Association of Revised WIC Food Package With Perinatal and Birth Outcomes: A Quasi-Experimental Study

Key Points

Question  Did revisions to the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) food package in October 2009, which increased access to whole grains, fruits, vegetables, and low-fat milk, result in improvements in maternal or infant health?

Findings  In this quasi-experimental study of 2 897 537 infants, the revised WIC food package was associated with reduced maternal preeclampsia and gestational weight gain as well as improvements in infant gestational age and birth weight.

Meaning  This evaluation of a major US public health program suggests that improving women’s nutrition may be an important target of clinical interventions and health policy.

Importance  The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) serves more than one-quarter of pregnant and postpartum women. In October 2009, the WIC food package underwent revisions to improve nutritional content. No studies have investigated the downstream effects of this revision on maternal and infant health.

Objective  To investigate whether the revised WIC food package improved perinatal and birth outcomes among recipients.

Design, Setting, and Participants  We conducted a quasi-experimental difference-in-differences analysis, comparing WIC recipients (the treatment group) before and after the package revisions while accounting for temporal trends among nonrecipients (the control group). Multivariable linear regressions were adjusted for sociodemographic covariates. This study was conducted using linked birth certificate and hospital discharge data from California from January 2007 to December 2012. Analysis began July 2018.

Exposures  Whether pregnant women received the revised WIC package, which included more whole grains, fruit, vegetables, and low-fat milk.

Main Outcomes and Measures  Measures of maternal and infant health, including maternal preeclampsia, gestational diabetes, and gestational weight gain as well as infant gestational age, birth weight, and hospitalizations.

Results  The sample included 2 897 537 infants born to 2 441 658 mothers. WIC recipients were more likely to be Hispanic, less educated, of greater parity, and younger than nonrecipients. The revised WIC food package was associated with reductions in maternal preeclampsia (−0.6% points; 95% CI, −0.8 to −0.4) and more than recommended gestational weight gain (−3.2% points; 95% CI, −3.6 to −2.7), increased likelihood of as recommended (2.3% points; 95% CI, 1.8 to 2.8) and less than recommended (0.9% points; 95% CI, 0.5 to 1.2) gestational weight gain, and longer gestational age (0.2 weeks; 95% CI, 0.001 to 0.034). Among infants, an increased likelihood of birth weight that was appropriate for gestational age was observed (0.9% points; 95% CI, 0.5 to 1.3). Although birth weight itself was reduced (−0.009 SDs; 95% CI, −0.016 to −0.001), this was accompanied by reductions in small for gestational age (−0.4% points; 95% CI, −0.7 to −0.1), large for gestational age (−0.5% points; 95% CI, −0.8 to −0.2), and low-birth-weight infants (−0.2% points; 95% CI, −0.4 to −0.004), suggesting that the revised food package improved distributions of birth weight.

Conclusions and Relevance  The revised WIC food package, intended to improve women’s nutrition during pregnancy, was associated with beneficial impacts on maternal and child health. This suggests that WIC policy may be an important lever to reduce health disparities among high-risk women and children at a critical juncture in the life course.

Maternal nutrition is an important determinant of fetal development.¹ Evidence points to the long-term effects of fetal programming on children’s health,^2,3 likely due to changes in fetal hormones, metabolism, and other physiologic processes.⁴ This transmission of disadvantage during pregnancy may explain the intergenerational persistence of health disparities in the United States.⁵ Researchers have called for additional studies to identify policies and interventions targeting maternal nutrition to reduce the long-term adverse effects among children.²

One of the main US programs focused on nutrition among vulnerable families is the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC). WIC is a federal program for low-income pregnant and postpartum women and children younger than 5 years that provides nutritional support, nutrition education, breastfeeding support, and referrals to health and social services.⁶ More than a quarter of pregnant and postpartum women receive WIC,⁷ highlighting its potential to reduce disparities in perinatal and later-life health. The introduction of WIC in the mid-1970s resulted in higher birth weights, particularly for less-educated mothers,⁸ with similar findings in more recent years.^9,10

