Video Abstract
The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), one of the largest US safety net programs, was revised in 2009 to be more congruent with dietary guidelines. We hypothesize that this revision led to improvements in child development.
Data were drawn from a cohort of women and children enrolled in the Conditions Affecting Neurocognitive Development and Learning in Early Childhood study from 2006 to 2011 (Shelby County, TN; N = 1222). Using quasi-experimental difference-in-differences analysis, we compared measures of growth, cognitive, and socioemotional development between WIC recipients and nonrecipients before and after the policy revision.
The revised WIC food package led to increased length-for-age z scores at 12 months among infants whose mothers received the revised food package during pregnancy (β = .33, 95% confidence interval: 0.05 to 0.61) and improved Bayley Scales of Infant Development cognitive composite scores at 24 months (β = 4.34, 95% confidence interval: 1.11 to 7.57). We observed no effects on growth at age 24 months or age 4 to 6 years or cognitive development at age 4 to 6 years.
This study provides some of the first evidence that children of mothers who received the revised WIC food package during pregnancy had improved developmental outcomes in the first 2 years of life. These findings highlight the value of WIC in improving early developmental outcomes among vulnerable children. The need to implement and expand policies supporting the health of marginalized groups has never been more salient, particularly given the nation’s rising economic and social disparities.
Researchers in previous studies have demonstrated that revisions to the Special Supplemental Nutrition Program for Women, Infants, and Children in October 2009 led to improved healthy food access, dietary quality, and birth outcomes among low-income families.
Using unique cohort data and a quasi-experimental design, we found that the 2009 revisions to the Special Supplemental Nutrition Program for Women, Infants, and Children also improved early childhood developmental outcomes, suggesting that nutritional interventions during pregnancy may reduce intergenerational transmission of health disparities.
Maternal nutrition plays a critical role in fetal growth and development.1–5 Suboptimal maternal nutrition during pregnancy may shape the programming of fetal tissues, which is thought to increase chronic disease risk in childhood and adulthood.6–9 To ensure optimal birth outcomes, child growth, and brain development, the nutritional demands of fetal development need to be met. In the United States, the unequal distribution of health care, social support, and access to healthy food has led to racial and socioeconomic disparities in maternal nutrition.10–12 Evidence suggests that nutritional interventions during pregnancy could substantially improve birth outcomes and subsequent child health and development, including cognitive development and physical growth.13–17 As such, targeted policies and programs that improve access to healthy foods in at-risk communities could be a promising approach for addressing social disparities in child outcomes that have been associated with maternal nutrition during pregnancy.16,17
Established in the 1970s, the US Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) was created to safeguard the health of women, infants, and children up to the age of 5 living in low-income households18 in part through the provision of healthy foods. WIC serves approximately half of all infants, 30% of all children aged <5, and approximately one-third of all pregnant and postpartum women.19 In 2019, >$5 billion was spent to aid ∼6.4 million women, infants, and children enrolled in WIC nationwide.20 WIC serves marginalized communities, with ∼50% of children being Hispanic or Latino and 20% being African American.21 In 2009, the WIC food packages for both women and children were revised for the first time to better align it with national dietary guidelines.22,23 This revision included healthier food options (eg, fruits and vegetables), restrictions to less-healthy options (eg, requiring milk to be low-fat and bread to be whole grain), and incentives for breastfeeding.19 Importantly, the revised WIC food package introduced $10-per-month vouchers for purchase of fresh fruits and vegetables.
Extant literature has found that the revised WIC food package improved healthy food availability,24–27 maternal and child dietary quality,28–31 perinatal and birth outcomes,32 and breastfeeding.33–35 However, no researchers, to our knowledge, have evaluated effects of this revision on downstream child health or development. In this study, we address this gap in our understanding of one of the largest US safety net programs by examining the association between the revised food package and several early childhood developmental outcomes. By employing a quasi-experimental difference-in-differences (DiD) design, we estimated effects of the revised food package on child growth, socioemotional, and neurodevelopmental outcomes. We hypothesized that children whose mothers received the revised WIC food package during pregnancy would have better developmental outcomes.
Methods
Data
The sample consisted of mothers and their children from the Conditions Affecting Neurocognitive Development and Learning in Early Childhood (CANDLE) longitudinal cohort study, which enrolled 1503 pregnant women from Memphis and surrounding Shelby County, Tennessee during 2006–2011.36 Longitudinal data collection occurred during pregnancy and throughout childhood. We included data from the baseline survey during pregnancy to capture key covariates and later survey waves that included the outcomes of interest. The sample was restricted to children for whom valid data were collected on these outcomes (Fig 1).
Variables
Exposure
The main exposure was whether women received the revised WIC food package during pregnancy. Because the revision was implemented in October 2009, women who gave birth after this date were considered to be exposed during their pregnancy.37 Participation in WIC during pregnancy was self-reported retrospectively during postpartum survey waves. Unfortunately, CANDLE did not ask women about their receipt of WIC during the postpartum period, during which they would have also been eligible for breastfeeding assistance, so in this study we focused on WIC receipt during pregnancy as the primary exposure.
Outcomes
We considered 3 core child developmental outcome domains measured at 1, 2, and 4 to 6 years of age, each of which has been previously linked with maternal nutrition: child growth, cognitive development, and socioemotional development. Below, we describe in detail those outcomes that fulfilled the assumptions of our statistical methods and were therefore included in our analyses (see Supplemental Information for more details on outcomes not included).
Child Growth, calculated by using Centers for Disease Control and Prevention growth charts,38 was assessed by using length-for-age z scores at 12 months, weight-for-age and weight-for-length z scores at 24 months, and height-for-age z scores at 4 to 6 years. Researchers in several studies have documented potential long-term consequences of failure to thrive in infancy, including adverse cognitive outcomes.39,40 Early childhood growth has also been linked to subsequent growth and height in adulthood,41,42 which has been found to be correlated with improved adult outcomes even in high-income countries.43,44 Other anthropometric measures (eg, length for age at 24 months) were not included because they did not meet the assumptions of statistical models (see Supplemental Information).
