unpublished Ph.D. dissertation, Michigan State University, 1970 . llinson, J. , and Herr, C. E. A. "A Sociomedical View of Neighbor- hood Health Centers." Medical Care, No. 8 (March—April 1970) pp. 97-103. lling, Ray H. "Health Planning in International Perspective." Medical Care, Vol. 9, No. 3 (May 1971), pp. 2114-2311. rang, K. "Contributions Towards the PhiIOSOphy of Health." International Journal of Health Services, Vol. I, No. 2 (1971), pp. 98#105. sin, Rashi. The Doctor Shortage: An Economic Diagnosis. Washington, D. C.: Brookings Institution, 1967. 1dstein, Martin S. Economic Analysis for Health ServicetEfficiency: Econometric Studies of the British National Health Service. Chicago: Markham Publishing Co., 1968. "The Contribution of Health Services to the American 3113, Victor R. Economy." The Milbank Memorial Fund Quarterly. Health Services Research II,Tol. 144, No. U, Part 2 (October 1966), pp. 65-103. —7— ... -..... 1814 Fuchs, Victor R. "Health Care and the United States Economic System." The Milbank Memorial Fund Quarterly. Health Services Research I, V01. 1, No. 2, rare I (April 1972), pp. 211—237. Ginzberg, Eli , and the Conservation of Human Resource staff. Urban Health Services: The Case of New York. New York: Columbia university Press, 1971. Griffith, D. H. 8., Banana, D. V., and Mashaal, H. "Contribution of Health to Development." International Journal of Health Services. vol. I, No. 3 (1971), pp. 2532270. Kane, Robert L. "Detemination of Health Care Priorities and Expec- tations Among Rural Consumers." Health Services Research, Vol. 1! (Summer 1969), pp. 142—151. Klarman, Herbert E. The Economics of Health. New York: Columbia University Press, 1965. Klarman, Herbert E. "What School Can Teach About Health Services Planning." International Journal of Health Services, Vol. I , mo. 2 (1971), pp. 15A-16S. Kosa, John, Antonovsky, Aaron, and Zola, Irving Kenneth, eds. Poverty and Health: A Sociological Analysis. Cambridge, Mass.: Harvard University Press, 1969. MacMahon, Brian, and Pugl, Thomas F. Epidemiolog: Principles and Methods. Boston: Association of American Medical Book Publishers, Little Brown and Company, 1970. 'echanic, David. Medical Sociologi: A Selective View. New York: Free Press, 1968. "Redefining the Health Problem and Implications lavarro, Vicente. HSMHA Health Reports, for Planning Personal Health Services." 86 (August 1971), pp. 711-7214. auly, M. V. Medical Care at Public Expense. New York: Praeger Publishers, Inc., 1971. lyne, Marie Britt, and Praiss, Israel L. "Patterns of Medical Care: Comprehensive Care vs. Categorical Disease Programs." Medical Care, Vol. 10, No. 1 (January-February 1972), pp. 1-7. >sem1an, Cyril. "Problems and PrOSpects for Comprehensive Health Planning." American Journal of Public Health (January 1972), pp. 16-19. I'kar, Shyamelendu. "The Copper Country Medical Industry of Michigan as It Serves Rural People." Unpublished Ph.D. dissertation, Michigan State University, 1969. 185 Saward, Ernest W., and Greenlick, NErwyn R "Health Polic . y and the HMO." The Milbank mebrial Fund Quarter . Health Services Research, Vol. 1, No. 2, Part I (April 1972), pp. 1&7-176. Schultz, T. W. Investment and Human Capital. New York- The Free Press, 1971. Simon, Julian L. "Family Planning Prospects in Less—Developed Countries, and a Cost-Benefit Analysis of various Alternatives." The Economic Journal, 80 (March, 1970), pp. 58-71. Somers, Anne R. Health Care in Transition: Directions fer the Future. Chicago: Hospital Research and Education Trust, 1971. State of Michigan. Executive Office of the Governor. The Right of Michigan Citizens to Health Care. Lansing, Michigan, 1972. Tbrrens, Paul R. "Administrative Problems of Neighborhood Health Centers." Nbdical Care, V01. 9, No. 6 (Novembethecember 1971) , pp. 7— 97. University of Michigan. The Economics of Health and Medical Care. 1 Proceedings of the Conference on the Economics of Health and Nbdical Care, Why 10-12, 1962, Ann Arbor, Mich.: University of Michigan, 1964. U.S. Department of Health, Education and Welfare. Public Health 1 Service. A Program fbr Research in Health Economics. Health Economics Series No. 7. washington, D. 0.: Government Printing Office, 1967. Wéinerman, Richard E. "Research on Comparative Health Service Systems." Nbdical Care, Vol. 9, No. 3 (May-June 1971), pp. 272-290. Weisbrod, B. A. "Investing in Human Capital." Journal of Human Resources, V01. 1, No. 1 (1966), pp. 5-21. D) Research on Health Outputs Deshaies, John C., and Seidman, David R. "Health Infbrmation Systems." Socio—Economic Planning Science, Vol. 5 (1971), pp. 515-533- FEnshel, S., and Bush, J. W. "A Health Status Index and Its Applicap tion to Health Services OutcomesJ' Operations Research, Vol. 18 (November-December, 1970), p. 1021. Flax, MiGhael J. Blacks and Whites: An Experiment in Racial Indicators. Washington, D. C.: The Urban Institute, 1971. Jones, Martin V., and Flax, Michael J. The Quality of Life in Metro— politan Washington, D. C.: Some Statistical Benchmarks. Washington, D. C.: The Urban Institute, 1970. *7 186 Jones, Martin V., and Flax, Michael J. The Quality of Urban Life in the United States: A Study of Urbgnfi Indicator Methodolog and Presentation. Working Paper 13612. Washington, D. C.: The UrbanT'lstitute, 1970. Katz, 8., Downs, T. D., Cash, H. R., and Grotz, R. C. "Progress in Development of the Index of ADL." Gerontoloa, 10(1970), pp. 20—30. Kisch, A. 1., et. al. "A New Proxy Measure for Health Status." Health Services Research, Vol. 14 (Fall 1969), p. 223. Moriyama, Iwao M. "Problems in the Measurement of Health Status." Indicators of Social Change. Edited by Eleanor Bernet Sheldon and Iwao M. Moriyama. Noble, John H. Jr. "Designing Information Systems for Comprehensive Health Planning." I_nguir_'y, (December 1970), p. 314. Olson, M. , Jr. "Social Indicators and Social Accounts." Socio- Economric Planning Sciences, Vol. 2 (1969). Reeves, Philip N. , "Data for Health Planning." American Journal of Public Health, Vol. 62, No. 6 (1972), pp. 87—14-76. Sheldon, E. B., and Moore, W. E., ed. Indicators of Social Change. New York: Russell Sage Foundation, 1968. F U.S. Department of Health, Education and Welfare. Toward a Social Report, Washington, D. C.: Government Printing Office, 1969. U.S. Department of Health, Education and Welfare. Public Health Service. Origin, ProgramL and ggeration of the U.S. National Health Survey. Vital and Health Statistics, Series 1, No. 1. Washington, D. C.: Government Printing Office, 1963. LS. Department of Health, Education, and Welfare. Public Health Service. Health Survey Procedure: Concepts ,_Questionnaire Deve10pment ,_and Definitions in the Health Interview Survgy. Vital and Health Statistics, Series 1, No. 2. Washington, D. C.: Government Printing Office, 1961!. .S. Department of Health, Education, and Welfare. National Center for Health Statistics. An Index of Health: Mathematical Problems. Vital and Health Statistics, Series 2, No. 5, Washington, D. C.: Government Printing Office, 1965. .S. Department of Health, Education, and Welfare. Public Health Service. Conceptual Problems in Developing an Index of Health. Vital and Health Statistics, Series 2, No. 17. Washington, D. C.: Government Printing Office, 1966. 