N,“ '1‘) W f, 3‘3“? I ‘J 07‘ M. .477 7‘7"“ . I bus I I I . "1 7'" ‘VI '7.’I'7II “#777777 7715* 7 71'1“. ‘ ._ 171111, I7I 77. ll 17 7%" I.“ ”p.717“ I77II77III7( ‘I I77II'7 7'77 7.77777777737- 7"; '71.: W37III1I’7IIII7I 777” .777' 77 771M777I .17 ' $717. 7777717 In". ":17'171'7“ 7... 7‘I7r7 )7 7'II \7 I 7 I77 I ”777717-77 7.77 II fin’ I“). I" 111 g; I 7 . 1 1 1.1117177773777517 .. . I7 7 II' 7,77 .1 II." 777; 4.;7 “6727.; In" fifafl‘ I ‘IH M1. hQWI I 1 .7IL7I‘I flEIhéI “77777“ 7‘7“.) 777 ‘.‘I‘I77I ' 1 III II" 272732;. ; “I'- . *7. " 777‘“ 7 7'I7,I'1'.."IO'I.'1+'11/1T 73.9.1.1.“ i777 I Iw. ,.‘ ‘7 “Ia-.37" 47:- I‘M ZIP“: . IT ‘ '~ 2‘5” ‘_ -. ‘ I I . ‘.I17;7I'.1.‘.1fi ”1,11 ‘.I 17.1. -)\7 #7776141-7 . P11 II .' 77:7 .. “I777 ’17“; ($\J IIII)’ I.» I “77.11 $57“ If)“ ”"17 I I7Ivhl)7'7..f\l 7.717I7I::.II177 17.711.17.571111 1 #1)“ ,1. 7'77“: I. ‘ .- JVLI ‘ 1777‘ ' 7'?" :' ‘ I r I ‘ " 7k.‘.-'~YII.E77':FI' 777‘ 7 I‘ pm 'I ‘I I'I “4711.732. 7757:7773. I u? 'I \ II'I . 1“. 7'5‘7'7 7.1.7.1757 77(7' 7'I~'-;II “~ "7 77‘77' 4575,7545 .; .713.~:-I.Y~'“«I‘II.I"" . - if.) .- '.- l .-.;.-- .I . “I 7.24:“ RN)“ inf .. 13?. ~I-:‘.'I.'.'.‘..~I.“§.£; IIIIJI'I‘IT‘EI‘I".I‘-” ~ . - Il"‘. I II _.. I..‘ .L'. I‘ In]. -. ‘ '9' ‘ I :15 (-1371: .' . .‘7 1.1.1 I f a; .I’rm‘, 'I..I ”II;- 77:31" w: f7'71....'h‘.'I3% . 1.751., IV‘Ig-I-Jgjmf -_ 1-, -1 4' ’3 Inf' 31.". ‘771II7 I,» . .,I '..-:-.;I I. ‘..-. . ‘... ’ ‘ .f. ”9112» 1"0.~.;1.I..-'vI '7". '« \ . K 1 1 1 .1 1 1_$,J1117I.I - 9. 1.. I N71 “377.1".37231 1.1.1,.“ ‘1'. ”11"!“ngth ‘7'. n .1 1 '. "“- “ .I' ‘ In . M 7 7533.7 70M I‘M: -)....7m JIM-77' 'I"‘I)I.K77‘z§t 373$ 7"” 0 ‘III II'I'.I. , I .I.‘\I‘7‘II' . . .I“ '~ u .III .I .I‘u .3 ('11-'11 'L-J \. I _- I . . I I . It‘ I rm. 2; .7 . .I 1." Jan 1711‘? 77 II mu; . , 7.77 .I‘ 1 I77. . 1| 1 ‘1 1 11 ”1‘. I, .. 1% ' 1 ”sun J Ca ‘ I)... m: {$14 r" .r 1 r .1 1111111111111“ :71 -. .77 II ‘7. -I 7 77’1IIIJ 7 II.' ‘77,“; ‘3.“727599‘ 7 7 2771511‘1 ‘V 1.135;}(«1731‘7' '. 3‘. *S'I ”a I I777 I 1 1 ..1 1.. I1_ ... .1 .1. I u' I “In.” . .I p... ... 1771777.” 7'7" 1' '- 77:77'I" 'm' P' .. .1 777' ‘.‘"’ WIMP 1.9.: 11:1111111113117‘1121 I.‘L'r-. 1777‘- I.‘ (77! ”J: I ‘ ' ‘7 .. ' I ‘.I 7 ‘I77' j'." ‘7 ‘777L7. .77 "I‘7‘77'3é’7k-I‘»'7'7’It\-"I”".‘II' I'M-"777:7“ ' “7' "-' Q77“ 77‘ 1 1: I 1 I ~ .I I . 1- 1n|11:31..1‘1, - ‘EJII‘I- _, r .. 1- .‘I }.I11.1.11,"\‘.1,‘|1 1, I013.“ m. 171', ,1 v". ..; ‘14}- i‘ : 7' ' I . I 1'; 1. m ,IH- - ‘I .-~'. 777-‘7 ..~I - . 1-. ’ . x; ‘1', 7 77 7 I 77 )II -|'777 77.777. 7‘7. I7- 7757“. II ."7'7¥..:.77.5;fi17£:7 1’7 7 (I W‘:7( 7 ‘0) “77.;73 I “737' s 7 II 7 K I II, 7". 7- '77-77 QM 7' lb.‘ 7:70 :7 '777‘. 'I’. h 7.“.xt’7 :7 II #:3'77J‘b“ 7 7 I I I 17 7I 7.77‘7 I t7 I I. I 7' . 1.1-. 1.1.4.. I,” 0 ~ ‘ .' "'7’ . I ' . 17 7 I 7 1' 71 I 1"I'II7'7 II ‘77 7'77 1MH7IR‘I7,7I 77.". {7757' I ‘ ' I 77~II'7IIJ-',':T-' “7:07 7.7“;77'7 I ‘7 ' . I I 1 :1" .1. 1 1 ‘ >1“. u q"-7_ . t '1. . I | . I. l '01 I 1 I7; ’ ' 1 I 7 '1I11: . uv'wnh I111: $7.1:ka #:7177711 In I I 7 ’L - .'.I" I' I ' I , ‘ 7. I 1 I ‘ I, .1 I 1 7 ' \ ~ I.“ I '7 7 7777 ~'I'7I~I7 IIII II 7 . I 7777 7 I. 7 II I ‘ I (707 'III V1.1‘7I': I .7 I c 7 7“;- ”3‘7 ' 7II' 1N" . 7I‘I'II‘I77..7/777 7' .7777 I I 7 ,',. 1.17777 7.777 I III .7 I. ‘ '7 7.77 517' .1 ‘1- ‘ “77“I77I77‘7‘717r1711'77'7177 5711777" ' .7I 777!‘ ’ 77"“ .773. ““777; 1,717.5... I 115777 ’ 1' I: 7 II 7 71 '7.) : 17-II1E1IIII) [.3 Hr 7 7 7I7 II' 7777\7‘- (7‘? 7 %"_ kn: $1.11!. A ‘ . 7“". . _'_. “.7 .... ~.-..I',, I..(..' “-4275 «mu! . - ‘ H .— I h)‘ .. 17 I|I77‘ ‘ I ' I ' .1 ' . "I" IN III .‘7 ' I , Yu' I" 1‘ 1 11 . I 11111 I III 7 7.77:7 I I1 '11" . .11121‘. 777.5; Lg)!“ ‘ 11(1‘7‘4‘4’1 J7‘1 ‘1 II}: II .7 I , .III ":'.'. ' .' ’- I I: '2 I917 ~ :I'c’I I'm :‘ I ' ,7 ' .H‘ 1 I :1 I 7| I: I 1 77 1 7 ”1" 7'. 7." l1_".'. 11"", 7-1.1- 11.11.11 17' I‘ .1 ¢ 0 I I 7 I7I I 'I 7 l I II . 7I. 1 1“, II 1', I17 : 1) .1 :I 7 1 > 1 . , .1 1..“ ,,I I 7 7 1 1'. I. I' 7 ' .\' I». l '. .. ' 7 II_- 11(11171117 Ml. . 7: ~‘; I. I. I ' "I. .77 ' III'sIII'T-I: I 77." 777*,‘77IHI ..' 'l ' ‘ 1 ‘ 1 I. lII.I 7 II ‘ 7 ... 1 7 ' 77 717 7. I1 7 7777771111”: III.1I1I.7.77 7 17M 7I 777)] 7I7'II“77777II I .. 1‘7". 171“” unmv * Inwa- so.» M w This is to certify that the dissertation entitled CHARACTERISTICS OF HOUSEHOLDS 0F INFANTS WHO LIVE OR DIE IN A NEONATAL INTENSIVE CARE UNIT: AN ECOLOGICAL ANALYSIS presented by Ann Stirling Mullen Johnson has been accepted towards fulfillment of the requirements for Ph.D. degreein Family and Child Ecology ’ Major professor fl Date August III! 1985 MCI/ita- Alf .' - ' 1‘ 0-12771 Ill llllllllllllllllllllllllll 3 1293 01109 9029 RETURNING MATERIALS: MSU Place in book drop tof lemmas remove this checkout ram .1.”- your record. FINES mil be charged if book is returned after the date stamped below. CHARACTERISTICS OF HOJSEHQDS OF INFANTS WHO LIVE OR DIE IN A NEONATAL INTENSIVE CARE UNIT: AN E'PIEICALLNJLWS By Ann Stirling Mullen Johnson A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of. DOCTOR OF PHILOSOPHY Department of Family and Child Ecology 1985 @1986 ANN STIRLING MULLEN JOHNSON All Rights Reserved ABSTRACT CHARACTERISTICS OF HOUSEHOLDS OF INFANTS NHO LIVE OR DIE IN A NEONATAL INTENSIVE CARE UNIT: AN ECOLOGICAL ANALYSIS By Ann Stirling Mullen Johnson This research concentrated on the development and testing of household variables in a study of factors related to infant mortality. Twenty-seven mothers of infants who died in a neonatal intensive care unit and 30 mothers whose infants survived were surveyed through use of a mail questionnaire in the spring of 1985. The factors that charac- terize the households of infants who die in the general population did not characterize the infants who died in a neonatal intensive care unit. Furthermore. the households of infants who survived in a neo- natal intensive care unit did not appear to be particularly "at risk" before their birth. A path analysis was conducted on a population of infants who died and a matched sample of infants who survived. The results of the path analysis accounted for 57% of the variance in birth status. Survival was enhanced by a pattern of low household number and high income or high household number and mother working more. These combinations yielded a high number of sources of tangible assistance and support to the household during pregnancy. High levels of support Ann Stirling Mullen Johnson yielded high levels of information about pregnancy which. in spite of anxieties over being laid off from work. resulted in the survival of a vulnerable infant. A separate path. seemingly unaffected by the present variables. indicated that high birthweight also resulted in the survival of a vulnerable infant. Smoking and adequacy of prenatal medical care had no effect on outcome. The data supported the critical importance of support and information on outcome and the value of the 'household as a source of data. To my mother. Alberta Blanche Thom Mullen. born a two-pound. full- term baby in the Canadian Northwest; to my grandmother. Kathryn Blanche Stirling Thom. who died in childbirth. longing for her native Michigan; and to my great-grandmother. Ann Collins Stirling. whose passion for knowledge has transcended four generations. ACKN (M LEDGMEN TS Many people have contributed to the completion of this study. I would like to mention a few of them. though all are appreciated. I am deeply grateful to my advisors: Dr; Margaret Bubolz for her continued guidance and support; Dr. Linda Nelson for her expertise in helping me edit the manuscript and for her abundant encouragement; Dr. David J. Kallen and Dr. Robert Griffore. with whom I worked in the development of the project;lhu Beatrice Paolucci. who. before her death. was a source of inspiration and encouragement; and Dr. Andrea Doughty. who assisted with the questionnaire development. I particu- larly want to thank Dr. Kallen and Dr. Sue Kallen for their expertise. guidance. and hospitality during my visit to California to plan the data analysis. I want to express my appreciation to the United States Department of Health and Social Services for the Biomedical Research Support Grant IRR 05656-16. and to the College of Human Ecology for a Dissertation Award and the Alice J. Thorpe Award. Dr. Eugene DOlanski. FLO" Medical Director of the E. W. Sparrow Neonatal Intensive Care Unit. and his staff were most helpful and supportive. Thanks are also due to the women who responded to the questionnaire. Without their help. the project could not have been completed. iii A number of others contributed valued services. Paul Nelson did an excellent Job of printing the questionnaire. Sue Cooley. my typist. provided fine technical assistance on the manuscript. Alice Kalush. Ginny Powell. and Beverly Fico—Brown were helpful to me in setting up and maintaining the computer program. The staffs of several depart- ments were also helpful in a variety of ways: The Departments of Family and Child Ecology. Pediatrics and Human Development. the Instructional Media Center. as well as the Michigan and Ingham 00unty Departments of Public Health. I am grateful to my family and friends. I want to thank my mother and father. Alberta Blanche and Donald M. Mullen. for their love and nurturance through the years; and all my family for their tangible assistance and loving support. I am especially grateful to my daughter. Stirling Johnson. who made it all possible. Her understand- ing and companionship are deeply remembered and treasured. Many friends also contributed to the completion of this study and my enjoyment of it. Several deserve particular mention: Norman Frisch; Martha Napior; Bunee Kayne; Fred Fico; Lawain Nillett; Barb. Dave. and Renee Cummings; and Marah and Wayne Grist. They. and others. contributed in important ways to a rich learning experience. TABLE OF CONTENTS LIST OF TELES O O O 0 O 0 LIST OF FIGURES Chapter I. INTRODUCTION Statement of the Problem Research Questions . . Importance of the Study Conceptual Framework. A Family Ecosystems Perspective on Infant Mortality . . . . Def1n1t1ons O O O O O O O O O O O O O O 0 Summary and Overview II. REVIEW OF THE LITERATURE 0N INFANT MORTALITY Dependent Variables . . . . . . . . Independent Variables . . . . . . . Mother's Pregnancy History Prenatal Care . . . . . . . . . . Mother's Health During Pregnancy Economic Resources Summary of the Literature . . . . . III. CONCEPTUAL FRAMEWORK AND METHODOLOGY The Rationale . . . . . . . . . . . . . . A Reconceptualization of Some Traditional Pregnancy History . . . . . . . . Prenatal Care . . . . . . . . . . Mother's Health During Pregnancy Economic Resources . Sumary....... The Study . . . . . . . Questionnaire . . . . . Variables Page vii viii l I Statistical Measures Limitations . . . . . Assumptions . . . . . IV. RESULTS AND ANALYSIS Response Rate . . . . Results of Chi-Square Deficiencies in the High-Risk Approach The Ecological Model Scales Used Procedure . . Results . . . Summary . . . Discussion . in Path Analysis and Student Rationale for Path Analysis . . Requirements for a Causal Model V. SUMMARY AND RECOMMENDATIONS Significance Limitations . . . . . . . . The Path Model in Ecosystem Implications Recommendations . . . APPENDIGS O O O O O O O O O O A. HUMAN SUBJECTS APPROVAL B. NICU PROJECT INSTRUMENT C. COVER LETTERS . . . . . REFERENCES I O I O O O O O 0 0 vi t-Tests Perspective Page 128 130 131 135 155 158 Table 3.l 4.1 4.2 4.3 4.4 4.5 LIST OF TABLES Page Traditional and Household Variables Used in NICU Project Study . . . . . . . . . . . . . . . . . . . . . 70 Birth Status by Birthweight . . . . . . . . . . . . . . . 9l Birth Status by Gestational Age . . . . . . . . . . . . . 92 Birth Status by Planning for Pregnancy . . . . . . . . . 93 Zero-Order Correlations Among 14 Scales Used in the Path Ana] ytic M0d61 O O O O O O O O O I O O O O O 0 O O 106 Regressions Performed for Path Analysis . . . . . . . . . ll2 vii Figure 1.1 2.1 3.1 3.2 4.1 4.2 5.1 LIST OF FIGURES Diagram of Family Ecological Framework . . . . . . . . Schematic Representation of Factors Tested in Major Birthweight Studies Cited by Institute of M9d1Ci "e (1985) O O O O O O O O O O O O O O O O O 0 Schematic Representation of the Literature on Infant Marta} 1ty O O O O O O O O O O O O O O O O O 0 Schematic Representation of the Relationship of Factors Concerning the Household and the Mother . . The Proposed Path Model . . . . . . . . . . . . . . . Final Path Analytic Model Showing the Influence of Prior Fetal or Infant Loss. Number of Persons in Household. Amount of Time Mother Worked. Per Capita Income. Number of Sources of Tangible Assistance and Support. Number of Sources Of Information About Pregnancy. Fear of Layoff. and Actual Layoff. Birthweight. and Birth Status . . . . . . . . . . . Final Path Analytic Model in Ecosystems Perspective: Some Hypothetical Relationships Between the Macro- and Micro-Systems. and Within Micro-Systems . viii Page 10 48 64 69 105 111 125 CHAPTER I INTRODUCTION Infant mortality in the United States has declined from 162.4 per thousand in 1900 to lO.5 per thousand in l984 (U.S. Dept. of Commerce. 1976). In the past 5 years alone. the rate has dropped by 3 per thousand (Advance Report. 1984b). It is generally agreed that the decline is a result of improved health and sanitation and general economic conditions in the country. the availability of family planning services and abortion. and advances in medical care (Fuchs. l983). These factors are very much interdependent. Economic conditions made possible improved general hygiene and sanitation. Economic prosperity has been credited with the decline of the number of children per household. When families no longer needed children to help them Inake a living. they began to Choose to have fewer Children. The reduc- tions in infant death assured families that they would not have to have more children than they desired to reach their expected family size. Family planning services and abortion made it possible for families to control if and when they had children and to terminate pregnancies that were unwanted. Medical advances helped create safe methods for abor- tion and for the treatment of newborn infants who were having diffi- culties as a result of being born prematurely or with congenital anomalies. In addition. medical technology helped to reduce the infec- tions associated with postneonatal death. In spite of the decline nationally. there remain high rates of infant mortality for black infants that compare to US. rates for 1970 (Advance Report. 1984b). Some cities presently have infant mortality rates that rival thi rd-world nations (Children's Defense Fund. 1984). The gap between blacks and whites persists. and there is some indica- tion that it is increasing (Advanced Report. 1984b; Kessel. Rooks. & Cushner. 1976; Michigan Health Statistics. 1981; Moriyama. 1960; Shin. 1975a). It is this continued disparity that has prompted investigators to search for the causes of infant mortality. An abundance of documenta- tion over the last century has produced much data on demographic and social factors and a small number of behavioral and psychological factors. Using this information to intervene. program specialists have created risk scales for geographic areas and for mothers by rating them on the basis of the findings from research. These risk factors include mother's age; parity; prior fetal and infant losses; race; socioeco- nomic. nutritional. and marital status; smoking; drug consumption; and prenatal care. The problem with the risk approach is that even where these fac- tors are compounded. most of the babies of these mothers survive. For example. very young women with little education had an infant mortality rate in 1968 in New York City of 33.5 per thousand. Women 25 to 29 who had 4 or more years of college had a rate of 8.4 infant deaths per thousand births (Kessner. Singer. Kalk. & Schlesinger. 1973). While this meant a difference in 25.1 infants per thousand. there were 966.5 surviving babies per thousand born of women 15 to 19 with an elementary education or less. What factors impinge on those 25.1 per thousand that make the difference between an acceptable rate of mortality and an unacceptable one? On the other hand. we might well be concerned about those 8.4 who should be experiencing optimum conditions. In spite of the decline. the rate of infant mortality in the United States remains an embarrassment. The U.S. has ranked between 14th and 18th in the rate of infant mortality among nations and in 1984 was tied with West Germany for 14th place. Taiwan. Spain. Luxembourg and Ice- land. Japan. Sweden. and Finland have lower infant mortality rates than the U.S. (Population Reference Bureau. 1984). The United States is frequently compared with Sweden. whose rate Of infant mortality is considerably lower than that of the U.S. This has frequently been explained by the fact that Sweden has a small. homogeneous population. with a high level of education. a high standard of housing. a more even distribution of wealth. national health insurance. and free medical care. The fundamental difference in the mortality rates is the higher proportion of low birth weight infants in the United States. Geijerstam (1969) has shown that the differences in neonatal mortality between the two countries decreased from 42 to 12% when the U.S. incidence of low birth weight infants and the Swedish weight-specific mortality were used in the equation. Low birth weight is a major predictor of neonatal mortality (infant death within the first 28 days of life). Six percent of the births in the U.S. are low birth weight. This group. however. comprises 75% of all the neonatal mortality. Among babies of less than 1.000 grams (2 pounds 3 ounces). three of four died during the neonatal period. Babies over 4.500 grams (9 pounds 15 ounces) also died at increased rates. Due to increased surgical skill and technology. there has been a reduction in the rates of neonatal mortality. but no significant reduction in the population of low birth weight infants (Brooten. 1983; Lee. Paneth. Gartner. Pearlman. & Gruss. 1980). The impact of low birth weight goes beyond the neonatal period. Birth weight is said to affect postneonatal mortality and results in significant increases in congenital anomalies. developmental delay. and rehospitalization. Low birth weight babies are said to be twice as likely to be rehospital ized; very low birth weight babies are 4.5 times as likely to be rehospitalized. McCormick. Shapiro. and Starfield (1984) claimed that this increased vulnerability is not entirely explained by the increased prevalence of congenital anomalies and developmental delay in low birth weight infants or interaction with the adverse sociodemographic correlates of low birth weight. Even low birth weight infants in advantaged families are more likely to be rehospitalized. Estimates of handicap indicate that among babies of 751 to 1.000 grams (1 pound 10 ounces to 2 pounds 3 ounces) one Of four will have a major handicap. The prevalence of handicap for very low birth weight infants has remained at 6 to 8%. even though the chances of survival of these tiny infants have increased from the early 19605 (Stewart. Reynolds. & Lipscomb. 1981). According to Cohen. Sigman. Parmalee. & Beckwith (1982) studies analyzing both biological and social factors have concluded that the major impact on developmental outcomes is made by the environmental factors. not by medical complications. "The results indicated that early illness cannot be ignored. although it does not predict outcome" (p. 337). Thus it appears that the factors that affect low birth weight also have an impact on the child's survival of the first year as well as health during the first years Of childhood. whether this is a function of the medical problems at birth or the continued effects of precipitating factors of low birth weight. Recent data show that low birth weight is the major cause of death for black infants. while congenital anomalies are the major cause of death for white infants (Advanced Report. 1984b). While low birth weight and congenital anomalies are frequently associated. it is likely that because black infants have a better chance of survival at very low birth weights (Yerushalmy. vandenBerg. Erhardt. a Jacobziner. 1965). white surviving babies with congenital anomalies are more likely to be largen. Thus the causes of death of white babies are not listed as low birth weight. but congenital anomalies. W The problem of infant mortality continues to mystify research- ers. One reason may be that researchers continue to use similar variables and similar operational measures and similar methodologies. In some instances it appears that the heavy emphasis on assessing the relationship between prenatal medical care and outcome of pregnancy is self-serving; that is. it mobilizes social efforts to pay physicians for medical care. The focus of prenatal care may also be explained by the relative ability to intervene. Involving women in prenatal care may appear easier than redistributing incometor assuring adequate employment or nutrition to all households. factors that may ultimately affect outcome of pregnancy more than prenatal care. It appears that some new approaches are needed. For that reason. this study concentrated on the development and testing of household characteristics and family relationships. Furthermore. it focused on the most vulnerable infants: those who were in a neonatal intensive care unit at or shortly after birth. The question thus became whether household characteristics and family relationships are associated with a vulnerable infant's death or survival before leaving the neonatal intensive care unit. Winn: The research questions in this study involved comparisons of the two groups of infants who were treated in a neonatal intensive care unit: the households of the infants who died prior to leaving the NICU and the households of the infants who lived following treatment in a NICU. To what extent did the two groups differ on traditional measures such as conditions prior to pregnancy. prenatal care. health conditions during pregnancy. and economic resources? To what extent did the two groups differ on household characteristics and family relationships? The study attempted to assess the relative importance of household characteristics and family relationships in comparison to the tradi- tional factors that distinguish the households of the infants who die from the households of the infants who live in the general population. Wm An understanding of the household characteristics and relation- ships relating to the death or survival of a vulnerable infant could greatly enhance the development of a social policy that adequately intervenes to protect the health of babies. Since medical advances have enabled many vulnerable infants to survive. the focus now needs to be on preventing low birthweight and debilitating birth condition. Present social policy. although acknowledging the contributions of income. education. and other factors. has focused on the providing of prenatal care. This has been promulgated in spite of evidence that the correlations between prenatal care and outcomeedo not infer causa- tion. In fact. in many instances. there is no correlation between prenatal care and outcome when gestational age. income. and other factors are controlled. Changes in social policy would affect the nature of profes- sional intervention. Should household crowding. social support. and job anxieties prove to be critical in the survival of a vulnerable infant. two approaches might be more beneficial: societal efforts toward redistributing income»in a more equitable manner. such that the amount of poverty in this country is reduced and helping individual households generate needed resources. It would also indicate the importance of a family specialist in assessing household problems. generating resources. alleviating stress and crowding. and helping to restore household relationships. Thus. family advocates both at pro- gram and household levels of intervention would play more important roles in reducing infant mortality. Indications that household characteristics and relationships are important would prompt reanalysis of present studies and develop- ment of new studies focusing on further uncovering the role of the household in the outcome of pregnancy. Wm WOW This project recasts the literature available into a family ecosystems framework and. in doing so. conceptualizes the household as the unit of analysis. The fetus is seen as the "active growing human being" (Bronfenbrenner. 1979. p. 21) involved in a microsystem-~a pattern of activities. roles. and interpersonal relationships. The only microsystem that the fetus is involved with is the household. The exosystems—-parent's work. kin and friendship networks. prenatal care (all those systems that impinge on the fetus but which have no direct involvement with it)--are investigated to discover the links between them and the household. Some consideration will be given to the macro- system--the "consi stenci es. in the form and content of lower-order systems (micro— and exo-)-that exist. or could exist. at the level of the subculture or the culture as a whole. along with any belief systems or ideology underlying such consistencies" (Bronfenbrenner. 1979. p. 26). Since the fetus is not involved with any other microsystems. there are no links between microsystems (mesosystem). To further elaborate a model Of the relationship between fetus- household and the environment. the schema developed by Paolucci. Bubolz. and Rainey (1976) is used (Figure 1.1). These authors saw the family as individuals with personal attributes organized as a corporate unit: the family. It is conceived as the site of human resource formation and transformation. It interacts with three other systems: the physi- cal environment (composed of land. raw materials. climate. water. air. fuel. plants. and animals). the human-built physical environment. and the socio-institutional environment. The family is conceptualized as interacting and interdependent with these systems. In the process of interaction. the family becomes involved in market-economy activities generating income for the family and nonmarket activities consisting of household production. management of resources. farming and fishing for home use. individual maintenance. human resource development. community participation. and voluntary assistance. The family likewise provides an integrative function between the market and nonmarket activities culminating in four social and economic development functions: increased well-being (health. competence. skills. literacy). fertility management. increased income production. and increased social equity. Each of these factors is fed back into the family directly and into each of the defined environments. thus creating both change and IQ a.>u_mco>_c: Cumum cmm_;u_z ”mc_mcm4 ummm .mEmLmoLmlm:_ccmo_ .mELomucoc tam «mow—”Emu .coeoz .Lmnsoooo .mmm. .>oc_m¢ .z w .~_on:m .z .z ._uu:_omm .m ECLmV .xLCZOEme .mu_mo_ouo >_.Emm mo Emcmm_au-._._ Ocammu 11 stability that affect the institution's behavior and culture and the physical environment and. through them. the family. Definitions The terms used in this study are presented here in their conventional usage. W3 Death of a child before his or her first birthday. Wm: Death of an infant before 28 days of life. W: Stillbirth of an infant of 20 weeks or more of gestation. W: Still birth or death of an infant or fetus stillbirth or death of an infant from 20 weeks of gestation until 1 week following birth. Wm: Death of an infant from 28 days until the first birthday. W: A term that generally refers to medical monitoring of a child before birth. In some studies. it has included care during labor and delivery. More recently. the term has come to include services of a nutritionist. social worker. and other personnel associated with a medical clinic. W: The term "household" refers to the term "family" in the sense of persons related by blood or adoption. or a "bonded unit of interacting and interdependent persons" (Andrews. Bubolz. & Paol ucci. 1980. p. 32). The intention is to limit the unit to persons occupying the same living space. 12 .lnggme: A term that refers to all sources of revenue for a household. Sometimes it refers only to father's income; sometimes to household income. This study includes per capita income: the amount of income from all sources in the household divided by the number of people in the household. .Nggna1a1_1n1gn511g_gare_unjj;(NICU1: A unit of a hospital that specializes in care of neonates. This includes special staff as well as highly technical equipment. These units are found throughout the United States on a regional basis. Transportation is sometimes pro- vided to the NICU from outlying areas. W In response to the continued need to understand the factors behind why some babies live and some die. this study investigates the characteristics and relationships of households of vulnerable infants who live or die. The ultimate purpose of this study is to provide information to enhance survival of infants. A more immediate purpose is to identify variables that are worthy of study on a larger scale. It should be emphasized that this study is an exploration of many variables. in the hope of identifying the role of the household in infant mortality. Chapter II reviews the literature on infant mortality. It includes studies as early as 1949 and as late as 1985. It also covers a broad range of factors including characteristics of the mother before pregnancy. prenatal care. health factors. and economic resources. 13 Chapter III outlines in more detail the purposes of the study and provides a rationale for the development of several new variables and outcome measures. Chapter IV provides the data from the study and the data analysis. The significance and implications of the study are assessed in Chapter V. CHAPTER II REVIEW OF THE LITERATURE ON INFANT MORTALITY This literature review attempts to cover a broad range of factors that influence infant mortality. The focus here is on prenatal factors and does not cover the intrapartum period. delivery. or type of care. Some references are made to morbidity to infants as it appears in the literature on mortality. A few references are made to studies done outside the United States where the U.S. literature is lacking. This review will define. briefly. both outcome and independent variables and summarize the literature pertaining to them. Wank: Several outcome measures are found in the literature on infant mortality. They are neonatal mortality (death under 28 days) and postneonatal mortality (death from 28 days through 11 months). The term "infant mortality" combines these two measures. Fetal death is the death of an infant more than 20 weeks of gestation and stillborn. Perinatal combines fetal death and death occurring within 1 week of birth. Rates are calculated by dividing infant deaths by the number of births (plus fetal deaths in the case of stillborns) and multiplying by 100. 