|
|
| |||||||||||||||||||||||||||||||||
Background Studies have shown that the birth weight of infants is correlated with the birth weights of their siblings and their mothers. We investigated whether the birth weights of mothers and index children were jointly associated with the risk of low birth weight in the siblings of the index children.
Methods We used data on the live-birth cohort of the 1988 National Maternal and Infant Health Survey. The analysis included 1691 white and 1461 black mothers, each of whom had two or more live-born, singleton children. Multiple logistic regression with generalized-estimation equations was used to assess the risk of low birth weight among an index child's siblings. Four groups were studied: that in which neither the mother nor the index child had low birth weight (group 1), that in which only the mother had low birth weight (group 2), that in which only the index child had low birth weight (group 3), and that in which both the mother and the index child had low birth weight (group 4). There was adjustment for other maternal and infant covariates.
Results In groups 1, 2, 3, and 4, respectively, 3.6, 8.3, 21.2, and 38.9 percent of white siblings had low birth weights, as compared with 8.0, 19.0, 31.1, and 57.1 percent of black siblings. When group 1 was used as the reference group, the adjusted odds ratios (and 95 percent confidence intervals) for low birth weight in groups 2, 3, and 4 were 2.5 (1.4 to 4.3), 6.8 (4.7 to 9.8), and 15.4 (9.2 to 25.5), respectively, among white siblings and 2.6 (1.8 to 3.8), 4.7 (3.5 to 6.4), and 13.9 (9.2 to 20.9) among black siblings. These associations were consistently found for birth weights below 1500 g and those ranging from 1500 to 2499 g in both races and after stratification for the mother's age, parity, education, cigarette-smoking status, and weight and height before pregnancy and the infant's sex.
Conclusions Although the possibility of selection and recall biases cannot be excluded with certainty, our data suggest a strong familial aggregation of low birth weight among both whites and blacks in the United States.
Studies have shown that women tend to bear children whose birth weights are similar to each other. The correlation coefficients for birth weight among siblings range from 0.36 to 0.62 (median, approximately 0.50).6 This similarity persists even after adjustment for the length of gestation.7 The correlation among siblings is affected by characteristics of both the mother and the infant.8 Other studies suggest that a mother's own birth weight is also an important determinant of the birth weights of her infants. Infants born to mothers who had low birth weight themselves have a lower mean birth weight and are more likely to have low birth weight than those born to mothers with normal birth weights, even after adjustment for relevant covariates.9,10,11,12 The correlation between the birth weight of a mother and those of her children (0.15 to 0.25)12 is lower than that between the birth weights of siblings.
We studied mothers, white and black, who were included in the live-birth cohort of the National Maternal and Infant Health Survey, conducted in 1988, who had had two or more live-born, singleton infants at the time of the survey. Our primary goals were to assess the association between the birth weights of the mother and the index child and the risk of low birth weight among the siblings of that child and to see whether the association was influenced by characteristics of the mother and the infant, including the mother's race, age, parity, education, weight and height before the index pregnancy, and cigarette-smoking status, as well as the infant's sex. Furthermore, we evaluated whether the association involving these birth weights was mediated by the rate of intrauterine growth or the duration of gestation.
Methods
Study Population
A detailed description of the National Maternal and Infant Health Survey has been published elsewhere.13 Briefly, the survey was conducted by the National Center for Health Statistics to study factors related to the outcomes of pregnancy in the United States in 1988. The survey involved a representative, nationwide sample including 9953 women who had live births, 5332 women whose infants died within the first year, and 3309 women who lost their fetuses in 1988. This study was based on the survey's live-birth cohort, in which the response rate was 74.4 percent. The survey oversampled blacks and infants with very low or moderately low birth weights. Approximately half the respondents sampled were black, and 30 percent of the infants had low birth weights. The data used in this analysis were taken from birth certificates and questionnaires completed by the mothers.
The present study included white and black women who had had two or more singleton live births by 1988 and for whom there were no missing data on the birth weights of the mothers and their infants, or on other specified covariates. Of the 9953 women in the cohort, 9055 were black or white; 2192 (24 percent) did not report their own birth weights. Among the 6863 remaining women, 3219 (47 percent) had had two or more singleton live births by the time of the survey. Sixty-seven of these women were excluded because data on either the mother or the infants were missing or because the infants had birth weights below 500 g. These additional exclusions reduced the final sample to 3152 women.