Socioemotional development was measured with the Brief Infant Toddler Social Emotional Assessment (BITSEA).45 Assessments were conducted by a licensed psychologist or advanced graduate student in psychology at both 12 and 24 months of age; we include each of these time points as a separate outcome. The BITSEA consists of 2 multiitem scales: the competence score for socioemotional abilities and the problem total scale for behavioral problems and emotional dysregulation. Socioemotional development at age 4 to 6 years, measured by the Child Behavior Checklist, was also assessed but not included because of violations of the statistical model assumptions (see Supplemental Information).
Cognitive development was assessed at age 24 months by a licensed psychologist using the Bayley Scales of Infant Development (Bayley).46 The Bayley at 24 months is correlated with full-scale IQ and school readiness during preschool.47 Correlations are especially strong among children at developmental risk.48,49 We included 3 Bayley scores: the cognitive composite score, receptive communication subtest scaled score, and total language subtest composite score. We also included a standardized full-scale child IQ score, assessed at 4 to 6 years of age by trained psychologists using the Stanford-Binet Intelligence Scales (SB-5).50 The SB-5 has been validated in large, diverse populations and serves as a reliable assessment of intellectual and cognitive abilities. The Bayley expressive communication subtest was also assessed but not included because of violations to the statistical models (see Supplemental Information).
Covariates
Models included covariates that might confound the relationship between receipt of the revised WIC food package and the outcomes: mother’s age, educational attainment, marital status, race, receipt of Medicaid during pregnancy, household size, household income, and total number of pregnancies and child’s sex. We included indicator variables for birthyear in all models to account for secular trends.
Main Analysis
First, we calculated descriptive statistics stratified by WIC receipt during pregnancy. Then, we estimated the effects of the revised food package using DiD analysis. DiD analysis is a quasi-experimental technique, ideally suited to examine the effects of policy changes while accounting for secular trends.51 We leveraged the fact that revisions to the WIC food package in October 2009 were unlikely to be confounded by participant characteristics. Thus, we compared outcomes before and after the policy revision among WIC recipients, “differencing out” secular trends in outcomes among nonrecipients, which included both noneligible individuals and eligible nonrecipients. We chose this as our primary comparison group rather than WIC-eligible participants because CANDLE collected income as a crude categorical rather than a continuous variable; this would have introduced significant measurement error in identifying WIC-eligible families. We included a comparison with families likely to be WIC eligible as a sensitivity analysis (see below). Additional details on the DiD model, including the equation and testing of standard assumptions, are included in the Supplemental Information.
Sensitivity Analyses
We conducted several analyses to test the robustness of the results. First, we used WIC eligibility as the exposure, rather than WIC receipt (n = 967, n = 937, and n = 879 for outcomes measured at age 12 months, 24 months, and 4–6 years, respectively). WIC eligibility is defined by income standards up to 185% of the federal poverty threshold, enrollment in other social assistance programs, and nutrition risk. Roughly 50% of pregnant women in the United States who are eligible actually receive WIC benefits,52 so this analysis is analogous to an intent-to-treat approach. This analysis allowed us to correct for potential bias from “nonadherence,” because eligible women who enroll in WIC could differ from eligible women who do not enroll. By using CANDLE’s categorical income data, eligibility was computed by using WIC eligibility guidelines for income, household size, and Medicaid receipt. Second, we restricted the sample to women whose pregnancies did not include the WIC revision date (October 2009) to examine associations among women exposed to the revised WIC package throughout their entire pregnancy (n = 749, n = 727, n = 668 for outcomes measured at age 12 months, 24 months, and 4–6 years, respectively). Third, we restricted our analysis to women who made <$75 000 per year to be more reflective of women eligible for WIC (n = 777, n = 758, and n = 701 for outcomes measured at age 12 months, 24 months, and 4–6 years, respectively). Fourth, visual inspection of the parallel trends plots indicated that trends in outcomes between WIC recipients and nonrecipients were most comparable in the 20 months before the WIC revision. Thus, we restricted the sample to observations occurring within 20 months of the WIC revision (n = 918, n = 894, n = 818 for outcomes measured at age 12 months, 24 months, and 4–6 years, respectively).
Results
Sample Characteristics
More than half (56.9%) those who received WIC during pregnancy gave birth after October 2009 (Table 1). Overall, 87.2% of WIC recipients and 33.1% of nonrecipients were Black. Additionally, 24.2% of WIC recipients had more than a high school education, compared with 70.8% of nonrecipients. WIC recipients were more likely to be younger (24.6, SD: 5.1) compared with nonrecipients (29.0, SD: 4.9).