14 15 Prematurity refers to weight and gestation. Low bi rthweight (less than 5 pounds 8 ounces) and very low birthweight (3 pounds 5 ounces) technically refer only to weight. Since most infants who are low birthweight are also low gestation (less than 37 weeks). the terms have become synonymous. However. there are a few babies who are small for dates; that is. they have experienced intrauterine growth retarda- tion (IUGR). and although they are fully mature. they are smaller than normal. Some authors separate the two (Yerushal my et al.. 1965). Less frequently. outcome measures used are Apgar scores (a rating of the infant's vital signs). pregnancy complications. obstetric complications. congenital anomalies. and developmental delay. IndepmdenLVaniabJes The factors that affect outcome for the infant. based on the categories in this study. are: 1. Mother's pregnancy history. This includes factors that affect the mother prior to her pregnancy: age. parity. prior fetal and infant loss. marital status. and race. 2. Prenatal care. This has been defined in the literature almost exclusively in terms of the number of visits to a physician. the trimester of the start of care. "adequacy" of care. a combination of the two types of delivery service. and "adequacy" of care which combines these measures. Some efforts have been made to assess the quality of care. 3. Mother's health during pregnancy. This includes some of the major factors said to affect a woman's health during pregnancy: 16 weight. cigarette smoking. drug consumption. disease. and psychological factors. 4. Economic resources. This includes resources available to the household that affect the health and development of the fetus: education of the mother. occupation of father and mother. work status of father and mother. income. and crowding. Other factors such as housing conditions and receipt of public assistance have been used to create indices of socioeconomic status. .MQ1hfi£L§_E£flflnin§¥_H1§IQ£¥ Age has consistently been shown to affect the outcome of pregnancy (McCormick et ah» 1984; Perinatal Association. 1983; TOkuhata. Digon. & Mann. 1973). Both Heady. Daly. and Morris (1955) and Robinson (1967) have found more stillbirths among older women. On the other hand. as age increases. the tendency to have a low weight baby diminishes (Crump. Horton. Masuoka. & Ryan. 1957). Others have found no associa- tion with age. when other factors are controlled (Shah & Abbey. 1971). Sellig (1984) found that black women between the ages of 20 and 24 had a higher rate of infant mortality than younger or older women. Multiparity increases the risk of stillbirth (Heady et ah» 1955). neonatal mortality (Wallace. 1970). and perinatal mortality (Chamber- lain. 1979; Ross. 1964). Parity dramatically increases postneonatal mortality (Heady et al.. 1955). Crump etzal. (1957). however. found that the greater the parity. the heavier the infant at birth. But the lower the socioeconomic status (SES). the lower the birthweight. Thus. 17 grand multiparas (who tend to be lower SES) remain at higher risk for lower birthweight. Prior fetal and infant loss has frequently been found to be associated with adverse outcome. Boone (1982) found both types of losses associated with low birthweight and very low birthweight infants when compared with control infants of normal weight. In addi- tion. low birthweight was associated with previous therapeutic abor- tion. Shapiro. McCormick. Starfield. Krischer. & Bross (1980) found that prior fetal loss was associated with significantly higher neonatal loss. congenital anomalies. and developmental delay. However. among those survivors. there was a decreased risk for postneonatal death. Other studies have also found a relationship of fetal loss and the increased presence of congenital anomalies and developmental delay (Myrianthopaulos & Chung. 1974; Shapiro. Schlesinger. & Nesbitt. 1978L There is some’question as to the interpretation of the association of fetal loss and mortality. Gray (1984) suggested that the loss may lead to a short interval until the next pregnancy (also considered a risk factor in itself) because mothers tend to replace a lost child and go on to further pregnancies. He suggested that this may introduce a bias in longitudinal or cross-sectional studies when observation is truncated before family formation is complete since the progression to higher-order births may selectively occur among women with prior adverse outcomes. Illegitimacy is frequently found to be related to adverse outcome. Brooks (1980). using hierarchical multiple regression analysis to a 18 proposed causal ordering of variables. found that family incomeeand racial composition almost perfectly predicted legitimacy. Legitimacy and racial composition predicted low birthweight ratios. Neonatal and postneonatal mortality were strongly determined by low birthweight levels. Chase and Nelson (1973) and Kessner et a1. (1973) found rates Of mortality higher for illegitimate children. Stickle and Ma (1977) found that unmarried women over 20 had relatively low rates of mortal- ity and low birthweight. but in the category of under age 20. unmar- ried women had rates about six times as high as the other two age categories. Boone (1982) found no differences among very low birth- weight. low birthweight. and control groups because of marital status. This was. however. a population of black women in a Washington. 04:. hospital. where there was a likelihood of a very high rate of illegiti- macy. Race. likewise. has consistently been considered an important factor in infant mortality. Brooks (1980) found that racial composi- tion of the population strongly predicted low birthweight ratios. which in turn were found to directly affect the neonatal and postneo- natal mortality rates. Low birthweight accounts for large differences in the death rates for black and white infants (Erhardt & Chase. 1973; Shah & Abbey. 1971; Shapiro et ah» 1980; Yankauer. 1950). Shah and Abbey further elaborated that white babies had a lower risk of dying than did nonwhite babies. but when adjusted for birthweight. white babies had only a slightly better chance than did nonwhites. The adjustment of economic status. maternal age. and birth order did not 19 significantly affect the percentage deviations of either white or nonwhite babies. That is. these factors did not influence the effect Of race on neonatal mortality. 0n the other hand. Yerushal my et a1. (1965) found that when bi rth- weight was adjusted for gestational age. blacks had lower rates of neonatal mortal ity than whites in the low bi rthweight categories. However. among mature infants of normal birthweight. whites had lower rates than blacks. Naeye (1979a). in a large prospective study. found that he could account for racial differences in perinatal death by prepregnancy body weight and medical care. He concluded that race. per se. is not responsible for the differences. Both neonatal and infant mortality are higher for males than for females in the U.S. This pattern holds for both blacks and whites. However. Scrimshaw (1978) has noted the opposite trend for some countries. Little effort has been made in U.S. research undertakings to understand this phenomenon. 12mm Three major issues rel ate to prenatal care. First is the question of its value. To what extent does the provision of prenatal care make a difference in the outcome of pregnancy? The second question rel ates to barriers to that care. Why do women not seek or continue with prenatal care? Third. how is the research on prenatal care being used? Several public documents are examined from both state (Michigan) and 20 national levels to see how research is being used to justify proposed programs. The term "prenatal care" has comerto be synonymous with medical care during pregnancy. Since childbirth has come to be seen as a medical event (MacIntyre. 1977; Rothman. 1982; Stewart & Erickson. 1977). care during pregnancy has also come to be defined as the domain of physicians. The American College of Obstetricians and Gynecologists (Standards. 1974) defined prenatal care as "a planned program of medi- cal evaluation and management. observation and education of the preg- nant woman directed toward making pregnancy. labor. delivery and the postpartum recovery a safe and satisfying experience" (p. 35). The standards for care espoused by ACOG include at least one visit in the first trimester of pregnancy. one visit per month until the seventh month of pregnancy. two visits per month until the ninth month. and four visits per month thereafter until delivery. Most measures which have been used for research derive from this standard. Operational definitions of prenatal care have consisted of trimester of start of care. number of visits. type of delivery service. or measures of "adequacy" of care based on a combination of these measures. Many women do not have care that matches the standard. Gortmaker (1979) noted that only 45.8% of native white women in his sample (New York City) got "adequate" care. Chase and Nelson (1973) found that only 20% of their sample (also New York City) got "adequate" care. 21 If one considers only the proportion of women starting care in the first trimester. the number is somewhat higher. According to the U.S. natality statistics (Advance Report. 1984a). the proportion of mothers who received prenatal care in the first trimester increased steadily from 68% in 1969 to 76% in 1979. Since that time the propor- tion of women getting fi rst-trimester care has remained the same. While the proportion of white mothers remained the same. the proportion of black women getting fi rst-trimester care dropped slightly. There were increases in the proportions of white and black mothers receiving delayed or no care in 1982. Between 1981 and 1982. more women 25-29 and fewer women 30 and older received early care. With regard to socioeconomic status. the only group that showed an increase in the receipt of early care was mothers who had completed college. There were increased levels of delayed or no care for women with 9-11 and 12 years of schooling. The number of prenatal visits remained largely the same. 11.4 visits. in 1982. The median number of visits for whites was up slightly; for blacks it remained the same. There was a slight decline in the median number of visits for women beginning care in their third trimester. In terms of age. women between 25 and 34 were more likely to receive care in the first trimester than were younger women (82% compared to 53%). Women pregnant with their first child were less likely than women pregnant with their second child to get first- trimester care. The proportion of women getting f irst-tri mester care 22 declined after the second birth. Women who were unmarried. women who had not completed high school. and rural women got less care than their counterparts (Advance Report. 1984a). In order to assess the relationship between prenatal care and the outcome of pregnancy. a variety of operational measures have been used. Several studies have utilized the timing of care as a measure of prenatal care (Elster. 1984; Harris. 1982; Lewit. 1983; Shah & Abbey. 1971). Others have used prenatal visits (Dott & Fort. 1975; Ryan. Sweeney. 8. SOlOla. 1980). Most. however. have used combinations of these measures to create an index of care. Schwartz and Vinyard (1965) assessed the timing of care in relation to gestational age. Terris and Gold (1969) used time of first care and the ratio of observed to expected number of visits. Kessner et al. (1973) created an index of adequacy based on the A006 standards for care. which included the type of delivery service. To be adequate. a woman must have had care in the first trimester and a number of visits based on length of gestation and private delivery service. Chase and Nelson (1973) also used that index. Gortmaker (1979) modified the index. eliminating the type of delivery service. While many authors have made a distinction between care and no care. Boone (1982) used that as her sole measure of care. While most researchers have used only quantitative measures. Hall. Chng. and MacGillivray'il980) compared diagnosed versus actual outcome of pregnancy in a sample of mothers. Many of the studies on prenatal care have suffered from method- 01 091 cal problems. 23 l. The quantitative measures cannot assess the quality of care. Physicians are assumed to be equally competent in the care they provide and provide comparable care. 2. The number of visits is the least reliable measure. A given number of visits may be scattered throughout the period of pregnancy. may occur at the beginning or at the end of care. They are considered as equivalent; Too few visits has been associated with poor outcome. The assumption is that it is the lack of care that results in poor outcome. However. an excessive number of visits has also been associated with poor outcome (Perinatal Association. 1983). No one. however. has suggested that too much prenatal care is responsible for that outcome. 3. The timing of the start of care may be problematic. too. It-has been assumed that once care is started. it is continued. This may not always be the case. Birth records. the source of most of these data. may be inaccurate. Anecdotal evidence (Institute of Medicine. 1985) suggests that when birth records are completed. the trimester start of care may be dated from the beginning of the care received by the physician who was treating the woman just before delivery. Pre- vious care at another facility may not be counted. Watkins (1968) found that among black. married. low-income women. 46% of those she would have categorized as starting in their third trimester actually sought care in the first trimester but did not continue. Furthermore. a woman may be seen by a physician early in pregnancy to verify preg- nancy and assess the expected date Of delivery. thus qualifying a woman 24 for program benefits. such as Aid to Dependent Children or Women. Infants and Children (WIC) food supplements. She may or may not con- tinue. but might be counted as fi rst-trimester start of care. 4. There has been an implicit redefinition of prenatal care from medical observation (the basis for epidemiological studies) to mul ti-service delivery (the basis for programs and program evaluation). The meaning has changed. but the actual interventions that are said to be responsible for the improved outcome remain unclear. The emphasis in these programs appears to be counseling and education. With the exception of health problems such as diabetes and heart disease. most medical interventions are diagnostic. There is little evidence that physicians can prolong gestation or increase bi rthweight. Even where ocytocin is used to forestall delivery when it is considered premature. there is no assurance that contractions were not from false labor and thus would have stopped anyway. There seem to be two basic questions: Did the counseling actually take place? And if so. did it have any effect? For example. only 35% of cigarette smokers in one study reported getting any advice from physicians about smoking (Harris. 1982). A second example concerns food. To what extent. if one is counseled on proper diet. is the mother willing and able to follow instructions? Diets prescribed by physicians (and nutritional counselors) are frequently culturally biased (Institute of Medicine. 1985). Even where appropriate to the mother's life style. the counseling does not mean she is able to pur- chase and prepare the foods accordingly. Where food supplements are 25 part of the program. are they actually consumed by the mother? There is evidence that food supplements are actually distributed to the remaining family members (Jacobson. 1980). These questions. however. have not been systematically addressed. 5. The general izability of the result of program evaluations is questionable. Programs usually service high-risk women who may be at risk for a variety of reasons: age. parity. socioeconomic status. or medical problems. While epidemiologic studies are correlational but more generalizable. program evaluations would seem to indicate causa- tive or ameliorative effects but are not generalizable. What is needed are random clinical trials to avoid the problem of self-selection into care. This has been considered unacceptable and unethical precisely because medical care has been considered to be of value in affecting outcome. In spite of acknowledged methodological difficulties. Gray (1984) referred to the literature as a methodological mine field. Research studies continue to infer that prenatal care makes a difference in both prenaturi ty and mortality. Wm. Prematurity is an important outcome measure because birthweight and length of gestation are factors that strongly influence whether or not a child survives. Its relation to prenatal care has been the subject of many studies. Kessner et a1. (1973) found a strong association between prenatal care and bi rthweight. Gortmaker (1979) likewise found an association between care and outcome when he reviewed Kessner's data. modifying 26 Kessner's measure by eliminating the type of delivery service and using only a measure of "adequacy" based on trimester of start of care and number of visits. However. prenatal care made little difference among white women who delivered on a private service. Prenatal care did make a difference for white women who delivered on a general service and all black women. He suggested that socioeconomic conditions may account for this difference. Elster (1984) found that poor care had more effect on primiparas. whereas age had a greater effect on multiparas. Very young primiparas with early prenatal care were not at greater risk than adult primiparas in the same category. Multipara teenagers were at increased risk regardless of whether they had prenatal care. Terris and Gold (1969) and Ryan et a1. (1980) investigated the relationship of prenatal care to bi rthweight in low-income popula- tions. They came to different conclusions. Terris and Gold found that when mothers of premature and mature babies were matched for sex of infant. birth order. age. and marital status. there was no difference in the amount of prenatal care. Ryan et al. found that among women in a hospital who had zero to three or four or more prenatal visits and where they were similar in terms of education. medical complications. age. parity. and other sociodemographic factors. there was a signifi- cant difference in the amount of low bi rthweight. as well as fetal death. neonatal and perinatal death. Terris and Gold measured prema- turity as being birth before 37 weeks gestation; Ryan et al. measured prematurity in terms of birthweight. Whether their respective defini- tions made a difference is not known. One likely influential factor is 27 that Terris and Gold measured prenatal care by the timing of the first visit and the ratio of observed to expected number of prenatal visits. Ryan et al. measured only the number of visits. The differential between the mean number Of visits of the zero to three visits group was 'L43; the mean number of visits of the four or more visits group was 12.78. It is likely that in the Ryan et a1. study. termination of pregnancy (resulting from early delivery) interrupted prenatal care. thus creating a spurious association between the two variables. Moreover. although Ryan et al. claimed that there was little difference in socioeconomic status. they found that the group with zero to three prenatal visits contained half again as many women with less than eighth-grade education as the four or more visits group. Thus. in the first group. there could be considerably more women with less than high school education and in the second group many more women with college degrees. Since this measure was their only measure of SES. their claim Of no difference in SE5 should be questioned. Schwartz and Vinyard (1965) found that the influence of the timing of prenatal care was limited to uncomplicated pregnancies over 36 weeks gestation. In this interval. the prematurity rate (small for gesta- tional age) is significantly higher in babies of women with very late or no care than in those with earlier care. This association is inde- pendent Of maternal age. parity. race. and income level. Unknown components of prenatal care unrelated to the medical aspects of care may be responsible. they claimed. The authors suggested that for babies born at less than 36 weeks. care is interrupted by termination 28 of pregnancy rather than prematurity being affected by the lack of care. The trimester start of care is responsible for bi rthweight and obstetric abnormalities except pregnancy complications. according to Tokuhata et al. (1973). This was true regardless of mother's race. off— spring's sex. mother's age. the number of previous pregnancies. and the legitimacy of birth and month of delivery. They did not. however. control for socioeconomic status. Eisner. Brazie. Pratt. and Hexter (1979) controlled for education. age. parity. and race and attributed differences to lack of prenatal care. Differences. however. were between those with care and no care. There were no appreciable di ffer- ences between those with care. regardless of the timing of the visit and the number of visits. When the population was all black. low- income women. having at least one prenatal care visit did distinguish the women having a mature baby and women having a very low bi rthweight baby (Boone. 1982). Several authors attempted to calculate the relative importance of the effect of prenatal care. Harris (1982) claimed that the effect of prenatal care on bi rthweight was through increasing gestation. He could account for 2 to 4 ounces increase due to prenatal care. Green- berg (1983) found prenatal care to be important even when education and race were controlled. Five percent of low bi rthwei ght among blacks and 1% of low bi rthweight among whites could be attributed to care. Lewit (1983) found prenatal care to be associated with a 140 gram (5 ounce) increase in bi rthweight. He noted the possibility that 29 prenatal care may serve as a proxy measure for other behaviors that may be more causally related to birth outcome. He suggested that it is important to begin to separate out the effects of maternal diet. smoking. and alcohol consumption. factors he was unable to control because this information is not included in birth records. The extent to which prenatal care contributes to favorable outcomes by promoting healthful behavior needs to be investigated. However. Rosenzweig and Schultz (1982). controlling for smoking. found that delay in seeking care accounted for only 1.5 ounces. where smoking a pack of cigarettes per day accounted for a reduction in weight of 10 ounces. Other studies have shown an inverse relationship between the amount and timing of prenatal care to mortality. Infant mortality was related to the adequacy of prenatal care for infants over 36 weeks gestation (Chase & Nelson. 1973). Those with "inadequate" care had a rate of mortality three times as high. In spite of marked differences between ethnic groups. the relationship with care remained. In another study (Dott & Fort. 1975). the rate of infant mortality was eight times greater for infants whose mothers received no care compared tO*women who had nine or more visits. The association remained when birthweightrspecific mortality was considered. Fetal death was not considered. Thus. those at any given birthweight who were stillborn were not considered in the calculations. Infant mortality was inversely related to prenatal visits in both Charity (low- income) and major urban hospitals. However. the rate of mortality to infants born in major urban hospitals with one to four visits was 30 higher than the rate for infants born in charity hospitals with no visits. This phenomenon is probably due to the lower rates of death at very low birthweight for black infants than for white infants (North & MacDonald. 1977). suggesting that a substantial number of women with few visits deliver prematurely. Stickle and Ma (1977) showed that the death rate for infants was higher where mothers had late or no prenatal care. McCormick. Shapiro. and Horn's (1979) study of infant mortality in Chile concluded: Infant mortality rates were negatively associated with antenatal services personnel and utilization. medical services personnel and utilization and indicators of increasing socioeconomic status. The hypotheses not supported were those pertaining to immunization and nutritional status. (p. 151) However. in the multiple regression. number of university-trained Inidwives per 100.000 population was dropped from the equation because it had no predictive value. The only variable that had predictive value was professional attention at birth. Other variables (ambulatory visits. pediatric visits. illiteracy. and urbanization) added very little to the variance explained. Professional attention at birth explained between 70 and 90% of the variance. indicating that high percentages of births with professional attention were strongly associated with low values of infant mortality rates for the years studied. It may be that those who had a physician-attended delivery also had prenatal care. but that was not explicit. Because of the clear evidence of influence of obstetric care at birth and the recent emphasis on intervention during the interuterine 31 period. professional attention at birth should not be equated with prenatal care. Both Dott and Fort (1975) and Shah and Abbey (1971) found that postneonatal mortality was influenced by prenatal care. This suggests that prenatal care is acting as a proxy for other vari- ables. perhaps including postnatal care. The most valid conclusions with regard to prenatal care would seem to come from mortality more closely associated with the interuterine period: fetal and neonatal mortality. Fetal death. according to Robinson (1967). is more likely to be found among whites. Black babies have a higher survival rate at very low weights. but at higher weights (above 2.500 grams) black babies are more likely to die. For the women with the most economic and social problems. those with no prenatal care had a fetal death rate half again as much as in the best cared for and most affluent group. When multiparity was controlled (higher in the poor-care group). there was less evidence of the influence of prenatal care on the fetus. Furthermore. there was no difference in mean fetal weight in the care groups. Infants whose mothers had no prenatal care had much higher than average rates of neonatal mortality than those who had care. even when controlling for socioeconomic status. race. birthweight. maternal age. and live birth order (Shah 8. Abbey. 1971). When comparison was made between prenatal care groups. those who started prenatal care in the first trimester did not fare as well as those who started in the second and third trimesters. This may be accounted for by the premature births interrupting the care of some of those who started in the first 32 trimester. whereas all of those who started in the third trimester were still pregnant. In Shah and Abbey's study. those in the no-care group continued to have the highest risk for neonatal mortality. while those starting in the third trimester had the lowest risk. when controlling only for bi rthweight. However. when other factors are controlled. the negative impact on the no-care group is reduced by 50%. Kessner et a1. (1973). widely quoted as supporting the rel ation- ship between prenatal care and infant death. noted that the addition of a medical care "adequacy" and six other independent variables had no explanatory power beyond birthweight alone. Most studi es that have shown a correlation between prenatal care and outcome also included a reminder that the association does not signify causation (Gortmaker. 1979; Kessner et al.. 1973; Shah 8. Abbey. 1971). Tokuhata et a1. (1973). having claimed that "no conclusions can be drawn as to 'cause and effect' relationship of the statistical associations" (p. 183). suggested that the provision of care is ameli- orative: "Our study . . . has shown some interesting and significant results which support the importance of early prenatal care to reduce the risk of obstetric abnormal ities" (p. 163). In spite of the cautions about confusing correlation with causal- ity and the frequent findings of relationships between prenatal care and socioeconomic status and other factors such as age and parity. there has been a great deal of effort to develop and evaluate programs of care for high-risk women. The assumption is that the relationship 33 between prenatal care and outcome is causative and ameliorative. This assumption should be questioned in light of the evidence. Winds. Several program evalua- tions are reviewed here. dating from the 19605 to the present time. One of the studies compared the high-risk population served with pre- vious pregnancies in that group (Aubry & Nesbitt. 1969). Some compared a population served with a population unserved (Korenbrot. 1984; Levi. Wilkinson. a Marine. 1971; Peoples & Siegel. 1983; Sokol. Woolf. Rosen. & Weingarden. 1980). Others attempted to show the improvement in the rate of adverse outcome in the general population that a program serves (Peoples. Grimson. a Daughtry. 1984; Wallace. 1978). Aubry and Nesbitt screened 1.000 ward prenatal care recipients for low income. psychological adjustment. age. marital status. medical conditions. and past problems. .A team composed of social workers. psychologists. nutritionists. and medical consultants were provided to the women as needed. The outcome was assessed in comparison to the women's previous pregnancies. Outcomes included: a lower abortion rate. lower rate of low birthweight. and lower rates Of neonatal mortality. The fact that there was a higher rate of low birthweight among infants born after'36 weeks gestation was accounted for as the program's effect of extending the gestational age of the infants. The authors concluded that the special program for high-risk mothers was effective. 34 The comparison between present pregnancies with a known risk level and previous pregnancies (an average of four) with no assessment of risk level may be inappropriate. Levy et aL.