187 U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Disability Components for an Index of Health. Vital and Health Statistics, Series 2, No. 112. Washington, D. C.: Government Printing Office, 1971. U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Quality Control in a National Health Examination Survey. Vital and Health Statistics, Series 2, No. 1414. Washington, D. C.: Government Printing Office, 1972. U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Re ort Health Events in Household Interviews: Effects of Reinforcement , Question length, and Reinterviews. Vital and Health Statistics, Series 2, No. 45, Washington, D. C.: Government Printing Office, 1972 . U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Acute Condi- tions: Incidence and Associated Disability, United States- Jpgy 1968-June 1969. Vital and Health Statistics, Series 10, No. 69. Washington, D. C.: Government Printing Office, 1972. U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Age Patterns _ig Medical Care A Illness, and Disability, United States , 1968— 1969. Vital and Health Statistics, Series 10, No. 70. wash- ington, D. C.: Government Printing Office, 1972. U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Time Lost from Work Among the Currently . olpyed Population J United States 1968. Vital and Health Statistics, Series 10, No. 71. Wash- ington, D. C.: Government Printing Office, 1972. ~ U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Current Estimates from the Health Interview Survey, United States-1970. Vital and Health Statistics, Series 10, No. 72, Washington, D. C.: Government Printing Office, 1972. U.S. Department of Health, Education, and Welfare. National Center for Health Statistics. Annotated Bibliography on Vital and Health Statistics. Washington, D. C.: Government Printing Office, 1970. World Health Organization. "Constitution of the World Health Organi- zation, Annex I." The First Ten Years of the World Health Organization, Geneva: WHO, 1958. |||||||l||IIIlllIIII-----::——---———rre 188 Wylie, Charles M. "The Definition and Measurement of Health and Disease." Public Health Reports, V01. 85 (February 1970) pp. 100-10“. E) Sampling, Interviewing, Analysis Cochran, William G. , and Cox, Gertrude M. Experimental: Designs. 2nd Ed. New York: John Wiley 8: Sons, Inc. , 1957. Federer, W. T. Experimental Design. New York: The MacMillan Company, 1955. Fisher, R. A. The Desggn 19; Experiments. 11th Ed. Edinburgh: Oliver and Boyd, Ltd.~, 19147. Kish, Leslie. Survey Samplipg. New York: John Wiley and Sons, Inc., 1965. Lansing, John B., and Morgan, James N. Economic Survey Methods. Ann Arbor, Mich.: Institute for Social Research, University of Michigan, 1971. Loewenstein, Regina. Two Approaches to Health Interview Surveys. New York: School of Public Health and Administrative Medicifi, Columbia University, 1969. iichigan Department of Public Health. Center for Health Statistics. Michigan Health Survey: Reference and Procedures Manual. Lansing, Mich., 1970. ser, C. A., and Kalton, G. Survey Methods in Social Investigation. 2nd ed. New York: Basic Books, Inc., 1972. edecor, George W. , and Cochran, William G. Statistical Methods. 6th ed. Ames, Iowa: The Iowa State University Press, 1967. ate of Michigan. Office of Planning Coordination. Bureau of Policies and Programs. Social Reporting in Michigan: Problems and Issues. Lansing, Mich., 1970. :eel, Robert G. D. , and Torrie, James H. Principles and Procedures of Statistics—With Special Reference to the Biological Sciences. New York: McGraw-Hill Book Company, Inc. , 1960. tchman, E. A., Phillips, B. S., and Streib, G. E., "An Analysis of Validity of Health Questionnaires." Social Forces, 36 3. Dapartment of Commerce. Bureau of The Census. Census Use Study, Report No. 6, Family Health Survey. Washington, D. C.: Government Printing Office, 1969. IIIIIIIIIIIIIIIllllllllllI---—— 189 U.S. Department of Commerce. Bureau of the Census. Census Use Study, Report No. [L Health Information System. Washington, : Government Printing Office, 1969. U.S. Department of Commerce. Bureau of the Census. Census Use Study, Report No. 12 Health Information System-II. Washington, D. C .: Government Printing Office, 1971. U.S. Department of Health, Education, and Welfare. Public Health Service. Communityélock Survey: Technigues of Community Health Analysis. Atlanta, Ga.: Center for Disease Control, (n.d.). U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Effect of Some Experimental Interviewing Techniques on Reporting in the Health Interview Survey. Vital and Health Statistics, Series 2, No. 741. Washington, D. C.: Government Printing Office, 1971. U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Sample Design and Estimation Procedures for a National Health Examination ‘ Survey of Children. Vital and Health Statistics, Series 2, No. 143. Washington, D. C.: Government Printing Office, 1971. U.S. Department of Health, Education, and Welfare. Public Health Service. National Center for Health Statistics. Interviewing Methods in the Health Interview Survey. Vital and Health Statistics, Series 2, No. 1&8 Washington, D. C.: Government Printing Office , 1972. Wonnacott , Thomas H. , and Wonnacott, Ronald J. Introductory Statistics for Business and Economics . New York: John Wiley & Sons , Inc. , 1972. F) Concentration » Alger, D. J. , and Heins, A. J. "A Social Welfare View of the Measure- ment of Income Equality." Review of Income and Wealth, 13 (March 1967), pp. 12-25. Alker, Hayward R. , and Russell, Bruce M. "On Measuring Inequality." Behavioral Science, Vol. 9, No. 3(July 1964), pp. 207-218. Benson, Richard A. "Gini Ratios: Some Considerations Affecting Their Interpretation." American Journal of Agricultural Economics, Vol. 52, No. 3 (August 19,70) pp. Mill-1447. Berry , Ralph E. Jr. "Product Heterogeneity and Hospital Cost Analysis." E98331, Vol. 7 (March 1970), pp. 67-75. 190 Bonnen, James T. "The Distribution of Benefits from Selected U.S. Farm Programs." Rural Poverty in the United States. A Report by the President '3 National Advisory Commission on Rural Poverty. Washington, D.C., May 1968, pp. 1461-505. Bowman, Mary Jean. "A Graphical Analysis of Personal Income Distri- bution in the United States." American Economic Review, 35 (September l9u5), pp. 607-628. Ginzberg, E. , and Rogatz, P. Plannixg for Better Hospital Care. New York: King's Crown, 1961. Grossack, I. M. "Toward an Integration of Static and Dynamic Measures of Industry Concentration." Review of Economics and Statistics (August 1965), pp. 301-308. Gurfield, R. M. , Quantitative Measures for Outpatient Care. Paper presented at annual meeting. Ins Angeles: Institute for Management Science, 1970. Kaluzny, Arnold D., et. a1. "Scalability of Health Services: An Empirical Test." Health Services Research (Fall 1971), pp. 214-223. Magruder, C. Donaldson, and London, C. David. "Time Study of Doctors and Nurses at Two Swedish Health Care Centers: Swedish Health Center Doctors and Nurses." Medical Care, Vol. 9, No. 6 (November-December 1971) , pp. 1157-7468. Tiller, Herman P. Trends in the Income of Families and Persggs in the United States: 19E7-19fl). Bureau of the Census Tech. Paper 8, ‘— Washington, D. C.: Government Printing Office, 1963. >rgan, James. "The Anatomy of Income Distribution." Review of Economics and Statistic_S_, 1414 (August 1962), pp. 270-283. :e, Dorothy P. , and Cooper, Barbara S. National Health Expenditures, 1929-1970. Social Security Bulletin, 1971. [1, Henri. "The Measurement of Income Inequality." Economics and Information Theory, Chapter 14. Amsterdam: North-Holland Publishing Co. , 1967. G) Michigan and Lake County Data Founty Planning Commission. Lake County Comprehensive Area-wide Plan for Water & Sewer Service. Grand Rapids, Mich.: Williams & Works, 1971. a: County: Economic Trends L Sugstions for the Future. Reports >f‘ class projects offisowce Deve10pment 816. Michigan State hiversity, Spring Term, 1965, (Mimeographed.) ’r—T m8" l9l Mas on County Overall Economic Development Committee . Overall Economic Deve10pment Progem. Grand Rapids, Mich.: Williams & Works , 196 6 Michigan Department of Public Health. Michigan Center for Health Statistics. Preliminary Report of Population Description Environmental Characteristics Health Complaints and Medical Care Reported for the Area of the Counties of lake, Newavgp, Mason, Manistee, and Oceana. lensing, Mich., 1968. Michigan Department