Only singleton live births were considered in the analysis. An index child was defined as the child who served to bring a family (or group of siblings) into the sample and who was studied independently of all the other index children.14 Thus, in this study the index child was defined as the index child included in the 1988 survey — that is, the youngest child in the family. The outcomes of interest were the birth weights of infants and their gestational ages. Low birth weight was defined as birth weight under 2500 g and was subdivided into very low birth weight (<1500 g) and moderately low birth weight (1500 to 2499 g). Preterm birth was defined as the birth of an infant at a gestational age under 37 weeks. An infant with intrauterine growth retardation was defined as one having a birth weight below the 10th percentile for gestational age among all singleton infants born to mothers of the same race in the live-birth cohort of the 1988 survey.
The major covariates studied included the following characteristics of the mother: race (white or black), age (less than 20, 20 to 29, or 30 or more years), parity (one, two, three, or four or more offspring), education (less than 12, exactly 12, or more than 12 years), place of birth (in the United States or elsewhere), cigarette smoking (yes or no) during the pregnancy that led to the birth of the infant in question, and weight and height before that pregnancy. In addition, the covariates included the year and season of the infant's birth and the infant's sex. We also considered two intervals: the interval between the birth of the infant in question and the birth of the index child and the interval between pregnancies, as defined elsewhere.15 Of these intervals, only the former was included in the final models, because the latter did not affect the outcomes of interest. In this analysis, the factors that could vary from one pregnancy to the next in a given mother included age, parity, and cigarette smoking; the remaining factors were assumed to remain constant over time. Although this assumption may not be accurate with respect to the mother's education and prepregnancy weight, that information was available only for the index pregnancy.
Statistical Analysis
Multiple logistic-regression analysis was used to assess the combined association of the birth weights of the mother and the index child with the risk of low birth weight in the index child's older siblings, after adjustment for covariates. The analysis included multiple offspring of the same mother rather than only one child for each mother, as was done in previous studies. This approach not only increased the statistical power of the study but also provided an opportunity to study a sequence of births. The standard errors were estimated by a method using generalized-estimation equations to accommodate correlations among siblings in birth weight and gestational age.16 A similar approach was used in assessing the risk of very low or moderately low birth weight, preterm birth, and intrauterine growth retardation. When very low birth weight and moderately low birth weight were considered, the reference group used was that of infants with birth weights of 2500 g or more.
The risks of low birth weight, intrauterine growth retardation, and prematurity among the siblings of the index children were studied in four groups defined as follows, according to the birth weights of the mother and the index child: the group in which neither the mother nor the index child had low birth weight (group 1), the group in which only the mother had low birth weight (group 2), the group in which only the index child had low birth weight (group 3), and the group in which both the mother and the index child had low birth weight (group 4). Our assessment of the interaction between the mother's birth weight and that of the index child was based on the assumption that there was no interaction on an additive scale. The statistical significance of this interactive effect was tested with the z statistic as described by Hogan and colleagues.17 When we addressed reductions in the frequency of disease, identifying deviations from additive effects appeared to be the most reasonable approach.18
No weighting was used, since all the analyses were stratified according to race and the birth weight of the index child. The analyses were also performed with stratification according to the mother's age, parity, education, cigarette-smoking status, and weight and height before the pregnancy and the infant's sex, to test whether these factors modified the association between the birth weights of the mother and the index child and the risk of low birth weight among the siblings of the index child. All P values were two-tailed.