(197l) studied the effect of a nurse-midwife program on infant deaths. Nurse-midwives were assigned the care of low-risk mothers. while physicians were assigned the high-risk mothers in a county hospital. In the 2 years of the program. the rate of prematurity (defined as low birthweight) dropped. as did the fetal and neonatal deaths. When the project ended. the rates of low birthweight and fetal and neonatal deaths increased dramatically. In contrast. in the rest of the county. the rate of low birthweight increased slightly in the program period. the rate of fetal mortality remained about the same. and the rate of neonatal mortality increased. While the authors attributed the changes to the special care received. they acknowledged that there were changes in the populations of the county hospital group during the two time periods that would better account for the changes in outcome. There were also indications that during the program period. the social worker was actively involved with mothers in determining eligibility for public-income programs. The effect of these interventions was not identified. The program evaluation also identified a dramatic increase in the postneonatal mortality rate. The rate was 6.0 before the program. increased to 16.5 during the program. and decreased to 8.0 after the program. Although this might be accounted for by a change in popula- tion served by the hospital. it is likely that the reduction in the 35 neonatal mortality rate did not represent as dramatic a reduction in death rate as is apparent. Rather. the effect of the specialized medical care may have been to prolong the lives of vulnerable infants past the 28-day cut-off for neonatal mortality. Adding the rates of neonatal and postneonatal deaths. there was a reduction of 3.1 from the baseline time period to the program period and an increase of 13.4 from the program period to the postprogram period. Given that there was a 10.3 increase in postneonatal deaths from the baseline period to the postprogram period. it is likely that there was a significant change in the population served. Wallace (1978) evaluated the effect of the U.S. Public Health Department's Maternal and Infant Programs (MIC). MIC programs include care to women during pregnancy. labor. delivery. and postpartum. Services include medical. dental. social. nutritional. educational. family planning. nursing. transportation. and child care. The studies attributed the decline in mortality and low birthweight to the pres- ence of MIC projects. Although there were rather continuous declines in adverse outcome during the study periods. it appears that the pro- ject was not cl early responsible for the changes. Wallace acknowledged the conclusions of Gendell and Hellegers (1973). Morris. Udry. and Chase (1975). Pakter and Nelson (1974). and Westoff (1972). who indi- cated that the decline in infant mortality was due to the shifting age- parity distribution of births. contraceptive use. and abortion. Sokol et a1. (1980) also evaluated MICs. comparing women from MIC neighborhoods and non-MIC neighborhoods who used the same prenatal 36 medical monitoring and delivery services. MIC participants had a lower incidence of preterm deliveries and of infants weighing under 2.500 grams than women in the control group. Since the women came from neighborhoods with similar socioeconomic characteristics. he attributed the differences to the ancillary services of the MIC project: educa- tion. home visitation. nutrition assessment and counseling. social service assessment and intervention. and dental care. Peoples and Siegel (1983) assessed an MIC project in North Carolina. 'They found a reduction in low birthweight infants for the high-risk but not the low-risk population. They suggested that the reasons for the differences in effect on the two subgroups may be that high-risk clients are more responsive to MIC services or that MIC services are provided differentially to the two groups. 0B Access. a special program funded and administered by the California Medi-Cal and the Maternal and Child Health programs. was found to reduce the incidence of low birthweight (Korenbrot. 1984L. 08 Access participants who received medical. educational. and psycho- social services and vitamin supplements were compared to Medi-Cal participants who received only medical visits. Groups were matched by race/ethnicity. maternal age. parity. plurality of births. gender of infant. and county of residence. 'The incidence of low birthweight was 4.7% for OB Access and 7.1% for the Medi-Cal births. The rate of very low bi rthweight (under 1.500 grams) was 61% lower in the OB Access group. The implication is that outcome is substantially influenced by nonmedical services. 37 It may also be that there were differences in the populations. The OB Access project. although designed to serve the Medi-Cal efligible population. was set up to serve women who did not have access to medical care. It may be that even though they were all "medically indigent." there may have been unidentified differences in the popula- tion that affected the differential rates. Another project in North Carolina. the Improved Pregnancy Outcome Projects studied by Peoples et al. (1984). found an increase in the number of black women. especially teenagers. receiving prenatal care. but did not find a difference in outcome. They attributed the lack of differences to the possibility of unidentified differences in control groups. lack of protocols for dealing with high-risk women. or lack of information about the etiology of the problem of poor pregnancy out- come. Several scholars have challenged the pervasive belief that prenatal medical care makes a difference in outcome. A review of the literature (Abramowicz & Kass. 1966) revealed no compelling evidence to support the direct influence of prenatal care in the outcome of pregnancy. The authors concluded: It is uncontested that certain factors in pregnancy. such as toxemia. placenta previa. incompetent cervix and Rh incompatibil- ity. may be factors in prematurity. and are to some degree made less Operative as causes Of prematurity if effective preventive obstetrics is practiced. However. these are numerically small causes of prematurity. and it is therefore not yet clear how prenatal care can be expected to play a major part in preventing the majority of premature deliveries. The fact that rates of prematurity have not changed significantly in major obstetric institutions during the past thirty years also suggests that at present prenatal care may have only a small role in preventing prematurity. (p. 1003) 38 Hall et a1. (1980) have shown that physicians overdiagnose and underdiagnose. leaving only a small portion who are appropriately treated. Furthermore. there are only a small number of conditions that are treatable. so screening procedures such as prenatal medical care may not be as useful as it has been considered. A substantial propor- tion of the conditions that resulted in emergency procedures during pregnancy were not identified during routine office visits. Some of the screening may be helpful for obstetric care. in which case monitor- ing could be reduced to two or three visits during the course of a pregnancy (Hall et al.. 1980; Holland. 1974). Wang. The questions "Why don't some women get prenatal medical care?" and "Why don't some women get ade- quate prenatal cared" presume that a pregnant woman should get medical care. Although the data on the value of prenatal care are inconclu- sive. these questions continue to be posed by those searching for reduction of infant mortality and low birthweight. While much is written on why women do not get prenatal care. little has been said about why women do get prenatal care. One study that focused on the latter question (Perinatal Association. 1983) found that women are prompted by concern for the baby's and own health and are responding to social pressures and social norms to get care. Many women who do get what has been labeled "inadequate" care also are responding to those same concerns. Refusal to spend money to get to a clinic--even when the care is free—-may represent decisions to use 39 transportation money for rent and food. which would also indicate concerns for self and baby. Much of the literature on the reasons why women do not seek or continue care attributed the problems to characteristics of the women themselves. Women may lack information about the importance of prenatal care (Perinatal Association. 1983) or the facilities that are available to them. They may be unwilling to be examined by a male physician or have fears about physician examinations (Institute of Medicine. 1985). Some women have expressed the belief that no care is needed because they were part of a large family and observed their mothers pregnant many times. Others believe that prenatal care is for wealthy women (Wright. 1985). Poverty. lack of transportation. and lack of child care are further barriers to caret(Fox. Goldman. & Brumfield. 1968; Institute of Medicine. 1985; O'Brien 8. Smith. 1981). Monahan and Spencer (1962) characterized women who did not get prenatal care as largely uneducated. older. users Of public transportation. and women who lacked prenatal care in a previous pregnancy. The fact that a woman worried about the impending birth or had medical problems did not always motivate her to seek care. Attendance at the clinic was adversely affected by language problems. employment Of mothers. and problems in caring for other children. (rBrien and Smith compared prenatal care in hospital clinics and with general practitioners of 100 mothers of legitimate children. Those who had attended a hospital clinic were more likely to feel they had waited longer than a reasonable ti me and spent more ti me undressed 4O waiting for the examination than those who had a general practitioner. These women spent more time traveling because they had to use public transportation. They found clinic hours inconvenient. Women who received care from a general practitioner experienced fewer different physicians: 92% versus 38% of the clinic population had only one or two physicians. Women who had a general practitioner were more likely to express satisfaction with care. Watkins (1968). interviewing black. low-income married women who had at least one child living at the time the study was conducted. found that early initiators of prenatal care were younger. had carried fewer pregnancies to term. had fewer living children. and were less likely to have a child less than 1 year old. She rated each in terms of kinds and number of stresses and found no differences among the early and late initiators of care. She did. however. find differences in the timing of care. depending on their attitude toward pregnancy. .A significantly higher proportion of women who sought care in their first trimester were rated as positive or ambivalent about being pregnant. Early initiators in the Watkins study were significantly more likely to believe they were susceptible to illness or disability. Among the early initiators were the women who had experienced a fetal or infant loss. which might have prompted them to suspect that they were vulnerable and could be helped by a physician. There was. however. no difference in the proportion of early and late initiators who sought care because they felt sick. 41 In the literature on help seeking. but not specifically with reference to prenatal care. the nature of a person's social support was identified as influencing the seeking of professional services. Women with denser social networks (where network members know each other) attended fewer parent groups (Bi rkel & Reppucci. 1983) than did those whose networks were less dense. The more contact a woman had with her kin. the fewer parent-group sessions she attended. Bi rkel also found that sel f-descri bed low-competence women in dense networks were least likely to participate in food programs. Low-competence women in low- density networks were most likely to participate. with high-competence women in between. Liberman and Mullan (1978) claimed that there was no evidence that those who seek professional help are better Off than nonseekers or those who seek help from informal sources. If these assessments are correct in relation to help seeking in pregnancy. some women may be brought into care at the cost of disturb- ing some of the protective relationships without obvious benefit. However. because these groups were sel f-sel ected. and because the studies were not done in relation to outcome of pregnancy. caution should be exercised in applying these conclusions. Other literature has identified subgroups in the population that do not share common (or scientifically supported) views of how the human body functions (Johnson & Snow. 1982; Snow. 1983) and different beliefs about healing (Chrisman & Kleinman. 1980; Snow. 198). Chris- man and Kleinman claimed that members of groups who 42 come into contact with orthodox medicine that differs signifi- cantly from their traditional practices will Often react by ignor- ing the treatment prescribed. or misusing it. or complaining about the quality of care they are getting. (p. 452) There appears to be a growing number of women. principally middle- and upper-c1 ass educated women. who are getting prenatal care from midwives. through attending pregnancy and childbirth classes and by reading books on the subject. Since "prenatal" care on birth registrations has been presumed to be medical care but is still vaguely defined. it may be that women who receive midwife care are indicating that they received prenatal care by trimester start of care and number of visits. Some of these women are also giving birth at home with husband or midwife as attendant at birth. The fact that the definition of prenatal care is ambiguous may account for the fact that there has been no decline in prenatal care. whereas there has been a significant decline in the number of hospital. physician-attended. births (Advance Report. 1984a). Structural barriers to care have been identified as time delay in certifying a woman for Medicaid (Perinatal Association. 1983; Institute for Medicine. 1985). lack of providers. ineligibility of some women where loss of employment leaves them without health insurance coverage yet amount of resources leaves them ineligible for Medicaid (Prenatal Care. 1984). high malpractice insurance costs for some providers. lack of legal authority for midwives to practice and collect third- party payments. and lack of resources and facilities by health depart- ments. Another major barrier is the lack of cultural sensitivity of medical personnel. Some women do not seek or continue in care because 43 of lack of understanding of their needs and customs (Institute for Medicine. 1985). Another barrier is the lack of legitimacy for alter- native forms of prenatal care outside the medical system: classes. midwife care. family care. and television and radio counseling. Access to care has been enhanced by assertive outreach: radio and television advertising in several languages. free hot-line for women with problems. and policies of requiring prenatal care to receive WIC food supplements. layettes and other tangible benefits. free transpor- tation and child care. and support groups. It seems clear that the target group is low income. Research is needed to verify if getting women into care will make a difference in outcome. or if substantial enhancement of living conditions will be required. Wags. Program documents assessing the need for programs of prenatal care are included in this review to show how the literature on infant mortality and low bi rthweight is being used. Documents serving as a basis for policy for the State of Michi- gan include "Barriers to Early Prenatal Care in Michigan." published by the Perinatal Association of Michigan; "Meeting the Problem of Infant Mortality: A Plan of Action for Michigan" and "Prenatal Care: A Healthy Beginning for Michigan's Children." both published by the Michigan Department of Public Health; and "Estimating the Need for Prenatal Care." prepared by University Associates. Lansing. Michigan. The last includes an extensive literature review. "Current Perspectives on Pre- natal Care" (Bergquist. 1984). 44 "Barriers to Early Prenatal Care in Michigan" is a 1983 study compiling vital statistics about prenatal care in the state. The study showed that: --The rate of low birthweight is slightly higher for mothers under 18 than mothers 20-29 with no prenatal visits (9.8% compared to 8.5%). For 13-14 visits. the percentages are 3.6 and 1.9. respec- tively. --Unmarried women have a higher percentage of low bi rthweight babies at every level of prenatal care. --There are more black and unmarried women whose mothers had fewer than five prenatal visits. --The mode for prenatal visits was 12. with 59.6% of all women having between 10 and 14 visits. . --Fetal death rate and low birthweight varied inversely with the amount of prenatal care. --A higher percentage of women with an eighth-grade education or less received no prenatal care or began care after the first trimester. --Fifty percent Of the women under 15 did not receive prenatal care or got no prenatal care in the first trimester. --The amount of care generally decreased with increasing birth order. The study cited the March of Dimes report. which concluded: Women at greatest risk of having low-weight babies are non-white. age 17 or younger. have less than eight years of schooling and are unmarried. Yet these groups characteristically have the fewest prenatal visits. 45 The only major research study cited to show a correlation between prenatal care and outcomes was that of Tokuhata et ah» (1973L Tokuhata et al.. as mentioned above. did not control for income or socioeconomic status. a major factor that is frequently associated with both poor outcome and the receipt of prenatal care. "Meeting the Problem of Infant Mortality: A Plan of Action for Michigan" (19$) described the situation with regard to unemployment in Michigan and the association of unemployment with infant mortality. The document suggested that prenatal care was a mediating factor between unemployment andlnortality because unemployment results in a large number of people lacking medical insurance. The unemployed. it explained. frequently have too many resources to qualify for Medicaid. During the early 19805. many unemployed did not seek care or were turned away from hospitals for lack of ability to pay. The number of women arriving at hospitals as "walk-ins" for delivery having had no prenatal care increased. Funds to some centers were cut and services to some hospitals were jeopardized because so many were unable to pay. The document assumed the usefulness of prenatal care. then proceeded to provide evidence that there are many women lacking the service and that it would be cost effective for the state to provide it. "Prenatal Care: A Healthy Beginning for Michigan's Children" asserted the importance of prenatal care: There really is no mystery concerning what is needed to increase birth weights and reduce infant mortality. Since the 19205. the medical literature has been replete with references to the fact that early comprehensive prenatal care increases birth weight and decreases infant morality. (p. 18) 46 Thus what has appeared in the literature as correlation becomes causation. "Estimating the Need for Prenatal Care" (1984) is an assessment of the number of women in Michigan who are without prenatal care because of lack of insurance. low income. and inaccessibility of services. The author estimated that there are 9.500 women in Michigan who do not get care and are in need economically and medically. The only source of evidence of a relationship between care and outcome is Ryan et a1. (19$). whose research supported a correlation between low birth- weight. still birth. neonatal mortality. and the number of visits to a physician. The literature review that accompanied the University Associ ates' report. like "Barriers to Prenatal Care." showed that Michigan women at greater risk of poor outcome are less likely to start prenatal care early and have fewer visits to a physician. In addition. two studies were Ci ted by the author as evidence of the relationship between prenatal care and outcome (Greenberg. 1983; Ryan et al.. 1980). The author of the review included the cautionary notes from Ryan et a1. and Greenberg. who reminded the reader that those who do not receive prenatal care are those already at risk for poor outcome. Greenberg's study controlled for race and education. However. he acknowledged that he did not control for age. smoking. parity. prior low birthweight. or other possibly confounding factors. Two documents that are likely to have an impact on the direction of programs at the federal level are reviewed here. They are 47 "Preventing Low Birth Weight." the report of a study of the Institute of Medicine (1985). and the report on Community Health Centers (Schwartz 8. Poppen. 1982). Nine of the 12 major studies on low bi rthweight reviewed here (see Figure 2.1) are cited as evidence of the strong relationship between prenatal care and bi rthweight. Except for Rosenzweig (1982). none of the studies controlled for two important variables. alcohol and smok- ing. Five of the studies did not control for parity. Each of these studies was qualified in this review to show that. at best. there is a correlation. but not causation. implied in these studies. At worst. prenatal care is a proxy for something else. It is. perhaps. a more sensitive indicator of socioeconomic status. Program evaluations would seem to be better evidence of a causative or amel iorative relationship between prenatal care and outcome. The studies by Korenbrot (1984). Levy et al. (1971). Peoples and Siegel (19$). and Sokol et a1. (1980) identified in the Institute study as supporting the ameliorative effect of prenatal care were discussed above. Although the document clearly stated the problems with the methodology of prenatal care and outcome studies. and although the document cited literature identifying qualifying factors in the correlations between prenatal care and outcome. the authors concluded: The overwhelming weight of the evidence is that prenatal care reduces low bi rthweight. This finding is strong enough to support a broad. national commitment to ensuring that all pregnant women. especially those at medical or socioeconomic risk. receive high- quality care. (p. 146) 48 >~ U m m «H H 3 ~- L U' c m m o O 0 "' U '0 '- 01 0- E > m < u > t: O O I l I m u .— .c L. u u u n u .- .x o O o «H U U U D G) L. 2 U U .C 3 Z Z Z 'O 01 (O '- (U H O (L o. o. u.) < O. (I) < o: O Eisner (1979) BW x x x x x x Elster (1984) GA x x x x x Gortmaker (1979) BW x x x Greenberg (1983) BW xa x x Harris (1982) BW x x x Kessner (1973) ALL x x x x x x Lewit BW (1983) GA x x x x x Rosenzweig (1982) BW x x x x x Showstack (1984) BW x x x x Key: BW = birthweight; GA = gestational age; ALL = birthweight. neonatal. postneonatal. and infant mortaltiy. aGreenberg measured prenatal care by some/no prenatal care. Figure 2.1. Schematic representation of factors tested in major birthweight studies cited by Institute of Medicine (1985). 49 The authors added: Because content of prenatal care is not defined carefully in many of the studies reviewed. it is not possible to trace the benefits of care to specific aspects of the total care package. (p. 146) The authors acknowledge promising interventions in other fields. Income-maintenance programs have been shown to reduce the incidence of low birthweight (Kehrer & Wolin. 1979); early childhood education programs (Weikart. 19$) have been shown to reduce teenage pregnancy. which has been shown to be a factor in the reduction Of mortality. These kinds of solutions. they maintained. lie outside the purview of the investigation committee (11 of the 12 members were either physicians or associated with university hospitals or medical institutionsL "The inquiry has been limited to interventions in the general domain of health services--an appropriate focus given the composition of the committee and its charge" (p. 115). The committee equated health with medical services. rather than considering any programs that enhance outcome of pregnancy as a health outcome. Reducing poverty and improving education could do much to decrease low birthweight. given the strong associations among birthweight. socioeconomic status and education. . . . but such perspectives are not explored in this report. (p. 115) The authors of the report on Community Health Centers (Schwartz & Poppen. 1982) were more cautious: . . . it is commonly asserted that prenatal care has an impact on pregnancy outcome. the literature shows that it has been difficult to demonstrate a strong relationship between care and outcome. This is the case for a number of reasons: there are many vari- ables which affect outcome; studies have had problems adequately controlling for these variables; data have not been routinely 50 available on variables such as alcohol abuse. smoking. and nutri- tion; and. finally. multi-variate analysis has only recently come into widespread use. (p. B-34) Russell (1982) in her examination of the "Short Report" (Second Report. 1980) noted some parallels. The Short Report. a British study. claimed it became involved "because of mounting public concern that babies were unnecessarily dying or suffering permanent damage during the latter part of pregnancy and the earliest part of infancy" (p. 1). Russell responded that there is a concern. but that it is not about death or handicap. She said: Dissatisfaction with existing patterns of reproductive heal th-care has been widely expressed. not only amongst the articulate middle class. but. given chance of expression. amongst working-class women too. The development of the current debate on perinatal mortality at the same ti me as these movements toward social change is not. I would argue. a coincidence. nor are apparent contrac- tions inexplicable. They are the result of a deliberate pol icy. not about babies. but rather the role of women. (p. 315) She claimed that the data presented to support the claims that babies are "unnecessarily dying or suffering . . . may be unhelpful. and at worst. misleading" (p. 304). If the issue were babies' health. the interventions could be different. she claimed. The commission that wrote the report acknowledged that the interventions might be directed elsewhere: Perinatal and neonatal mortality are twice as high in the lowest socioeconomic classes as in the highest. . . . We recognize that in the long term raising the standard of living is the best remedy for overcoming the problems of babies of socially disadvantaged mothers . . . but suggest that much of the death and handicap suffered by these babies . . . can be overcome by well applied medical intervention. (pp. 158-159) If there were a causal relationship between prenatal care and outcome. or if prenatal care. particularly with nutritional and parent 51 education and social services. could overcome the handicap of socioeco- nomic conditions. age parity. nutritional deficits. and whether the programs are implemented. then babies and families would benefit in terms of better health. If the relationship is correlational. with prenatal care acting as a proxy. perhaps as a more sensitive indicator than education or income. then the medical facilities and public health departments will benefit. Unemployed people may be employed to transport women to public facilities. to dispense WIC coupons. and to provide child care. Social work services may become better established within health programs. WWW Other factors said to influence the health of the fetus and infant are mother's weight and weight gain. maternal smoking. consumption of alcohol and drugs. health conditions. attitude toward pregnancy. and stress. They are briefly reviewed here. 1319111. Prematurity was lowest among women whose pregravid weight showed them to be overweight. highest among women whose pregravid weight showed them to be underweight (Terris 8 Gold. 1969). Naeye (1979b) and Eastman and Jackson (1968) claimed that weight gain in the second and third trimesters is critical. Naeye (1979b) investigated the relationship of nutritional status to perinatal mortality with a formula of optimal weight gain. He found that women who were at lower than desirable pregravid weight had the lowest rate of infant mortality if their pregnancy weights were 80 to 52 120% of optimal values and the highest infant mortality when they were less than 25% of optimun. Overweight women had the lowest infant mortality when they were at 24 to 54% of Optimal values. While outcome was less affected by variations in weight gain. overweight mothers had higher infant mortality rates in every category of weight gain. Smoking. Maternal smoking of cigarettes has been found to to be associated with increased risk of perinatal mortality. spontane- ous abortions. congenital anomalies. birthweight. and a number of pregnancy complications (abruptio placentae. placenta previa. bleeding. and ruptured membranes) (Davies. Gray. Ellwood. 8 Abernathy. 1976; Deibel. 1980; Meyer 8 Tonascia. 1977). Infants of smokers are not only lighter but.have smaller head circumference and are shorter. and have lower Apgar and Bayley scores. Rosenzweig and Schultz (1982) estimated that smoking a pack of cigarettes a day during pregnancy lowers birth- weight by 10 ounces. Since the infants of smokers have a normal layer of subcutaneous fat. smoking is thought to affect the infant through hypoxia (deficiency of oxygen reaching the fetus) and not through nutritional deficiencies (D'Souza 8. Richards. 1978). ‘DLugs. Unfortunately. studies on smoking have not controlled for alcohol consumption. which may account for some of the adverse effects. Alcohol. according to Harlap and Shiono (1980). may harm the fetus not only when alcohol is abused. but when taken in moderation. Two drinks per week raise the likelihood of spontaneous abortion by 25%. Drugs have been implicated in congenital anomalies and mortality. Prescription drugs. nonprescription drugs. and illegal drugs have all 53 been found to have teratogenic effects on the fetus (Hill 8 Stern. 1979; Moore. 1981). Drugs affect the fetus through the placenta. A number of factors affect the way in which a specific drug affects the fetus. Molecular weight. fat solubility. and the degree to which ionization occurs and the degree to which the drug binds to albumin in maternal blood affect the transfer of drugs through the placenta. Maternal acid-base balance and placental blood flow also may influence transfer. If the mother hyperventilates and blood pH rises. the fetus becomes acidotic. Drugs crossing the placenta may become trapped inside the fetus. Genetic factors may also affect the transfer of drugs (Moore. 1983). Disease. Two maternal conditions of particular concern are heart disease and diabetes. Hypertension is present in 7% of pregnancies in the United States (Wilson 8 Matzke. 1981). It is one of the leading causes of maternal death. fetal morbidity. and mortality. Friedman and Neff (1978). in a study of 38.636 cases. found that a maternal dia- stolic pressure of 841nm Hg at any time during pregnancy increased the risk of fetal mortality. Hypertension and protenuria combined increased the risk further (Naeye 8 Friedman. 1979L Diabetes is a systemic disease characterized by abnormal glucose metabolisnu Like hypertension. it is a chronic condition. but may also develop as a result of pregnancy. At one time. a majority of infants of diabetic mothers died (Weingold. 