Sample Characteristics
| Characteristic | No WIC in Pregnancy, n = 525 | WIC in Pregnancy, n = 700 | ||
|---|---|---|---|---|
| Mean (SD) | % | Mean (SD) | % | |
| Gave birth after October 2009 | — | 43.1 | — | 56.9 |
| Sociodemographic | ||||
| Mother’s age | 29.0 (4.9) | 24.6 (5.1) | ||
| Mother’s marital status | — | 86.3 | — | 37.1 |
| Annual household income | ||||
| $0–$9999 | — | 6.3 | — | 31.7 |
| $10 000–$24 999 | — | 7.3 | — | 32.3 |
| $25 000–$44 999 | — | 15.3 | — | 21.0 |
| $45 000–$74 999 | — | 33.3 | — | 12.7 |
| $75 000+ | — | 37.8 | — | 2.0 |
| Female child | — | 51.0 | — | 49.6 |
| Mother’s race | ||||
| Black | — | 33.1 | — | 87.2 |
| White or othera | — | 66.9 | — | 12.9 |
| Mother’s education | ||||
| Less than high school | — | 4.3 | — | 14.0 |
| High school | — | 25.0 | — | 61.8 |
| More than high school | — | 70.8 | — | 24.2 |
| Medicaid or TennCare | — | 21.6 | — | 82.2 |
| Total No. pregnancies | 2.4 (1.4) | — | 2.7 (1.8) | — |
| No. people in household | 4.0 (1.2) | — | 4.6 (1.6) | — |
| Child health outcomes at age 12 mo | ||||
| Length-for-age z score | 0.05 (1.2) | — | 0.04 (1.2) | — |
| BITSEA: problem total | 8.0 (4.9) | — | 10.9 (6.1) | — |
| Child health outcomes at age 24 mo | ||||
| Wt-for-age z score | 2.1 (1.1) | — | 2.0 (1.1) | — |
| Wt-for-length z score | 0.4 (1.2) | — | 0.4 (1.2) | — |
| BITSEA: problem total | 8.3 (5.5) | — | 11.3 (6.7) | — |
| BITSEA: competency total | 18.5 (2.4) | — | 17.7 (2.7) | — |
| Bayley: cognitive composite score | 102.5 (14.1) | — | 94.4 (10.9) | — |
| Bayley: receptive communication subtest | 10.5 (3.1) | — | 8.8 (2.4) | — |
| Bayley: total language subtest | 103.3 (17.0) | — | 94.8 (12.6) | — |
| Child health outcome at 4–6 y of age | ||||
| Height-for-age z score | 0.41 (1.0) | — | 0.48 (1.3) | — |
| Cognitive test score | 106.3 (14.7) | — | 96.2 (13.7) | — |
| Characteristic | No WIC in Pregnancy, n = 525 | WIC in Pregnancy, n = 700 | ||
|---|---|---|---|---|
| Mean (SD) | % | Mean (SD) | % | |
| Gave birth after October 2009 | — | 43.1 | — | 56.9 |
| Sociodemographic | ||||
| Mother’s age | 29.0 (4.9) | 24.6 (5.1) | ||
| Mother’s marital status | — | 86.3 | — | 37.1 |
| Annual household income | ||||
| $0–$9999 | — | 6.3 | — | 31.7 |
| $10 000–$24 999 | — | 7.3 | — | 32.3 |
| $25 000–$44 999 | — | 15.3 | — | 21.0 |
| $45 000–$74 999 | — | 33.3 | — | 12.7 |
| $75 000+ | — | 37.8 | — | 2.0 |
| Female child | — | 51.0 | — | 49.6 |
| Mother’s race | ||||
| Black | — | 33.1 | — | 87.2 |
| White or othera | — | 66.9 | — | 12.9 |
| Mother’s education | ||||
| Less than high school | — | 4.3 | — | 14.0 |
| High school | — | 25.0 | — | 61.8 |
| More than high school | — | 70.8 | — | 24.2 |
| Medicaid or TennCare | — | 21.6 | — | 82.2 |
| Total No. pregnancies | 2.4 (1.4) | — | 2.7 (1.8) | — |
| No. people in household | 4.0 (1.2) | — | 4.6 (1.6) | — |
| Child health outcomes at age 12 mo | ||||
| Length-for-age z score | 0.05 (1.2) | — | 0.04 (1.2) | — |
| BITSEA: problem total | 8.0 (4.9) | — | 10.9 (6.1) | — |
| Child health outcomes at age 24 mo | ||||
| Wt-for-age z score | 2.1 (1.1) | — | 2.0 (1.1) | — |
| Wt-for-length z score | 0.4 (1.2) | — | 0.4 (1.2) | — |
| BITSEA: problem total | 8.3 (5.5) | — | 11.3 (6.7) | — |
| BITSEA: competency total | 18.5 (2.4) | — | 17.7 (2.7) | — |
| Bayley: cognitive composite score | 102.5 (14.1) | — | 94.4 (10.9) | — |
| Bayley: receptive communication subtest | 10.5 (3.1) | — | 8.8 (2.4) | — |
| Bayley: total language subtest | 103.3 (17.0) | — | 94.8 (12.6) | — |
| Child health outcome at 4–6 y of age | ||||
| Height-for-age z score | 0.41 (1.0) | — | 0.48 (1.3) | — |
| Cognitive test score | 106.3 (14.7) | — | 96.2 (13.7) | — |
N = 1287 (data on WIC receipt were missing for 62 participants). Sample includes children enrolled in the CANDLE Study, Shelby County, Tennessee, 2006–2011. —, not applicable.
Less than 2% of the sample were Hispanic or other race or ethnicity. Thus, they were included in the “white” category to avoid unstable estimates due to small cell sizes.
Association of Revised WIC Package With Child Development
Compared with infants whose mothers did not receive the revised WIC food package during pregnancy, those whose mothers received the revised WIC food package had greater length-for-age z scores at age 12 months (β = .33, 95% confidence interval [CI]: −0.05 to 0.61) and higher Bayley cognitive composite score at age 24 months (β = 4.34, 95% CI:1.11 to 7.57) (Table 2). We observed no statistically significant association between receipt of the revised WIC package with child growth at 24 months or 4 to 6 years of age, measures of socioemotional development at both 12 and 24 months, or measures of cognitive development at 4 to 6 years of age.