of Public Health. County Health Statistics Profiles: State Planning and Development; Region 8 and 9. Lansing, Mich.: Center for Health Statistics, 1971. Michigan Department of Public Health. Western MicEg-Q Comprehensive Health Services Project, Inc.: Application 1971-72. Lansing, Mich. , I971. Nelson, Bettie L. Profile of Lake County and the Five-Cap Area. Lansing, Mich.: Michigan Department of Public Health, 1972. (Mimeographed.) Newaygo County Community Development. Committee Report, 1967. (Mimeographed.) Office of Economic Opportunity. Office for Health Affairs. Site Visit Appraisal of Lake County Comprehensive Health Center. Washington, D. C., 1969. >erkinson, Leon B. Heplth Service Differentials in Michigan. Agricul- tural Economics Report No. 213. Department of Agricultural Economics, Michigan State University, 1972. :ate of Michigan. Land Use Planning Section. A Social and Economic Study of the Negro Problem in Lake Count , Michigan, by John Malcus Ellison. Lansing, Mich.: Land Use Planning Section, Resettlement Administration, Region II, 1936. ~.te of Michigan. Executive Office of the Governor. Publicly Funded Family Planning Programs in Michigan: An Assessment of Effec- tiveness. Lansing, Mich. , 1972. evoogt, W. E. Rural Poverty in Michigan. Rural Manpower Center, Report No. 21. East Lansing, Mich.: Michigan State University, 1970. H) Rural Development and Health ma State University. Research Application in Rural Economic ngvelopment and Planning. Research Report P-665. Agricultural Experiment Station, Oklahoma State University, 1972. 192 The President's National Advisory Commission on Rural Poverty. A Report by the Commission. Rural Poverty in the United States. Washington, D. C.: Government Printing Office, 1968. Tweeten, Luther G. Rural Poverty: Incidence, Causes and Cures. Still- water, Okla.: Dept . of Agricultural Economics, Oklahoma State University, 1968. U.S. Congress. Senate. Committee on Government Operations. m Economic and Social Conditions of Rural America in the 19705. 92nd Congress, lst Sess., Part l,Washington, D. C.: Government Printing Office, 1971. U.S. Department of Agriculture. Economic Research Service. Rurality, Poverty, and Health: Medical Pr0blems in Rural Areas . Ag'icul- tural Economic Report No. 172. Washington, D. C.: Government Printing Office, 1970. U.S. Department of Agriculture. Economic Research Service. Availa- bility and Use of Health Services. . .Rural-Urban Common. Agricultural Economic Report No. 139. Washington, D. C.: Government Printing Office, 1968. APPENDICES «..--.. .. r”... .. . ....“ APPENDIX A .--~-~ - - ‘r-n'rw'". HF, ~. - ‘ v . .rA. ._ _ .. . APPENDIX A HEALTH SERVICES UTILIZATION AND MEASURES OF CONCENTRATION An increasing amount of literature has appeared during the last decade pointing to the fact that different popula- tion groups enjoy different standards of health. The health Status of a population has been usually measured in "nega- tive" terms of death and morbidity. Only recently efforts have been extended in measuring day—to—day activities which can be performed by the individual or by a group of indivi- duals—~in other words, a more "positive" measure of health status. Since all these outcome measures of the "health produc— tion process" impose the numerous conceptual, methodological and empirical difficulties listed in other parts of this Study, policy makers, planners and administrators most often prefer to support their arguments for or against certain programs and projects with the "hard" facts of health ser- Vices provided, i.e. the "inputs" into the health production process. Usually one finds this input-information displayed as average ratios, e.g. X thousand services of kind A per Y thousand recipients. Almost no attention, however, is given to the concentration of services within the group of reCipients. A literature search for the application of con- centration measures in the health services field produced 193 —i— 194 only a few efforts in that direction (Gurfieldl, Ginzberg and Rogatz2). This portion of the study should develop the rationale for concentration measures of health services and then per- form such an analysis using actual data. 1. Why Measures of Concentration or Distribution for Health Services Utilization? a) Program and Project Evaluation The nature of the health production process with its long investment periods and its great element of risk and uncertainty makes it hard for an administrator to evaluate his projects on grounds of outcomes or "outputs" of the I system. For judgment based on outcomes he needs either a large population or a long time series of observations and preferably both. On the project level he usually lacks both. He is left with observing his inputs (or intermediary outputs) and has to infer from their behavior on the success of his project. The more relevant and meaningful those "input- measures" the more appropriate will his decision be. Most of the available evaluation studies stratify the research sample carefully according to sex-race-age charac- teristics and compare average utilization rates with out— comes. No systematic treatment of concentration of services is available in the literature. This paper suggests that concentration measures are useful additions to the set of statistics presently used in IIIIIIIIIIIIIIIIIIIIIIlIII-""::_________-___— __ m“*_"______——_—“WI 195 evaluating the contribution of various health services inputs into the health production process. The ultimate goal, of course, would be to relate differences in con— centration measures to differences in health services out— comes. This, however, is a long term and elaborate under— taking and is outside the scope of this study. b) Project Administration Project evaluations are usually of a longer term nature. Yet, a review of reports on OEO and HEW health facilities indicates that many projects lack relevant infor— mation which could guide the local administrators in the short run. It is argued that also here concentration measures are an improvement over the reporting of plain averages. The administrator's problem is sometimes rooted in a phenom- enon which the theoretical literature calls "moral hazard."3 0) Mpral Hazard The health services planning and evaluation literature is full with discussions of overuse, possible overuse and abuse of services in the absence of a direct payment mechanism. Studies which provide empirical evidence of moral hazard report mainly average figures but fail to pin down the characteristics of overusers and do not show how this overuse phenomenon is distributed throughout specific populations. d) lpternational Comparisons Almost every study dealing with the United States "health crisis" contains a section contrasting American data *7— . 196 with those from Sweden or the United Kingdom, etc. With alarm it is then usually pointed out that in terms of mortality rates (infant and others) the U.S. ranks higher than other nations which conventionally are considered "less developed" or at least poorer. Most studies proceed then to contrast this face with the figures on per capita expendi— tures on health services. Since, with this ranking, the U.S. ranks on the top it is concluded that the U.S. health services system (or "nonsystem") is suffering from gross inefficiencies. It should here not be argued that inefficiency does not exist but rather that part of the inefficiency is attributed to the wrong causes and is being researched and consequently attacked with misguided emphasis. Besides suffering from different standards for data and collection systems international comparisons of expendi— tures are subject to the problem of purchasing power dif— ferentials. The difference in purchasing power is extremely crucial in highly labor intensive production such as the production of health in its present organization. The pOlicy recommendations given to remedy this situation are Correct: increase the productivity of the highly trained medical manpower by furnishing them with better organization, technology and assistance. But, will this close the health status gap observed in international statistical comparisons? Probably it will not, unless the distributional impact of health services IIIIIIIIIIIIIIIIIllll----::T_________'I__”____________—_IHI 197 utilization is taken into consideration. A rather casually observed phenomenon impresses the student of international comparisons of health care. Countries with otherwise similar characteristics which rank high in quality of income distribution also rank high in terms of health status indices. Although this is not a universal rule, it points toward some relationship between income distribution and national health status indices. There are no studies known to the author which pystemati— cally investigate this phenomenon. 