1981). Mortality has been reduced to 15% due to insulin diabetes management. Unrecognized gestational 54 diabetes. about 3% of the pregnant population. is a major problem (Barden 8 Knowles. 1981). While most of the literature on health and adverse outcome of pregnancy deals with physical health. a smaller number of studies investigate the relationship of mental health and outcome. They are categorized here as attitude and stress. We. Several studies have attempted to measure the attitude toward pregnancy and to assess its influence on outcome. Laucaran (1980) found that mother's attitude toward pregnancy was asso- ciated with perinatal death. congenital anomalies. postpartum hemor- rhage or infection. injuries. and psychosocial conditions of the mother. Birthweight and prematurity seemed to be unaffected by atti- tude. Frenzen and Hogan (1982). investigating families and mortality in Thailand. found that parents' attitude toward pregnancy was signifi- cant even after controlling for social class. Where parents wanted the child. there was less risk of mortality regardless of the ability of the family to provide. Life change events have been shown to rel ate to illness (Holmes 8 Rahe. 1967; Lipowski. 1973; Masuda 8 Holmes. 1967) and to pregnancy outcome (Nuckolls. Cassel. 8 Kaplan. 1972; Schwartz. 1979). Nuckolls found that neither life change nor psychosocial assets were significantly related to pregnancy complications. However. when life change score was high both before and during pregnancy. women with high psychosocial asset scores had only one-third of the complications as the women with low psychosocial asset scores. Women who had a low 55 birthweight baby were found to have experienced considerably more minor and major life changes compared to a control group of infants of normal birthweight (Schwartz. 1979). Much of the information gathered about the conditions of*women during pregnancy is not based on birth registrations because informa- tion on weight. smoking. drug consumption. and disease is not recorded. Much of the information. therefore. comes either from mother's report or from medical charts. Unfortunately. most studies. perhaps because these data require much work to obtain information. do not control for some major factors. Thus. answers to questions of whether the effect Of drugs causes reduction in mother's weight. or whether alcohol acts in combination or acts as proxy for smoking. are unknown. Even if a random clinical trial could be implemented. these major factors would have to be considered. W With a few exceptions. the literature has identified a strong relationship between socioeconomic status and outcome of pregnancy in spite of a wide variety Of operational measures and a lack of clarity in the conceptual definitions. Education. occupation. employment status. and income have been consistently identified as affecting mortality and birthweight. Education is the most used measure. It is relatively stable and available. Not only is maternal and paternal education noted on birth certificates. but it is available in public records. Maternal educa- tion has been found to be associated with outcome by Adelman (1963). 56 Chase and Nelson (1973). Ekanem (1972). Kessner et a1. (1973). and Stickle and Ma (1977). Erhardt and Chase (1973) noted that there is a sharp drop in the proportion of low birthweight babies of women who have between 1 and 3 years and those who have 4 or more years of high school. Foreign-born white*women have a lower rate of low birthweight than native-born white women at each level of educational attainment except for 4 or more years of college. where the rate is similar. Brooks (1975) found. in a comparison of factors related to infant mortality in 1938 and 1968. that the importance of the number of school years completed had not changed. Education may not be an accurate measure for very young women. Since most young adolescents have not completed their education. this measure may not represent their family (parental) income or socioeco- nomic level. Thus some not-at-risk women may be miscategorized as at- risk. Occupation has been measured by percentage of white—collar workers in a given population (Brooks. 1975) or percentage of technical workers (McCormick et al.. 1979). In Frenzen and Hogan's (1982) study on mortality in rural Thailand. social class was divided into three groups by whether the father was in a nonagricultural occupation (highest class) or whether he was a farm owner Uniddle class) or a farm laborer (lowest class). Willie and Rothney (1962). studying mortality in Syracuse. used occupation. as well as education and housing informa- tion. to create an index of socioeconomic status by census tract. 57 Donabedian. Rosenfelt. and Southern (1965) included income. education. and occupation. Occupation alone and occupation as a contributing factor in social class are considered to be associated with adverse outcome (Frenzen 8 Hogan. 1982; Hemminki. Niemi. Saloniemi. Vainio. 8 Hemminki. 1980). Willie and Rothney (1962) found that neonatal mortality is more asso- ciated with family income characteristics. but postneonatal mortality rates with socioeconomic status. The correlation of coefficients was low. however. and accounted for less than 15% of the variance in the distribution of mortality rates. McCormick et a1. (1979) found that the percentage of technical workers in the population accounted for very little of the variance in mortality. Willie and Rothney (1962) and Donabedian et a1. (1965) found little association of SES and infant mortality. The employment of women. although at one time considered to be an important factor in outcome (Dyhouse. 1982). is found now to be more a function of social class rather than employment per se. Gofin (1979) found a higher rate of women in heavy employment activities experienced abnormal conditions. but he considered this more a function of heavy physical strain than employment per se. It may also be the added stress of coping with larger families on smaller incomes. Unemployment. however. has been found to be associated with infant mortality. Brenner (1973) found that there are differential Tags in fetal and infant mortality. Increases in fetal mortality rates generally occur within the same year in which the national economic 58 indicator declines. Increase in infant mortality under 1 day tends to lag 1 year behind any given year of an economic downturn. Postneonatal mortality. however. tends to lag from 3 to 5 years behind any given year of an economic downturn. Shin (1975b) had similar findings in a study of 63 countries. Stockwell (1962) divided the census tracts of Providence. Rhode Island. into socioeconomic areas based on occupation. education. and income. He found that although income and crowding do correlate modestly with infant mortality. based on his measures. the highest SES area had the highest infant mortality. However. although Stockwell did not note this. the two highest areas. I and II. combined had lower rates than Areas II. IV. and V. He claimed that changes are occurring in the established inverse relationships because of medical care. Willie and Rothney (1962) also found no association between neo- natal mortality and SES (based on occupation. median school years com- pleted in adult population. average monthly rental. average market value of own homes. and percentage of single-family dwelling units). However. within the lowest socioeconomic area. they found that income distin- guishes survival and mortality. In the very lowest income levels. they found high rates of mortality. These lowest levels were all white. but of different ethnic backgrounds. The Polish and Italians had higher incomes and lower rates of mortality. The native-white area not asso- ciated with any ethnic group had lower incomes and higher rates of mortality. This would appear to be related to income. but they also found that among census tracts of the same income level. native blacks 59 did not have a higher rate of neonatal mortality than native whites. Willie and Rothney conjectured that there is a critical level of income. below which an infant is at considerably higher risk. It is possible. howewer. that the critical factor among whites was the tight community bonds among Polish and Italians compared to native whites. Theronly epidemiological study located that used per capita income was that of Altenderfer and Crowther (1949). They found that in cities of 100.000 or more. infant mortality varies inversely with per capita income and with the percentage of white persons in the population. They suggested. however. that this association is a consequence of the association of mortality with the percentage of births hospitalized for labor and delivery. They concluded that hospitalization at delivery appears to be important in preventing infant deaths.' Negative income transfers have been shown to be associated with birthweight (Kehrer 8 W01 in. 1979). While most studies do not con- sider family income in relation to family size. this program guaranteed each family a certain income above the poverty line. based on number of family members. Mothers who were at high risk for experiencing adverse outcome of pregnancy had infants who were higher in weight than a control baby. The weight gain was 0.3 to 1.2 pounds. Because there did not appear to be a significant increase in prenatal care. the impact was attributed to improved nutrition of the mother. The exact mechanism cannot be ascertained. The guaranteed income may have. in fact. acted to reduce stress. 6O Crowding may be a factor that relates closely to economic resources. Some studies in animals have indicated that crowding causes physical harm. self-destruction. or failure to reproduce. Studies in humans have had mixed findings. Both Cassel (1970. 1971) and Dubos (1965. 1968) suggested that crowding is a factor in the interplay of human social and physical factors and the way that stress affects the organism. making a person more susceptible to disease. Other studies have shown that humans adapt to crowded conditions. Immediate reac- tions to overcrowdedness may differ from later reactions. 'Thus. studies that measure only the former may conclude a negative effect of crowding. while longitudinal studies may show coping mechanisms. Overcrowding (measured by ratio of persons to rooms in social planning units in New York City) was shown to be a factor in both low birthweight and infant mortality (Streuning. Rabkin. Cohen. Raabe. Muhline. 8 Cohen. 1973). Galle. Gove. and McPherson (1972) found crowding to be directly associated with infant mortality in Chicago. Streuning et a1. asserted that the association with poverty and over- crowdedness is greater in white neighborhoods than in heterogeneous neighborhoods. suggesting that other factors may be important in non- white areas. The degree to which crowdedness is independent of poverty needs to be clarified. The measures of SES and education tell us little about the actual familial and social processes that affect the child (Kitchen et a1» 1982). Osofsky and Kendall (1973) summarized the literature on eco- nomic security in the following way: 61 Low income individuals. who are more likely to have early. fre- quent. and more numerous pregnancies extending into later life. are more likely to have premature deliveries with the increased risk of perinatal mortality. and are more likely to have the medical complications of pregnancy which predispose either to perinatal mortality directly or to premature and small-for-dates infants with their increased risk of mortality. (p. 108) SummLtbLLnsLatuts The literature on infant mortality tends to find that women whose infants die are at the extremes of the childbearing ages. have had more pregnancies. and are more likely to be unmarried. They. likewise. have had less prenatal care--if only because the gestational ages of the infants are shorter. These women start prenatal care later and have fewer than the recommended number of prenatal visits. It also appears that mothers who have the most prenatal care visits are more likely to lose a child. perhaps because the infant is already in danger and the woman is under close medical supervision. Women who lose their babies are said to be more likely to be underweight or overweight prior to pregnancy. and are more likely to smoke. take drugs. or have major health conditions such as diabetes or hypertension. Finally. hoseholds of infants who die are more likely to be economically deprived. Par- ents are likely to have less education than parents of infants who live. They are more likely to be lower socioeconomic status. are more likely to have lower status jobs. and have less income. In spite of numerous challenges to the findings. this remains the profile of the households of the infants who die. Challenges to these findings include reassessments of the critical factors when factors such as maternal age. parity. and prenatal care are controlled for 62 socioeconomic status. suggesting that it is economic deprivation that remains the critical factor in infant mortality. Other researchers have suggested that the importance of socioeconomic status is declining due to general health and well being of the general population. Birth- weight remains one of the most critical factors that differentiate the infants who die from the infants who live. At very low birthweights. infants die at about the same rate. regardless of their socioeconomic status. CHAPTER III CONCEPTUAL FRAMEWORK AND METHODOLOGY This chapter provides a rationale for the development of a new conceptual framework for the study of infant mortality. This framework is based on the household and consists of variables representing char- acteristics of and relationships within the household. The conceptual frameworks of Bronfenbrenner (1979) and Paol ucci et a1. (1976) described in Chapter I provide the basis for the reconceptualization. In addition. this chapter identifies the methodology of the study designed to test the household variables. .Ihfi.BflIiQnil§ There are two general types of studies on infant mortality. based on the source of the data: geographic entities and individuals. .Geeatanbis_snt1ties. These studies consider the attributes of blocks (Lewis. Charles. 8 Patwary. 1973). census tracts.(Donabedian et al.. 1965; Shah 8 Abbey. 1971; Stockwell. 1962; Willie 8 Rothney. 1962). cities (Altenderfer 8 Crowther. 1949). states (Brenner. 1973; Brooks. 1975. 1980; Moriyama. 1960). and countries (Shin. 1975a). Census data and vital and health statistics are frequently used. The advantage of this type of study is that broad geographic areas of risk can be identified. Individuals. however. cannot be identified. 63 64 Geo raphic Enfiles Individual Household Figure 3.l.--Schematic representation of the literature on infant mortality. Individuals. These studies consider the attributes of individ- uals. usually the mother who has given birth. To some extent the attributes of the fathers and infants are described. The most usual source of the information about individuals is birth registrations and birth-death records. Frequently these studies include thousands or hundreds of thousands of births (Amante 8 Brandt. 1978; Chase 8 Nelson. 1973; Eisner et al.. 1979; Erhardt 8 Chase. 1973; Kessner et al.. 1973; McCormick et aL» 1984; Rosenwaike. 1972; Stickle 8 Ma. 1977; Yerushal my et al.. 1965). Naeye's (1979a. 1979b) work on weight gain came from the Collaborative Perinatal Project. a prospective study of 33.318 pregnancies in 12 lLS. hospitals between 1959 and 1966. The 65 Rosenzweig and Schultz (1982) study came from the National Natality Followback Survey in the years 1967-1969. Other studies have combined birth certificate data and interviews (Shapiro et aha 1980). Knodel and Hermalin (1984) used birth records from 18th and 19th century German villages. Data from these studies have been used to create profiles of individuals at high risk for pregnancy outcome (Hobel. 1978; Sokol. Rosen. Stojkov. 8 Chik. 1977). Unfortunately. birth records are not always complete. While some data (such as birth weight) are usually in the records and accurate. other data are not. Information about the father is frequently missing. Gestational age. if given. may be inaccu- rate due to the difficulty of assessing the weeks of gestation through observation and the inaccuracies associated with the reporting of the last menstrual period. Missing data are sometimes counted with no prenatal care or low socioeconomic status (Gortmaker. 1979; Harris. 1982). In other cases they are counted separately or excluded (Chase 8 Nelson. 1973; Shapiro et al.. 1980). Inaccuracies in the reporting of prenatal care have been mentioned previously. While there is abundant information on the characteristics of geographic entities and individuals. there are no studies known that have used the household as the source of information and as the basis of variables in studies on infant mortality (Figure 3.1). Two studies are described here to illustrate how household information might be conceptualized and used as variables in such studies. Osterweis. Bush. and Zuckerman (1979) found that the family unit is a key determinant of 66 individual health and illness behaviors. In a study of 2.700 individu- als living in 800 families. using hierarchical and stepwise multi- variate regression. they found that the set of family context vari- ables--size of family. income. percentage females. percentage children in family. average family morbidity. and average other's nonprescribed and prescribed medicine use--is a better predictor of individual medi- cine use than the set of individual variables--age. sex. race. and morbidity. Some variables. such as family size. that have been shown previously to be significantly related to medicine use are not impor- tant in the presence of other variables in multivariate analysis. A second study concerns the illnesses of school children. Using components of the construct of competence developed by Foote and Cottrell (1955). Boardman. Zyzanski. and Cottrell (1975) found a strong negative association between school absences (proxy for illness) and the selected measures of family competence: mother's assessment of family participation. creativity. sel f-confi dence. commitment. judgment. pride in family. and communication. These factors remained important even when social position. chronic illness. father in the home. family size. age of mother. belief in education. and race were considered. Whereas most of the literature deals with attributes or statuses of geographic entities or individuals. there is little reference to processes. or. in Bronfenbrenner's terms. role and relationships. Boardman et al. used family processes (described above) as variables to predict illness. Several other studies are presented here to 67 illustrate the kinds of processes that might be relevant in the inves- tigation of infant mortality. Zuckerman. Winsmore. and Alpert (1979) compared primiparous ado- lescents and older mothers on their attitudes. maternal self-image. and social support systems at 2 weeks and 3 months postpartum. They noted their receptivity to advice or help from social workers. nurses. tele- vision. radio. and books. They found that. in comparing primiparous adolescents. primiparous women over'Zl. and multiparous women over 21. there were no differences in their attitudes toward their baby. disci- pline. attachment. or investment in their babies. At Time'l. adoles- cent mothers expressed insecurity about their maternal role and self- image. but at Time 2. no differences were noted. The adolescent mothers typically lived in an extended family. used family members as support. and were more likely to seek help from their mothers rather than professional help. At Time 2. adolescent mothers were more likely to have someone to share infant care. There was no difference in the willingness to receive advice or help from social workers. nurses. television. radio. or books. or their plans on how they were going to raise their children. Unfortunately. only women with good outcome of pregnancy were included. Podell (1972) surveyed mothers on welfare in New York City. He found that. regardless of ethnic group. women with more education. organizational participation. and media exposure were more likely to use preventive care services for young children. Although Puerto Ricans and Blacks were less likely to receive prenatal care during the 68 initial trimester. they were more likely to make use of preventive care services than Caucasians. Warnecke. Graham. Mosher. Montgomery. and Schotz (1975) found that. among blacks in Buffalo. New York. those who discussed health with close relatives were more likely to use available health services than those who did not discuss health with others. Besides discussion. need was also important. 'Those with children were more likely to use health services. Attendance at public meetings appeared to be limited to those with specific needs and concerns. The authors cautioned that it may also be that those most concerned about health were those most likely to have discussed health matters with health guides. W In this section. some traditional variables are reconceptualized in terms of the household unit of analysis and in keeping with a family ecological approach. (See Figure 3.2.) Conceptual changes in the variables related to pregnancy history. prenatal care. and economic resources are made. 'Traditional variables as well as household vari- ables in all four areas are tested in the study. (See Table 3.1J Wm Some of the variables relating to pregnancy history are trans- formed into household characteristics. While all of the traditional variables--age. parity. marital status. prior fetal and infant losses. race. and sex--are retained in this study. parity and marital status 69 .Lonuoe ecu new v_o£om:oc ecu mc_ccou:ou mLOuomm mo awcmco_um_oL on» we co_umuc0moLaoL u_umsocum "N.m ucam_u toaonm .0 9:09: wooScw $250 Lea coszLoLE EmE>oEEm Lo cozmoacw wo_:o>> woosom 920m“. .mo_mo_ozo>mo se.. 8:...me ommoflo two... 60% .0 RE: 2982, seems. maumum 5.3m 235 .8622 Emcoco 50.305 meccmcmoco Lo .8832 mmcu E25 Lo .99. Lord owe. B5992 waEmcozflmm mum 202330... 20:98:01 30:82.0...— 70 Table 3.1: Traditional and household variables used in NICU project study. Traditional Household HEBREW Age Number of pregnancies Prior fetal 8 infant losses Marital status Race (controlled) W Number of medical visits Trimester of start of medical care Manama W Weight Cigarette smoking Drug consumption Disease Psychological factors W Education Job status Income Number of persons in household Number of preschool children Number of school-age children Presence of husband/partner Planning of pregnancy Change of residence Crowdedness of household Presence of pets Mother got instruction on diet. exercise Mother got instruction on drugs. smoking Mother followed physician instructions Number of different physicians Helpfulness of physicians Barriers to prenatal care Number of sources of support Number of sources of information Support from husband/partner Separation from husband/partner Arguing with husband/partner Arguing with others Life satisfaction Lay-off from work Receipt of benefits Childhood economic deprivation Per capita household income 71 are transformed into household composi ti or. Not only are the concepts changed. but the operational measures as well. The basis for the change in the concept of parity from the number of live births a woman has had to the concept of children in the household is research by Knodel and Hermalin (1984). Their finding that it is not birth order (parity) but completed family size that best predicted survival or mortality in 18th and 19th century German fami- lies suggests that the amount of resources available per child may be more important in predicting outcome than the biological wear and tear on the mother. which is implied by the parity concept. It is anticipated that. even'though completed family size cannot be controlled in a cross-sectional study such as this. present house- hold size may be a better predictor of outcome than parity. Divorce. remarriage. and out-of-wedlock births often mean that household size is not synonymous with parity. A nulliparous woman can live in a family of six because she is unmarried and living with her family of origin. A woman who has had six children may be divorced and have custody of three children. or she may be remarried and have eight children in the household. Marital status. highly predictive of infant outcome. may repre- sent more than a legal document. Millard (1985) reasoned that marital status represents economic status. Single status at first pregnancy probably reflects the inability of a Mexican family to provide the requisite wedding ceremonies. Single status at the time she inter- viewed the mother (post-reproductive age) may represent the loss of a 72 wage earner. It may also reflect the presence or absence of a loving and supportive partner. Since economic status is already controlled. the variable is conceptualized in terms of presence of a husband/ partner and the degree of separation from him. In the absence of a husband/partner. extended family is included in this concept. Legal status is expected to be relatively unimportant compared to the absence of a significant household other. or economic variables. Several new variables emerge from the concept of household: planning the pregnancy. pets in the household. the degree of crowdedness of the living space. whether or not the household has moved. and the quality of relationship with the husband/partner. The planning of pregnancy is seen as the desire to add to the household. IIt is assumed that a household would not plan a child unless resources were expected to be available to nourish the fetus as well as the entire household. Thus the households that did not plan the pregnancy are expected to have the less desirable outcome. Pets are frequently part of the household. Although they. too. use household resources. they may serve as social support. helping to reduce stress. Thus it is expected that the presence Of a pet will enhance outcome. The degree of crowdedness indicates a lack of resources for household members. The operational measure for crowdedness is the number of bedrooms and bathrooms divided by the number of persons in the home. Although it may be a proxy for deprivation or lack of income. the lack of space itself may negatively affect outcome. Moving 73 from one residence to another. even where conditions are better. is expected to be stressful and adversely affect outcome. This study views the household composition not only in terms of who is there and how much physical space they have. but in the quality of the relationships. This study asks how much members argued about money and other things and asks the mother to rate her relationship with her husband/partner. W In this set of variables. it is the concept of prenatal care and health that is being changed. rather than individual variables within the set. Most of the literature on prenatal care is concerned with quantitative aspects of care: trimester of start of care and the number of visits to a physician. This study expands the notion of prenatal care to include the qualitative aspects of medical care. The woman is asked if she got instruction on diet. smoking and drug con- sumption. and the degree to which she adhered to her doctor's instruc- tions. She is also asked how many different doctors in how many settings provided care as a way of measuring continuity of care. Sev- eral doctors in different settings would likely not have access to each other's records. It may be that even the examinations by different doctors within the same settings may not provide the same quality of care as having the same physician. Her rating of the usefulness of her physician is likewise a measure of quality of care. 7h Another ramification of the quality of care is the difficulty or ease with which she was able to secure prenatal medical care. Thus the search for factors that impeded this care may be useful. This project explores the concept of prenatal care as the care and nurturance of the mother and fetus by household and community. The noninstitutional sources of care--husband/partner. kin. friends. and other support persons--are investigated. In addition. sources of information regarding pregnancy--radio. television. magazines. classes. friends. neighbors. and medical and nonmedical professionals--may likewise be important. .MOthaLLs_uealtb_0un1na_Enfinnnnsx The concept Of health. traditionally measured by weight. smoking. drugs. disease. and psychological factors. is reconceptualized in terms of household factors. Rather than focusing on the mother's physical and emotional health. the focus is on the emotional climate in the household. Thus. the extent of separation of the woman from her husband/partner. the amount of help and support she received from him. her assessment of her relationship with him. the amount of arguing about money and other things with her husband/partner and others. and her assessment of her satisfaction with life are assessed to determine the emotional health of the household. A low degree of separation and arguing. a high degree of support and life satisfaction. and a good relationship with husband/partner are expected to be associated with survival of the infant. 75 W In this set of variables. the concept of "socioeconomic status" frequently found in the literature is limited to economic resources in the household. Income'is conceptualized as household income per capita. The traditional variables--father's and household income-- place households of any number of members into the same risk category just because they have the same income. Father's income is also inade— quate because the household may have other sources Of income. Using thisuneasure. households would be placed in the same risk category whether the mother did not work or whether she had substantial earn- ings. It is apparent that. in attempting to measure economic resources. studies using father's or household income may have grossly miscategorized respondents. The literature on unemployment does not tell us.if it is the unemployed households that are at risk or if unemployment in the community leaves some employed more destitute. while many of the unem- ployed have the cushion of unemployment insurance to shield them from adverse outcome of pregnancy. This study seeks to know if it is the unemployed who are at risk of adverse pregnancy outcomeP-if it is the absolute loss of income or a relative loss of incomelthattcreates the risk. In addition. we inquire if the mother feared a lay-off. or if she worried or argued about money with her husband/partner. predicting that stress relating to finances would contribute to the risk of mor- tality. 