Effects of the Revised WIC Food Package on Child Health Outcomes
| Outcomes at Age 12 mo | Outcomes at Age 24 mo | Outcomes at Age 4–6 y | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Length-for-Age z Score | BITSEA: Problem Total | Wt-for-Age z Score | Wt-for-Length z Score | BITSEA: Problem Total | BITSEA: Competency Total | Bayley: Cognitive Composite Score | Bayley: Receptive Communication Subtest | Bayley: Total Language Subtest | Height-for-Age z Score | Cognitive Test Score | |
| WIC × Postrevision | 0.33** (0.05 to 0.61) | −0.46 (−1.85 to 0.94) | 0.20 (−0.10 to 0.51) | 0.18 (−0.16 to 0.52) | −0.49 (−2.11 to 1.12) | −0.12 (−0.80 to 0.57) | 4.34*** (1.11 to 7.57) | −0.21 (−0.91 to 0.49) | −0.39 (−4.13 to 3.35) | 0.10 (−0.20 to 0.39) | −0.07 (−3.59 to 3.45) |
| WIC | −0.18 (−0.44 to 0.08) | 0.29 (−1.02 to 1.60) | −0.15 (−0.44 to 0.14) | −0.09 (−0.41 to 0.23) | 0.61 (−0.93 to 2.15) | −0.04 (−0.69 to 0.61) | −3.53**(−6.60 to −0.46) | 0.26 (−0.40 to 0.92) | 0.84 (−2.71 to 4.40) | 0.03 (−0.25 to 0.31) | 0.43 (−2.86 to 3.71) |
| Postrevision | 0.16 (−0.16 to 0.48) | 0.26 (−1.34 to 1.85) | 0.23 (−0.10 to 0.57) | 0.37* (−0.00 to 0.75) | −0.82 (−2.62 to 0.99) | −0.02 (−0.79 to 0.74) | −2.54 (−6.14 to 1.06) | 0.06 (−0.71 to 0.84) | 0.37 (−3.80 to 4.54) | 0.11 (−0.24 to 0.45) | 0.09 (−3.92 to 4.09) |
| Outcomes at Age 12 mo | Outcomes at Age 24 mo | Outcomes at Age 4–6 y | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Length-for-Age z Score | BITSEA: Problem Total | Wt-for-Age z Score | Wt-for-Length z Score | BITSEA: Problem Total | BITSEA: Competency Total | Bayley: Cognitive Composite Score | Bayley: Receptive Communication Subtest | Bayley: Total Language Subtest | Height-for-Age z Score | Cognitive Test Score | |
| WIC × Postrevision | 0.33** (0.05 to 0.61) | −0.46 (−1.85 to 0.94) | 0.20 (−0.10 to 0.51) | 0.18 (−0.16 to 0.52) | −0.49 (−2.11 to 1.12) | −0.12 (−0.80 to 0.57) | 4.34*** (1.11 to 7.57) | −0.21 (−0.91 to 0.49) | −0.39 (−4.13 to 3.35) | 0.10 (−0.20 to 0.39) | −0.07 (−3.59 to 3.45) |
| WIC | −0.18 (−0.44 to 0.08) | 0.29 (−1.02 to 1.60) | −0.15 (−0.44 to 0.14) | −0.09 (−0.41 to 0.23) | 0.61 (−0.93 to 2.15) | −0.04 (−0.69 to 0.61) | −3.53**(−6.60 to −0.46) | 0.26 (−0.40 to 0.92) | 0.84 (−2.71 to 4.40) | 0.03 (−0.25 to 0.31) | 0.43 (−2.86 to 3.71) |
| Postrevision | 0.16 (−0.16 to 0.48) | 0.26 (−1.34 to 1.85) | 0.23 (−0.10 to 0.57) | 0.37* (−0.00 to 0.75) | −0.82 (−2.62 to 0.99) | −0.02 (−0.79 to 0.74) | −2.54 (−6.14 to 1.06) | 0.06 (−0.71 to 0.84) | 0.37 (−3.80 to 4.54) | 0.11 (−0.24 to 0.45) | 0.09 (−3.92 to 4.09) |
Sample includes children enrolled in the CANDLE Study, Shelby County, Tennessee, 2006–2011. n = 956 for outcomes assessed at age 12-months, n = 929 for outcomes assessed at age 24-months, n = 844 for outcome assessed at 4–6 years old. The above estimates represent β coefficients with 95% CIs in parentheses. The effect estimate of interest is the coefficient on the interaction term. Analyses involved regressing each of the outcomes listed above on an indicator for WIC receipt, an indicator for whether the birth occurred after October 2009, and the interaction between the two. The models additionally adjusted for mother’s age, educational attainment, marital status, race, receipt of Medicaid during pregnancy, household size, household income, and total number of pregnancies and child’s sex.
P < .01; ** P < .05; * P < .10
Sensitivity Analyses
When we considered WIC-eligible individuals as the treatment group (rather than recipients) (Table 3, model 1), and when we excluded pregnancies including October 2009 (model 2), results for length for age remained similar to the primary analysis. However, we noted an attenuated effect size and a CI that included the null for the Bayley cognitive composite score. When we excluded women with incomes of $75 000 or more (model 3), the association with the cognitive score remained similar to the primary analysis. However, we noted a decreased effect size and CI that included the null for length for age. When we excluded observations occurring >20 months before the revision, the results remained similar to the primary analysis (model 4).
