Birth weight, maternal age, and education: new observations from Connecticut and Virginia.

It has been well established that increased maternal education, income, and social status contribute to increased birth weight, as well as reduced risk for low or very low birth weight offspring. However, there remains controversy about the mechanism(s) for this effect, as well as the interactions between these factors, maternal age, and race. Presented here is the analysis of a large, recent sample of over 20,000 consecutive live births in 12 hospitals, about half in Connecticut and half in Virginia, including a maternal population that is educationally and racially diverse. Although information on potentially relevant details such as prenatal care, smoking, occupation, and neighborhood is lacking the data set, there is sufficient information to explore the previously noted strong effect of maternal education on birth weight, as well as the large racial difference in outcome at every educational level after adjustment for the effects of age, marital status, state of residence, and gender of the offspring. However, this relationship was not monotonic, and there were differences in the effect between the white and black families, with black women showing a linear and consistent benefit from education across the range, while whites show a sharp benefit from completion of primary education, less from subsequent schooling. A surprising result was the apparent negative impact of very advanced education (>16 years), with lowered birth weights and higher risk of low birth weight offspring in the women with post-college training. The data also shed some addition light on the effect of age and birth weight. Whites show established improvement in birth outcome to about age 30, with slight decline thereafter, whereas in blacks there was progressive decline in birth weight with rising age starting in adolescence, as previously demonstrated by Geronimus. An additional unexpected observation was a sizable difference between births in Connecticut (larger, fewer low birth weight) than Virginia, correcting for all other covariates. It is hypothesized that this may reflect differences in services used, prenatal care in particular given similarities in smoking rates and other predictors. Because of the non-representativeness of and the limited information available in the present study, the conclusions should be taken as hypotheses for further research rather than definitive.