Amencan Journal of Epidemiology Copyright© 1992 by The Johns Hopkins University School of Hygiene and Pubic Health M rights reserved
Vol. 136. No 3 Printed in U.S.A.
Preterm Delivery and Low Birth Weight among First-Born Infants of Black and White College Graduates
Gene A. McGrady,1 John F. C. Sung,1 Diane L. Rowley,2 and Carol J. R. Hogue2
Reproductive outcomes were investigated in black and white female college graduates, presumed to be of similar socioeconomic status and similar risk profile with respect to environmental factors. Data were gathered by mail survey from graduates (19731985) of four Atlanta, Georgia, colleges between February and June 1988. Of 6,867 alumnae to whom questionnaires were mailed, 3,084 responded. A follow-up study of black nonrespondents yielded responses from 14% (335) of those who did not respond to the mail survey. For all graduates with a first live bom at the time of survey (n = 1,089), the rates of preterm delivery, low birth weight, and infant mortality were 80.8, 82.6, and 14.6 per thousand births (primigravida), respectively. Compared with white graduates, black graduates had 1.67 times the risk of preterm delivery and 2.48 times the risk of low birth weight. Measures of social and economic status differed significantly by race. However, adjustment for these variables did not reduce the estimated risk for black graduates compared with whites. Analysis of the nonresponder survey suggested that respondent data alone overestimates the incidence of adverse outcomes in blacks; using nonresponder data, relative risks of 1.28 (preterm delivery) and 1.75 (low birth weight) were calculated as lower limits of the increased risk for blacks. Am J Epidemiol 1992;136:266-76. data collection; education; infant, low birth weight; infant, premature; labor, premature; students
The disparities in rates of infant mortality and rates of low birth weight in offspring of black and white mothers in the United States define a phenomenon which is increasingly a public health concern (1-3). As yet, there is no clear understanding of the forces that cause the mortality and birth weight differences. Instead, conflicting results in regard to the relative contribution of unspecified
Received for publication Juty 24,1991, and infinalform January 21, 1992. 1 Department of Community Health and Preventive Medicine, Morehouse School of Medicine, Atlanta, GA. 2 Division of Reproductive Health, Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control, Atlanta, GA. Reprint requests to Dr. Gene A McGrady, Department of Community Health and Preventive Medicine, Morehouse School of Medicine, 720 Westview Drive, S.W., Atlanta, GA 30310. This work was supported by grant no. CCR-R.O.-1403020 from the Centers for Disease Control
environmental factors have promoted a continuing debate (1, 2, 4, 5). One aspect of this phenomenon which deserves more attention is the evidence that, even at high levels of socioeconomic status, increased risks of infant mortality and low birth weight for black infants are still manifest. This appears to support the view that factors other than those that reflect socioeconomic status are important in producing the differentials noted. Prior work suggests that examination of reproductive outcomes, socioeconomic status measures, and related variables in populations of women of high socioeconomic status may be a useful strategy to investigate the problem. This strategy was adopted to study reproductive outcomes in the first live births of a population of black and white college graduates. The results of a survey of this population are reported here.