Sensitivity Analyses: Effect of the Revised WIC Food Packages on Child Health Outcomes
| Outcomes at Age 12 mo | Outcomes at Age 24 mo | Outcomes at Age 4–6 y | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Length-for-Age z Score | BITSEA: Problem Total | Wt-for-Age z Score | Wt-for-Length z Score | BITSEA: Problem Total | BITSEA: Competency Total | Bayley: Cognitive Composite Score | Bayley: Receptive Communication Subtest | Bayley: Total Language Subtest | Height-for-Age z Score | Cognitive Test Score | |
| Model 1: WIC eligible | |||||||||||
| WIC × Postrevision | 0.39*** (0.10 to 0.67) | −0.08 (−1.51 to 1.35) | 0.29* (−0.02 to 0.60) | 0.12 (−0.23 to 0.46) | −0.68 (−2.33 to 0.97) | −0.03 (−0.73 to 0.66) | 1.87 (−1.43 to 5.17) | −0.28 (−0.99 to 0.43) | −0.33 (−4.13 to 3.47) | 0.13 (−0.18 to 0.43) | −1.75 (−5.30 to 1.81) |
| WIC | −0.03 (−0.57 to 0.51) | 0.85 (−1.84 to 3.53) | 0.04 (−0.53 to 0.62) | 0.16 (−0.49 to 0.81) | 2.07 (−1.03 to 5.17) | 0.46 (−0.86 to 1.77) | 3.68 (−2.52 to 9.88) | 0.89 (−0.45 to 2.22) | 4.97 (−2.17 to 12.11) | −0.04 (−0.59 to 0.52) | 5.16 (−1.35 to 11.67) |
| Postrevision | 0.15 (−0.16 to 0.46) | 0.07 (−1.50 to 1.64) | 0.21 (−0.12 to 0.54) | 0.41** (0.04 to 0.78) | −0.73 (−2.50 to 1.04) | −0.07 (−0.82 to 0.68) | −1.27 (−4.81 to 2.28) | 0.10 (−0.67 to 0.86) | 0.39 (−3.68 to 4.47) | 0.10 (−0.25 to 0.44) | 0.87 (−3.08 to 4.82) |
| Observations | 967 | 967 | 866 | 912 | 937 | 937 | 937 | 937 | 937 | 879 | 865 |
| Model 2: exclude October 2009 pregnancies | |||||||||||
| WIC × Postrevision | 0.30* (−0.01 to 0.61) | −0.02 (−1.54 to 1.49) | 0.23 (−0.09 to 0.56) | 0.13 (−0.25 to 0.51) | −0.41 (−2.12 to 1.31) | −0.17 (−0.92 to 0.59) | 3.64** (0.07 to 7.20) | −0.49 (−1.26 to 0.28) | −1.42 (−5.56 to 2.71) | 0.15 (−0.18 to 0.48) | 0.52 (−3.41 to 4.44) |
| WIC | −0.21 (−0.49 to 0.07) | 0.23 (−1.14 to 1.59) | −0.18 (−0.48 to 0.12) | −0.10 (−0.44 to 0.25) | 0.38 (−1.17 to 1.93) | 0.13 (−0.55 to 0.81) | −3.52** (−6.74 to −0.29) | 0.39 (−0.31 to 1.08) | 1.56 (−2.18 to 5.30) | −0.03 (−0.32 to 0.27) | 0.32 (−3.11 to 3.76) |
| Postrevision | −0.01 (−0.50 to 0.47) | −1.96 (−4.31 to 0.39) | −0.04 (−0.57 to 0.50) | 0.42 (−0.19 to 1.02) | −3.79*** (−6.49 to −1.10) | 0.68 (−0.52 to 1.87) | −2.30 (−7.92 to 3.31) | 1.69*** (0.48 to 2.91) | 6.89** (0.38 to 13.41) | 0.03 (−0.47 to 0.52) | 4.91* (−0.93 to 10.75) |
| Observations | 746 | 747 | 666 | 703 | 726 | 726 | 726 | 726 | 726 | 668 | 654 |
| Model 3: exclude income ≥$75 000 | |||||||||||
| WIC × Postrevision | 0.26 (−0.06 to 0.57) | −0.18 (−1.87 to 1.50) | 0.18 (−0.18 to 0.53) | 0.15 (−0.24 to 0.54) | −0.65 (−2.63 to 1.33) | 0.18 (−0.63 to 0.99) | 4.72** (1.03 to 8.41) | −0.27 (−1.04 to 0.50) | −0.20 (−4.33 to 3.93) | 0.01 (−0.34 to 0.35) | −0.46 (−4.56 to 3.64) |
| WIC | −0.09 (−0.37 to 0.19) | 0.06 (−1.43 to 1.55) | −0.10 (−0.41 to 0.22) | −0.01 (−0.36 to 0.34) | 0.72 (−1.04 to 2.49) | −0.25 (−0.97 to 0.48) | −4.09** (−7.39 to −0.79) | 0.27 (−0.42 to 0.95) | 0.29 (−3.40 to 3.97) | 0.04 (−0.27 to 0.34) | 0.83 (−2.73 to 4.40) |
| Postrevision | 0.14 (−0.24 to 0.51) | −0.10 (−2.10 to 1.91) | 0.28 (−0.13 to 0.69) | 0.50** (0.05 to 0.96) | −0.92 (−3.23 to 1.39) | −0.37 (−1.32 to 0.58) | −1.62 (−5.93 to 2.69) | 0.26 (−0.64 to 1.15) | 0.52 (−4.30 to 5.35) | 0.13 (−0.29 to 0.54) | 1.13 (−3.73 to 5.98) |
| Observations | 775 | 774 | 694 | 733 | 755 | 755 | 755 | 755 | 755 | 701 | 690 |
| Model 4: exclude observations >20 mo prerevision | |||||||||||
| WIC × Postrevision | 0.39*** (0.10 to 0.67) | −0.41 (−1.81 to 1.00) | 0.25 (−0.07 to 0.56) | 0.22 (−0.12 to 0.56) | −0.77 (−2.40 to 0.86) | −0.03 (−0.72 to 0.67) | 4.46*** (1.13 to 7.80) | −0.21 (−0.93 to 0.51) | −0.63 (−4.46 to 3.20) | 0.15 (−0.16 to 0.45) | −0.58 (−4.14 to 2.97) |
| WIC | −0.23 (−0.50 to 0.05) | 0.21 (−1.13 to 1.54) | −0.18 (−0.48 to 0.11) | −0.13 (−0.46 to 0.20) | 0.95 (−0.61 to 2.51) | −0.16 (−0.83 to 0.50) | −3.70** (−6.89 to −0.51) | 0.26 (−0.43 to 0.94) | 0.99 (−2.67 to 4.65) | −0.00 (−0.29 to 0.29) | 0.68 (−2.67 to 4.03) |
| Postrevision | 0.13 (−0.19 to 0.45) | 0.25 (−1.32 to 1.82) | 0.22 (−0.12 to 0.55) | 0.36* (−0.02 to 0.73) | −0.67 (−2.46 to 1.12) | −0.07 (−0.83 to 0.70) | −2.57 (−6.23 to 1.08) | 0.06 (−0.72 to 0.85) | 0.45 (−3.75 to 4.65) | 0.08 (−0.26 to 0.43) | 0.37 (−3.61 to 4.35) |