2. Concentration Measures 1 Several publications have appeared utilizing and de— veloping concentration measures. Some deal with income dis- tribution such as those by Bonnenu, Bowman5, Miller6 and 7 Morgan , some deal with problems of industry concentration as summarized by Grossack8. An exploration of various kinds 9 of concentration measures is found in Alker and Russett , in Aigner and Heinslo, and in Theilll. Although there are several measures of concentration available, only two re— lated measures are discussed in this study. They are Lorenz curves and Gini ratios. a) Lprenz Curves One obtains Lorenz Curves by plotting the cumulative percentage of recipients against the cumulative percentage of receipts (Figure Al). In the case of completely equal distribution, the resulting Lorenz Curve coincides with the diagonal. The less equal the distribution the further the ]98 T % cummulated percentage of receipts 3 -- thO y’//”/,/ Lorenz Curve .50 A . ~—— cummulated percentage 0 50 100 % of recipients Figure Al. Lorenz Curve Figure A2. Crossing Lorenz Curves 199 Lorenz Curve moves away from the diagonal. In the extreme case where one recipient gets the total amount to be distributed the Lorenz Curve coincides with the coordinates. b) Gini Ratios To get a numerical measure of concentration, Gini suggested the ratio (G) of the area between the Lorenz Curve and the diagonal divided by the area between the coordinates and the diagonal, i.e. G =—A%B_ . Completely equal distribution results in a Gini ratio of 0, complete concentration yields a coefficient of 1, thus defining the upper and lower bound of possible concentration ratios. 0) Crossing Lorenz Curves In establishing Gini ratios one is sometimes con- fronted with the problem of crossing Lorenz curves (Figure A2). Both Lorenz curve (LI) and (LII) might include the same area (A), thus resulting in identical Gini ratios although obviously describing different situations. Thus far no satisfying method could be developed to ameliorate this problem. In fact, most researchers hOpe that the phenomenon will not occur and if it occurs assume that it does not impose a serious problem. This paper is pp exception to this "rule." w . _ m N.“ ‘4 -'~. ~« -'4 .7.. _ _ 200 3. Calculations of Gini Ratios To introduce the reader to the procedure it might be helpful to present the steps which are undertaken in in- come distribution studies. First, the range of income has to be divided in rele- vant and manageable strata and grouped at a decreasing or increasing order. The number of strata to be selected is of considerable significance since a small number of strata will result in large line segments and will thus result in a smaller value for area A, or in other words, will lead to an underestimation of the Gini coefficient. Several approaches to deal with this situation are avail— able in the literature (Bensonl2). For computational purposes the Gini coefficients are calculated in the following way.13 Take the area of a square (A + B) x 2 to be equal to l and that of the triangle (A + B) to be equal to l/2 (Figure A3). The Gini ratio Can now be interpreted as G =—i§%§:—§ = l - 2 (B)- The area (B) is now calculated by assuming that the Lorenz Curve can be approximated by straight lines and by calcula- ting the area of the resulting quadrangular segments. The area of each segment (i) is denoted by F1’ Y + Y. +1 Where (1) _ _ l 1 ) F1 ‘ (Xi+l Xi)( 2 AI‘ea B is the summation over all 1, i.e. (2) Y. + Yin) B = z (x1+1 — X1)( 1 2 2()l cummulated percentage of recipts 1+2 iil cummulated percentage X0 X1 X1+1 X1+2 Figure A3. Computation of Gini Coefficient 4 Figure AA. Concentration of Ph ysical Unit ’ of recipients cummulated percentage of physical units received (e.g. health services) /’/”///,Inrenz Curve cummulated percentage r, of recipients 8 Received. ii, 202 This leads to the following numerical expression of the Gini coefficient. G = l — 2 Z (X1+1 - X1 0]." (3) G = l " 2 ”(1+1 " Xi) (Y1 + Yi+l) A. Concentration of Health Services Utilization a) In the Presence of Market Prices If the researcher knows the market prices for services consumed it is a rather simple process to extend the income distribution analysis to that of health services utilization. This procedure assumes that the costs of services reflect the amount and quality of the service rendered. It is also assumed that market prices are known to the investigator. b) In the Absence of Market Prices Much of present day research is concerned with the provision and evaluation of social infrastructure which is Organized outside the constraints of the market mechanism with its guiding parameters——prices and costs. Health ser— vices are but one example ofrnflflfil="interference" with the market. To apply Lorenz Curves to such a situation one has to rely mainly on the quantities of services provided. A "strong" assumption, however, has to be introduced, in pursuing this strategy: services rendered have to be similar in quality and should cost the provider similar amounts of 203 resources (time and other resources). In the case of health services, only a sufficient large sample will warrant these assumptions. Since this particular study is concerned with the evaluation of health services delivery in a largely free-service setting, the following empirical investigation will apply the "non- market" approach by using data from the Western Michigan Comprehensive Health Services Project. Lorenz curves and Gini ratios will be established as depicted in Figure AA. This graph differs from the usual (income) distri- bution framework in that the physical units of the trans- ferred commodities are recorded and not their monetary value. 5. Empirical Analysis a) Hypotheses To Be Tested Three main hypotheses were the starting point of this part of the study: Hypothesis (1): There are differences in concentra- tion rates of health services utili— zation for different user groups and different service categories. Hypothesis (2): Differences in concentration rates are related to differences in health status and other socioeconomic out- comes of population groups. Hypothesis (3): Concentration rates can be used for evaluation and planning purposes. 204 Only Hypothesis (1) can be tested at this stage of the research, although it is anticipated that the informa- tion necessary to examine the validity of the other two hypotheses can be developed at subsequent stages. (Especial— ly a better understanding of outcomes is needed.) b) Data Preparation The services provided by the Health Center were grouped into 44 different services categories using partly the Cen— ter's categories and partly combining Center categories into blocks of services, in order to reduce the amount of ser— vices to a manageable size. A computer routine was developed to extract a ten per— cent sample from the files of approximately 7,000 regis— trants. For each of the 699 sample registrants a separate printout was obtained listing their case number, sex, age, county of residence plus their utilization of the services categories. The program was written in such a way as to display a person's receipt of the AA services over the period of one quarter of a year. The services were then aggregated and appeared as totals on every data sheet. The same type 0f information was produced for three consecutive quarters. 