Preterm Delivery and Low Birth Weight in College Graduates
MATERIALS AND METHODS Mail survey
Four colleges (colleges A, B, C, and D) in metropolitan Atlanta, Georgia, were selected as the source of women college graduates. Three of the four colleges (A, C, and D) have traditionally had a predominantly black enrollment, and one of these colleges matriculates women only. The other selected college (college B) has a predominantly white enrollment and matriculates women only. Women graduates of these institutions were eligible to participate in the study if they satisfied the following three criteria: 1) graduated no earlier than 1973 and no later than 1985; 2) resident of the continental United States at the time of the survey; and, 3), if parity >0, her first live-born offspring was a singleton birth born no earlier than the graduation year of the respondent. We ascertained the third criterion after receiving a response to the survey. The alumni offices of the four colleges used their computer files of active alumni—alumni for whom the recorded address was not known to be incorrect, and the alumnus was not known to be deceased—to prepare lists of potentially eligible graduates. The survey protocol, adapted from Dillman (6), consisted of the following steps: 1) an initial mailing, including questionnaire and cover letter; 2) a postcard reminder mailed one week after the initial mailing; 3) a second mailing to nonrespondents that consisted of a questionnaire and a revised cover letter mailed approximately 3 weeks after the reminder; and 4) a final mailing, by certified mail, to those not responding by 8 weeks after the initial mailing. The survey questionnaire consisted of 38 items. Questionnaire responses provided measures of characteristics of the respondent's first live born (birth weight, gestational age, and survival), her pregnancy (number of prenatal care visits, specialty of provider, and presence of medical complications), her social circumstances (income level, occupation, marital status, educational attainment, and usual occupation of her parents), and her status with respect to known risk factors for
low birth weight and preterm delivery (maternal height, pre-pregnancy weight, age at delivery, and cigarette and alcohol usage). Survey of nonrespondents
Nonrespondents were followed up by telephone survey. Eligibility criteria were as follows: 1) nonresponse 12 weeks after start of the mail survey; 2) nonresponse not attributable to incorrect mailing address (e.g., moved, with no forwarding address); and 3) a published telephone number available from directory assistance, given the alumna's name and address as listed with the alumni association. Eligible subjects were stratified by college of attendance and year of graduation, and a 15 percent random sample was selected from each stratum. Details of the sampling scheme are shown in figure 1. Trained interviewers made at least six attempts to administer a brief, 12-item questionnaire that ascertained the following: year of graduation, parity, presence or absence of a low birth weight or preterm first born infant, demographic information, and reasons for nonresponse to the mail survey. Analysis
Cumulative incidence was used to describe reproductive outcomes and relative risk to measure association. The analysis was performed in two stages. First, the data from respondents to the mail survey were analyzed without consideration of the data gathered in the telephone survey of nonresponders. Frequency distributions of selected characteristics of black and white respondents were compared using either x2 or Kolmogorov-Smirnov statistics (7). Stepwise logistic regression analysis (BMDP statistical package, BMDP Statistical Software, Inc., Los Angeles, California) was used to model the odds of low birth weight and preterm delivery. In the second stage of analysis, nonresponse bias was analyzed using data gathered in the telephone survey of nonresponders. Estimates of the cumulative incidence of low birth weight and preterm delivery in blacks
Eligible N = 2743
N = 499
N= 3 1 37
Returned Mail N = 168
Undercove raffc N : 1870
RGURE 1. Sampling scheme: mail survey of female graduates of Atlanta, Georgia, colleges, February-July 1988. *Reachable, telephone number available from directory assistance given name/address. tNot Reachable, telephone number not available. $Non-Participating, nonrespondents from one college did not participate in nonresponder survey.
N = 3084
fO O) 09
Preterm Delivery and Low Birth Weight in College Graduates
were corrected for nonresponse; responders to the mail survey, nonresponders who could be reached, and nonresponders who could not be reached (figure 1) were considered as separate strata. An adjusted estimate was calculated as the weighted sum of stratum-specific estimates (8, 9). Stratumspecific estimates for reachable nonresponders were based on data from interviewed nonresponders; stratum-specific estimates for unreachable nonresponders were assumed to equal estimates for white respondents. Birth weight was reported in pounds and ounces; low birth weight was defined as a birth weight less than 2,500 g. Estimated gestational age was calculated from the respondent's report of the number of weeks her first born was delivered early, and preterm delivery was defined as delivered early by 4 or more weeks. RESULTS Survey response