Results
In all, our study included 1691 white mothers and 1461 black mothers with 4353 and 3984 singleton, live-born children, respectively, including the index children. Table 1 shows race-specific characteristics of the mothers, grouped according to the birth weights of the mother and the index child. The characteristics of the mothers differed among groups. Among the white mothers, those in group 4 (both mother and child with low birth weight) were more likely than those in group 1 (neither mother nor child with low birth weight) to be teenagers, were less well educated, and were more likely to have smoked during the pregnancy; the mothers in group 4 also weighed less before pregnancy and had shorter stature. The characteristics of the mothers in groups 2 and 3 were intermediate, but the characteristics of group 3 resembled those of group 4 more closely. Similar patterns were found among the black mothers, but black mothers in general were more likely than white mothers to be teenagers and to have had fewer years of education.
|
There were large differences among study groups in the mean birth weights of the siblings of the index child. The percentages of siblings with very low birth weight, moderately low birth weight, and overall low birth weight also varied greatly according to study group (Table 2), with the lowest rates found in group 1 and the highest rates found in group 4. The rates for groups 2 and 3 were intermediate, but group 3 had higher rates of these conditions than group 2.
|
These associations were quantified further in logistic-regression models. The crude odds ratios for low birth weight in groups 2, 3, and 4, with group 1 as the reference category, were 2.5, 7.3, and 17.2, respectively, for whites and 2.7, 5.2, and 15.3 for blacks. Table 3 shows the adjusted odds ratios and 95 percent confidence intervals for the risk of low birth weight, after adjustment for other covariates. Again, the lowest odds ratios were found in group 1, and the highest in group 4. The associations were similar in whites and blacks. The magnitude of both the adjusted and the crude odds ratios was similar. More important, the odds ratio for group 4 could be approximated by multiplying the odds ratio for group 2 by the odds ratio for group 3. This product was much greater than the sum of the individual odds ratios (for whites, z = 2.43, P = 0.015; for blacks, z = 3.42, P<0.001), indicating a statistically significant interaction between the mother's birth weight and the index child's birth weight on an additive scale. A stratified analysis was also performed to examine heterogeneity in the associations among the risk strata for the mother and the infant. The interaction between the mother's birth weight and that of the index child persisted across the strata of the mother's cigarette-smoking status and the infant's sex (Table 3). Similar results were obtained when the analysis was stratified according to the mother's age, parity, education, weight before the pregnancy, and height (data not shown).
|
|
In findings consistent with those of previous studies,6,8,9,10,11,12 this study shows that the birth weight of the mother and that of the index child are both significant and independent predictors of low birth weight in the siblings of the index child. More important, the data suggest that this effect is interactive. In contrast to the odds ratios for low birth weight in the siblings of 2.6 in group 2 and 5.4 in group 3, the odds ratio of 14.1 in group 4 was among the highest for the risk factors known to be associated with low birth weight. This interactive effect was observed even after adjustment for the mother's race, age, parity, education, cigarette-smoking status, weight before the pregnancy, and height and for the year and season of the sibling's birth, the interval between the sibling's birth and that of the index child, and the sibling's sex.
Our data support the notion, suggested previously, that when infants are studied individually the birth weight is correlated with both the length of gestation and rates of intrauterine growth,19 whereas maternal birth weight is more strongly associated with the infant's intrauterine growth than with the duration of gestation.11 Furthermore, the combined effect of the mother's birth weight and that of the index child on the risk of preterm birth and intrauterine growth retardation in a sibling of the index child is either additive or interactive, which suggests that the combined effect on the risk of low birth weight is mediated by both attenuated intrauterine growth and shortened gestation.
The strong familial aggregation of low birth weight in this study may result from genetics, environmental factors, or both. Genetic studies of normal birth weight estimate that 10 percent is determined by the fetal genotype and 24 percent by the maternal genotype.6 Epidemiologic studies to date have identified a number of sociodemographic, environmental, and behavioral risk factors associated with low birth weight.2,4 Our data confirm previous findings that in mothers, low weight before pregnancy, short stature, and cigarette smoking are significant predictors of low birth weight in their infants.4 Although these risk factors are more prevalent among both white and black mothers who had low birth weights themselves than among those with higher birth weights, they do not explain the clustering of low birth weights in their infants. In addition, although black mothers have both more risk factors and higher rates of low birth weight than white mothers, there is no significant difference between the races with regard to the strength of the combined association of the birth weight of the mother and the index child with the risk of low birth weight among the siblings.
