Most research examining child health outcomes for Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) participants beyond infancy has focused on nutrient deficiencies, immunization, and health care service use. In order to examine the effectiveness of WIC as a multidimensional food and health assistance program for income-eligible families, it is imperative that research move beyond these examinations of limited health outcomes of children at very young ages that usually focus on nonrepresentative samples. Instead, research is needed that focuses on how WIC may serve as a mechanism to reduce disparities in early childhood health outcomes for low-income children by offering food, health, and social service resources to income-eligible participants. Additionally, research must consider appropriate comparison groups when examining associations between WIC participation and child health outcomes. Selection bias and inappropriate comparisons with children who do not meet income and categorical WIC requirements could lead to incorrect associations that mislead policy discussions. In this study, we examine associations between WIC participation and child health outcomes by using appropriate comparison groups and statistical methods with a nationally representative sample of children.
Research has shown the impact that poverty, or low-income status, can have on childhood health and developmental outcomes. However, one large dimension missing from the policy-relevant empirical work at the national level is the role that child WIC participation may play in helping to reduce poor health and nutrition outcomes for low-income children and their families. Poverty has been shown to generate various health risks for children, including a variety of morbidity conditions and increased mortality. By adding empirical information about WIC eligibility and child participation when testing the relationship between low-income status and childhood health, we may be better able to explain why this relationship persists or is reduced for certain at-risk families. This more complex relationship is important to this study because higher morbidity risks have been found for middle ear infections, high blood lead levels, asthma, and lower respiratory illness among children living in poverty than for those illnesses among nonpoor children. However, the association between child WIC participation compared with nonparticipation and specific child morbidities is not known from a population perspective among WIC-eligible families. This research helps to fill this gap.
Data for this analysis were taken from the Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), Longitudinal 9-month-Preschool Restricted-Use Data File. The ECLS-B is a current and comprehensive national data source that provides information on family poverty status, child WIC participation, diverse racial/ethnic groups, and childhood morbidities. The ECLS-B follows a nationally representative probability sample of U.S. children born between January and December 2001. For this analysis, approximately 6,100 children of the total 10,700 complete 9-month sample were used. The sample was restricted to include only children who were eligible to receive WIC benefits. Four variables in the ECLS-B were used to determine WIC eligibility, including whether the child’s family income was at or below 185 percent of poverty at the 9-month interview or an adult in the household had used one of the following social service programs in the year since the focal child’s birth: Medicaid, the Food Stamp Program (FSP)—now the Supplemental Nutrition Assistance Program (SNAP)—or Temporary Assistance for Needy Families (TANF).
Five child health measures were used as dependent variables in the analysis, including whether a doctor gave the child a diagnosis of asthma, a respiratory illness (bronchitis, pneumonia, or bronchiolitis), a severe gastrointestinal illness (frequent vomiting, diarrhea, or dehydration), or an ear infection and a parental rating of the child’s health as fair or poor at the 9-month interview. Additional variables used in the analysis included focal child use of WIC vouchers for formula or food in the past 30 days at the 9-month interview, maternal race/ethnicity, family poverty status, adult in household used Medicaid in past year, adult in household used food stamps in past year, adult in household used TANF in past year, maternal age at focal child’s birth, maternal educational level, mother employed full-time at 9-month interview, mother married at focal child’s birth, maternal poor self-rated health, and urban residence.
In order to compare child health outcomes among WIC participants and nonparticipants, statistical techniques were employed that allowed for an explicit comparison between children that used WIC benefits in the past 30 days and those that did not use WIC benefits but were eligible to participate. Propensity-score-matching methods were used to create these comparison groups. To estimate the propensity scores, the maternal sociodemographic variables listed above were used as predictors in a logistic regression model to determine child WIC participation. The PSMATCH2 module, prepared for use in Stata statistical software, was used to estimate the propensity scores and conduct the matching. Once the propensity scores were estimated, children participating in the program were matched to children not participating in the program based on their propensity score. The average effect of the treatment on the treated (ATT) then was estimated, which provides the average effect of child WIC participation on the childhood morbidity diagnoses and poor health rating for those children who participate in the program. The ATT is a rate ratio obtained from comparing the percentage of children with a specific diagnosis among children participating in WIC with what the percentage of children with a diagnosis or poor health rating would have been if the child did not participate.
Results from this empirical analysis indicate that WIC does not lead to poorer child health outcomes for participants than for nonparticipants, once appropriate comparison groups are generated based on observed maternal characteristics. Stated differently, once the characteristics that have been shown to be associated with both WIC participation and poor child health outcomes are balanced between child WIC participants and nonparticipants, no significant differences in childhood morbidity diagnoses of asthma, gastrointestinal illnesses, respiratory illnesses, or ear infection are observed. Similarly, child WIC participants and nonparticipants have similar odds of their mothers rating their health as poor, once appropriately matched comparison groups are created. However, prior to matching, children using WIC benefits in the past 30 days apparently have much higher odds of having an asthma, respiratory illness, or ear infection diagnosis or having their mothers rate their health as poor compared with children eligible to receive WIC benefits but not participating. These significant differences are likely due to differences in maternal sociodemographic characteristics because mothers of children not participating in WIC but eligible for benefits tend to be racial/ethnic minorities, have lower levels of education, be younger than 20 when the child is born, live in poverty, and be unmarried when the child is born.
Direct inquiries about this study to the Project Contact listed above.