College B (predominantly white enrollment) did not participate in the telephone follow-up of nonresponders or in the third phase mailing; college D did not participate in the third phase mailing. Registrars of the participating colleges provided totals for the number of women who graduated from 1973 through 1985; during this period, 8,091 women graduated from the four colleges. Alumni offices listed 6,867 active alumnae for the same period. Inactive alumnae consisted of those known to be deceased and those whose mailing
address was out of date (incorrect) at the most recent attempt to contact the alumna. Responses to the mail survey were received from 3,084 of the 6,867 alumnae who were sent surveys (44.9 percent). Of the 3,084 alumnae who responded, 2,743 satisfied the eligibility criteria for inclusion in the study. Response rates, the quotient of all responders to listed alumnae, varied markedly by college, ranging from 52.9 percent to 19.6 percent. Also, the proportion of responders who satisfied eligibility criteria varied by college, but in a narrower range (table 1). Mail survey: characteristics of eligible responders
Eligible responders ranged in age from 23 to 38 years. Mean ages were 31.4 years for white responders and 30.6 years for black responders. The age distributions of black and white responders were significantly different. The distribution of current income (at the time of the survey) also differed significantly by race. Compared with blacks, fewer whites reported incomes of $10,000$30,000 per year, and more reported incomes of $50,000 or more. Educational attainment of the respondent's parents was reported in nine categories, ranging from no formal education to completion of a graduate degree. Approximately 46.4 percent of black graduates but only 10.9 percent of white graduates were raised in families where the father's education did not include any college; 36.0 percent of black graduates and 16.1 percent of white graduates had mothers with no college education. Within each race,
TABLE 1. Survey of women graduates (1973-1985) of selected Atlanta, Georgia, colleges: survey coverage and response rates CoBege A
Female graduates Not covered Listed alumnae Responders Eligible responders % response
3,018 851 2,167 1,147 1,074 52.9
-646t 2,222 1,077 897 48.5
2,207 523 1,684 704 648 41.8
1,290 496 794 156 124 19.6
• In college B only, listed alumnae (alumni association) exceeded graduates (registrar), t Many alumnae of college B were nongrsduates.
8,091 1,870 6,867 3,084 2,743 44.9
McGrady et al.
the distribution of educational attainment achieved by parents appeared to be bimodal. For both fathers and mothers of black graduates, the modal categories were 1) completed high school (23.7 and 22.4 percent, respectively) and 2) completed graduate study (24.1 and 23.8 percent, respectively). The modal categories for the parents of white graduates were 1) completed college (20.4 and 36.3 percent for fathers and mothers, respectively) and 2) completed graduate study for fathers, 38.0 percent, or some college for mothers, 24.3 percent (table 2). Mother's age at birth of the first live-born child was calculated for the 39.8 percent of eligible responders who reported at least one live birth at the time of survey. The range of age at first live birth was 21 to 37 years; the mean age for black graduates was 27.4 years, compared with 28.9 years for whites (p < 0.01 Kolmogorov-Smirnov). Approximately 7.4 percent of women were single during the pregnancy that resulted in their first live birth (table 3)—12 percent of black women and 0.5 percent of white women. Conversely, 84.8 percent of blacks and 98.5 percent of whites were married and living with spouse (x2 = 62.8, degrees of freedom (df) = 5, p < 0.001). Among eligible responders with a first live birth, income during the pregnancy differed significantly by race. More black graduates than white graduates reported incomes in the $ 10,000-$20,000 range, and significantly fewer reported incomes of $50,000 or more ( x 2 = 32.9, d f = 5 , p « 0 . 0 0 1 ) . Mail survey: reproductive outcomes
Birth weight, estimated gestational age, and survival in the first year of life were the outcomes of interest. A total of 1,089 eligible respondents had delivered a first born at the time of survey. Ninety infants weighed less than 2,500 g at birth. Eighty-eight were delivered 4 or more weeks before term, and 16 infants did not live to their first birthday. The incidence of low birth weight was 82.7 per thousand births (primigravida); the incidence of preterm delivery was 80.9 per thousand, and the infant mortality rate was
TABLE 2. Characteristics of eligible respondents—percent distribution by race (n = 2,743): survey of women graduates of Atlanta, Georgia, colleges, 1988 Race
Age (years) at survey 20-24 25-29 30-34 35-39 Current marital status Single Married Separated Divorced Widowed Other Current yearly income