The Use of Twins To Understand the Effect of WIC on Birth Outcomes
Research Center: Institute for Research on Poverty, University of Wisconsin-Madison
Investigator: Joyce, Ted, and Diane Gibson
Institution: Baruch College, City University of New York
Department of Economics and Finance
Baruch College, City University of New York
One Bernard Baruch Way
NY, NY 10010
There is little evidence in the clinical literature to suggest that supplemental
nutrition for pregnant women in developed countries has an important effect on
birth outcomes. By contrast, studies in the social science literature indicate that
the Supplemental Nutrition Program for Women, Infants, and Children (WIC)
has had a major impact on the incidence of preterm birth. Clinicians view this
finding as suspect, since few interventions tested in randomized trials have
proven effective at preventing preterm birth. This study takes a first step
towards reconciling the belief among policy analysts of the efficacy of WIC
with the skepticism among medical researchers that nutritional supplementation
in the United States is a meaningful determinant of birth outcomes.
The original study objective was to analyze the effect of prenatal WIC
participation on birth outcomes among twins. However, during the research,
it became evident that the overall association between prenatal WIC participation
and birth outcomes had weakened considerably in New York City
between 1988 and 2001. Accordingly, the scope of the project was expanded
to analyze why the association between WIC and infant health among all
births had become less robust over time. The analysis of twins was an
important auxiliary project to support this larger research question.
Associations in the social science studies between WIC and preterm birth
may be due to omitted variable bias. Therefore, this research considered
measures of fetal growth in addition to preterm birth as an outcome.
Second, cross-sectional birth certificate data were utilized to examine the
extent to which the association between WIC and birth outcomes had changed
over time, and if so, whether the changing composition of participants could
explain the difference. The growth in WIC coincided with the expansion in
Medicaid eligibility thresholds that occurred in the late 1980s and early 1990s.
Women on Medical Assistance are automatically eligible for WIC, even if the
income thresholds for Medicaid exceed those for WIC. As a result, the growth
in WIC among pregnant women was likely to include proportionately fewer
women at risk for adverse birth outcomes. With 14 years of data and over
800,000 births to women on Medicaid in New York City, the data provided a
unique opportunity to test whether the changing composition of women on
Medicaid and WIC from the period before the Medicaid eligibility expansions
through the most recent expansions for pregnant women under State
Children's Health Insurance Program (SCHIP) could explain the weakening
association with infant health.
Third, many studies reported that improvements in birth outcomes associated
with WIC were greater among women at medical risk such as smokers,
teenagers and those with a previous premature delivery. Such results were
interpreted as evidence that WIC was more beneficial for women in need of
nutritional assistance. But these risk factors tended to be correlated with
other, harder to measure, determinants of birth outcomes such as substance
abuse, sexually transmitted disease, and stressful home environments.
Effects associated with WIC among a sample of women who smoke, for
instance, may reflect greater unobserved heterogeneity between WIC and
non-WIC participants than was found among lower-risk groups such as
nonsmokers. As a potentially less biased test of whether the association
between WIC and birth outcomes was greater among women at nutritional
risk, the analysis compared outcomes between WIC and non-WIC participants
who deliver twins. Multiple gestations represent a random health
shock that increases the risk of anemia, inadequate weight gain and adverse
birth outcomes, but should be orthogonal to other risky behaviors.
Finally, a universal concern among WIC analysts has been selection bias.
Do women who participate in WIC differ from non-participants in ways that
are hard to measure but that are correlated with the outcomes of interest? As
with almost all previous studies, this research lacked a truly exogenous
instrument or quasi-experimental design to address selection bias directly.
However, the analysis allowed the effects of WIC on fetal growth to vary
over time and within relatively homogenous groups of women. Therefore,
treatment effects that were clinically implausible and differed substantially
over time and across groups should be interpreted cautiously and viewed as
possible evidence of selection bias.
Findings indicate no statistical association between WIC and fetal growth,
except for a positive association among US-born blacks who deliver twins.
One implication may be that targeted nutritional supplementation during
pregnancy might be useful. The broadly held notion that WIC improves
birth outcomes emanates from its association with preterm birth, which the
clinical literature has suggested is implausible. The analysis showed that
prenatal WIC participation had at best a modest impact on fetal growth.
Previous assessments of WIC by social scientists have tended to overlook
the rather weak association with fetal growth.