In October 2009, the US Department of Agriculture implemented major revisions to the standard WIC food package after criticism that it had been essentially unchanged for more than 30 years.¹¹ The revised food package was intended to be more nutritious, thereby addressing obesity and chronic disease.¹¹ It included more whole grains, fruits, vegetables, and low-fat milk.¹¹ The revised food package increased household expenditures on whole grains and decreased purchasing of refined grains and whole milk among WIC recipients,^12-16 and it led to improvements in maternal and child nutrition.^17,18 Little is known regarding the downstream effects of these changes in maternal nutrition on perinatal outcomes. Evaluating public policies is critical, as a handful of prior studies have found that related safety net programs may have unintended negative consequences.^19,20 Also, WIC benefits may be redirected to other household members,²¹ such that changes in purchasing may not translate into improvements in maternal and infant health.

The present study examined the association between the revised WIC food package and perinatal outcomes, using data from more than 2 million births in California. Leveraging a quasi-experimental difference-in-differences (DID) design, we compared pre-post policy changes among WIC recipients with pre-post changes in a control group of nonrecipients to estimate the associations of program revisions with outcomes. Because prior work has shown racial/ethnic disparities in counseling received by WIC recipients and in nutritional and breastfeeding practices prior to the implementation of the revised food package,^22-24 we also conducted subgroup analyses by race/ethnicity. At a time when federal funding for WIC is increasingly challenged,²⁵ evidence on the intergenerational impacts of WIC may help inform policy making and other clinical and community-level interventions targeting maternal nutrition.

This study used data from January 2007 to December 2012 from the California Office of State Health Planning and Development database. The database contains linked birth certificates and hospital discharge records for mothers and infants from 1 year before birth to 1 year after birth. We included California singleton infants live-born from 2007 to 2012 with a gestational age of 20 to 44 weeks at delivery (Figure and eMethods in the Supplement). The study was approved by the Committee for the Protection of Human Subjects within the Health and Human Services Agency of California (protocol #12-09-0702l). Informed consent was waived by the institutional review board because this study used administrative birth certificate data. Analysis began in July 2018.

Beginning in 2007, California birth certificates included a question indicating whether the mother received WIC benefits during the current pregnancy. Among WIC recipients, we considered a woman to have received the revised WIC food package if her pregnancy occurred after the October 2009 revisions. Women whose pregnancies included October 2009 were classified as having received the revised package if they gave birth in February 2010 or later (ie, at least half the pregnancy occurred after the package revisions).

We selected maternal and infant outcomes that could be affected by the mother’s receipt of the revised food package. Maternal outcomes included whether the mother was diagnosed as having preeclampsia or gestational diabetes. We also calculated whether the mother gained less weight, more weight, or was within recommended gestational weight gain (GWG) according to guidelines from the Institute of Medicine (now the National Academy of Medicine).²⁶ These categories were determined based on height, prepregnancy weight, and delivery weight.

Infant outcomes included a z score for birth weight (mean [SD], 0 [1]), weeks of gestation at delivery, and binary variables representing whether the infant was preterm (ie, born before 37 weeks’ gestation); was appropriate for gestational age (AGA), small for gestational age (SGA), large for gestational age (LGA), had low birth weight (LBW, <2500 g) or very LBW (<1500 g); had a longer than expected hospital admission at birth (>2 days for vaginal delivery, >4 days for cesarean delivery); or was readmitted to the hospital within 1 year after birth.

Covariates included sociodemographic variables on the birth certificate. We adjusted models for mother’s education, age, race/ethnicity, and parity and infant’s sex and year of birth.