| Observations | 916 | 916 | 829 | 868 | 892 | 892 | 892 | 892 | 892 | 818 | 805 |
| Outcomes at Age 12 mo | Outcomes at Age 24 mo | Outcomes at Age 4–6 y | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Length-for-Age z Score | BITSEA: Problem Total | Wt-for-Age z Score | Wt-for-Length z Score | BITSEA: Problem Total | BITSEA: Competency Total | Bayley: Cognitive Composite Score | Bayley: Receptive Communication Subtest | Bayley: Total Language Subtest | Height-for-Age z Score | Cognitive Test Score | |
| Model 1: WIC eligible | |||||||||||
| WIC × Postrevision | 0.39*** (0.10 to 0.67) | −0.08 (−1.51 to 1.35) | 0.29* (−0.02 to 0.60) | 0.12 (−0.23 to 0.46) | −0.68 (−2.33 to 0.97) | −0.03 (−0.73 to 0.66) | 1.87 (−1.43 to 5.17) | −0.28 (−0.99 to 0.43) | −0.33 (−4.13 to 3.47) | 0.13 (−0.18 to 0.43) | −1.75 (−5.30 to 1.81) |
| WIC | −0.03 (−0.57 to 0.51) | 0.85 (−1.84 to 3.53) | 0.04 (−0.53 to 0.62) | 0.16 (−0.49 to 0.81) | 2.07 (−1.03 to 5.17) | 0.46 (−0.86 to 1.77) | 3.68 (−2.52 to 9.88) | 0.89 (−0.45 to 2.22) | 4.97 (−2.17 to 12.11) | −0.04 (−0.59 to 0.52) | 5.16 (−1.35 to 11.67) |
| Postrevision | 0.15 (−0.16 to 0.46) | 0.07 (−1.50 to 1.64) | 0.21 (−0.12 to 0.54) | 0.41** (0.04 to 0.78) | −0.73 (−2.50 to 1.04) | −0.07 (−0.82 to 0.68) | −1.27 (−4.81 to 2.28) | 0.10 (−0.67 to 0.86) | 0.39 (−3.68 to 4.47) | 0.10 (−0.25 to 0.44) | 0.87 (−3.08 to 4.82) |
| Observations | 967 | 967 | 866 | 912 | 937 | 937 | 937 | 937 | 937 | 879 | 865 |
| Model 2: exclude October 2009 pregnancies | |||||||||||
| WIC × Postrevision | 0.30* (−0.01 to 0.61) | −0.02 (−1.54 to 1.49) | 0.23 (−0.09 to 0.56) | 0.13 (−0.25 to 0.51) | −0.41 (−2.12 to 1.31) | −0.17 (−0.92 to 0.59) | 3.64** (0.07 to 7.20) | −0.49 (−1.26 to 0.28) | −1.42 (−5.56 to 2.71) | 0.15 (−0.18 to 0.48) | 0.52 (−3.41 to 4.44) |
| WIC | −0.21 (−0.49 to 0.07) | 0.23 (−1.14 to 1.59) | −0.18 (−0.48 to 0.12) | −0.10 (−0.44 to 0.25) | 0.38 (−1.17 to 1.93) | 0.13 (−0.55 to 0.81) | −3.52** (−6.74 to −0.29) | 0.39 (−0.31 to 1.08) | 1.56 (−2.18 to 5.30) | −0.03 (−0.32 to 0.27) | 0.32 (−3.11 to 3.76) |
| Postrevision | −0.01 (−0.50 to 0.47) | −1.96 (−4.31 to 0.39) | −0.04 (−0.57 to 0.50) | 0.42 (−0.19 to 1.02) | −3.79*** (−6.49 to −1.10) | 0.68 (−0.52 to 1.87) | −2.30 (−7.92 to 3.31) | 1.69*** (0.48 to 2.91) | 6.89** (0.38 to 13.41) | 0.03 (−0.47 to 0.52) | 4.91* (−0.93 to 10.75) |
| Observations | 746 | 747 | 666 | 703 | 726 | 726 | 726 | 726 | 726 | 668 | 654 |
| Model 3: exclude income ≥$75 000 | |||||||||||
| WIC × Postrevision | 0.26 (−0.06 to 0.57) | −0.18 (−1.87 to 1.50) | 0.18 (−0.18 to 0.53) | 0.15 (−0.24 to 0.54) | −0.65 (−2.63 to 1.33) | 0.18 (−0.63 to 0.99) | 4.72** (1.03 to 8.41) | −0.27 (−1.04 to 0.50) | −0.20 (−4.33 to 3.93) | 0.01 (−0.34 to 0.35) | −0.46 (−4.56 to 3.64) |
| WIC | −0.09 (−0.37 to 0.19) | 0.06 (−1.43 to 1.55) | −0.10 (−0.41 to 0.22) | −0.01 (−0.36 to 0.34) | 0.72 (−1.04 to 2.49) | −0.25 (−0.97 to 0.48) | −4.09** (−7.39 to −0.79) | 0.27 (−0.42 to 0.95) | 0.29 (−3.40 to 3.97) | 0.04 (−0.27 to 0.34) | 0.83 (−2.73 to 4.40) |
| Postrevision | 0.14 (−0.24 to 0.51) | −0.10 (−2.10 to 1.91) | 0.28 (−0.13 to 0.69) | 0.50** (0.05 to 0.96) | −0.92 (−3.23 to 1.39) | −0.37 (−1.32 to 0.58) | −1.62 (−5.93 to 2.69) | 0.26 (−0.64 to 1.15) | 0.52 (−4.30 to 5.35) | 0.13 (−0.29 to 0.54) | 1.13 (−3.73 to 5.98) |
| Observations | 775 | 774 | 694 | 733 | 755 | 755 | 755 | 755 | 755 | 701 | 690 |
| Model 4: exclude observations >20 mo prerevision | |||||||||||
| WIC × Postrevision | 0.39*** (0.10 to 0.67) | −0.41 (−1.81 to 1.00) | 0.25 (−0.07 to 0.56) | 0.22 (−0.12 to 0.56) | −0.77 (−2.40 to 0.86) | −0.03 (−0.72 to 0.67) | 4.46*** (1.13 to 7.80) | −0.21 (−0.93 to 0.51) | −0.63 (−4.46 to 3.20) | 0.15 (−0.16 to 0.45) | −0.58 (−4.14 to 2.97) |
| WIC | −0.23 (−0.50 to 0.05) | 0.21 (−1.13 to 1.54) | −0.18 (−0.48 to 0.11) | −0.13 (−0.46 to 0.20) | 0.95 (−0.61 to 2.51) | −0.16 (−0.83 to 0.50) | −3.70** (−6.89 to −0.51) | 0.26 (−0.43 to 0.94) | 0.99 (−2.67 to 4.65) | −0.00 (−0.29 to 0.29) | 0.68 (−2.67 to 4.03) |