0) Stratification of Population To test Hypothesis (1) ("There are differences in con— centration rates of health services utilization for dif— ferent user_groups and different services categories.“) the pOpulation was categorized into different strata: —i— 205 Female-Male. Available health statistics give clear evidence that there are significant sex-specific differences in health services utilization. This study should add to this existing knowledge information on con- centration of utilization. 'BlackAWhite. Because racial differences are often associated with differences in health care availability it is advantageous to stratify the population along racial characteristics. There were only ten "Other" (mainly American Indian and Mexican Americans) out of the 699 sample registrants. Although this is still 1.4 percent of the population, it was decided to put them together with the group "Black." ‘Age Groups. The age groups of O-A; S-lA; 15-44; 45969; 65+ reflect important stages in a person's life in relation to the health services system. (School entry, onset of fertility and entry of labor force, menopause, retirement). These or very similar breakdowns are used in most health related studies. ReSidenqy. Since distance is one of the crucial factors in health services planning it is of interest to examine variations in concentration rates in terms of patient-to—provider distance. Originally it was planned to account_for the presence of satellite facilities by basing this analysis on township data. Unfortunately, the small sample size and inaccurate reporting of residency precluded this more refined analysis. This study compares, i 206 therefore, only Lake County (where the main clinic is located) with the other three counties served by the project. The other three counties, Mason, Manistee, Newaygo are called "Satellite Counties." d) Results ’Equation (3) formed the basis for a computer program which calculated the concentration coefficients for the various groups of recipients and services. An attempt at developing concentration rates for the rather high amount (44) of services being investigated conflicted with the comparatively small sample size of 699 recipients. Only in a few cases curves could be obtained, in many cases only points or no estimates at all could be calculated. Two ways were open to circumvent this statistical problem of "limited" degrees of freedom. One was to group the services into larger categories-~compounding the problem of incompatibility (and non-additivity) of services, the second was to examine the data at a simple level of cross stratification. A combination of both alternatives was taken here. On one hand all AA services were grouped into three basic classes: Medical, Dental, Family Services; on the other hand only a simple stratification of clients was employed. The results of this analysis are displayed in Table A1. Each cell contains two numbers: the integer represents the average amount of services utilized during the given time period. The number in parentheses indicates the degree of 2?()'7 p.303 8804 page; .30 mm mops E “seem no access «magmas one Mwmwmmwmmmumwmwmcmaunwaemm use .eoaaoa use» cm>fim use Acmemm.v . . moss: oz» mcampeoo demo comm Am "ouoz nemeee v Awmemm.v lemmee.v A . m m m a :mwmm o Anemoo.o Aommem.o Amemm:.v AmmmomJ AmszwJ A .3 a 3 03.25 .3909 3a coo . . . H ma mm o Ameomz v Aaommm v lemmem.v Asoaee.v Amammq.v Ammawm.v Ammmm ma 3 m +wm “ma Ha me o Asseme.c Amemmm.o Ammeem.c Amwemm.v leeaem.v Ammaom.c e ma ea e e menme ANH Aaoomm.v Amemme.v Amhmwm.v . e e m Awemee v Aeeeem.c Aeeaee.c Ameeem.c lomeee.c A a a ma OH 3 : mausa AAA 0 so” V Aememe.v hommom.v Asmmmm.v Amoeme.v Aeneae.v Aeeemm.v neaeem.v m m m a a m cane Aoa Amwmmm V Awmem~.v Asmmmm.o Amaoae.v Aammse.o Amomom.v “Hemmm.v Amqmee.v m a m m m m m mno Am flammaw.v Ammmam.v Ammmmm.v Asamme.v Awmemo.v Azammm.v Azszom.v Ammm~=.v m m 2 NH NH 3 m on; 3 Asshmo.v Hemeoo.v Aoomom.v Asammz.v Aomsmm.v Awsomm.v Aommmm.v Aazaom.v 5.. ma 3 m ma m m z xomam C Amamem.v Aaaemo.v Asmmsm.v Aommw:.v Amcmmm.v Ammma .v sm.v Aowmmz.v m m m m ma mm Acme: m madame Ae Azsmw:.v Ammeam.v Amoesm.v Afimwo:.v Aesmmo.v Aommso.v Aomonm.v Aoewpz.v a a m a an ea 3 m «Hesse mean: Am Amhmmm.v Ammmam.v Ameamm.v Amamam.v Amowom.v Asmmom.v Azfiamm.v Aqwoom.v em on a a ma OH m a panama nomHm as ea mm. m . Hmmmm.v cpmm=.v Ammmmm.v Aommao.v Aazmem.v Ammms:.v A m e c lee we I A m l e on a e e tam: Am Ameomm.v lessee.c leeeMe.o Ammmue.v nememm.v Ammewo.v ammamm.o Ammeme.o was: mean: Am w m page same new: lemmas new; arm? Emmi arm: as is s m mooa>uow moofi>now mooa>hom mooa>nom mmofisnmw mmoa>nwm 9M 8358 333% 3&8 83 355 .8 e eofismeeeweeoo see meeassmm cease: co eoaeeuaaap: ewesm>< B Sausage»: .8 SEE Ad. 0&3 43:39... 208 concentration of the consumed service, where 1.000 would be completely concentrated and 0.000 would be perfectly equally distributed. An inspection of Table Al shows that the concentra— tion of services varies widely between the various cells. For instance, black males in Lake County receive medical services with a concentration ratio of .A9A98. Their white counterparts receive medical services with a lesser con- centration (.41758). Similarly, black males in Lake County receive dental services at a higher concentration (.44375) than white males (.33155). The significance of these data can be emphasized by 1 examining the average rates of the first four cells of the Lake County sample (medical and dental services for black and white males). Although both medical app dental services are consumed at higher concentration rates by the male black group, average rates of blacks are only larger for medical services while they are lower for dental services. Similar observations can be made for other cells of Table Al. 6. Conclusions The analysis of the available data indicates that in the case of the Western Michigan Comprehensive Health Center services are differently distributed among the selected groups. Whether this is "normal" or not cannot be deter- mined without relating concentration to measures of health status outcomes and physiological and mental states of users. The data in Table A1 prove that average utilization data do 209 not describe utilization completely and are insufficient indicators in evaluating the Western Michigan Comprehen- sive Health Center Project. The main objective of this presentation was not to prove or disprove a particular hypothesis but to explore possible uses of concentration measures. This exploration was done in light of the fact that although both OEO and HEW have created programs and projects to improve the dis- tribution of health services, no formal research has been proposed to explain the mathematical relationship between changes in the health services distribution and changes in the established health status indices of various popula- tions. It seems, however, to be vital to understand the "mechanics" of an indicator which is being used not only as an argument for political discussions but also as an actual measuring stick for success or failure of national or local policies and programs. This section was intended to explore one type of concentration measure, i.e. Gini Ratios, in relation to health services utilization. Experimenting with other measures is recommended as an agenda for further research in this area. l. 10. ll. l2. l3. 