Several recent studies have linked low birth weight with the occurrence of impaired glucose tolerance, hypertension, and ischemic heart disease in adulthood.20 Thus, it is biologically plausible that low birth weight may be associated with subsequent abnormalities of growth and development and of the functional capacity of one or more organ systems, including the reproductive system. This study supports the view that research on low birth weight should go beyond focusing on individual pregnancies and factors pertaining to those specific periods.2,4,5 Our data underscore the need for a genetic and epidemiologic approach to elucidating the links between biomedical, social, and environmental factors and low birth weight, both within and across generations.
When the results of this study are interpreted, several methodologic limitations should be taken into account. Because the study was limited to singleton live births, the generalizability of its findings to multiple births and stillbirths is unknown. Birth weights were not reported for 14 percent of the white mothers and 30 percent of the black mothers, a circumstance that may have led to a selection bias. Mothers who did not report their birth weights were more likely not to have completed high school, to have been born outside the United States, and to have higher parity. Nevertheless, our analysis indicated no significant association between a mother's knowledge of her own birth weight and the risk that her infant would have low birth weight or be born prematurely.
The accuracy of the birth weights reported by mothers for themselves and their children was unknown in this sample. A study in Washington State found 78 percent agreement between the weights reported in interviews by mothers as compared with the weights shown on their birth certificates (absolute mean deviation, 4.3 oz [120 g]).21 Previous studies22,23,24 have documented that mothers recall their own children's birth weights accurately. In the Collaborative Perinatal Project,23 57 percent of mothers gave the exact weights for their children, and 77 percent gave weights accurate to within 1 oz (28 g). The generalizability of those findings to our sample is uncertain because of the different characteristics of the populations. If the mothers of index children with low birth weight systematically underestimated their own birth weights or those of their older children, the risk estimates would tend to be spuriously inflated, whereas systematic overestimation of birth weights would spuriously reduce the estimates. In this study, the estimates of the effect of maternal birth weight on the birth weights of offspring are consistent with those in previous studies11,12 in which the maternal birth weights were obtained from birth certificates. This similarity suggests that any recall bias was not sufficient to alter the study conclusions substantially.
The gestational ages of the mothers were not available in this study, so it could not be determined whether the mothers with low birth weights were small at birth because of shortened gestation, intrauterine growth retardation, or both. Data on several characteristics known to affect birth weight, including prenatal care25 and any use of illicit drugs by the mother during pregnancy,26 were not available for the siblings. Adjustment for these variables in the analysis that was limited to the index children did not significantly alter the observed associations between a mother's birth weight and that of the child. Finally, from a clinical and a public health perspective, it is important to have information on prior outcomes in order to predict future outcomes. This study was limited by its retrospective nature.
In summary, although selection and recall biases cannot be excluded with certainty, the data suggest strong familial aggregation of low birth weight among both whites and blacks. Health professionals should recognize that the risk of a recurrence of low birth weight in the same generation is related to the birth weights of both the mother and the index child. More important, women who themselves had low birth weight and have ever delivered a baby with low birth weight are at disproportionally high risk for having another low-birth-weight child. This information should allow women and their physicians to decide on a more informed course of prenatal and postnatal assessment and management, one consistent with each woman's reproductive risk. Researchers should be encouraged to elucidate the biomedical, social, and environmental pathways that contribute to the familial aggregation of low birth weight. The identification of these pathways may help us to understand better the causes of low birth weight and may lead to better strategies for its prevention.
Supported in part by a grant (MCJ-259501) under Title V of the Social Security Act from the Maternal and Child Health Bureau, Health Resources and Services Administration, Department of health and Human Services, and by a National Research Service Award (HRSA 5 T32 PE10014) from the Division of Medicine, Bureau of Health Professions.
We are indebted to Drs. Howard Bauchner, Allen Mitchell, Theodore Colton, Joel Alpert, Michael Kramer, Mark A. Klebanoff, and Kenneth C. Schoendorf for constructive comments, and to Jackie Ashba for her assistance in the data analysis.
Source Information
From the Department of Pediatrics, Boston University School of Medicine and Boston City Hospital (X.W., B.Z., M.J.C.), and the Boston University School of Public Health (G.A.C.) — both in Boston.
Address reprint requests to Dr. Wang at the Department of Pediatrics, Boston University School of Medicine, 818 Harrison Ave., Boston, MA 02118.
References
| |||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