76 In addition to the lack or loss of income. a household may be at risk for low birth weight orinortality because Of a lack of medical insurance. It is suspected that. in Michigan. unemployment during the early 19805 and the consequent loss of medical coverage affected the household's ability to get prenatal medical care and obstetric care during labor and delivery ("Meeting the Problem." 1983). Another new variable in this study is parents' school attendance. It is anticipated that this would indicate both low income (especially if both are in school).and stress. which are expected to affect out- come. A final new variable is mother's childhood economic situation. While most of the literature on infant mortality assumes that the conditions or interventions would result in an outcome during the same calendar year. Brenner (1973) and Shin (1975b) have indicated that there is a lag effect. such that the effects are not felt until some time later. Some English studies have suggested that childhood nutritional status is associated with pregnancy outcome (Terris 8 Gold. 1969). Mother's childhood economic situation is measured by her own assessment of adequacy of household resources‘when she was under 12 years old. .A51an adjunct to this hypothesis. an estimate of the amount of time the mother has experienced poverty may help to establish the relevance of lifetime deprivation to pregnancy outcome. Sumac: The intention of reconceptual izing the variables from the tradi- tional literature on infant mortality into a household unit of analysis 77 with a family ecosystems perspective is to be able to compare and con- trast the predictability of these variables. both traditional and new. to neonatal mortality among high-risk babies. 'The new variables are expected to add predictive value and assert the validity of the house- hold unit of analysis and a family ecosystems approach to the study of infant mortality. Kaplan (1975) posed the question aptly: The need to develop better methods of assessing family func- tioning and its relationship to family and/or individual well- being is still a central problem in the sociology of the family and health (p. 89) 1115.5de This study explored the differences between the households of the infants who died in the Neonatal Intensive Care Unit of E. W. Sparrow Hospital in Lansing. Michigan. between July 1982 and June 1984. and the infants who lived. The project was partially funded by the Biomedical Research Support Grant (RR 05656-16. Human subjects approval was granted by the E. W. Sparrow Hospital Institutional Review Committee and by the Michigan State University Committee on Research Involving Human Subjects (Appendix A). This study compared the households of the entire population of infants who died in the neonatal intensive care unit (index group) during a 2-year period with a sample of the surviving NICU infants. The procedure involved matching the household of each infant who died with the household of the next surviving infant of the same race and sex as listed in the NICU log bok. a chronological account of the infants as they are admitted to the unit. Of the 916 infants who were 78 listed in the NICU log book for that time period. 69 died during their hospitalization. The households of 13 multiple births and 5 whose records were incomplete were removed from the index group. leaving a total of 51. An equal number of households of surviving infants brought the study group to 102. Medical records elicited the following information for each child: Birth day. month. year. and hour Admission day. month. and year Discharge day. month. and year Expiration day. month. year. and hour Birth weight in grams Gestational age in weeks (where range was given. lower limit was recorded) Age on admission to NICU Birth size (small. normal. or large for gestational age) Apgar scores Presence of congenital anomalies City of birth Hospital of birth Maternal age in years Gravidity Anaesthetic used in delivery Type of delivery Address of household Questionnaire The questionnaire (Appendix B) was a 17-page instrument designed to elicit direct information from the mothers in the households rele- vant to the hypotheses developed and presented here. It covered the three categories identified in the literature: pregnancy history. pre- natal care and health. and economic resources. The questionnaire was coded to insure anonymity of the respondents. The final page of the questionnaire provided a consent form so that those who were willing to be interviewed could be identified. In 79 addition. a space was left for those who wished to have a written summary of the results to indicate their interest. The Total Design Method (TDM) developed by Dillman (1978) was used to maximize the return of the completed instrument. A cover letter (Appendix C) to the respondent embodied three principles of eliciting the cooperation of the respondent: (a) minimize the costs for respond- ing. (b) maximize the rewards for responding. and (3) establish trust that those rewards will be delivered. Specifically. the respondent was involved as a consultant in an important undertaking. The instrument was simplified so that she could complete it in a short time with little effort. A postage-free enve- lope was provided. as well as a complimentary pen that she retained as a token of appreciation for her efforts. Trust was established through the affiliation with the NICU and its medical director. whom some of the respondents knew personally. The mailing procedure entailed mailing the questionnaire with a letter personally addressed to her and signed in ball-point pen (so as to confirm its importance and personal interest). After 1 week a postcard was sent to all respondents thanking them if they had mailed their questionnaire and reminding them if they had not done so. A second letter (Appendix C) was sent at 3 weeks to those who had not responded. including a second copy of the questionnaire. Respondents were assured of complete confidentiality and were given a telephone number to call if they had questions. 80 W The households of the infants who died and the households of the infants who survived were compared on the basis of pregnancy history. household characteristics. prenatal care. and health and economic resources. The dependent variable of interest was mortality prior to leaving the NICU. Chi-square goodness-of-fit tests and student .t-tests were performed to determine if there were differences between the characteristics and relationships of households of the infants who lived and those who died. Factor analysis was conducted on the 300 variables to assess the factors that clustered together. The scales that emerged from the factor analysis as well as those emerging directly from the instrument were correlated with each other to determine which scales in the same domain were correlated with each other. Scales that appeared to represent the same factors as other scales were removed. Fourteen scales remained and were subjected to path analysis to test a hypothetical causal order of factors relating to mortality or survival of a newborn. ”mm 1. The study compared the households of the most vulnerable infants. those who were cared for in the Neonatal Intensive Care Unit of E. W. Sparrow Hospital between July 1. 1982. and June 30. 1984. While the care given there was the most skilled available and the facilities the best in the area. some infants were born in other hospi- tals. There was. then. some variation in care at birth. It may be that the infants who were transported were hardi er than those who were 81 born at E. W. Sparrow Hospital. It may also be that the process of transporting the infants adversely affected them. Nevertheless. this study considered that the infants had comparable treatment at birth. 2. The comparisons being made were between those babies who lived or died in the NICU. It may be that the babies who lived or died in the NICU are more similar than different and that a non-NICU control group would have been preferable. Such a group was not available for study at that time. 3. A control group consisting of infants of the same birth weight might have been desirable. This could have been done sacrificing age and sex. but combining the three in this small study was virtually impossible because of the small number of nonwhite infants. 4. Because of the time span. recall of information about pregnancy may be less for those who delivered 2 years previously. Thus we com- pared women whose recall may have varied considerably. 5. This study does not include fetal deaths. Since all the infants were born outside theiNICU. those who were dead at delivery were never admitted to the NICU. Assumptions 1. It was assumed that women can with reasonable accuracy recall the events prior to and during their pregnancy. 2. It was assumed that women can distinguish with reasonable accuracy what happened prior to and during their pregnancy. 82 3. It was assumed that. in comparing information between groups. the actual duration Of pregnancy was not important. but rather the period just prior to and during pregnancy. Women in both groups have ranges of duration of pregnancy from 6 to 9 months. 4. It was assumed that the differences between an infant living or dying are not due to any differences in obstetrical care during or following delivery. CHAPTER IV RESULTS AND ANALYS IS Chapter IV is the analysis of the data from the NICU Research Project. The first part is a comparison of respondents and nonrespond- ents in terms of the data available for every case found in the medical records. The data from the respondents are then analyzed through the use of chi-square goodness-Of-fit tests and student t-tests. The groups are compared on each variable in the study. Finally. the path analysis is discussed. This includes a description of the scales constructed to represent the major factors of the hypothesis. a state- ment of the hypothetical path model. and a testing of the hypothesis with the data from the study. The hypothetical model. the path ana- lytic model. and examples of the paths to survival and mortality are illustrated in Figures 4.1 to 4.4. Bosoonss_Bate One hundred two questionnaires were mailed to the mothers of the infants in the study: 51 to mothers of babies who died before leaving the NICU and 51 to mothers of the control infants who survived. Dillman's Total Design Method (TOM) was used to enhance the return of the questionnaire. By the end of the first week. 21 questionnaires had been returned. Of those. 13 were from mothers of infants who died and 8 from mothers of infants who survived. A postcard was mailed in the 83 8h second week to all participants (except those whose questionnaires had been returned undeliverable). thanking them if they had already responded and reminding them if they had not responded. During the second week. 12 more questionnaires were received. five from mothers of infants who died and seven from mothers of surviving infants. A¢.the end of 3 weeks. a second letter was sent to women who had not previously responded (Appendix B). Enclosed with this letter was a duplicate questionnaire. Probably as a direct response to this effort. 24 more questionnaires were received. for a total of 57. Unlike the first week. when many more mothers of nonsurviving babies responded. the final weeks included responses from more mothers of surviving babies. Three women refused to participate. Two of them explained that the questionnaire seemed too long; a third objected to the purpose of the study. A fourth woman telephoned to say she was having difficulty completing the questionnaire because she was still feeling grief over the loss of her child 2 years ago. Her permission was obtained to make a referral to the hospital social worker. Eleven questionnaires were undeliverable because of lack of forwarding address. The total return by mothers of dead infants was 27 or 53%; by mothers Of surviving infants. 30 or 59%. Adjusted for questionnaires delivered. the rate Of return was 61% and 64%. respectively. It appeared that the first wave of responses represented a good deal Of emotion. both gratitude and anger. A number of women. particu- larly among those who responded immediately. took advantage of the 85 invitation to comment on their experiences. Most of the comments related not to the pregnancy. but to their experiences in the NICU. Several indicated that they wished they and their husbands could have had some time alone with a baby who had died. One woman wrote her baby's name. birth date. and death date on the front of the question- naire. making it appear to be an epitaph. It appeared. in any case. that there was a residue of feelings about the birth. even among the women who had given birth more than 2 years ago. More mothers of dead babies (77% compared to 66%) indicated they were interested in receiving a copy of the project report. Consider- ably more mothers of dead babies consented to be interviewed (52% compared to 36%). Of the households that responded. 48 were white. 9 nonwhite; 21 had female infants. 36 had male infants. For 21 households. this was first child; 29 women had had two to four pregnancies; five had had more than four pregnancies. While about 40% of the women were having their first babies. only one of the respondents was less than 16 when she had the child. Fifty-two of the respondents were between 16 and 35. and only one was over 35.1 An analysis was performed to determine if there were significant differences among those who responded to the questionnaire and those who did not respond. The comparison included the information derived from the medical records. This information included: IData are missing for sane of the cases. 86 Birth status Birth year (1982. 1983. or 1984) Gestational age Maternal age Gravidity Prior prematurity Prior multiple birth Prior abortion History of diabetes History of hypertension Other family history of disease Anesthetic used in delivery Type of delivery Sex of the infant Race Of the infant Age on admission to the NICU Size of the child (small. normal. or large for gestational age) APGAR score at 1 minute APGAR score at 5 minutes Presence of congenital anomalies Place of birth Hospital of birth A chi-square analysis was performed on these variables. None of these factors was significant at the .05 level. Whites were more likely than non-whites to respond. but the difference was not signifi- cant. Bg5“]15 9f Obi-Square and Student $-135I5 A first task in analyzing the data was to determine if the infants who died could be distinguished from the infants who survived in terms of the factors that are considered to be major predictors in epidemio- logical studies. Chi-square goodness-of-fit tests were used on the nominal variables. and student t-tests on variables where interval data were available. There were no differences between the two groups in terms of marital status. maternal age. number of pregnancies. the trimester 87 start of care. the number of physician visits. major health conditions. father's or mother's education. or father's or household income. Existing studies have suggested that marital status is a major factor in determining outcome of pregnancy. Illegitimate children are more likely to die in infancy than legitimate ones. Among the most vulnerable babies. marital status of the mother made no difference in outcome. In fact. there were slightly more mothers of surviving babies who were either divorced or single (seven compared to four). While some of the women who were single at 6 months before pregnancy got married during pregnancy. status at 6 months before pregnancy did not distinguish the groups either. Mothers in both groups tended to be between 16 and 35 and to have had from two to four pregnancies. Curiously. there were twice as many mothers of surviving babies who were having their first baby (14 as compared to 7% Feur mothers of dead babies compared to one mother of a surviving baby had had more than four pregnancies. The differences were not significant. Thus. with regard to age and gravidity. mothers of vulnerable infants did not fit the pattern of infant mortality and survival in the larger population. There were likewise no significant differences in the two groups with regard to prior loss. According to the hospital records. no woman in either group had had a prior infant who was premature. There were no differences between the two groups with regard to the traditional measures of prenatal medical care. Most of the mothers in the two groups started care in the first trimester. Of the four 88 women who started care in the second trimester. three had infants who survived. There were no differences with regard to the number of physician visits. in spite of the fact that the babies who died tended to be of lower birthweight and lower gestational age. and delivery would. in that case. have preempted prenatal care and resulted in fewer visits. The two groups could not be distinguished on the basis of the source of prenatal care: private physician. clinic. or health mainte- nance organization. Women who are overweight. underweight. have diabetes or heart disease. and smoke and take drugs have been considered in the literature as at risk for infant loss. The findings here were that these characteristics did not distinguish the groups with and without infant loss. Not unexpectedly. with this small number of respondents. there were no women at all in some of the categories: those with a family history of heart disease and diabetes. other diseases. women with prior prematurity. and those with prior multiple births. There were no differences between the two groups in the amount of weight gained during pregnancy. Pounds gained by the index-group mothers ranged from 3 to 40. and in the matched group from 6 to 50. The index group had a mean weight gain of 26 pounds; the matched group had a mean weight gain of 28 pounds. but the difference was not significant. Mother's pregnancy weight in the two groups was similar: 156 for index mothers and 153 for matched-group mothers. Again. the difference was not significant. 89 Ouetlet's index (weight divided by height squared) was used to compute weight adjusted for height to determine if there was a difference in the two groups (Khosla 8 Lowe. 1967). The mean adjusted weight resulting from this equation was .0380 for mothers of dead babies and .0389 for mothers of surviving babies. The student t—test indicated there was no difference in the adjusted weight. Smoking. considered to be a risk factor in pregnancy. made no difference in the outcome of the pregnancy. nor did the consumption of drugs. One might expect that bi rthweight of babies of smokers would be less than that of infants of nonsmokers. In fact. the mean birthweight of the babies of nonsmokers was less than the bi rthweight of babies of smokers (1.959 grams compared to 2.440 grams--a difference of more than a pound). The difference was not significant. In the epidemiological studies. there is frequently a difference found between the groups in terms of mother's education. father's income. and household income. This was not so among these vulnerable babies. Educational level of mothers of surviving babies (13.6) was almost identical to that of mothers of dead babies (13.4). For fathers. educational levels were 13.0 and 14.1. respectively. The number of mothers and "other person" working full time both before and during pregnancy was similar. The mothers of dead babies did not worry more or less about someone in the household being laid off than mothers of surviving babies. nor were there differences in the number of persons unemployed. The groups were similar in the receipt of public assistance. food stamps. unemployment benefits. and in having 90 health and pregnancy insurance coverage. On the whole. parents of NICU babies. whether they lived or died. were not recipients of public assistance. food stamps. or unemployment benefits. About a third of the households in both groups did not have pregnancy insurance. Income was similar in the two groups. While studies using income as a measure have found households of dead babies more deprived. neither household income nor father's income differentiated between the groups. Most studies have found birthweight to be a major predictor of outcome of pregnancy. In this comparison of the most vulnerable infants. birthweight did distinguish the groups of babies who lived and those who died. The mean birthweight of the infants who lived was 2.622 grams; of the infants who died. the mean birthweight was 1.704 (p = .001). a difference of about 2 pounds (Table 4.1). As might be expected. gestational age also was significantly different in the two groups (x2 = 17.95. or = 4. p = .001). As can be seen in Table 4.2. many more babies of less than 28 weeks gestation died. and more survivors can be found in the upper-birthweight categories. According to census records. boys are more vulnerable than girls. and nonwhites are more vulnerable than whites. But chi-square analysis indicated that in this study there were no differences between females and males or between racial groups in terms of outcome of pregnancy. The mean bi rthweight for girls was less than for boys (2.039 grams compared to 2.274 grams. a difference of about 8 ounces). Nonwhite 91 Nm. n o ooo. n.m . «.mm oo..N . Nx wo.oo. wm... wm... wm.o. wad. woe: w...N wm.m ww.m. wmd No m m o o o N. N o m .eeo» wo. Nm woo. woo we . mo ww . oo wm . 8 wm . mm wom on m N a o N N . o o es..< wit. woo wm.mm wm. mm me. we. 3 woo woo. woo. N o m N m . m _ m m eeeo ooo.a oom.m ooo.m oom.N ooo.N oom.. ooo.. .aeo» ooo... -.om.m -.oo.m -.om.N -.oo.N -.om._ -_oo.. -.om oomv Amsmcu 2.. u;m_ozcuc.m .ucm.ozcuc_p >3 msumum suo.m "..o e.on. 92 infants had a slightly lower birthweight than whites: 1.680 grams compared to 2.282 grams. a difference of 1 pound. 5 ounces. Student i-tests indicated there were no significant differences. Table 4.2: Birth status by gestational age. Gestational Age (in Weeks) Total <28 28-32 33-37 38-41 >41 Dead 13 2 3 8 O 26 92.9% 33.3% 25.0% 38.0% 46.4% Alive 1 4 9 l3 3 30 7.1% 66.7% 75.0% 61.9% 100% 53.6% Total 14 6 12 21 3 56 25.0% 10.7% 21.4% 37.5% 5.4% 100.0% x2 = 17.95 at = 4 p = .001 s = .59 While the traditional measures failed to predict substantial dif- ferences between the two groups. householdineasures did not fare any better. Only two of the household variables differentiated between the two groups: household size during pregnancy (but not before pregnancy) and the planning of pregnancy. Student t-test on household size indi- cated that the difference in the mean number of persons in the house- hold for each group was significant (p < .05). The mean number of persons was 2.7 for households of dead infants and 3.0 for households of surviving infants. The small number in the households is probably 93 accounted for by the large proportion of babies in both groups born to women who were pregnant for the first time. Although household number distinguished the two groups. number of children (both preschool and school age) did not. Nor did the presence of the father or other person in the household or the amount Of separa- tion the mother had from him during pregnancy. Presence of another adult in the household did not differentiate the two groups. The literature has suggested that attitude toward pregnancy and the planning of pregnancy make contributions to the outcomerof preg- nancy. As shown in Table 4.3. chi-square analysis showed significant differences between the two groups (X2 = 4.88. if = l. p = .03). It appears that the households of the babies who died were more likely to plan the pregnancy than those of the babies who survived. the opposite of what might be expected. Table 4.3: Birth status by planning for pregnancy. Planning for Pregnancy Total Planned Not Planned Dead 22 5 27 59.5% 25.0% 47.4% Alive 15 15 30 40.5% 75.0% 52.6% Total 37 20 57 64.9% 35.1% 100.0% X2=4.88 .d_f=l ,p=.03 9’4 The measures of prenatal care incorporated the notion of quality of care and. departing from the idea that "prenatal care" is the province of physicians. developed measures of household and community support for the pregnant woman and measures of sources of information about pregnancy. In terms of the variables that reflected quality of care. there were no group differences in the number of women who were refused prenatal care. in the number who were worried about paying for prenatal care. or in the number of different physicians examining the mothers during pregnancy. ‘The number of sources of assistance and support was similar in both groups (3(- = 2.63 for households of dead infants; Y = 2.27 for the mothers of surviving infants). Number of sources of information was similar for the two groups ()(n= 3.96 and 4.00. respec- tively). The household measure. per capita income. like household and father's income. showed no differences between the two groups. Asked about the amount of poverty in their lives. members of both groups responded that they had experienced no poverty in their lives. and that when they were children. their families could afford almost everything they needed. Thus. only two traditional variables--birthweight and gestational age-were different in the two groups. Likewise. only two of the household variables showed significant differences. This is somewhat surprising because. of 300 variables. 15 could be expected to be 95 significantly different by chance. Household variables. therefore. failed to be better predictors of outcome of pregnancy. It appears from these data that the households of the infants who lived or died in a neonatal intensive care unit did not fit the pattern of the infants described in epidemiological studies. The literature. however. dealt with the general population of infants. most of whom experience no difficulties in the first few weeks of life. It appears that. on the basis of these variables. both traditional and household NICU infants may have more in common than babies who live or die in the larger population. Furthermore. household variables were not better predictors of mortality and survival. as had been hypothesized. .Dsiiolsnsiss_1n_tns_Hishzflisk_boonoaoh Neither traditional nor household variables appear to be impor- tant in predicting NICU mortality. But. in fact. the high-risk approach which derives from studies of infants in the general popula- tion who lived or died has not been shown to be effective in reducing infant mortality. Studies identifying higher rates of mortality among younger and 01 der mothers. women at the extremes of child-bearing ages. high parity. low education. and low income have revealed that most of the babies survive. even when the risks are compounded. The high-risk approach requires that a great many women who are technically at risk by these standards must be served at great expense to salvage a few babies whoimight otherwise have died. Even so. this assumes that prenatal medical care (the major approach to intervention in this 96 problem) has ameliorative effects on a high-risk pregnancy. This assertion has been questioned. as indicated in the literature review. The linear approach. from which the "high-risk" model evolved. has failed to demonstrate how infants come to die within the first weeks of life. The discovery that the difference between survivors and nonsur- vivors in the general population does not hold for survivors and non- survivors in the NICU population further indicates that there are some deficiencies in our understanding of the etiology of infant mortality. If the "risk" model were adequate. the households of the NICU survivors would have been classifiable as at risk. whereas the households of the NICU fatalities should have been classifiable as at high risk. But in the NICU population. there were a substantial number who were not at risk in the sense of such factors as extremes of child-bearing age. high parity. low education. and low income. Most women in both groups were between 16 and 34 and had had between two and four pregnancies. The mean educational level for women and men in both groups was between 13 and 14. Four households in each group had incomes in excess of 340.000 per year. The deficiencies in the usefulness of the "risk"inodel may derive from the fact that the right variables have not been identified or that the linear model is inadequate to describe the process by which a fetus develops into a vulnerable. nonsurviving infant. It is clear that another approach is needed that describes the interrelationships among the factors that affect the well-being of the fetus and newborn infant. 97 While it is possible that other variables might be identified that offered more power in terms of predicting mortality. it is more likely that this single-variable approach will remain inadequate to describe complex processes. It is even more likely that there are a variety of paths to infant mortality. rather than a single path. W What happens in the macrosystem (government. culture. and belief systems) and the exosystem (in this case. parents' jobs. friendship. and informational networks) is linked to the fetus through the house- hold microsystem (Bronfenbrenner. 1979). More specifically. what hap- pens in the physical environment. the human built physical environment. and the socio-institutional environments affects the characteristics and relationships in the household. The household. through its income- producing capacities (market and nonmarket) and through its market and nonmarket activities. affects its own well-being. fertility management. income production. and social equity (Paolucci et al.. 1976). This study is not a test of these models. but rather a use of parts Of the ecological model to test some hypotheses about the inter- relationships of the various factors involved in the death or survival of an NICU infant. Specifically. the health and well-being of an individual (in this case. the fetus) are affected by the events. activities. and relationships within the household. To the extent that the stresses occurring within a household affect the mother physiologi- cally and emotionally. the fetus is likely to be affected. To the extent that stresses are mitigated by adequate income. support. 