210 FOOTNOTES R. M. Gurfield, Quantitative Measures for Ouppatient Care, Paper presented at Annual Meeting (Los Angeles: Institute for Management Science, 1970). E. Ginzberg and P. Rogatz, Planning for Better Hospital Care (New York: King's Crown, 1961). Mark V. Pauly, Medical Care at Public Expense (New York: Praeger Publishers, I97I), p. A2. James T. Bonnen, "The Distribution of Benefits from Selected U. S. Farm Programs," Rural Poverty in the United States, A Report by the President's National Advisory Commission on Rural Poverty (Washington, D. C., May, 1968), pp. A6l—505. Mary Jean Bowman, "A Graphical Analysis of Personal Income Distribution in the United States," American Economic Review, 35 (September, l9A5), pp. 607-28. Herman P. Miller, Trends in the Income of Families and Persons in the United States: l9A7—l960, Bureau of the Census Tech. Paper 8 (Washington D.C.: Govern— ment Printing Office, 1963). James Morgan, "The Anatomy of Income Distribution," Review of Economics and Statistics, AA (August, 1962), pp. 270-283. I. M. Grossack, "Toward an Integration of Static and Dynamic Measures of Industry Concentration," Review of Economics and Statistics, (August, 1965), pp. 301- 308. Hayward R. Alker and Bruce M. Russett, "On Measuring Inequality," Behavioral Science, Vol. 9, No. 3 (July 196A), pp. 207—18? D. J. Aigner and A. J. Heins, ”A Social Welfare View of the Measurement of Income Equality," Review of Income and Wealth, 13 (March, 1967), pp. 12-25. Henri Theil, "The Measurement of Income Inequality," Economics and Information Theory, Chapter A (Amsterdam: North Holland Publishing Co., 1967). Richard A. Benson, "Gini Ratios: Some Considerations Affecting Their Interpretation," American Journal of Agricultural Economics, Vol. 52, No. 3 (August, 1970) pp. AAA-A7. Bonnen, op. cit., p. A62. APPENDIX B 211 a Appendix Table 151. A Profile of the Four Counties Served by the Health Project and the Corresponding State Averages. ) son Age: All ages 22.612 8 777 560 65 and over n Sex: Male 11 157 3 882 868 Female Race: White 22 383 Nonwhite Urban places ( With 2,500 or more population) Population per household Median school years completed 8.7 General economy: Median income of families in dollars $3 158 Median value of housing units Percent of families with $3,000 or les Percent of families with 10,000 or mot Per capita personal income Public Assistance Payments(monthly average) Old age assistance Aid to dependent children Aid to blind Aid to the disabled Ccncral assistance _ . Employed persons b major industry grOUp: Total employczl’ 2 726 864 Agriculture, forestry and fisheries Mining Construction Manufacturing Utilities Trade Finance and real estate Business repair and personal services Other services Public administration Industry not reported Inventory of health and medical resources a. Manpower: M.D. * D.0. ** Dentists b. Hospitals and long-term facilities Licensed hospitals Licensed nursing homes Homes for aged County med. care fac. Livebirth rate per 1,000 population Infant death rates per 1000 livebirths: under 1 year under 1 day under 7 days under 28 days Perinatal death rate per 1000 total births Illegitimate ratio per 1000 livebirths Persons divorced pct 1000 p0pulation Crude death rate per 1000 population * Source: Michigan State Board of Reg. in Medicine ** Source: "Education for Health Care in Michigan" 1970 a) Adapted from: Michigan Department of Public Health, Qggpflptive Summary of Western Michigan Comprehensive Health Services Proiecf, Inc., Baldwin, Michigan (Lansing, Mlchlqan: MDPH, I970). (Mimeographed.) 221.2 Appendix Table B2. Demographic and Econcmic Profile: POPULATION AND AREA 1970 1960 Number £22233: Total Population 5,661 8 Percent of State .06 5’?37 323 County Density/square mile 9.9 9.3 Land Area in square miles 571 Net Migration 1960 to 1970 (b) 416 AGE DISTRIBUTION - 1970 Male Female Under 18 912 91“ 18-1114 650 673 “5—65 621 723 65 and over 595 573 Total 2 , 778 2 , 883 INCGWE 1969 l9§9 Total Personal Incane (thousands of dollars) $ll'900 $5’600 Percent of State .03 .03 Per Capita Income County (b) $ 2,120 $1,099 DiCOME BY MAJOR SOURCES ousan o o ars Total Personal Income Total Wage & Salary Disbursements + Other Labor Income Proprietor-5' Incane Property Income Transfer Payments Less Personal Contributibutions for Social Insurance Total Eamings Farm ‘ Total Non-Farm EaJnings Government Earnings Total Federal State and local Private Non-Farm Earnings Manufacturing Mining Contract Construction Trans. Comm. 8: Public Utilities Wholesale & Retail Trade Finance, Insurance & Real Estate Services Other Lake County a) Percent EEEEEE 6.1 Percent of Total Male 32.8 23. 22.14 21.4 100.0 Female 3107 23.3 25.1 19.9 100.0 Percent Ci_la_ng§_ 112.5 0.0 102.1 1967 $8,900 3,800 1,800 1,200 2,000 5,600 61: 5,706 1,901: 1.437 3,802 656 1417 247 1,138 190 1,077 73 a) Adapted from: Michigan Department of Comnerce, Economic Profile, (Michigan Department 0 Comerce, Office of Economic Expansion, Research Division, November 197 . (Mimeographed) . ~ c-___~~_ j..___-'_-._.~:-.---.wv-.--.r-5‘r 213 Appendix Table 83. Demographic and Economic Profile: Montmorency County a) POPULATION AND AREA Number Percent .2_1 70 __1960 Class ”its: Total Population Percent of State 5'237 lung: 823 18.6 County Density/square mile 9'5 80 Land Area in square miles '55 ° New Migration 1960 to 1970 (b) 685 AGE DISTRIEJ'I'ION - 1970 Percent of Total Male Female Male «Qua ] e Under 18 925 864 35.u 32.8 18—“8 623 658 23.8 25.0 45-65 623 725 23.8 27.5 65 & over “A“ 385 17.0 1N.7 TOTAL 2,615 2,632 100.0 100.0 INCOME l 6 19 9 Percent Change Total Personal Income (thousands of dollars) $11,900 $5,800 120.H Percent of State .03 . 0.0 Per Capita Income County (0) 2,29“ 1,234 85.9 INCOME BY MAJOR SOURCES (thousands of dollarsj 1967 Total Personal Income Total Wage and Salary Disbursements 9:300 + Other Labor Income 11,800 Proprietors' Inccme 1,800 Property Income 1,100 Transfer Payments lass Personal Contributions . for Social Insurance 1,000 Total Earnings 6,600 Farm ‘ - 11 Total Non-Farm Earnings 6,630 Government Earnings 1 ,791 Total Federal 282 State and local 1,509 Private Non-Farm Earnings “.839 Manufacturing 1,1411 Mining - Contract Construction uuo Trans. Comm. & Public Utilities 108 Wholesale & Retail Trade 1,676 Finance, Insurance, and Real Estate 231 Services 891i Other 79 a) Adapted from: Michigan Department of Commerce, Economic Profile (Michigan Department of Commerce, Office Of Economic Expansion, Research Division, November 1971). (mmographed) . v.1: .-.‘7-.!_".~' .‘ _ 2:141 Appendix Table B4. Stmmary of Budget of the Western Michigan Comprehensive Health Services Prodect fer Years B - E * B C D E Personnel Costs 948,957 1,577,335 2,708,950 2,384,384 Salaries and wages 78,207 168,135 1,849,000 1,707,291 Fringe Benefits 16,650 288,300 277,350 256,093 Consultants and Contract Services 592,600 421,000 Non-personnel costs 685,864 503.530 687,455 Travel 76,700 110,000 110,000 113,086 Space Costs and Rentals 124,197 61,480 92,200 Consumable Supplies 28,183 208,000 294,500 334,200 Equipment 456,784 60,650 87,755 75,778 Other Costs 380,432 63,500 103,000 247,520 Total Costs 2,015,253 2,080,965 3,396,405 3,154,968 i * Year B - January 1, 1969 - December 31, 1969 1 Year c - NOvember 1, 1969 - October 31, 1970 ‘ Year D - November 1, 1970 - October 31, 1971 Year E - November 1, 1971 - October 31, 1972 Source: Bettie L. Nelson, Profile of Lake County and the Five-Cap Area (Lansing: Michigan Departmen o c a , , p. . grap ed). 