| Postrevision | 0.13 (−0.19 to 0.45) | 0.25 (−1.32 to 1.82) | 0.22 (−0.12 to 0.55) | 0.36* (−0.02 to 0.73) | −0.67 (−2.46 to 1.12) | −0.07 (−0.83 to 0.70) | −2.57 (−6.23 to 1.08) | 0.06 (−0.72 to 0.85) | 0.45 (−3.75 to 4.65) | 0.08 (−0.26 to 0.43) | 0.37 (−3.61 to 4.35) |
| Observations | 916 | 916 | 829 | 868 | 892 | 892 | 892 | 892 | 892 | 818 | 805 |
Sample includes children enrolled in the CANDLE Study, Shelby County, Tennessee, 2006–2011. The above estimates represent β coefficients with 95% CIs in parentheses. The effect estimate of interest is the coefficient on the interaction term. Analyses involved regressing each of the outcomes listed above on an indicator for WIC receipt, an indicator for whether the birth occurred after October 2009, and the interaction between the two. The models additionally adjusted for mother’s age, educational attainment, marital status, race, receipt of Medicaid during pregnancy, household size, household income, and total number of pregnancies and child’s sex.
P < .10; ** P < .05; *** P < .01.
Discussion
In this study, we leveraged a quasi-experimental design to provide some of the first evidence of the effects of maternal receipt of the revised WIC food package during pregnancy on developmental outcomes in early childhood. These results suggest that the revised food package resulted in improved length for age at age 12 months and cognitive development at age 24 months.
Maternal nutrition could lead to lasting epigenetic modifications to the fetal genome through a process termed “fetal programming” and may result in developmental adaptations that change the physiology of the offspring in a lasting manner.53–55 Previous literature has shown that the WIC revision affected outcomes during several stages of fetal growth and development. Namely, the revised food package was found to improve dietary quality and nutrient intake during pregnancy28,56 as well as downstream perinatal and birth outcomes.32 With our study, we suggest that effects of a mother’s receipt of the WIC revision during pregnancy could be observed into early childhood, up to 24 months of age.
The magnitude of the findings in this study represents clinically relevant effect sizes and provides evidence that one of the largest US safety net policies improves developmental outcomes among low-income and marginalized children. We found a 4.3-point increase in the Bayley cognitive composite score, which represents approximately one-third of an SD. These results are consistent with previous studies, in which researchers found improvements in neurodevelopmental outcomes for children whose families participated in WIC.57–59 However, previous work was cross-sectional in nature or evaluated effects of WIC participation more generally rather than the WIC revision. With our study, we advance this understanding by providing quasi-experimental evidence to strengthen the causal interpretation of this association. This provides evidence that policies to improve maternal nutrition during pregnancy can have long-term consequences for child health. Moreover, shifts in the population distribution of Bayley scores, which may be observed in response to far-reaching safety net programs such as WIC, could ultimately lead to reductions in the number of children at risk for developmental delay. We found that there was no association of the revised WIC food package with cognitive scores at ages 4 to 6 years, suggesting that the effects of the revised WIC food package in pregnancy may wane over time. This is also consistent with previous literature suggesting that socioeconomic gaps in cognitive performance widen with age and that even gains in cognitive test scores in response to interventions designed specifically to improve cognitive performance fade over time (although other benefits of early intervention are often observed).60 Furthermore, there are notable measurement differences between the Bayley and SB-5 that reflect large developmental differences between ages 2 to 3 and 4 to 6; correlations between these scores are at best modest within the same, typically developing children.61 Thus, it is possible that improvements in the WIC package do have longer-term impacts on early childhood developmental outcomes not measured in this study.