98 information about pregnancy. and prenatal care. the fetus is likely to be affected. Smoking and prior losses. thought to represent biological processes. may also be prompted by events. activities. and relation- ships within the household. lhnfll the interrelationships between these factors can be understood. efforts to intervene will remain thwarted. BationaloioLEathnalxsls The literature on infant mortality has tended to infer causality without using an appropriate statistical model. Particularly with regard to prenatal care. the literature has frequently suggested that prenatal care makes a difference. Furthermore. policy makers and program developers have relied on correlations of outcome with prenatal care tojjustify massive outlay of public funds to pay for medical care on the grounds that it will save money now being spent in paying for neonatal intensive care facilities and services. One problem with inferring causation from correlation is that there are a number of factors that correlate with infant mortality. More- over. there is little evidence that the provision of prenatal care is ameliorative; that is. given problems of poverty. poor nutrition. and very early childbearing. that prenatal care can exert an ameliorative effect. Chamberlain (1984) pointed out the consequences of using this "risk" model. that a great many women have to have special prenatal care in order to identify a few who are in need. The author suggested 99 that an "ecological" model would shift the focus from physician- intensive programs to communitywide programs that reach all families and children. just as sanitation and disease programs have done. Regardless of the type Of program. a basic understanding is needed about the relationship of the factors that interact and influence the well-being and survival of neonates. To that end. a model is proposed here that provides a hypothesis about the relationship between the factors that influence outcome of pregnancy. WW There are three criteria for inferring a causal relationship between variables (Lazarsfeld. 19591. The first criterion is that cause precede the effect in time. The second criterion is that the two variables be empirically correlated with one anothen. The third criterion is that the two variables cannot be explained by a third variable that influences both of them. Path analysis. a method developed by Sewall Wright (1934. 1960) and Duncan (1966). is a means of decomposing and interpreting linear relationships among a set of variables. It has two basic assumptions. It assumes. first. a causal order. One variable may affect a second. but the second cannot affect the first. ‘This implies that the two are correlated. and that there is a time sequence. The second assumption is that the relationships among the variables are causally closed; that is. the variables in the analysis are not affected by other factors outside the analysis (Kim 8 Kohout. 1975). 100 Path analysis represents a linear relationship between variables and is uncharacteristic of the relationships found in an ecosystems analysis. in which it is assumed that components of the system mutually influence each other. However. causal relationships do exist within an ecosystem. Path analysis is used here. in spite of its limitations. as aineans to advance the present knowledge about infant mortality by showing the interrelationships among the variables that are said to affect. or exert a causal effect on. the outcome of pregnancy. From a theoretical viewpoint. the relationships represent correlation; that is. a change in one variable is said to correspond to a Change in another variable. The development of risk scales and the creation of prenatal care programs infer that these factors are causative in that the manipulation of the independent variables is said to result in improved outcome of pregnancy. In spite of its limitations. the path model advances the understanding of the etiology of infant mortality by showing the relationships between variables in a hypothetical model. SoflsLUssdALEatthalsts Fourteen scales were used in this study to represent the factors in the hypothetical path model. They include the following variables: BIRTH STATUS. The condition of the infant prior to leaving the NICU. (1) Death of infant. (2) Survival. BIRTHWEIGHT. This variable represents the actual birthweight in grams of the infant. WORRYI. This scale represents worry resulting from mother being unemployed and looking for work. The information derives from the question: 101 Did you (the mother) have a job? (Check) Yes. a full-time job. Yes. a part-time job. No. but I was looking for work. No. and I was not looking for work. Those who indicated they were either working full time. part time. or not.looking for work were assigned a 1; those who were looking for work were assigned a 2. WORRYZ. This scale represents worry resulting from an actual job loss of the mother. It derives from the question about job changes. Yes. I was laid off or lost a job. If the mother did not check this statement. she was assigned a 1; if she checked this statement. she was assigned a 2. WORRY3. This scale represents worry from a potential job loss. Was there a time when you were worried that you might be laid off whether or not you actually were laid off? (Check) Yes No Those who checked "No" were assigned a 1; those who checked "Yes" were assigned a 2. WORRY4. This scale represents the mother's feelings about her job. Good Neither good nor bad Bad If she checked "Good" or "Neither good nor bad." she was assigned a 1. If she checked "Bad." she was assigned a 2. 102 INFO. INFO is the actual number of sources of information the mother had during pregnancy. The question was: How did you learn about what you should and should not do during pregnancy? Please check all that apply. Already knew from previous experience Husband or partner ._____Members of your household _____Family outside your household Friends Neighbors ___Physici ans Other medical people (nurses. nutrition counselor. health educators) ______Social workers Radio and television Preparation for childbirth class Other classes _____Books. magazines or newspapers Other The score on INFO was the total number of checks. WORK. WORK is represented by the proportion of the period in which the mother was working: Were you working for the full six-month period? (Check) Yes _____No. only part of the six-month period If the mother had indicated she was not working. she was assigned a 1; if she was working part of the 6 months. she was assigned a 2; if she was working full time. she was assigned a 3. ADECARE. ADECARE indicates whether or not the mother received adequate prenatal medical care. Prenatal medical care was considered adequate if the mother started prenatal care in the first trimester and had the appropriate number of visits. as judged by gestational age. 103 Specifically. the number of visits was adequate if it met the following criteria: gestational age < 28 weeks and number of visits = 1-4 or more gestational age 28-32 weeks and number of visits 3-8 or more gestational age 33-37 weeks and number of visits 9-12 or more gestational age 38-41 weeks and number of visits 9-12 or more gestational age > 41 weeks and number of visits = 12-16 or more Lack of adequate care was assigned 1. adequate care. 2. LOSS. LOSS represents an actual fetal or infant loss through miscarriage. stillbirth. or death. NO loss was represented by 1. a loss by 2. HOUSEHOLD. This variable represents the actual number of persons in the household. This may include husband/partner. children. other family. or unrelated persons if they participated in the household activities and shared income and food preparation. SUPPORT. SUPPORT is the actual number of sources of tangible assistance and support during pregnancy. Put ons_check by all the people who assisted you in any way during this pregnancy (meals. food. money. health care. transportation. child care. someone to talk with. advice. etc.) Wat Jabon_and_delixeny. Put two_ghegks by all the people whom you found very helpful and supportive. ___Husband or partner ___Members of your household ___Family outside your household ___Friends .___Neighbors ___Physicians ___Other medical people (nurses. nutrition counselor. health educators) ___Social workers ,___Counselor or psychologist .___Support group ,___Other 104 Respondents were given a 0 for no checks. 1 for 1 check. 2 points for 2 checks. The number for each respondent was the actual total of checks. PER CAPITA INCOME. This variable represents per capita income 6 months prior to pregnancy. Per capita income equals household income prior to pregnancy divided by the number of persons in the household prior to pregnancy. SMOKE. This variable represents the effect of smoke on the fetus. If mother did not smoke and was not around other persons who smoked a lot during pregnancy. she was assigned a 1; if she smoked or was around people who smoked a lot. she was assigned a 2. Enocsduss The procedure for conducting a path analysis involves the state- ment of the hypothetical model (Figure 4.1). This model is guided by both the literature on infant mortality and the zero-order correlations between the variables considered important in this study. The zero- order correlations are particularly important because of the fact that the population under study--the households of the infants who died following treatment in a neonatal intensive care unit and a matched sample of infants who survived--may differ from the households of infants who died or lived in the general population from which most of the literature derives. An assessment of the zero-order correlations (Table 4.4) suggested that bi rthwei ght was the key factor to outcome of pregnancy. A rel a- tionship of .42 (p = .001) suggested that low birthweight strongly 105 585 5.5 ..Ouos sumo cumoooLo ask u..: o.:m.n 9.95 v >552, u .653 In EBBAIIIIICE. III taosmI 282:. Eye; 5.5 P E52. 6800:. mzomo Lon to; 20:60:01 30.. Loco 106 . .coCLoaoL Com mo. v um ucmu_$.cm.m mco.um.occou >_co "Ouoz om. Nu. : m N on. m o m. Nm. M. m. N NN.| om. mm. m:.- Nu. mu. :m. mm.1 mm. mm. mm. NN.| N . N . N 4' wxozm mzouz_ Pacemam oaozmmao: mmO. wm¢moz m>¢¢oz N>¢¢03 _>¢¢oz h=c_w3:h¢_m m2h.mcm sumo Ono c. cum: mo_mum a. mcosm mco.um_0ocou coucoi0co~ 2.... e33 107 predicted infant death. .Associations were also shown with WORRYZ. representing being laid off or fired from a job (j: = .22. p < .05; being fired was associated with survival) and with per capita income (n = -.25. p < .05; survival here was associated with lower per capita income). Per capita income was related to WORRYI. job status (1; = -.30. p < .05). but not with the other "job worry" categories. WORRY2 and WORRY3 were correlated with each other (_n = .86. p = .001). WORRY3 was also associated with INFO (number of sources of information about pregnancy; J: = .26. p < .05) and WORRY4 (mother's feelings about her job; L = .32. p < .01). Thus feeling bad about one's job was related to worry about being laid off and appeared to be related to the number of sources of information about pregnancy. Put another way. the worse a mother felt about her job. the more likely she was to lose her job and the more sources of information she was likely to have. The number of sources of information was associated with the number of persons in the household. That is. the more persons in the household. the fewer the sources of information (1: = -.46. p = .001). Whether this was actually an effect of a negative relationship of number of persons in the family with per capita income (J; = -.52. p = .001) or whether it suggested that a larger family (with presumably 108 more children) has a limiting effect on a mother's contacts with other people and her access to books. classes. and media is not clear. In addition to being associated with HOUSEHOLD. INFO was asso- ciated with SUPPORT and PER CAPITA INmME. The more support. the more sources of information (I = .50. p = .001). and the more per capita income. the more sources of information (I: = .32. p < .01). The amount of work (whether mother was not working. working part of the time. or working full time) had a negative relationship with per capita income and a positive relationship with number in the household. That is. the lower the per capita household income. the greater propor- tion of the time she was working I: = -.33. p < .01); the more persons there were in the household. the greater proportion of the time she was working (1 = .32. ,p < .01). Number of persons in the household and per capita income were associated with the number of sources of assistance and support (I = -.22. ,p < .05; 1: = .34. .p < .01. respectively). The smaller the household and the larger the income. the greater number of sources of support. Adequacy of prenatal care was associated with INFO (I; = .23. p < .05) but not with birthweight or birth status. However. the literature would suggest that there is a strong association with outcome. either 109 birthweight or survival (Boone. 1982; Kessner et alN. 1973; Ryan et al.. 1980). Prior fetal or infant loss is said to be associated with outcome (Boone. 1982; Shapiro et ah. 1980L Loss. in the zero-order correla- tions. was associated with birthweight (.t = -.22. p < .05). Some literature (Rosenzweig 8 Schultz. 1982) has shown a strong association of smoking cigarettes with bi rthweight and outcome of pregnancy. Smoking was also seen in the zero-order correlations to be associated with bi rthweight (1; = .22. p < .05). but the positive correlation suggests smoking is associated with higher birthweight. The inclusion of another person smoking frequently in the mother's presence in the SMOKE scale may have confounded the association. The Michigan documents (Meeting. 19$; Prenatal Care. 1984) have inferred that unemployment directly and through its effects on prenatal care has affected the rate of infant mortality. It is hypothesized here that mother's being unemployed and looking for work directly affects the outcome of pregnancy (BIRTH STATUS). Low PER CAPITA INCOME tends to reduce prenatal care (ADECARE) and this. in turn. tends to increase the likelihood of infant death. LOSS. representing prior fetal or infant loss. is hypothesized to increase the likelihood of low birthweight and this. in turn. increases the likelihood of infant death. Smoking (and being closely associated with a smoker) is predicted to increase the likelihood of an infant dying. Per capita income. through its effects on support. and amount mother worked and her feelings about her work are expected to affect 110 the number of sources of information. Her per capita incomeiis expected to result in high number of sources of support during pregnancy. and this is expected to contribute to a large number of sources of information. 'The more the mother worked. the more sources of information she is expected to have; the worse she felt about her job. the fewer sources of information she is expected to have. House- hold number is predicted to affect per capita income and amount of mother's work. Few persons in the household is expected to increase the per capita income and to decrease the amount a mother works. The procedure for implementing the path model is a series of multiple regressions whereby a dependent variable is regressed against the remaining variables in the hypothetical model. The variables which are shown to be significantly correlated with the dependent variable become'the dependent variables for the remaining variables in the equation. They are. in turn. regressed against the remaining variables in the hypothetical model. This process is continued until no vari- ables in the model are significantly correlated with the dependent variable. The multiple regressions provide more useful information than the zero-order correlations because each variable in the regres- sion is controlled for the other variables in the equation. A sche- matic representation of the relationships between variables. although not essential (Duncan. 1966). is usually a part of the path analysis. Bosuln The actual path analytic model (Figure 4.2) suggests a need for revision of the hypothetical model. ‘TwO paths to survival or mortality 111 .maumum cuL.n use .uzm.ozcuc.n .wmo>m_ _msuom cam .mmo>m_ .0 Leo. .>ucmcmoco uaonm co_umE.omc. mo moucaom mo Con55c .ucooasm cam oucmum_mmm o.n.mcmu we moocsom .0 Loose: .oEOoc. mu_amu Loo .coxcoz Locuos OE.u we ucsosm .u_ozomao; c. mnemcoo .0 Conan: .mmC. uence. co _muom Co.oo mo mucus—mo. ecu mc.3o;m _oeos u.u>_mcm sumo _mc.n "~.: Ocam.m 29.356... 2:2 I .36 I .50 ‘IIocAIII can: ‘III 088:. w em. m 2. N. . m 3 3.. . . en. t w mm. 2.30 can eta m Egososm moo. 112 are indicated by the first regression of BIRTH STATUS on the remaining variables (Table 4.5). The first path is through BIRTHWEIGHT. with a path coefficient of.53 indicating that infants with higher birthweight are more likely to survive. This path accounts for 28% of the variance in BIRTH STATUS. When the remaining variables in the'path were regressed on BIRTHWEIGHT. none was significantly related. The second path was through WORRY2. mother's being fired or laid off from work. The relationship between BIRTH STATUS and WORRYZ was a positive one with a path coefficient of .54. thus accounting for 29% of the vari- ance in BIRTH STATUS. Together the two paths account for 57% of the variance in BIRTH STATUS. Table 4.5: Regressions performed for path analysis. Independent Variable Beta Weight Significance Dependent Variable: BIRTH STATUS BIRTHWEIGHT* .53 .000 INFO .26 .145 WORK -.23 .099 SMOKE -.02 .895 WORRY2* .54 .035 ADECARE -.16 .190 WORRYI .03 .804 LOSS .22 .128 WORRY4 -.19 .166 PER CAPITA INCOME -.16 .342 HOUSEHOLD .18 .286 SUPPORT -.09 .612 WORRY3 -.42 .112 113 Table 4.5: Continued. Independent Variable Beta Weight Significance Dependent Variable: BIRTHWEIGHT INFO -.O6 .788 WORK .00 .987 SMOKE .25 .109 ADECARE .09 .565 WORRYI -.02 .896 LOSS -.30 .065 WORRY4 .08 .630 PER CAPITA INCOME .05 .811 HQJSEHOLD .07 .713 SUPPORT -.08 .705 Dependent Variable: WORRY2 INFO .00 .974 WORK .05 .544 SMOKE .07 .361 ADECARE -.O6 .455 WORRYI -.03 .692 LOSS -.07 .424 WORRY4 -.12 .149 PER CAPITA INCOME -.05 .657 HOUSEHOLD -.10 .353 SUPPORT -.08 .448 WORRY3* .87 .000 Dependent Variable: WORRY3 INFO* .44 .028 WORK -.01 .973 SMOKE .04 .798 ADECARE -.15 .291 WORRYI -.07 .650 LOSS .16 .281 WORRY4 .25 .102 PER CAPITA INCOME -.18 .362 HOUSEHOLD -.07 .698 SUPPORT -.21 .269 11h Table 4.5: Continued. Independent Variable Beta Weight Significance Dependent Variable: INFO WORK .02 .869 SMOKE -.01 .920 ADECARE .10 .322 WORRYI -.18 .127 LOSS* -.30 .007 WORRY4 .02 .840 PER CAPITA INCOME -.14 .359 HOUSEHOLD* -.42 .002 SUPPORT* .54 .000 Dependent Variable: SUPPORT PER CAPITA INCOME* .38 .001 LOSS .09 .483 WORK* .33 .015 HOUSEHOLD -.12 .432 Dependent Variable: WORK LOSS -.05 .697 HOUSEHOLD" .33 .014 Dependent Variable: PER CAPITA INCOME LOSS .01 .912 HOUSEHOLD* -.52 .000 *Significant at < .05. 115 The path to BIRTH STATUS through WORRYZ suggests that the actuality of mother's losing her job precedes the survival of the infant. It is preceded in time by WORRY3. anxiety over the prospects of mother's losing her job. The path coefficient is .87. accounting for three-fourths of the variance in WORRY2. WORRY3 is linked to INFO with a path coefficient of .44. The number of sources of information about pregnancy played a key role in the second path to BIRTH STATUS. Prior loss affected the number of sources Of information. with a path coefficient Of -.30. suggesting that having no prior fetal or infant losses was associated with an increase in the number of sources of information. Household number tended to decrease the number of sources of information (path coefficient of -.42). and there was a path coefficient of .54 between SUPPCRT and INFO. Household number has a path coefficient of -.52 with PER CAPITA INCOME and a coefficient of .33 with WORK. suggesting that few in the household leads to a higher per capita income. and many in the house- hold leads to mother working more. Both high income and high work situations induce high levels of support. with no relationship between mother's work and household income. WORK has a path coefficient Of .33 with SUPPORT. and PER CAPITA INCOME has a path coefficient with SUPPORT of .38. 116 Stillman There appear to be two key elements in the path to survival: number of persons in the household and number of sources of information about pregnancy. The number in the household directly affects the number of sources of information. The fewer in the household. the more sources of information about pregnancy. Not experiencing a prior loss also results in a greater likelihood of having more sources of informa- tion. In addition to having a direct influence on the number Of sources of information. the number of persons in the household has an indirect influence on sources of infonmation. Not unexpectedly. having few persons in the household increases‘theiamount.of per capita income available to the household. Having a higher per capita income gen- erates more support. which increases the number of sources of informa- tion. Mother's working more is another pathway to increased support. increasing the assurance of survival. While having a large number of persons in the household depresses per capita income. it also increases the likelihood of mother's working full time. which in turn generates support for the household during the pregnancy. The number of sources of information results in worries about being fired and an actual layoff. This path results in the survival of an NICU infant. The second path. which accounts for 28% of the variance in BIRTH STATUS. is birthweight. The higher birthweight babies tend to survive. None of the factors tested in the present study influenced birthweight. 117 Discussion It appears that in looking at the "job loss" path to outcome of pregnancy. the NICU infants who die fit the profile of the infants who die in the general population; the NICU infants who live fit the pro- file of the infants who survive in the general population. It also appears that were it not for the job anxieties. the NICU survivors might have been born under normal circumstances. It also appears that. because of the importance of the SUPPORT-INFO link. were it not for working more. large households (with presumably less income)Inight have been unable to generate the critical support and information needed for survival following treatment in the NICU. A surprising finding is that smoking and adequacy of prenatal medical care. as defined in this study. play no part either in survival or death of a vulnerable infant. It may be. however. that these two factors are influenced by a different set of factors not appearing in this equation. Forty-three percent of the variance is unexplained in this path analysis. CHAPTER V SUMMARY AND RECOMMENDATIONS This research project sought to develop and test household variables related to infant mortality. Fifty-seven women whose infants were treated in a neonatal intensive care unit responded to a mail questionnaire in the spring of 1985. More than 300 variables were tested to see whether any of them related significantly to death or survival of an infant treated at birth in a neonatal intensive care unit. Birthweight and gestational age were significantly related to outcome. That is. infants who died before leaving the NICU were sig- nificantly more likely to be of low birthweight and low gestational age. In addition. two of the household variables were significantly related to outcome for the infant: 'the planning of pregnancy and the number of persons in the household during pregnancy. The differences were not in the expected direction: Households of the infants who died were more likely to have planned the pregnancy and tended to be smaller. The variables that have frequently been found to distinguish the two groups--maternal age. number of pregnancies. prior fetal or infant loss. prenatal medical care. maternal health conditions. income. 118 119 and education--did not differ among the households of the infants who lived or died following treatment in a neonatal intensive care unit. A path analysis was conducted to identify the relationships between selected traditional and household variables and birth status. The path analysis tests a hypothetical model of the causal relationship between selected variables. ‘The actual path analytic model showed that prior fetal and infant losses. the number of persons in the household. and the number of sources of social support during pregnancy affect the number of sources of information about pregnancy. The number of sources of information is significantly related to fears about job loss. which. in turn. is significantly related to an actual job loss. A large number of sources of information about pregnancy appears to relate to anxiety about job loss. Mother's fears about job loss preceding actual job loss tend to predict survival in a child receiving treatment in a neonatal intensive care unit. While these findings seem inconsistent with the association of deprivation and infant loss in the general population. they are consistent if one looks closely at the relationship between income. work. and sources of social support during pregnancy. Households with either few persons in the household and higher per capita income. or households with more persons and where mother is working more tend to have more sources of social support during pregnancy. 'The number of sources of social support during pregnancy is directly related to the number of sources of information about pregnancy. Thus it is the households with more economic and social resources that tend to have 120 infants who survive. in spite of maternal job losses. The households of infants who die following treatment in a neonatal intensive care unit tend to be those that have more persons in the household and fewer economic and social resources. 'These mothers have not experienced job losses. but may be employed in low-paying occupations. or may be unemployed by choice and have husbands or partners who have low incomes. Significance There are five findings of significance in the path analysis. 1. Although it is apparent that there are several paths to survival or mortality. two were identified in this study. One path. accounting for 28% of the variance. was through birthweight. None of the variables in this study appeared to influence birthweight. The second path was through job 1055. Together they accounted for 57% of the variance in outcome. Forty-three percent of the variance was not accounted for by these variables. 2. The paths through birthweight and job 1055 appeared to be uninfluenced by smoking or the adequacy of medical care. If these factors affected the outcome for the infants in a neonatal intensive care unit. it was through their association with variables not tested in this study. 3. The number of sources of information appeared to play a key role in one path to survival or mortality. The "job loss" path showed that the number of sources of information was significantly associated with prior loss. number of persons in the household. and the number of 121 sources of support. Not experiencing a prior fetal or infant 1055. having few persons in the household. and having a larger number of sources of support all appeared to contribute to an increase in the number of sources of information. This is particularly important in light of the fact that prenatal care programs have stressed the informational components of the care. Since doctor and other medical resources comprised two of the possible sources of information. this component appears to be included in this variable. Women whO>had only a physician and medical staff as sources of information could be more vulnerable to infant loss because of the critical importance of this variable. 4. Work and per capita income provided alternative paths for sources of social support. The more per capita income the household has. and the more the mother is working during her pregnancy. the more sources of support she has. Both work and per capita income appeared to be influenced by the number of persons in the household. The more persons in the household. the lower the per capita income and the more a mother tends to work. So while a larger household would tend to have fewer sources of information (a direct relationship between the two variables). a larger household would tend to prompt a mother to work more during her pregnancy. This would appear to help generate more sources of support. and through the sources of support would appear to increase the number of sources of information. 5. Household number. like number of sources of information. appeared to play a major role in determining survival or mortality 122 along this particular path. This supported the contention of authors cited in the literature review (Gendell 8 Hellegers. 1973; Morris. Udry. 8 Chase. 1975; Pakter 8 Nelson. 1974; Westoff. 1972) of the importance Of family planning and abortion in the reduction of infant mortality. Limitations Several limitations exist in the data that suggest that caution be exercised in the interpretation of the data. 1. The two paths to outcome for the infant accounted for 57% of the variance. Forty-three percent of the variance was not accounted for by the variables that were used in the hypothetical path model. The apparent lack of importance of two variables recognized in the literature--smoking and prenatal medical care--may be explained by their relative position when the variables in the hypothetical model were considered. There~may be a number of paths in which these two variables play a part. 2. Some important data on the mothers were available by cate- gories only. For instance. the available records provided maternal age in categories of under 16. 16-35. and over 35. Had the actual ages been used. some difference might have been found in the maternal ages for the infants whoilived or died. 'The questionnaire sought informa- tion as to whether a woman had ever had a stillborn child. an abortion. a miscarriage. or an infant loss. Had the actual number been avail- able. there might have been differences between the groups. 123 3. The support and information variables measured only the number of sources of support and information recalled by the mother. The number of sources of information cannot be equated with accuracy of information (and. in fact. sources could conflict in the information provided. causing conflict and worry). The number of sources of sup- port may reflect quality better than number of sources of information in that the question asks the mother for support that was considered useful to her. 4. The relationship of smoking with outcome may have been confounded by the inclusion of the presence of other persons smoking around the mother. It may also have been confounded by the inclusion of any amount of smoking. Had the smoking variable been limited to a certain amount of smoking. for example. more than a pack of cigarettes per day. this might have revealed differences between groups or influ- enced its appearance in the path model. 5. Father's job worries were not included in the analysis. Mother's job anxieties were related to survival of an NICU infant. Had father's job worries been included in the analysis. a different pattern might have emerged. In addition. the fact that mothers whose infants died did not have job anxieties did not mean that they did not have worries over money. Had the items regarding worries about money been included. a different pattern might have emerged. 6. Since this study was retrospective. the accuracy of informa- tion provided by the mothers may have been influenced by the passage of time and the outcome for the infant. 