2:155 Appendix Table BS. Comparison of Socio-Economic Indicators Between Lake County and Montmorency County Indicator Lake Montmorency Poverty Index b) 129.5 115.1 Poverty rank among 83 83 81 Michigan Counties Percent fUnctional illiterate 13.7 4.4 Median school years 8.6 9.8 Average annual unemployment rate: 1965 7.6 6.1 7 1966 9.7 5.9 1967 9.8 10.6 1968 13.6 8.8 Population/Physician Ratio 1968 1125 4200 (including M.D.'s and Osteopaths) a) Adapted from: w. E. Vredevoogd, Rural Povertg in Mic%gan, Report No. 21, Rural Manpower n r, c a e versity, November 1970 (East Lansing: Rural Manpower Center, 1970), pp. 15-65. b) Prepared from 1960 census data. The index consists of the sum of feur percentages, % earning $3000 or less, % unemployed, % functionally illiterate, % houses in bad repair. Highest possible score is 4 x 100% - 00. 216 Appendix Table B6. Eligibility Criteria for 0E0 - a), Foodstamps - b) and Commodity c) Programs. O E O Foodstamps d) Commodity 97 Annual Monthly Income Income Number in Household Nonfam Firm 1 2 . 000 l . 700 210 210 2 2.600 2. 100 250 250 3 3 . 300 2 . 800 307 290 4 4 . 000 3. 400 373 330 5 4 . 700 4 .000 440 370 6 5. 300 4.500 507 410 7 5 . 900 5. 000 573 450 a) b) C) d) e) Source: Health Center Records, Baldwin, Michigan Source: Social Services Dapt. Records, Baldwin, Michigan Source: Social Services Dept. Records, Atlanta, Michigan The maximum allowable resources of all members of a household may not exceed $1500. Exce tion: For households of two or more persons with a member or members age 60 or over, the allowable maximum is $3000. The maximum allowable liquid assets may not exceed $1000 for one-member households and $1500 for households with 2 or more members. 217 Appendix Table B7. Profile of Sample Households. Camarism between Lake County and Montmorency County (Values of the Mmtmmency County 3 1e are ain brackets while percentages are in sea) Sectim 1) Number of sample households according to their race and their public assistance status BLACKS (Lake 0113) WHIIFS (Lake (nly) TOTAL SAMPLE Not on assistance (NA) 20 (243 o? Sum—totaI) 39 (465 of3un total) 0 um to 01 public assistance(PA) 17 (20% of Sun total) 8 (10% of Sun total) 25 (30$ of Sum total) [35 (421)] Total 37 (44% of Sun total) 47 (56% of Sun total) Sum total - 84 [84] Section 2) Average number of household members (Household questionnaire #a) BLACKS (Lake (nly) WHITES (lake only) Total Sample Not on assistance (NA) 2.77 3.2 3.0 [2.9] On public assistance(PA) 3.26 3.6 3.4 [3.5] Total 3.01 ~3.4 43.2 [L21 Section 3) length of residency of sample households (Household questionnaire #c) BLACKS (Lake Q'fly) M‘II'I‘ES (Lake duh!) Total Sample Lived 3 rs. 8- in county 29 (78% of all blacks) 36 (77% of all whites) 65 (77% of total sample) Lived 1- yrs. in countL 8 (22% of all blacks) 11 (22% of all whites 19 (23% of total sample) NA Clients PA Clients Total Sample Lived 5 yrs. & in county 49 (831 of all NA) %6l(‘61(l%901)‘]a11 NA) 65 (77% of total sample [68(801)] 90 1)] 2 Lived 1-4 yrs. in county 10 17% of all PA) 9(361101‘ all PA) 19 (23% of total sample [16901)] 15(10111 [11 (31%)] Section 4) Availability of sanitary facilities (Household questiomaire li-e) BLACKS (Lake Only) mum‘s (Lake Only) Total (Lake Only) Eggletgéjcflities 27 (73% of all blacks) 28 (601 of all whites) 55 (66: of total sample) 72 Improved facilities 8 (221 of all blacks) 5 (101 of all whites) 13 (15: of total sample) since 1968 Lack facilities completely 1968- 2351 of all ol_a_c_ksL 14 (301 of all whites) 16 (l s of total 5 is; NA Clients PA Clients Total Sample C lete facilities 43 (721 of all NA) 12 (48% of all PA) 55 (66% of total sample) 196 . 1971/72 [33 ( 7%)] [25 (72%)] [58 6 1)] Irnproved facilities 7 (12% of all NA) 6 (24% of all PA) 13 (15% of total sample) since 1968 [10 (21%)) [6 (17m [16 (19%)] . Lack facilities 9 (15% of all NA) 7 (281 of all PA) 16 (191 of total sample) ggpleted 1968-72 [6 (121)] La (11%)] [10 (12:01 Section 5) Change in heading of hares of sample households as perceived by the respondents (Household questiomaire fim) BLACKS (Lake (my) WHI'IES (Lake Only) meme Only) Heading armed 13 (35% or all blacks) ll (23% of all whites) 24 (29% of total sample) since 19 Heading stayed the 23 (62% of all blacks) 32 (68% of all whites) 55 (65% of total sample) same since 1968 A Heading worsened l (3% of all blacks) 4 (9% of all whites) 5 (6% of total sample) since 1968 NA Clients PA Clients Total Sample Heading improved 17 (29% of all NA) 7 (281 of all PA) 24 (29% of total sample) since 1968 [20 (41%)] [15 (43%)] [34 (42%)] Heading stayed the 38 (64% of all NA) 17 (68: of all PA) 55 (65: of total sample) 3% since 1968 [26 (53%)] [19 (54 1)] [45 (53%)] [leading worsened 4 (71 of all NA) 1(4% of all PA) 5 (65 of total sample) since 1968 [3 (5%)] [1 (3%)] [4 (5%)] fif 218 Sectim Appendix Table 7 (can't) b) Cinge in nutrition of ample households as perceived by the respondents (household questionaire Ilp) slide; (Lake Only) mamas (Lake Only) Total (lake Only) Family nutrition 18 (49% of all blacks) 24 (51% of all. whites) 42 (50% of total sample) improved since 1968 Family nutrition did 19 (511 of all blacks) 23 (491 of all whites) 42 (505 of total sample) not improve since 1968 NA Clients PA Cléents Total Sample Family nutrition 29 (49% of all NA) 13 (52% of all PA) 42 (50% of total sample) 1mproved since 1968 [l4 (29%)] [13 (37%)] [27 (321 Family nutritioo did 30 (51% of all NA) 12 (48% of all PA) 42 (505 of total sample) L013 improve since 1953 [35 (71%)] [22 (53%)] [57 (68%)] Section 7) mspondents who indicated that they and their families would visit the doctor more often if they had more income or if doctor services were more readily available. (Household questionaire Ilg) BLACKS (lake Olly) WHI'IES (Lake 0113) Total (Lake Olly) Wofgld see doctor more 21 (57% or all blacks) 24 (51: of all whites) 45 (545 of total sample) 0 en Would n__ot see the 16 (43% of all Macks) 23 (491 of all whites) 39 (46% of total sample) doctor?" more often NA Clients PA Clients jotal Simple Would see doctor 31 (535 of all NA) 14 (56% of all PA) 45 (54% of total sample) more often [22 (45%)] [15 (43%)] [37 (449] Would not see doctor 28 (47% of all NA) 11 (44% of all PA) 39 (461 of total sample) more often J21 (555)] E20 1573] [41 (5QH1 Section 8) Enrollment of sample households in comprehensive health center (obtained from FIG records, lake County Only) BUCKS (Lake Only) mums (Lake Olly) 'Dotal (Lake Only) Ehrolled 34 (921 of all blocks) 29 (621 of all whites) 63 (75% of total sample) Not enrolled 3 (8: of all blacks) 18 (38: of all whites) 21 (255 of total sample) NA Clients (Lake Only) PA Clients (lake Cth) Total (Lake Olly) Enrolled 46 (78% of all NA) 17 (68% of all PA) 63 (75% of Total sample) Not enrolled 13 (22% of all NA) 8 (32% of all PA) 21 (251 of total sample) Section 9) Change in real income of sample households as perceived by the respondents (Household questionnaire #v) BLACKS (lake Qfly) METFS (Lake 0115’) Total (Lake Olly) Renal incgge has improved 16 (43% of all blacks) 17 (36 of all whites) 33 (395 of total sample) s cc 19 Real incare has stayed 8 (22% of all blacks) 10 (21 of all whites) 18 (221 of total sample) the same since 1968 Real income has worsened 13 (35% of all blacks) 20 (43 of all whites) 33 (393 of total sample since 1968 NA Clients PA Clients Total Sample Real income has improved 2111(263 of all NA) 2(482 of all PA) 3 ( 91% of Total sample) since 1968 [1122”] [9261)] 2‘0 heal income has stayed 13 (22% of all NA) 5 (20% of all PA) 18 (221% of total sample) the same since 1968 [22 (45 1)] [12 (341)] [34 (4 0%)] Real income has worsened 256 (42% of all NA) 8(32%o of all PA) 33 (381 of total sample) M968 J16Q31U [14 (40 5)] [30 ( 36%)] Section 10. Average of highest grade level of school completed by head of sample households (Household questionnaire Ilw) BLACKS (Lake Olly) WIII'I‘ES (Lake Chly) Total Sample NA 7.9 9.0 9.4 8. PA 10.5 10.0 10. E92] Total 9.7 ‘95 9.9 [9.1J Note: a Black and white comparisons are for Lake County sample only since there are no black families in the Mmtmorency county sample. 