We additionally found that the revised WIC food package was associated with increased length-for-age z scores at age 12 months by approximately one-fifth of an SD. There is a positive correlation between height in childhood and adulthood,62 so improvements in linear growth could potentially contribute to later life height potentials. Although body height is largely determined by genetic factors, environmental factors including nutrition could explain up to 20% of the height variation in high-income nations.63 There is a well-known association between height in adulthood and higher earnings even in high-income countries,43,44 and height is thought to be a proxy for improved socioeconomic status in childhood more generally. For example, taller children have been found to have higher average cognitive test scores64 and later life educational attainment.44
There was no observed effect of the revised WIC food package on child growth beyond age 12 months, cognitive development at age 4 to 6 years, or measures of socioemotional developoment for across all age ranges included. There are several potential explanations for why we were unable to detect an association. First, it may be that revisions to the WIC food package were too modest to influence these outcomes, which may be associated with more proximal aspects of the social environment, such as parental stress and mental health.65 Relatedly, child language is associated with the home language environment,66 which may overshadow more distal associations with prenatal nutrition. Second, it may be that effect sizes were too small to detect given of the sample size. Third, pregnant women’s receipt of the modest revision to WIC could have limited benefits to child development over time. Finally, although there are hypothesized mechanisms through which improved maternal nutrition in pregnancy could influence socioemotional development,17,56,67 the current evidence remains correlational and largely inconclusive,68–70 and it may be that there is in fact no effect for this particular outcome.
This study has several strengths. We are among the first to test the effects of the revised WIC food package on downstream child health. In this study, we leveraged a natural experiment to produce rigorous estimates of the effects of the revised food package on several outcomes, strengthening the growing body of evidence that supports the positive association between WIC and multiple health outcomes. Furthermore, by examining a variety of outcomes related to child growth and development, in this study we provide a more-comprehensive illustration of the different pathways through which improved maternal nutrition affects child growth and development. Finally, this study was conducted with data from the CANDLE cohort, providing rich longitudinal data on both pregnancy and later child health, which is rare in the United States.
This study has several limitations. Although the WIC food package is standardized at the federal level, food availability and access and the characteristics of women and children in Tennessee may differ from other states. Therefore, results may not be generalizable to other states. However, CANDLE participants share characteristics with other, majority-Black communities in the urban South that are underrepresented in studies of child development and face high levels of social disadvantage. Additionally, income, WIC receipt, and other covariates were self-reported and may suffer from reporting bias. Next, the findings were not uniformly robust to sensitivity analyses, although the sensitivity analyses were not considered to be the main analyses because of their own limitations (eg, the sensitivity analysis using WIC eligibility as the treatment group was subject to measurement error because income was collected as a crude categorical variable). Furthermore, because most sensitivity analyses led to reduced sample size, this may have resulted in reduced power to detect effects. Also, children whose mothers were eligible for WIC in pregnancy may themselves be eligible for WIC after birth, so part of the effect that we are capturing may reflect continued receipt of the revised WIC package during childhood. This is likely to bias the results toward the null, because some children born soon before the revision who technically fall into the control group could then receive the revised WIC food package as children, therefore making them more similar to the treatment group. This also makes it difficult to disentangle the specific aspect of the WIC revision that led to the results found in this study. Additionally, because we evaluate several outcomes, multiple hypothesis testing could have led to a false rejection of the null hypothesis. Finally, we observed potential differences in receipt of Medicaid or TennCare and household size among WIC participants and nonparticipants over time. Although the analyses were adjusted for these factors, we cannot rule out confounding by unmeasured factors. Finally, there may be additional effects of WIC on childhood development (eg, socioemotional development at 4–6 years) that were not evaluated in this current study because violations of the statistical models.
Conclusions
Findings from this study suggest that revisions to WIC, one of the largest United States public health programs, resulted in downstream child health benefits among a high-risk, understudied population. These findings provide timely and critical evidence for the role that WIC plays in improving the health of the nation’s most vulnerable populations, suggesting meaningful impacts of the revised WIC food package on child development. It provides rigorous scientific evidence that the WIC program may be essential to minimizing the burden of poor health experienced by the most socially disadvantaged members of society. Given the substantial reach of WIC among vulnerable populations, this study has implications for future WIC policy decisions and other related safety net programs, especially given the decrease in program funding for this upcoming fiscal year.71,72 Specifically, considering the relatively modest scope of the 2009 revision, more-substantial updates to the program based on up-to-date nutritional guidance may have substantial effects on improving the health of WIC recipients. Additionally, considering that half of eligible pregnant women do not participate in WIC, increasing enrollment may be a useful strategy for reducing the intergenerational transmission of health disparities in communities facing poverty and social disadvantage.
Ms Guan conceptualized this study, conducted the initial analysis, and drafted the initial manuscript; Dr Hamad conceptualized this study, supervised the analysis, and reviewed and revised the manuscript; Dr LeWinn conceptualized this study, designed the data collection instruments and coordinated and supervised data collection, and reviewed and revised the manuscript; Ms Batra contributed to data organization, data analysis, and code review; Drs Bush and Tylavsky designed the data collection instruments and coordinated and supervised data collection; and all authors critically reviewed the manuscript for important intellectual content and approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: Funded by the National Institutes of Health (National Heart, Lung, and Blood Institute grant K08HL132106, National Institute on Aging grant T32AG049663); the University of California, San Francisco, National Center of Excellence in Women’s Health; and the Urban Child Institute. The funders of this study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Funded by the National Institutes of Health (NIH).
- Bayley
Bayley Scales of Infant Development
- BITSEA
Brief Infant Toddler Social Emotional Assessment
- CANDLE
Conditions Affecting Neurocognitive Development and Learning in Early Childhood
- CI
confidence interval
- DiD
difference-in-differences
- SB-5
Stanford-Binet Intelligence Scales
- WIC
Special Supplemental Nutrition Program for Women, Infants, and Children
Comments