124 Wire Although the path analytic model adds considerably to the existing literature. the linear relationships are not characteristic of an ecosystem approach. An ecosystem model would take into consideration that there may be reciprocal as well as linear relationships between variables (see Figure 5.1). Within the path model. for example. the number of sources of support and the number of sources of information are likely to influence each other. rather than being a one-way effect. As a household receives tangible assistance and support. it is also likely to receive information from those same sources in the process of receiving support. Likewise. as a household finds information about pregnancy. it may identify sources of support. The relationships between per capita income and sources of support are likely to be reciprocal. The more per capita income a household has. the more social support it is likely to have during a pregnancy. But. likewise. these same sources of support may have helped household members iden- ti fy sources of income either through employment opportunities or through sources of public support. These relationships cannot be shown in a path analysis even though they might exist because the path analy- sis is based on a hypothetical. linear. model. It can only show rel a- tionships as they are hypothesized to occur. Furthermore. it can only show relationships in one direction. The outcome of pregnancy. whether a child survives or dies. may also be related to the survival or death of a subsequent child. To the extent that the literature on infant mortality is appropriate to this 125 .mEOum>mi0Lu.s c_;u_3 cam .msoum>mioLu.s 0cm nocums ecu coozuoa mo.;mco.am.OL _mo.uo;uoo>s 050m uo>_uuoomcoo mEOum>mouu c. .0005 u.u>_mcm sumo _mc.n "_.m o.:m.u 26.3w 30.9.0: I 30.05.00» 55.; 822233: .8858? D 8350 89:8 t o co 0 oEe.m>w.oLo..2 new 6822 :ZNmMmuWo «five—m c6922. 3222.26". 30:658.»: I 35:02“. .6 . . ‘II 5.3335 .0. 000.. H 295.0 .0 8093.02 2:250 ozwww. 305500 .0 on: \ / ., \v x _\ «305 5.3 IEO? I 0310. _II tone-6 All 8300 \ 50¢ c0.t00< 1/ $55.05.: W . gnaw. 22...» 5.3 to; / N.\\ ....:...Z.\ .32.... 30.23 EWFm>m 5:552 H EoE>o.oEm cozatoonca... 30.02.. .0 .0». 0(5 \ » \\ / + £55.22 00:22.34. 20:05 c0603 .0 5.3:... III 059.com 2.25 oozes 5.3 .500 o. .500 -8235 «0530.2 009» mo=noaw .e 85.8 __0 126 population. it appears that survival of the infant may contribute to the survival of a subsequent infant (lack of infant loss) or the death of a subsequent infant (through the increase in the number of persons in the household). Other factors are likely to be more closely related to outcome. By placing the entire path analytic model in an ecosystem perspec- tive. the effect of the macro-system (physical. human constructed. and human behavioral environments) on the micro-system (househol d. includ- ing mother and baby) can be hypothesized. Figure 5.1 identifies selected hypothetical reciprocal relationships among the factors in the macro— and micro-systems. Although the relationships indicated in Figure S.l are far from complete. they illustrate the complexity of effects on the household. which then affects the mother and fetus. One example shown in the figure is the potential effect of such macro- system variables as governmental measures to control inflation and the balance of trade on employment opportunities. Coupled with customs of discrimination against women and lower pay in female-dominated occupa- tions, household per capita income and mother's work are affected. Mother's working and per capita income are shown in the path analysis to affect the outcome of pregnancy through their effects on the number of sources of support. the number of sources of information. and Job worries. A second example is the effects of governmental spending on the availability of family planning services and abortion. Family planning services and abortion may be related to infant mortality rates through influencing the number of persons in the household and through 127 the number of sources of information. The number of sources of infor- mation is also likely to be related to the openness of the climate for discussion of pregnancy. religious beliefs. and the neighborhood. e.g.. proximity to friends and relatives or whether there are close or dis- tant relationships between people. Level of technology. e.g.. televi- sion and radio. and religious beliefs influence the openness of climate about discussing pregnancy. These examples serve to indicate the interrelationship between factors in the macro-system and micro-systems that ultimately may affect the outcome of pregnancy. Implications The potential for generalizing to a more general population or to other populations of NICU babies is limited. This suggests that more studies are needed to validate the path analytic model that was devel- oped here and to test further the usefulness of household variables. Should further studies support the conclusions. different approaches to assisting households during pregnancy may be needed. Support for the model suggests that social policy should be redirected toward redistribution of income and income maintenance. Job security. and Employee Assistance Programs. This may. in itsel f. help households develop wider sources of support and information. It also suggests that family specialists need to assist communities to develop social support and information systems. This may mean extension of childbirth classes. support groups. and peer counseling systems. The concept of labor coach could be extended to include the development and training of pregnancy coaches to provide both support and information. 128 In addition to creating an environment that is conducive to healthy growth and development of infants. prenatal care should be reconceptualized. The importance of household number and sources of information and support suggests that rather than having only indi- vidualized medical care and a standardized educational program. inter- vention should be made throughout the household. 4A family specialist could determine if there are household factors present that might lead to an infant's needing special care or being vulnerable to death in the first few days or weeks of life. Counseling to reduce stress. advocacy to provide assistance with Job finding. or the provision of social support may be.Just as critical as nutritional counseling or medical monitoring. Regardless of the specific implications. it appears that if the results of this study are validated through the further testing of infants in neonatal intensive care units and the further testing of the household variables. the exclusive reliance on prenatal medical care may be replaced by broader. community supports for households expecting a baby. madman: 1. In light of the apparent importance of the sources of support and information. clarification is needed on the importance of these factors in the functioning of a household expecting a baby. The developmeht of a theory regarding the relationship between these and other variables that appear to affect the outcome of pregnancy is r 129 needed. It is recommended that the mothers and fathers of the infants in the study be interviewed to assess their observations concerning some of these important variables. as a basis for theory development from a grounded perspective. 2. The relationship of household variables in the general popula- tion to the outcome of pregnancy needs to be assessed. This is a more difficult task than the usual epidemiological studies because the latter have been able to rely on public records such as birth registra- tions. Assessing household factors requires either a mail question- nai re or interviewing members of the household. It is recommended that studies of this nature be done through sampling households and making contact with mothers in hospitals shortly after a baby is born. iAn understanding of the relationship of household variables to outcome in the general population would lend support for a household intervention rather than relying exclusively on intervention through the mother. 3. The present study should be replicated with households of infants who have died in other neonatal intensive care units and matched samples of households of surviving infants. It is recommended that studies be done in low-incomeras well as heterogeneous populations to assess differences in the path analysis by income level. 4. Should support be found for the results of the present study. new approaches should be developed to reduce infant mortality and to prevent the need for intensive care of neonates. APPENDICES 130 APPENDIX A HUMAN SUBJECTS APPROVAL 13] 132 MICHIGAN STATE UNIVERSITY umvutsmr comma; ox sesame" mourns-c _ us: we - mans»: - um wuunssuanxxsumnuam m mmmnos auatousc fininsnu. April 5, i983 Dr. David J. Kallen - Pediatrics 8 Human Development Dear Dr. Kallen: Subject: Proposal Entitled, "Infant Mortality: High Risk infants Who Live or Die” UCRIHS review of the above referenced project has now been completed. I am pleased to advise that the rights and welfare of the human subjects appear to be adequately protected and the Committee, therefore. approved this project at its meeting on April A. 1933 . . You are reminded that UCRIHS approval is valid for one calendar year. If you plan to continue this project beyond one year. please make provisions for . obtaining appropriate UCRIHS approval prior to the anniversary date noted above. Any changes in procedures involving human subjects must be reviewed by the UCRIHS prior to initiation of the change. UCRIHS must also be notified‘ promptly of any problems (unexpected side effects, complaints. etc.) involving human subjects during the course of the work. Thank you for bringing this project to our attention. If we can be of any future help. please do not hesitate to let us know. Sincerely, Henry E. Bredeck Chairman, UCRIHS HEB/jms 133 FRANCIS A. BREEN, JR., M.D. PlOrtlIIONAL CORPORATION P.0. 00X 30400 P iZlS E. MICHIGAN AVENUE P LANSING. MICHIGAN 40909 «acne-mum: FRANCIS A. IREEN. JR.. M.D.. F.A.C.P. TELEPHONE (517) mmmmcuoucomav 483- 90 mmvmv ‘ 433-2891 PETER B. BULICK. 0.0. mmmmmv “mm September 5 , 1984 Ann Stirling Johnson, A.C.S.H. Dept. of Pediatrics/Human Development 8-240 Life Sciences Building Michigan State University East Lansing, HI 48824-1317 Dear Hs. Johnson, The Edward II. Sparrow Hospital Institutional Research Review Comittee at its regular monthly meeting on September 4. 1984, approved continuation of your study ”Infant Mortality: High Risk Infants IIho Live or Die". The Comittee - had some reservation that the length and complexity of your questionnaire night discourage response. Sincerely. fie/5.... Francis A. Breen, Jr.. II.D., Chairman Institutional Research Review Connittee 131. MICHIGAN STATE UNIVERSITY UNIVERSITY COMMITTEE ON RESEARCH INVOLVING EAST LANSING ' ”Dim ' “ISM wumx sumac“ iucnms; 238 ADMINISTRATION BUILDING unuusuu. November 6, l98h Dr. David J. Kallen Pediatrics 5 Human Development' Dear Dr. Kallen: Subject: Proposal Entitled, "Children Who Live and Die" UCRIHS review of the above referenced project has now been completed. I am pleased to advise that the rights and welfare of the human subjects appear to be adequately protected and the Committee, therefore, approved this project at its meeting on November 5, l98A. You are reminded that UCRIHS approval is valid for one calendar year. If you plan to centinue this project beyond one year, please make provisions for obtaining appropriate UCRIHS approval prior to November 5, l985. Any changes in procedures involving human subjects must be reviewed by the UCRIHS prior to initiation of the change. UCRIHS must also be notified promptly of any problems (unexpected side effects, complaints, etc.) involving human subjects during the course of the work. Thank you for bringing this project to our attention. If we can be of any future help, please do not hesitate to let us know. Sincerely, We Henry E. Dredeck Chairman, UCRIHS HEB/jms cc: Ann Johnson mt-WukWWh-afiulba APPENDIX B NICU PROJECT INSTRUMENT l35 l36 CODE NUMBER NEWATAL INTENSIVE CARE UNIT STUDY QJESI'IGINAIRE A cooperative effort between E. ll. Sparrow Hospital, Michigan State University and the fanilies of NICU babies. Hay, 1985 l37 THERE ARE A FEW IMPORTANT THINGS YOU NEED TO Klilm1 BEFORE BEINNING THIS QUESTIONNAIRE. PIEASE TAKE A WT TO READ THIS: 1. This questionnaire should be filled out by the woman whose infant was in the Neonatal Intensive Care Unit at Sparrow Hospital; 2. k'nen we refer to pregnancy, we are asking about our re nancy for the child who was in the Neonatal Intensive Care Unit at Sparrow Hospital. 3. khen we talk about household, we are asking about the people with whom you lived. This might include someone who was not part of your family. 4. Some questions refer to "the baby's father." Some questions refer to "your husband or partner." For some people this will be the sane person. For some people this will be a different person. 5. Some questions may appear to be the same. In one place we ask about the number of living children you have had. In another place we ask for the number of children in the household. 'Diese may be the same, but they may be different. There there have been remarriages or divorces, children also get separated. Try to answer the question just as it is written. 6. Answer every question unless you are directed otherwise. Put one answer for each question unless you are directed otherwise. 7. If you have any questions about the study, please call Ms. Joi'mson, (517) 355-6092 or (517) 353-5042. You may call collect to the first nunber if you live outside the Lansing area. 8. Your help is very much appreciated. PIEASETHLUSABOUI'll-IEPEDPIEINYGJRl-DUSB-IOID. WEWUIDLIKE'IDKMN ABOLHIHEPEDHEWIIHWYOULIVEDSDIIONDISBHDREYWRPREBNANCYAND DURING THE PREGNANCY FOR THIS OIILD. Household Information: Six lbnths Dm‘in more fiegnancy Eegnancy Ihat was the total mmber of people who lived in the household. Do not count renters unless they ate meals and were part of your household activities. (Fill in nunber). How many pre-school children lived in your household? (Fill in nuuber). How many school-age children lived in your household? (Fill in nunber). I38 Household Information: Six Months Durin more Eegnancv fiegnancv Your marital status: _1. Married _1. Phrried _2. Divorced __2 . Divorced _ . Single __3 Single __4 . Widowed _4 . Widowed _5. Separated _5. Separated Has the father of the baby living in the household? (Check). _1. Yes 1. Yes _2. No _2. No mat other adult was living in your household with you? (Check) __1. No one _1. No one _2. Parent _2. Parent _3. Other _3. Other Hhat pets did you have in the household? (Check). _1. None _1. None _ . Cat _ . Cat __ . Dog _ . Dog 4. Bird 4. Bird _5. Fish _5. Fish _6. Other _6. Other Did you change your residence? (Check). _ . Yes _ . Yes _ . No _2. No i-iowmanybedroomsdidyouhaveinyour apartment or house? (Fill in umber). How many bathrooms did you have in apartment or house? (Fill in umber . Here you going to school during these time periods? (Check). _1. Yes _1. Yes _2. No _2. No Has the baby's father going to school during these time periods? (Quack). _é. Yes —%. :35 . No . At the time this child was born, what was the highest level of education you had completed? (Circle nunber). School Oollege/‘l‘echn Graduate/Prof . 123456789101112 1234 12345678 At the time this child was born, what was the highest level of education the baby's father had completed? (Circle timber). ' School College/Team Gradua te/PrOf . 123456789101112 1234 12345678 l39 HE IUJID LIKE TO m Am YOU. lbw such did you weigh before this pregnancy? __pomds. How mach weight did you gain during this pregnancy? __potmds. How tall are you? feet inches Before this pregnancy, had you ever had (check): 1. A stillborn child . N1 abortion . A miscarriage 4. An infant who died . A handica child who lived :6. A child died Before this pregnancy, how many living children had you had? (timber). Has thislpr;gnancy a plamed pregnancy? . es _2. No How my months pregnant were you when you first noticed the pregnancy? (Check). _1. less than 1 month ___2. 1-3 months _3. 4-6 months _Aa 7-9 mum Mdidwairstgotoadoctorforthispregnancy? __1. less than 1 month _2. 1-3 months _3. 4—6 months 4. 7-9 months 5. During labor Pleasecheckall theplaces youwent forcedicalcarewhileymwerepregnant. 1. Fri—vita physician _2. Midwife _ . Sparrow Hospital Clinic . St. lawrenca Hospital Clinic _5. ibelth Central ___6. Other .1 llio Here you worried about being able to pay for prenatal care? (Qieck). _1. Yes :2. No Did anyone refuse to give you prenatal care? (Check). 1. Yes :2. No l(ias an one else in your household pregnant when you were pregnant with this child? (heck . _1. Yes _2. No Here are some reasons why people seek medical care durin their pregnancy. l-hich were true for you? Qieck a__l_l the reasons that app y to you. . Felt sick . Hanted to see if you were pregnant 'Ihink every woman should have prenatal medical care . Needed some information on pregnancy Hanted to make sure baby was healthy l-hsband or other adult wanted me to get care Other reason some reasons why peeple put off going for prenatal medical care. Please 1 the reasons that were true for you. 1. Didn't know where to go for prenatal medical care Did not want prenatal medical care Did not have enough money Did not have medical insurance Did not need medical care Did not have anyone to stay with other children Couldn' t get time off from work ihsband or other adult didn' t want you to go . Didn't like long wait in doctor' s office __10. Other reason EA: :5 I .JsrU UNI“ lljjyj. If]. I. l Plat onecheck by all the people who assisted you in any way during this pregnancy (mane s, , money, health care, transportation, child care, someone to talk withadVi 'se, etc). Do not count labor and deliverL Put {we chech by all the people who you found very helpful and supportive. band or partner Members of your household anily outside your household friends Neighbors Physicians Other medical people (nurses, nutrition counselor, health educators) Social Workers . Counselor or Psychologist 10. Support group 11. Other I IJJJJJINI O llIl lbw many different physicians examined you during this pregnancy, not countigg labor and delive ? (Check). _1. fine _2. Only one :4: Three _5. Fom: or more Did you have trouble finding a place to get medical prenatal care (care durigg mm (Check). 1. Yes _'—2. No How did you learn about what you should and should not do during pregnancy? Please check g_l_l that apply. ___1. Already knew fran previous experience __2. Husband or partner 3. tunbers of your household _4. Family outside your household 5. Friends 6. Neighbors . Physicians ~ 8. Other medical people (nurses, nutrition camselor, health educators) ___9. Social Workers _10. Radio and television _11. Preparation for childbirth class _12. Other classes _13. Boole, magazines or newspapers _14. Other MNDCIQJESIIONSASKYWIOSAYIWWORWOFIENWEGW GIMME. MEAREMEEIDRDSMTDECRIBEPOSSIBIEANSUERS. LNDERNEA‘IH'IHEIDRDSARESMERS. PLEASECIRCIE'Il-IENIHBER‘IHATMIUIES YCXJRANSWER. FORM: How helpful was your mther during this pregnancy? Not at all 2 Sage ' 4 Very gelpful 1 IFYUJRWNAS'WUIATAIL"HE1PFUL,YWWIDCIRC1£1. IFSHEHAS. WID"SOIE" 5mm, CIRCLE 3. IFSHEHAS'VERY HEIJ’FUL" C1303 5. ZANleAREINBENEEN THESEII'REEI’OINI'S. He had (enough money to get along on. Never Sane of the time Always 1 2 3 4 5 lie had enough food. Never Sane of the time Always 1 2 3 - 4 5 l42 lie had transportation . Never Sane of the time Always 1 2 3 4 5 lie had a telephone. Never Sane of the time Always 1 2 3 4 5 We had enough heat in the house. Never Sane of the time Always 1 - 2 3 4 5 We had a television Never Sane of the time Always 1 2 3 4 5 lie had enough clothing. Never Some of the time Always 1 2 3 4 5 We had trouble with members of our family outside the household. Never Sane of the time Always l 2 3 4 5 We had trouble with people in our neighborhood. Never Some of the time Always l 2 3 4 5 How hel ful were the ysicians who examined you duri your regnanc ? Do notpinclude laborfla‘nd deliverv). ng p y t p u erately helpful Very helpful 1 2 3 , 4 5 How much instruction did your doctor or other medical people give you on what you should eat? None Sane A lot 1 2 3 4 5 How much did you follow their instructions on what you should eat? Not at all Sane Calgletely 1 2 3 4 How ouch instruction did your doctor or other medical people give you on sucking and drugs? None Sane A lot 1 2 3 4 5 How much did you follow their instruction on smoking and drugs? Not at all Sane ' Completely 1 2 3 4 143 To what extent were you separated from your lusband or partner during your wearer‘s—y? Never separated Some Frequently 1 2 3 4 5 How much help and support did you get fran your husband or partner? None Sane A lot 1 2 3 4 5 How nuch did you worry about money during your pregnancy? No worries at all Some worries Horrigd a lot 1 2 3 4 How much did you argue with your husband/partner about money or other things during this pregnancy? Did not argue at all Saneaargunents Argueds a lot 1 2 4 How nuch did you argue with other people about money or other things during this pregnancy? Did not argue at all I Sane argunents Argued a lot 1 2 3 4 5 . How difficult was it for you to get medical care during this pregnancy? No difficulties 2 Sane3difficulties A lot of difficulty 1 4 How satisfied were you with your relationship with your lmsband/partner before your pregnancy? Very dissatisfied (K Very satisfied 1 2 3 4 5 How satisfied were you with your relationship with your husband/partner during this pregnancy? Very dissatisfied CK Very satisfied 1 2 3 4 5 Overall, how satisfied were you with your life before this pregnancy? Very dissatisfied 2 OK Very satisfied 3 4 5 Overall, how satisfied were you with your life during this pregnancy? Very dissatisfied QC Very satisfied 2 3 4 5 w. TIE NEXT QJESTIONS (INCERN TIE DIPIDYMENT SITUATION IN YOUR HOUSEHOLD. LET'S START WITH YOU AND YOUR WRK WIRIBICE IN TIE SIX I'DNTHS BEFORE YOU (171' PREIENANI VITI-l THIS CHILD. Did you have a job? (Quack). _1. Yes, a full-time job.l , _2. Yes, a part-time 'obJ .__3. No, but I was 100 ing for work. 4. No, and I was not looking for work. 5 If yes, were you working for the full six month period? (Check). 1. Yes :2. No, only part of the six months. During the six months prior to this pregnancy, did you get laid off, lose a job, quit a job, or get a new job or get a pranotion? (deck). _1. No, my employment status did not change. _2. Yes, I was laid off or lost a job. _ . Yes, I quit a job. 4. Yes, I got a new job. 5. Yes, I got a pranotion. If you were working at all before you got pregnant, what was your job title and what did you do? I Has there a time when you were worried that you might be laid off whether or not you actually were laid off? (Qieck). _1. Yes __2. No How did you feel about your work situation just before you got pregnantcwzith this child? (Check). 1 M, :2. Neither good nor bad 3. M m IURK DEERIENCE HIRING YOUR m: Did you have a job? (Check). _1. Yes, a full-time job.'L _2. Yes, a part-time job.J l ;__3. No, that I was looking for work. _4. No, and I was not looking for work. '~ If yes, were you working all during this pregnancy? (Check). 1. es :2. No During this pregnancy, did you get laid off, lose a job, quit a job, get a new job or get a pranotion? (Check). _1. No, my employment status did not change. _2. Yes, I was laid off or lost a job. _3. Yes, I quit a job. _4. Yes, I got a new job. V, __5. Yes, I got a pranotion. HIS If you were working at all during this pregnancy, what was your job title and what did you do? Has there a time when you were worried that you might be laid off whether or not you actually were laid off? _1. Yes _ 2. No How did you feel about your work situation during this pregnancy? (Check) . 1 Good V :2zibithergoodnorbad 3. Bad NGJWEIDUIDLIIETUWABOUIUIIERPEDPIEWYOURIDUSBDIDEIDWEREINTIE IABOREORCEINTIESRIDNIIISBETOREYOUGOTPREGNANIWITHTIIISGIID. who usually csngributed the most money to the household income? (Q'ieck). ' . I i :2. My husband/partner _3. Parent ___4. Other person in the household Not counting yourself, how many other people in your household were working? (Fill in umber). Not counting yourself, how many other le in your household were unemployed and looking for work? (Fill in wtberg. PLEASEANSWERTIEIDWQUFSTIGISAWTOEQEMPERSWWDUSUAILY Q‘NI'RIBUTED TIE MST TO TIE I-DUSEI-DLD INCIME. II" lbw-ELSE BESTDES YOURSELF OJNI'RIBUI'ER TO TIE musmow INCDE, SKIP TO TIE SECTION CALLED "DURING YOUR PRENANCY. Has this person working before this pregnancy? (Check). __1. Yes, a full time job\ 2. Yes, a part-time jobJ l .:3. No, but he/she was looking for work. _4. No, and he/she was not looking for work. If yes, was he/she working for the full six month period?(Check). 1. Yes _2. No Before this pregnancy, did he/she get laid off or lose a job, quit a job, et a new job or get a promotion? (Check). _1. No, hisgher employment status did not change. _2. Yes, he/she was laid off or lost a job. _3. Yes, he/she quit a job. _4. Yes, he/she got a new job. _5. Yes, he/she got a promotion. If he/she was working at all before the pregnancy, what was his/her job title and what did he/she do? \ M7 E‘EIDUIDLIKETOKNOWIFYOURECEIVEDANYUDERFORMSOFINMORSUPPORT BEI'OREANDDURINGYOURPRHZNANCYFORTHISOIIID. (CIRCLEYESORNO). Benefits: Six Months Duri —— Mammy 5.3—3 gnancy unemployment benefits Yes No Yes No Food stamps Yes No Yes No Public Assistance Yes No Yes No Health insurance (any kind) Yes No Yes No Did insurance cover pregnancy? ' Yes No Yes No WE WOULD LIKE TO KNW APPROXIMATELY l-W MUG-l INCGE YOUR IDUSDDLD RECEIVED DURING TIE TIJO TIME PERIODS. PLEASE INCLUDE INCIHE FRO"! ALL WEIS, COVERNMENI PAYMENTS , PAYNE-TS FRO“! CHER INDIVIDUALS SUCH AS CHILD SUPPORT AND REGULAR GIFTS. INCLUDE INTEREST W SAVINGS. INCUJDE RENI‘ f'RG‘l RENI'ERS AND BOARDERS, BUT NOT THEIR WIRE INCfl‘E UNLESS IT IS AVAILABLE FOR YGJR USE. INCLUDE SIUDENI LOANS AND SCHOLARSHIPS FOR LIVING EXPENSES. Please add up all the amounts and check which best describes your income. If it is easiest to figure monthly, lease use the "lncme Per Ibnth" chart. 'If it is easiest to figure yearly, p ease use the "Income Per Year" chart. Do not use both. ' Income Per Month: Six anths 9513-13 Efore egnancy Eegnancy Less than $400 per month Between $400 and $699 per month Between $700 and $999 per month Between $1,000 and $1,29942er month Between $1,300 and $1,599 per month Between $1L600 and $1,899 per month Between $1,900 and $2,199 per month Between $2,200 and $2,699jer month Between $2,500 and $2,799 per month Between $2,800 and $3,099 per month Between MOO and $3,399 per month More than $3,400 per month 148 Income Per Year: Six anths 11min Efore egnancv "Eegnancy Less than $4,800 per year Between $4,800 and $8,399 per year Between $8,400 and $11,999 per,year Between $12,000 and ,515,599 per year Between $15,600 and $19,199 per_year Between $19,200 and $22,799 per_year Between $22,800 and $26,3994per year Between $26,400 and $29,9994per vear Between $30,000 and $33,599 per year Between $33,600 and $37,199 per year Between $37,200 and $40,799 per year More than $40,800 per year UEWLIKETDKNWAPWOXIMATEYIWWO-l INCIIETIEMALEIEADOFIUJSBDLD MIBUTED TO TIE HOUSDIOLD INQNE DURING TIDSE “mo '1' . HEAD OF HOUSEIDLD IS NOT NECESSARILY THE SAME PERSW YOU IDENTIFIE IN TIE DEW—Tm W %%!I , . . IS DE STUDENT RECEIVING IT. WNOI‘INCLDDEINCDEERG‘IULIERSWRCES. ‘me following are the sure charts used for household income. hat a check in each of the two time periods next to the income level. If it is easiest to figure monthly, use "Income Per Month" chart. If it is easiest to figure yearly, use "Income Per Year" chart. Do not use both. whose income are you listing here? (Quack). 1. Husband 1. limb-and 2. Partner _2. Partner :3. Your :3. Your father father _4. No male _4. No male head head (If no mle head.of busehold, go to page 14) 1’49 Income Per Month: Less than $400 per month Six Months ore $51-95 gm, Between $400 and $699 Jer month Between $700 and $999 per month Between $1,000 and $1,299 per month Between $1,300 and $1,599 per unnth Between $1,600 and $1,899 per month Between $1,900 and $2,199 per month Between $2,200 and $2,499 per month Between $2,500 and $2,,799Jer month Between $2,800 and $3,099 per month Between $3,100 and $3,399 per month More than $3,400 jer month Income Per Year: less than $4,800 per year Six Months ore PEegnancy During m Between $4,800 and $8,399 per year Between $8,400 and $11,999 per year Between $12,000 and $15,599 per year Between $15,600 and $lil9iper year Between $19,200 and $22,799 per year Between $22,800 and $26,399 per year Between $26,400 and $29,999 per year Between $30,000 and $33,599 per year Between $33,600 and $37,199 per year Between $37, 200 and $40,799 per year More than $40,800 per year 150 NE WILD LIKE TO mow AWI' m GIN FAMILY"S INm‘E. l-low mach of your life have you been poor (not had enough food, clothing, shelter). (Check). ___1. Not at all 2. A quarter of my life _3. Half of my life _4. tbst of my life . Olly for brief periods J Hut—w; your fanily's financial situation when you were less than 12 years old? (Check) . 1. We couldn't by some things that were necessary. . We could buy necessities only. . We could afford sane of the things we wanted, but not everything. . He could afford about everything we wanted. . 9e IJJJJ could afford abwt everythim we wanted and hd sane left over. PLEASET'HLUSABOIHYWRSHIDURBYGTIETWTDEPERIODS:SD(WSBEFDRE Six Months Dung ore egnancy egnancy Did you smoke cigarettes (at least one, _1. Yes ___1. Yes but lessthanapackaday).(01eck). _.No _.No Didyousmokemorethanapackof _.Yes _.Yes cigarettes a day? (Check). _2. No _ . No Did yaromdywsmke frecpently? _. Yes _.Yes (am 0 _2. NO — 0 Lb Didyoulive or work inanarea of air _. Yes _. Yes pollution? (check). _2. No _2. No Did you take dilantin or any other medication prescribed for seizure _1. Yes _1. Yes disorders? (Check) _ . No _ . No Did you take insulin? (deck). _ . Yes _ . Yes _ . No __ . No Did you drink caffeinated coffee or _ . Yes _ . Yes sodapop? (Check). _.No _.No Did you have high blood pressure? _ . Yes _ . Yes (Q1eck). -_ . No _ . No 151 WEAREINTERESTEDINUIEII‘ERCERTAINWENI‘SHAPPM INYOURHOUSEHOLD. IFTIEYDIDHAPPB‘,UEWLIKETOWKNYOUFELIABOUTEACBOFW. IEREAREPOURSTATWI‘SNU'IBEREDPRO‘SITO4. AFTEREAGIEVENI‘BEJJN, CIKJEI OFTIEI'UJR NUMBERS IN FAG! SPACE, DEPENDINGONWHIG! STATEMENT IS TRUE. 1. It did not happen 2. It happened but it did not worry me. 3. It happened and it worried me somewhat 4. It happened and it worried me a lot. PORDIAMPLE, IFWEGOIREALLYSICXINYOURWSWOIDSDU‘DNIHSBEFORE YOURPREGNANCYANDITWRRIEDYOUALUI', PUT4. IF MCNEVASSICXDURINGYGJRPREBNANCX,PUT1. INCLUDEYOURSEIIANDENERYONE IN YOUR musmou). Six months . tor In our household. WV lDu :ingm I Hsaneone gotmarried 12341234 someone was born (do not count NICUbaby) 1234 1234 someone got really sick 1 2 3 4 1 2 3 4 someone smoked cgarettes a lot 1 2 3 4 1 2 3 4 someonehad their marriage breakup 1 234 1 234 someone stopped smoking_c_i_.garettes 1 2 3 4 1 2 3 4 someone got drunkoften , 1 234 1234 someonedied , 1234 1234 someone got pregnant (do not include _pgggnancyforchildwhomsinNICU) 123 1234 someone had trouble with the law 1 3 1 2 3 4 someone was using over-the-counter drugs such as aspirin 1";3 4 1 2 3 4 ‘ someonewasusin escrition 3 L234 1234 smeonewasusitghomersnedies 1234 1234 ’mlflgmm f" 1234 1234 someone's enployment benefits ran out 1 2 3 4 1 2 3 4 someonewasturneddownforfoodstamps 1234 1234 someonewasturneddownforADC 1234 1234, someone lost health insurance (am/kind) 1234 1234 152 IE IDULD LIKE TO KNCM WHAT HAPPENED AFTER YOUR BABY WAS BORN {hat was the outcome for the child born in the NICU? (Check) ___.1 It was born sick, and died. :2. It was born sick, but survived and it healthy. :3. It was born sick, but survived with some handicaps. Please describe handicaps How long was your baby in the Sparrow Hospital Neonatal Intensive Care Unit? (Write in numbers). months weeks days hours How often did you see your baby in the NICU? (Check). 