219 Appendix Table Cl: Codes of Household Characteristics of Individuals (As Listed on Computer Tape of Member Questionnaires) Column Location on Computer Tape Column 145 White:l ' B1ack:2 Column 146 Residency in Residency in County 5 yrs+ County 1-11 yrs Lake Not on Publ. Ass. (NA) ' l 2 County On Publ. Ass. (PA) 3 A Montmor'b ency Not on Publ. Ass. (NA) 5 6 County On Publ. Ass. (PA) 7 8 Column 147 All Sanitary Partial Sanitary No Sanitary Facilities Facilities in Facilities in 1968 and 1968 and in in 1968 and in 1971/72 1971/72 in 1971/72 Heating up since 1968 l 3 5 Heating same or down 2 ll 6 Column 148-149 Income Income Income UR Same Down Family Nutrition Would See Doctor Up Since 1968 More Often ll l2 13 Would See Doctor Not More Often 21 22 23 Family Nutrition Would See Doctor Not Up More Often 31 32 33 Would See Doctor Not More Often Lu 1&2 I43 Column 50-51 Education of Head of Household (Grades Completed) Column 52-53 Enrollment in Health Center (Lake County population only): If "Not Enrolled":l If "Enrolled" or in Montmorency County: blank. —>—m——‘:‘“ ’ " APPENDIX C a) b) e) d) e) 220 HOUSEHOLD Before we start, I'd like to fird out scmething about who lives in your household. Let us start with the head of the household: For each household matter, complete first line on "MEIER" questionnaire. ' 1) Fill in # of household matter. 2) First name. 3) Relationship: "head," "spouse,“ “children." ‘6) Age, sex. Is there anyone else \(ho usually lives with you? How long have you and the mariners of your household been in this county? EWtMnSyears —.(d) larger than 1 year —5 'WBER" questionnaire [:1 Shorter than 1 year—fl "EXIT" interview List those members of your household who moved into this county (hiring the last 5 years. Column location on computer tape of Household questionnaire I-BL «41* litigation to m where did ———————-— When . ______ misehold 1) :.2) 3) GO TO "MEMBER" questionnaire. Finish "W110“ questions at the end of the intewiew. 221 1') List those of your family who are presently in a hospital, home for the aged or extended care facility: NAME: Age Sex Relation to head What facility is Since when has of household in? been there? 1) 2) F4” 3) 9 g) List those manbers of your household who have died since 1968: NAME Age Sex Relation to head When died Did live with you of household dwim'fie last year of l_ his ye? 1) Q YES E_'_]N0 2) Dims 1 NO 9 3) I] ns [2 mo h) The next questions are related to housing. 1) Did you have a working bath tub or shower during the last year? [I] YES Inc J) Did you haveaworm bath tuborshowerinl968? mus @m k) Did you have a. worldng flush toilet during the last year? ID YES .NO 1) ma you have a working flush toilet in 1968? [I] res .mo m) If you compare last year with 1968, did your heating [flimpmve since 1968? Esta}! the same? Eworsen? n) Now I would like to ask you a few questions related to mtrition ard health services. c) If you compare this year with 1968, do you as an individual £41 eat Elmore meat Ekame E Emore milk flame m Elmore vegetableslzkame m p) Has your family's nutrition improved since 1968? EYE .NO q) Would members of your household visit the doctor more frequently if you had more income or if doctor services were more readily available? [3m [gm r) Durirg the last year, how would you rate the availability and quality of health care in this area? Choose a position on the scale. 1971/72 1968 +3 good +3 6 ‘” . '7. +2 +2 5' 0 20 +1 +1 #0 3: -1 -1 3 .. 2‘ -2 -2 20 1J— 2.2 ° -3 poor -3 23 5) Where would you place availability and quality of health care in 1968? better than now? worse than now? How may levels up or down would you place it in 1968? t) If there was a change between 1968 and 1971/72, what caused it? LISP u) With the next question we would like to find out how your income situation has changed since 1968. v) Do you think that since 1968 your household's inccme situation has 2." [D improved? [In stayed the same? wworsened? 1' he ho h 1d? LISI" as w) What was the highest level of education completed by the head 0 t use 0 . 26 OBTAIN FTDM RECORDS: x) Black White 3 Black White 21 I Not on Assistance Not on Assistance On Public Assistance On Public Assistance y) Not Enrolled in blank Enrolled in Health Center Health Center ..z ...-.- ,-— ,. ..,. _-.» . - . _ , , ‘ 222 Column location mmutcrupe MEMBER gmqmtim— re First Nam. :gtgmegld: Age: Sex: 1) SCI—1:“: 's muzfishggfngfizmgr his/her age and sex, how would you 1971/12 1968 8 +3 wove average +3 6 7, +2 +2 5 9 _ :1 _ _ +1 9 ‘9 -1 -1 3 IO -2 -2 2 A -3 bola: average -3 I I 2) How would you mm: 's hulth for 1968? Better than now? Verse than now? lbwmnwlsvelsupar—da'mwuudywplsce__rwl968? 3) It” therewassclwuebetwem 1968m1971/72ndntcmssd1t? LIST: 11) Did___Mvesgsmrelplwsicalexmm-smediealahschlpsinoel9687mm Elmo 5) Did__mvccda1tolotncmpmmtmpoot12mntm [Illa Em 6) Hmmtimesdidfigotosdmtistdm'irgthsputlamnfln? times 7) Hmmtflmdid__sotosdootororhea1thmdurimthepest12Inmths? tum :1 8) Is _'s activity unltodlnammbccmcc or disability or health? mm [gm-.03) I1 9) Mutan___fidobecsmeordissbiutyormalth? LISI‘: 10) inlet are the realth omditiom that caused this limdtatim? LISI‘: , nouns ' mm, M 0 'Y Abouthovla-ghss bemthstm?flcwmsmm1ths1 Hamyears? 12) Does need help from urther persm gettitg aroma inside a- cutsids the nurse? mm Elm ‘if 11 v ..p 13) ACE: 0.17 years—”(30 18 am overqufl) lu) lntmotmlth is eblstowurkatallam-dthetnuss? Um [Elmo 20 15) Is limited in the kind or hone activities became or health? Elm Duo 16) Is limited in the noun ment at base activities because or W mm mm 17)mm~mnydqsofthepast365dwudidillressorin1mkesp mouthing gl hs/she M does eras-1d the house? d”, 11 18) Age 66+ —>(m'r); otherviss $2.19. #19) Do health common- keep nu- heir; emplqed or alt-”lowed? genes C] no 2‘. 20) Is limited as to counties? or mm of work became of health? mm D m 21) Is linitcdlnthcammtotumthc/shc candobecsuseofhslthmm [316)"; ‘ 22) Was aployed at all Mix. the last year? U YES ONO—um) 23) Gahowmrvdwsot’thepast 3654sysdidillnessar-wtuyleep nmthethirgs 25' he/sheMdoesathansgmmJob 27 2n) 1. colt—aploum? mm . —-p (26) 28 25) 'Ihe test questions are about your Ilployunt. Please, thirlc or the ward "mloyed" ‘ to include the total or outside colors—1t plus ult-esploymnt. 26) Iatusmovertmpastyesr,m2thtdrmmth.sndl‘indcuthcwmweelcpernulth _, ' Wamcmlsstyelr. 27) WWW 2’8)me 29)th 30)Wrwwss not 31)L1strorssch was employed: s was _ queued—(Er mnth the (plus seIr- wee was so oyed (plus self-opioyed) runner or days upland) durim Uployel ult-arployed) full-tin durirc when illness the mh on (Elm self- cut-1m that that period? or injury kept unloved) tins? (paid munch - _ the during that employed) usual days _gaegt.l971 WI > I we? E5: M1 t 2+3) : 31-3'? 32) Speakiru in ml terms, how would you describe 's mloyment sni «fir-Tumult 35 durim the t ear? my FILL-mm: I3/a'rms Il/zm Il/u'ms: Items 33) Did have some «mloym'ent and self - - - » in 1971/(2 than in 1968 or less? 36 m than in 1968 i was then in 1968 -'3a) Is ablstotakspnrtatallinmdimyplaywithotherchildrsn?DYE Elm 35) Is he limdted in the kind or play because or his health? Elves Duo 37 36) Is he limited in the momt or plsy because of his health? E5] YES NO 37) D’AflE6—l macro—9M2) 38) In t or health, would be able to go to school? I-JYFs E] "0 38 39) Does have to go to a certain type or school because at health? m as [:1 N0 1:0) Is limited in school attendance because or health? I771 m: E] no .. “1) If one: to school. on how 15 or the past school year did illness or injury ‘3') keep from 93113 to school days _” q l #142) On how many ms or the past 365 days did 1111254 or lruury keep __ n-on playing as usual 1)} g from gcirg to school? Ed”! 1111' LIBRQRIES MICHIGQN STRTE UNIV. :. Raw 1. e v e. a .\ u.»;w;:r; ‘ . o . a 3 a \x :r. . . Rename. . , pad. .,, u e. e 3....» . . oe.ou,.maswpm. ..s e ...e a, 2.; .5 a