1. Not at all . Once a week 3. A few times per week . (rice a day . More than once a day 6. Other: describe Did you participate. in the HOPING group meetings? (Queck). _1 . Yes _2 O M Was your family helpful and supportive during the time your baby was in the NICU? _1. Yes _2. No _3. Somewhat Were the families of other NICU babies helpful and supportive during this time? . Yes . No Sanewhat . I had no contact with other families IJJJ_ 153 HERE ARE A NUMBER OF WAYS THAT PEOPLE SGIEI'IMES FEE. ABOUT HOSPITALS. FOR PAC}! ONE, PLEASE SAY MED-ER OR NOT YOU FELT THIS WAY WHEN YOUR GIILD WAS INTIENICU. GECKYESORNO. I was treated with dignity. _1.Yes _2.No I was made to feel I was not respected as a person. _1.Yes- _2.No I was made to feel I was behaving 'responsibly. _1.Yes _2.No I was made to feel the doctors were really interested in me as a person. _1.Yes: _2.No I was made to feel the nurses were really interested in me as a person. _ _1.Yes _2.No I was made to feel I was just another nunber in the system. _1.Yes _2.No I had a good chance to have my questions about treatment answered. _1.Yes __2.No I had a good chance to talk to people about how I felt about what was happening in the hospital. _1.Yes _2.No I really felt alone and isolated during the hospitalization. __1 .Yes _2.No .I felt upset and frightened a lot. _1.Yes _2.No Mast of the time I felt I could cope. _1.Yes __2.No ISTIEREANYIIIDBHSEYOU'DLIKETDTELUS? HIASEPEI'REETOUSE TIEBURASPACEWTI-IISSIEEI‘. TI-IANKSFORHEIIING: YOUAREWITIDTOREIIEIVEAWRITMREPORTOF'DIISSTUDY. IFYOUW LIKEAODPY,PIIASEFIILINYOURNA‘IEANDADDRESSWTIEIASTPAGE. IFYOUIDLTDUIETOBEDHHWIEHED,PLEASESIGNTIEWSPNTMMWTIE LASTPAGE. WWAREWE,PLLASEHITTIEQJISTIWNAIREINTIEHKIDSEDHWEIDPE ANDMAILIT. TIEWEIDPENESNOI‘NEEDANYPOSTAGE. 15‘: YES, IIDULDLIKET’ORECEIVEAURIT'I'ENREPORTONTIESIUDY. NAME: STREET AND NLMBER: CITY: STATE: ZIP (DOE: YES, I W BEWTLLINGTO BE INI'ERVIBJEDABOUI‘HYDCPERIBVCES. SIGNATURE: ME: STREET AND NLMBER: STATE: CITY: ZIP cons: T'ELEPIDNE NLMBER: APPENDIX C COVER LETTERS 155 156 MICHIGAN STATE UNIVERSITY mm 0! EXAM/HUMAN WENT EAST we 0 mm a “4.15.7 DO I!!! sauces May 6, 1985 we are seeking information about the situations of families whose babies were in the Neonatal Intensive Care Unit of l. H. Sparrow Hospital within the last three years. Sometimes it is unknown what events, if any, leading up to the birth of a child, make it necessary for an infant to need the special care of e Neonatal Intensive Care Unit. So we are asking a number of women whose babies were in the NICU many questions in the hopes that we can begin to find out what problems women and fami- lies hava during pregnancy that may make this care necessary. Records show that you had a child in the NICU during that time. Be hope you will help us understand the conditions that may result in a new-born baby's needing this special help. we realise the hours, days or weeks when your baby was in the NICU were probably very difficult for you. we ask that you spend a few minutes now so that we can help other mothers, families and babies. Of course, your participation in this study is voluntary and confidential. Your answers will be combined with those of other women so that no one except the re- searchers will ever know what your answers are or if you returned the questionnaire. The envelope provided for returning the questionnaire does not need postage. Va realise that you may have some information for us that is not on the question- naire. or you may have some questions about the study. If so, please call Ms. Johnson at ($17) 335-6092 or (517) 353-5042. You may call collect to the first number if you live outside the Lansing area. At the and of the questionnaire is a place for you to indicate if you are willing to be interviewed. You are entitled to receive a copy of a written report of the study. If you would like one, please put your name and address on the last page. we welcome your participation. Sincerely, ‘ (5:;15lo—IL. (:1. §:>° ‘LE‘JLF'TD Bugana A. Dolanski, M.D. Ann Johnson Medical Director, NICU Project Manager, NICU Study I. V. Sparrow Hospital Por David J. Kallen, Ph.D. Professor Department of Pediatrics and Human Development And Robert J. Oriffora, Ph.D. Associate Professor and Chair Department of Pamily and Child Ecology enclosure Wham-“WWW 157 MICHIGAN STATE UNIVERSITY ”Am 0’ mum/Hum DEVELOPMENT EAST LANSING 0 MICHIGAN 0 0124-1317 .1“ "FE SCIENCES May 30, 1985 About three weeks ago, I wrote you asking about your experiences during your pregnancy for a baby who was in the Neonatal Intensive Care Unit at E. W. Sparrow Hospital. The researchers involved have undertaken this study because we believe it is important to understand the circumstances of mothers and families during this time. I am writing to you again because of the importance of each response. ‘me information about your household will retain confidential. It will be combined with information about other households where an infant has been in the NICU. In case the questionnaire has been misplaced, a replacement has been enclosed with an envelope that does not need postage. Thank you so reach for your help. Sincerely , Ann Jotmson NICU Project l'hnager flw enclosure HSU'm - MAM/Eyed W)! hesitation REFERENCES 158 REFERENCES Abramowicz. H.. & Kass. E. H. (1966). Pathogenesis and prognosis of Prematurity- Ihe New England .Lcucnal of Medicine. 215. 878-885. 938-943. 1001-1007. 1053-1059. Adelman. I. (1963). An economic analysis of population growth. American Economic 821191: 53. 314-339. Admcefieeeueiflnalflatalinfiaiisiicsilm. (1984a. September 28). Vol. 3. no. 9. Washington. DC: U.S. Department of Health and Human Services. National Center for Health Statis- tics. mammnflmmmnnmmmz. (1984b. December). Vol. 3. no. 9. Supplement. Washington. DC: U.S. Depariment of Health and Hanan Services. National Center for Health Statistics. Altenderfer. M. E.. & Crowther. B. (1949). Relationship between infant mortality and socioeconomic factors in urban areas. .Euhlic .Hfiilih.3929£1§o fifiill). 331-339. Amante. D.. & Brandt. N. (l978). The social ecology of obstetrical competence- Social Science and Medicine. .12. 391-401. Andrews. M. P.. Bubolz. M. H.. & Paolucci. B. (1980). An ecological approach to study of the family. Haulage and Eamflx Bender. ail/2). 29-49. Aubry. R. H.. & Nesbitt. R. E. (1969). High risk obstetrics. I. Peri- natal outcome in relation to a broadened approach to obstetric care for patients at special risk. .Amezlcan.leunnal.eI.Qh§IeI£125 and mnecejeex: 241-247- Barden. T. P.. & Knowles. H. 0.. Jr. (1981). Diagnosis of diabetes in pregnancy. .Qlinical Obstetrics and sinecelecxo 24. 3-19. Bergquist. C. L. (1984. March)..Qurreni.2ersnectixes.cn.nrenaia1 .Qane. Lansing. MI: University Associates. 159 160 Birkel. R. C.. & Reppucci. N. D. (l983). Social networks. information- seeking. and the utilization of services. .Amenican.lounnal.oi Summit! Esxcholmx. 2(2). 185-205. Boardman. V.. Zyzanski. S. J.. & Cottrell. L. S. (1975). School absences. illness. and family competence. In B. H. Kaplan a J. c. Cassel (Eds.). Eamflxandhealthehneoidemolocioal .annnoach. Chapel Hill: University of North Carolina. Boone. M. S. (1982). A socio-medical study of infant mortality among disadvantaged blacks. .fluman.QLgan12atlon. 41(3). 227-236. Brenner. N. H. (1973). Fetal. infant and maternal mortality during periods of economic instability. .Intennationa1.lounnal.of health Seances. 3 (2). 145-159. Bronfenbrenner. U. (l979). Ihe acolggy of human deyelonment. Cambridge. MA: Harvard University Press. Brooks. 0. H. (1975). The changing relationship between socioeconomic status and infant mortality: An analysis of state characteris- tics. Journal of Health and Social Behavior. .16. 291-303. Brooks. 0. H. (l980). Social. economic and biologic correlates of infant mortality in city neighborhoods. ,Jounna1.of.flea1th.and Social BehaxJor. 21. 2-11. Brooten. D. (l983). Issues for research on alternative patterns of care for low birthweight infants. .Image;.Ihe.lou£na1.of.uunsing Scholarshio. 15(3). 80-83. Cassel. J. (l970). Physical illness in response to stress. In S. Levine 3 N. A. Scotch (Eds.). Social 51:25:. Chicago: Aldine. Cassel. J. (1971). Health consequences of population density and crowd- ing. In R. Revelle (Ed.)..Ba21d emulation sheath: and policy implications. Baltimore: Johns Hopkins University Press. Chamberlain. G. (l979). Background to perinatal health. .Ihe.Lancet. 2(8151). 1061-1063. Chamberlin. R. N. (l984). Strategies for disease prevention and health promotion in maternal and child health: The "ecologic" versus the "high risk" approach. lounnal of. Public Hamil E91131. 5(2). 185-197. Chase. H. C.. & Nelson. F. G. (1973). Education of mother. medical care and condition of infant. Mariam Jamal of Public Health. Supplement. o3. 27-40. 161 Children's Defense Fund. (1984). Amenigan.ohildnen.1n,noyorty. Chrisman. N.. 3 Kleinman. A. (1980). Health beliefs and practices. In S. Thernstrom (Ed.)..Harxard.£ncxc1ooedia.ot.American.Ethnic .Gnonpa (pp. 452-462). Cambridge. MA: Harvard University. Cohen. S. E.. Signan. H.. Parmalee. A. H.. 3 Beckwith. L. (l982). Perinatal risk and developmental outcome in preterm infants. Seminars in Eerinatoloox. 5(4). 334-339. Crump. E. P.. Horton. C. P.. Masuoka. J.. 3 Ryan. 0. (1957). Growth and development: Relation of birth weight in Negro infants to sex. maternal age. parity. prenatal care and socioeconomic status. Journal of. Pediatrics. 51. 678-697. Davies. 0. P.. Gray. 0. P.. Ellwood. P. 0.. 3 Abernethy. M. (l976). Cigarette smoking in pregnancy: Associations with maternal weight gain and fetal growth. .Ihe.Lan£et..1(7956). 385-387. Deibel. P. (l980). Effects of cigarette smoking on maternal nutrition and the fetus. Journal of mm W and Neonatal hiureino. 2(6). 333-336. Dillman. D. A. (1978). Haiiandteleohonesuriexsilhetotaldesion .method. New York: John Wiley 3 Sons. Donabedian. A.. Rosenfelt. L.. 3 Southern. E. N. (l965). Infant mortality and socioeconomic status in a metropolitan community. Euhiic Health Reports. 31202). 1083-1094. Dott. A. B.. 3 Fort. A. T. (l97S). The effect of availability and utilization of prenatal care and hospital services on infant mortality rates: Summary of the findings of the Louisiana Infant "ortallty Study- Part II. American Journal of Obstetrics and W. 123(8). 854-860. D'Souza. S. N.. 3 Richards. B. (1978). Neurological sequelae in newborn babies after perinatal asphyxia. .Arohixes.of.fliseases.oI.Ch11d: 1129.0: 53.. 564-569. Dubos. R. (l965). Man adapting. New Haven. CN: Yale University. Dubos. R. (1968). 59 human an animal. New York: Charles Scribner's Sons. Duncan. 0. D. (1966). Path analysis: Sociological examples. .American Journal of Sociology. 12. l-16. 162 Dyhouse. C. (1982). Working class mothers and infant mortality in England. 1895-1914. Journal of Social History. .12. 248-267. Eastman. N. J.. 3 Jackson. E. (l968). Weight relationships in preg- nancy: I. The bearing of maternal weight gain and pro-pregnancy weight on birth weight in full-term pregnancies. Obstetrics and fixnecologx Surrey. 23(2). 1003-1025. Eisner. V.. Brazie. J. V.. Pratt. N.. 3 Hexter. A. (1979). The risk of low birth weight. American Journal of Eubllo Health. $19). 887-8% e Ekanem. I. I. (l972). A further note on the relation between economic development and fertility. Demography. 9.. 383-398. Elster. A. B. (1984). The effect of maternal age. parity and prenatal care on perinatal outcome in adolescent mothers. American Journal of .thtetrics and fixnecclogx. .132. 845-847. Erhardt. C. L.. 3 Chase. H. C. (1973). Ethnic group. education of mother. and birth weight. American Journal of Euhiic Health. Supplement. .63.. l7-28. Foote. N.. 3 Cottrell. L. 8.. Jr. (l955). ldentiiyandinternerecnal cmpetence. Chicago: The University of Chicago Press. Fox. R. I.. Goldman. J. J.. 3 Brumfield. N. A. (1968). Determining the target population for prenatal and postnatal care. Public Health BM, .63., 3s Frenzen. P. D.. 3 Hogan. D. P. (1982). The impact of class. education. and health care on infant mortality in a developing society: The case of rural Thailand. Demography. 12(3). 391-408. Friedman. E. A.. 3 Neff. R. K. (l978). Hypertension-hypotension in $403033. Journal of the American Medical Association. 232(21). Frisancho. A. R.. Matos. J.. 3 Flegel. P. ”983). Maternal nutritional status and adolescent pregnancy outcome. American Journal of .Qllnloal Nutrition. so. 739-746. Fuchs. v. (1983). How we lire: An economic oerspectixe on Americans ircm birth to death. Cambridge: Harvard University Press. Galle. 0. R.. Gove. N. R.. 3 McPherson. J. M. (1972). Population density and pathology: What are the relations for man? Science. .125, 23-30e 163 Geijerstam. G. (l969). Low birth weight and perinatal mortality. Public Health Reports. 8.4. 939-948. Gendell. N.. 3 Hellegers. A. E. (1973). The influence of the change in maternal age. birth order. and color on the changing perinatal mortality. Baltimore. l96l-66. .Health Serxice Reports. 82. 733 -742 e Gofin. J. (l979). The effect on birth weight of employment during pregnancy. Journal of Biosocial Science. 22. 259-267. Gortmaker. S. L. (1979). The effects of prenatal care upon the health of the newborn. American Journal of Euhlic Health. 52(7). 653-660. Gray. R. H. (1984). Maternal reproduction and child survival. .American Journal of Eublic Health. 1.4.(10). 1080-108l. Greenberg. R. S. (1983). The impact of prenatal care in different social groups. American Journal of. Obstetrics and .Gxnecolodx. 155. 797-780. Hall. H. H.. Chng. P. K.. 3 MacGillivray. I. (1980). Is routine antenatal care worthwhile? .Ihe Lancet. 78-80. Harlap. S.. 3 Shiono. P. H. (1980). Alcohol. smoking. and incidence of spontaneous abortion in the first and second trimester. ,Ihe Lancet. 2(8l87) . 173-176. Harris. J. E. (1982). Prenatal medical care and infant mortality. In v. R. Ruchs (Ed.). Economic aspectsof health (pp. 15-52)- Chicago: University of Chicago Press. Heady. J. A.. Daly. C.. 3 Morris. J. N. (1955). Social and biological factors in infant mortality. II. Variation of mortality with mother's age and parity. .Ihe.Lancet. 1(6680). 395-397. Hemminki. K.. Niemi. M. L.. Saloniemi. I.. Vainio. H.. 3 Hemminki. E. (1980). Spontaneous abortions by occupation and social class in Finland. International Journal of Epidemiology. 2(2). 149-153. Hill. R. N.. 3 Stern. L. (1979). Drugs in pregnancy: Effects on the fetus and newborn. .Drugs..ll. 182-197. Hobel. C. J. (l978). Risk assessment in perinatal medicine. .Clinical obstetrics and .Gunecologx. 21(2). 287-295. Holland. N. N. (1974). Screening for disease: taking stock. .lhe .Lancet. 1494-1497. 164 Holmes. T. H.. 3 Rahe. R. H. (1967). The Social ReadJustment Rating Scale. Journal of Euchosomatic Research. .11. 213-218. Institute of Medicine. (1985). Ereyenting leg hirtnueignt. Washington. DC: National Academy Press. Jacobson. H. N. (1980). A randomized controlled trial of prenatal nutritional supplementation (editorial). ,Eediatrics. Q5. 835- 836. Johnson. C. D. (1984. March). Eetimating the need for prenatal care. Lansing. MI: University Associates. Johnson. S. M.. 3 Snow. L. F. (l982). Assessment of reproductive knowl- edge in an inner-city clinic. .Social.Science.and.Medicine..lo. 1657-1662. Kaplan. B. H. (1975). An epilogue: Toward further research on family and health. In B. H. Kaplan 3 J. C. Cassel (Eds.). Eamilx.and health“. An eoideniclmical approach. Chapel Hill: University of North Carolina. Kehrer. B.. 3 Wolin. C. (l979). Impact of income maintenance on low birthweight: Evidence from the Cary experiment. .qurnal.oi.fluman Resources. .14. 434-482. Kessel. S. 8.. Rooks. J. P.. 3 Cushner. I. M. (l976). A child's begin- ning. Betterhealthicrourchildreninnationalstrateox- Vol. 4. Washington. DC: U.S. Department of Health and Human Services. Kessner. D. M.. Singer. J.. Kalk. C. E.. 3 Schlesinger. E. R. (1973). Infant death: An analysis by maternal risk and health care. In .Contrasts.in.health.status. Vol. 1. Washington. 00: Institute of Medicine. Khosla. T.. 3 Lowe. C. R. (l967). Indices of obesity derived from body weight and height. British Journal of Ereuentiire Social Medicine. 21. 122-128. Kim. J.. 3 Kohout. F. J. (1975). Special topics in general linear models. In N. H. Nie. C. H. Hull. J. G. Jenkins. K. Steinbrenner. 8- D. H. Bent (Eds.). Statisticalflackaoeiortheiccialficionces (20d 6d.)s ppm 368-397s Kitchen. W. H.. Ryan. M. M.. Rickards. A.. Astbury. J.. Ford. 6.. Lissenden. J. V.. Keith. C. 0.. 3 Keir. E. H. (1982). Changing outcome over 13 years of very low birthweight infants. .Seminars in Eerinatoloox. om. 373-389. 165 Knodel. J.. 3 Hermalin. A. I. (1984). Effects of birth rank. maternal age. birth interval and sibship size on infant and child mortality: Evidence from 18th and 19th century reproductive histories. American Journal of EuinC Health. 15.. 1098-1106. Korenbrot. C. C. (1984). Risk reduction in pregnancies of low-income women: Comprehensive prenatal care through the 08 Access Project. .Mohius. A. 34-43. Laukaran. V. H.. 3 vandenBerg. B. J. (1980). The relationship of maternal attitude to pregnancy outcomes and obstetric complica- tions: A cohort study of unwanted pregnancy. .American.Journal.oI .Qbstetrics and Gxnecoloox. 135.. 374-379. Lazarsfeld. P. (1959). Problems in methodology. In R. K. Merton (Ed.). .Scciolocy.tcday. New York: Basic Books. Lee. K. S.. Paneth. N.. Gartner. L.. Pearlman. M.. 3 Gruss. L. (1980). Neonatal mortality: An analysis of the recent improvement in the United States. American Journal of Rubiic Health. 2.0.. 15. Lewis. R.. Charles. M.. 3 Patwary. K. M. (1973). Relationships between birth weight and selected social. environmental and medical care factors. American Journal of Euhlic Health. cam). 973-981. Levy. 8. 8.. Wilkinson. F. 3.. 3 Marine. W. M. (1971). Reducing neo- natal mortality rate with nurse-midwife program. .Medical.Care. 11; 491-500. Lewit. E. (1983). The demand for prenatal care and the production of healthy infants. In D. Salkever. I. Sirageldin. 3 A. Sorkin (Eds.). Greenwich. CN: JAI Press. Liberman. M. A.. 3 Mullan. J. T. (1978). Does help help? The adaptive consequences of obtaining help from professionals and social net-rorks. American Journal of Community Esycholodx. s(5). 499-517. Lipowski. Z. J. (1973). Life events as stressors: A methodological 10901 ('7. Ccmorehensixe Esxchiatrx. 14(3). 203-215. Masuda. M.. 3 Holmes. T. H. (1967). The Social Readjustment Rating Scale: A cross cultural study of Japanese and Americans. .Journal of Esichosomatic Research. i1. 227. MacIntyre. S. (1977). The management of childbirth: A review of sociological research issues. .Socia1.Science.and.Medicine..11. 477-484. 166 McCormick. M. C.. Shapiro. 8.. 3 Horn. 8. D. (1979). The relationship between infant mortality rates and medical care and socioeconomic variables. Chile 1960-1970. International Journal .91 Evident .9199 . 3(2). 145-154. McCormick. M. C.. Shapiro. 8.. 3 Starfield. B. (1984). High risk young mothers: Infant mortality and morbidity in four areas in the United States. 1973-1978. American Journal of. Rubiic Health. 14(1). 18-23. Meetinotheorchlmoiintantmortalituznolanofactionicrflichioan. (1983. February). Lansing: Michigan Department of Public Health. Meyer. M. B.. 3 Tonascia. J. A. (1977). Maternal smoking. pregnancy complications and perinatal mortality. ,American Journal of 911512112195 and W9 21(2): 297-310. .Michidan.health.statisticsi.1281. (1981). Lansing: Michigan Deparunent of Public Health. Millard. A. (1985). Child mortality and economic variation among rural Mexican households. Social Science and Medicine. Monahan. H. B.. 3 Spencer. E. C. (1962). Deterrents to prenatal care. .Children. 3(9). 114-119. Moore. M. L. (1981). Neuncrn family. Philadelphia: W. B. Saunders. Moore. M. L. (1983). Realities in childnearing. (2nd ed.). Philadelphia: W. B. Saunders. Moriyama. I. M. (1960). Recent change in infant mortality trend. Euinic Health Reports. 15(5). 391-405. Morris. N. M.. Udry. J. R.. 3 Chase. 0. L. (1975). Shifting age- parity distributions of births and the decrease in infant mor- tality. o5. 359-362. Myrianthopaulos. N. D.. 3 Chung. C. 8. (1974). Congenital malforma- tions in singletons: Epidemiologic survey (Report from the Col- laborative Perinatal Project). In D. Bergsma (Ed.)..Birth .uefectet.9riginal.articie.series. Miami: Symposia Specialists. The National Foundation March of Dimes. Naeye. R. L. (1979a). Causes of fetal and neonatal mortality by race in a selected U.S. population. .American.Journal.ci.Euhlic 1193.11.11, Q2(9) . 857-961. Naeye. R. L. (l979b). Weight gain and the outcome of pregnancy. AmericanJournaloi .thtetrics andexmoloox. 135(1). 3-9. 167 Naeye. R. L.. 3 Friedman. E. A. (1979). Causes of perinatal death associated with gestational hypertension and proteinuria. AmericanJournaloi .thtetrics and .Gxnecolodx. 133(1). 8-10. North. A. F.. Jr.. 3 MacDonald. H. M. (1977). Why are neonatal mortality rates lower in small black infants than in white infants in similar birth weight? .Journa1.oi.Eediatrics..29(5). 809-810. Nuckolls. K.. Cassel. J.. 3 Kaplan. B. (1972). Psychosocial assets. life crises. and the prognosis of pregnancy. .American.lournal.ot Eoidemiolodx. as. 1131-1141. O'Brien. M.. 3 Smith. C. (1981). Women's views and experiences of ante- natal care. .Ihe.Eractitioher. 225. 123-125. Osofsky. H. J.. 3 Kendall. N. (1973). Poverty as a criterion of risk. Clinical .thtetrics and .Gxnecologx. lo. 103-109. Osterweis. M.. Bush. P. J.. 3 Zuckerman. A. E. (1979). Family context as a predictor of individual medicine use. .Social.§cience.and Medicine. 13.3. 287-291. Pakter. J.. 3 Nelson. F. C. (1974). Factors in the unprecedented decline in infant mortality in New York City. .Bulletin.oi.thc New iork Academy of Medicine. 5!). 839—868. Paolucci. B.. Bubolz. M. M.. 3 Rainey. M. (1976. December). lumen; iamilies. and honzformal learning oroorams. East Lansing: Michigan State University. Perinatal Association of Michigan. (1983)..Barriers.to.earlx.prenatal ,care. Lansing: Michigan Department of Public Health. Peoples. M. D.. Grimson. R. C.. 3 Daughtry. G. L. (1984). Evaluation of the effects of the North Carolina improved pregnancy outcome prodect: Implication for state level decision making. .American Journal of Public Health. 1i(6). 549-554. Peoples. M. D.. 3 Siegel. E. (1983). Measuring the impact of programs for mothers and infants on prenatal care and low birthweight: The value of refined analyses. .Meuical.9ure. 21. 586-605. Podell. L. (1972). Utilization of prenatal and other preventive health care service by mothers on welfare. .American.Journa1.oi.thtet: rics and anecoloox. 113(7). 859-862. Population Reference Bureau. (1984). 1281 ucrld population data sheet. Washington. 00: Population Reference Bureau. 168 ErenatalCarezhhealthxheoinninotorMichioanicchildren. (1984). Report of the Director's Special Task Force. Michigan Department of Public Health. Lansing: Michigan Department of Public Health. Robinson. 0. (1967. April 1). Precedents of fetal death. ,American Journal at ehstetrics and fixnecoloox. 936-942. Rosenwaike. I. (1972). Sources of prenatal care of mothers having births in Maryland. American Journal of Rublic Health. o2. 186-190e Rosenzweig. M. R.. 3 Schultz. T. P. (1982). The behavior of mothers as inputs to child health: Determinants of birth weight. gesta- tion. and rate of fetal growth. In V. R. Ruchs (Ed.). Economic .aspects.oi health. Chicago: University of Chicago Press. Ross. R. A. (1964). Some important factors in perinatal mortality statistics in a rural state. .American.lournal.of.Qbstetrics.and Cunecoloox. 88(3) . 342-348. Rothman. B. K. (1982). 13.21.11.119 birth: Alternatiies in childbirth. New York: Penguin. Russell. J. (1982). Perinatal mortality: The current debate. .Scciclcgy of. Health and Illness. 4(3). 302-319. Ryan. G. M.. Sweeney. P. J.. 3 Solola. A. 8. (1980). Prenatal care and pregnancy outcome. American Journal of. thtetrics and Bunecoloox. 131(8). 876-881. Schwartz. J. L. (1979). A study of the relationship between maternal life-change ewents and premature delivery. In J. L. Schwartz 3 H. Lawrence (Eds.)..iulnerahle.iniants. New York: McGraw-Hill. Schwartz. R.. 3 Poppen. P. (l982)..Meaeuring.the.impact.of.CHQs on .pregnancy.outccmesi final report. Rockville. MD: Health Resources and Services Administration. Division of Evaluation and Analysis. Schwartz. 8.. 3 Vinyard. J. H. (1965). Prenatal care and prematurity. .Euulic.flea1th.Benerts. 80(3). 237-248. Scrimshaw. 8. C. M. (1978). Infant mortality and behavior in the regu- lation of family size. Population and Deuelopment Reuiew. A. 383-403. Second report from the Social Seruices committee. Eerinatal and Neonatal . (1980). House of Commons Paper 663-1 (Session 1979-80). London. HMSO (Chairman. Renee Short). 169 591119: s. (1984). Infant mortality filo-.1282; Ihefieneseefiountx .matohed record study. Flint. MI: Genesee Child Health Council. Shah. F. K.. 3 Abbey. H. (1971). Effects of some factors on neonatal and postneonatal mortality. .Milbank.Memor1a1.Eund.Quarterlx. 49,! 33-57 a Shapiro. 3.. Schlesinger. F. R.. 3 Nesbitt. R. E. I. (l978)..1n£ant. oerinataltmaternalandehflohoodmortalitxintheuniteditates. Cambridge. MA: Harvard University. Shapiro. S.. McCormick. M. C.. Starfield. B. H.. Krischer. J. P.. 3 Bross. D. (1980). Relevance of correlates of infant deaths for significant morbidity at 1 year of age. .Amerigan Journal of .Qbstetries and .Gxnecoloox. 135(3). 363-373. Shin. E. H. (1975a). Black-white differentials in infant mortality in the South. 1940-1970. .Qemoaranhx. 12(1). l-19. Shin. E. H. (1975b). Economic and social correlates of infant mortality: A cross-sectional and longitudinal analysis of 63 selected countries. .Sogial Biology. 22(4). 317-325. Showstack. J. A.. Budetti. P. P.. 3 Minkler. D. (1984). Factors asso- ciated with birthweight: An exploration of the roles of prenatal care and length of gestation. ‘Amerioan.Journal.oI.Euhlio.flealth. 15; 1003-1008. Snow. L. F. (1983). Traditional health beliefs and practices among lower class black Americans. .Ihe.fleetern.Journal.oI.Menicine. .132, 820-828 0 Sokol. R. J.. Rosen. M. 6.. Stodkov. J.. 3 Chik. L. (1977). Clinical application of high-risk scoring on an obstetric service. American Journal of .thtetrios and aneooloox. 125(6). 652-661. Sokol. R. J.. Wool f. R. B.. Rosen. M. 6.. 3 weingarden. K. (1980). Risk. antepartum care. and outcome: Impact of a maternity and infant care project. ,Qbstetrios.and.Gxneoologx. 56. 150-l56. Standards to: obstetrizoxneoologlt eerrices. (1974). The American College of Obstetricians and Gynecologists. Stewart. A. L.. Reynolds. E. 0. R.. 3 Lipscomb. A. P. (1981). Outcome for infants of very low birthweight: Survey of world literature. .Ibe.Lanoet..l(8228). 1038-1040. Stewart. M.. 3 Erickson. P. (1977). The sociology of birth: A critical assessment of theory and research. .Ihe.Social.Scienoe.Journal. 15(2). 34-47. I70 Stickle. G.. 3 Ma. P. (1977). Some social and medical correlates of pregnancy outcome. American Journal of Dostetries and Gunocolmx 121. 162-166. Stockwell. E. G. (1962). Infant mortality and socioeconomic status: A changing relationship. unlbank.uemorial.fiund.Quarterly.‘AQ. 101-111. Streuning. E. L.. Rabkin. J. G.. Cohen. P.. Raabe. G.. Muhlin. G. L.. 3 Cohen. J. (1973). Family. ethnic and economic indicators of low birth weight and infant mortality: A social area analysis. .Annals mmmmmumoim. 2.18.. 87-102. Terris. M.. 3 Gold. E. M. (1969). An epidemiological study of pre- maturity. American Journal of .Qbetetrlos and Gxneoologx» mm. 358-370. Tokuhata. G. K.. Digon. E.. 3 Mann. L. (1973). Prenatal care and obstetric abnormalities. .Journal.9I.Qh£9n19.21§§o§e§. 25. 163- 185. United States Department of Commerce. Bureau of the Census. (l9 ). Wmmmmmmmmmn Wallace. H. M. (1970). Factors associated with perinatal mortality and morbidity. .QJJnioal .Qbstetrios and Gxnecologxo 13(1). 13-430 Wallace. H. M. (1978). Status of infant and perinatal morbidity and mortality. .Euh11£.fleilth.BenoEI§. 23(4). 386-397. Warnecke. R. B.. Graham. 8.. Mosher. W.. Montgomery. E.. 3 Schotz. W. E. (1975). Contact with health guides and use of health services among blacks in Buffalo. .Euhlio.flealth.Beoorts..oo(3). 213-222. Watkins. E. L. (1968). Low-income Negro mothers-Their decision to seek prenatal care. American Journal of Eublic health. 58(4) . 655-667. Weikart. D. P. (1983. June 30)..Ihe.oost.eiieotixeneee.oi.hioh.oualitx .early.ohildhood.eoucation.programs. Testimony prepared for the Select Committee on Children. Youth and Families. Washington. DC: U.S. House of Representatives. Weingold. A. B. (1981). Diabetes in pregnancy. learbook. Chicago: Year Book Medical Publishers. Westoff. C. F. (1972). The modernization of U.S. contraceptive practice. family flannino Eersoettiues. A. 9-12. 17] Willie. C. V.. 3 Rothney. W. B. (1962). Racial. ethnic and income factors in the epidemiology of neonatal mortality. .American Sociological Beuieu. 21. 522-526. Wilson. A. L.. 3 Matzke. G. R. (1981). The treatment of hypertension $2 2ggegnancy. 12mg Intelligence and .CJinlsol Em. .15.. Wright. S. (1934). The method of path coefficients. .Annals.of.Mathe: matical Statistics. 5. 161-215- Wright. S. (1960). Path coefficients and path regressions: Alternative or complementary concepts? .Biometrics.,1§. 189-202. Wright. T. (1985). Director. Infant Health Promotion Coalition of Detroit. Michigan. Personal communication. Yankauer. A.. Jr. (1950). The relationship of fetal and infant mortal- ity to residential segregation. .American Sociological Boxiew. 15; 644-648. Yerushalmy. J.. vandenBerg. B. J.. Erhardt. C. L.. 3 Jacobziner. H. (1965). Birth weight and gestation as indices of immaturity. AmerioanJournalofDiseasesoIQhfloren. 102. 43-57. Zuckerman. B.. Winsmore. G.. 3 Alpert. J. J. (1979). A study of atti- tudes and support systems of inner-city adolescent mothers. Ihe Journal of Eediatrics. 25(1). 122-125. "uuuuui“