We first tabulated sample characteristics among WIC recipients and nonrecipients who gave birth before and after implementation of the revised food package. We then estimated the associations of the revised food package with outcomes using DID analysis. Difference-in-differences design is a quasi-experimental technique for examining the association of policy changes with outcomes while accounting for secular trends.^27,28 We took advantage of the fact that revisions to WIC food packages occurred in October 2009 and that this policy change was unlikely to be associated with mother or infant characteristics. This created a natural experiment in that a mother who gave birth in February 2010 or later was exposed to the revised food package for at least half of her pregnancy, while a mother who gave birth before February 2010 was exposed to the revised WIC food package for less than half of her pregnancy or not at all. We thus compared WIC recipients who gave birth on or after February 2010 with those who gave birth before February 2010. To factor out possible secular trends in outcomes over time, DID analysis “differences out” pre-post changes observed among individuals who did not receive WIC.

One underlying assumption is that pre-post differences in outcomes would have been similar between WIC recipients and nonrecipients in the absence of the revised package. While this counterfactual scenario fundamentally cannot be tested, we assessed whether trends in the outcomes during the prerevision period were parallel for WIC recipients and nonrecipients (eMethods, eFigure 1, and eFigure 2 in the Supplement). We also assessed whether the pre-post changes in observed covariates were similar among WIC recipients and nonrecipients to rule out differential compositional changes over time.

Practically speaking, DID analysis involved the implementation of multivariable linear regressions, including an interaction term between WIC status and an indicator for whether the birth occurred before or after the revised food package was implemented. These analyses adjusted for covariates listed above, and additional sensitivity tests included controls for month of birth. The eMethods and eTable 1 in the Supplement include additional details and equations. All tests were 2-tailed, and P values less than .05 were considered to be statistically significant.

First, we conducted subgroup analyses to test for heterogeneity in the response to the WIC revisions by race/ethnicity. To do so, we carried out stratified analyses to produce estimates for each racial/ethnic group, as well as regressions including an interaction term between race/ethnicity and the primary exposure variable to determine whether differences by race/ethnicity were statistically significantly different from one another. Second, because birth certificates do not include information on timing or duration of WIC receipt during pregnancy, we excluded women whose pregnancies included October 2009 to avoid misclassification (n = 383 483).

The final sample included 2 897 537 infants born to 2 441 658 mothers. Overall, 68.1% of WIC recipients were Hispanic, compared 26.0% of nonrecipients. In addition, 25.2% of WIC recipients had more than a high school education, compared with about 72.9% of nonrecipients. WIC recipients were also more likely to be of greater parity and younger than nonrecipients (Table 1). Maternal health was similar for WIC recipients and nonrecipients, although the health of WIC recipients’ infants was worse for several outcomes (Table 2). This included a lower birth weight z score (−0.002 vs 0.05), a higher proportion with SGA birth weight (8.9% vs 8.0%), longer than expected admission at birth (11.1% vs 9.6%), and more readmission in the first year of life (11.1% vs 8.7%).

All outcomes except preterm birth and gestational diabetes demonstrated roughly parallel trends during the prerevision period (eFigure 1 in the Supplement). The changes in key covariates between the prerevision and postrevision periods among WIC recipients were similar to pre-post changes among nonrecipients (Table 1), which supports the assumptions underlying the DID models.

The revised food package was associated with reduced likelihood of preeclampsia (−0.6% points; 95% CI, −0.8 to −0.4), reduced likelihood of more-than-recommended GWG (−3.2% points; 95% CI, −3.6 to −2.7), increased likelihood of less-than-recommended GWG (0.9% points; 95% CI, 0.5 to 1.2), and increased likelihood of within-recommended GWG (2.3% points; 95% CI, 1.8 to 2.8) (Table 3). We were unable to rule out the null hypothesis that there was no association with gestational diabetes. In the sensitivity analyses in which we excluded mothers whose pregnancies included October 2009, results were similar to the primary analyses (eTable 2 in the Supplement).

Infants of WIC recipients exposed to the revised food package had longer mean gestational age at birth (0.2 weeks; 95% CI, 0.001 to 0.034) and reduced birth weight z score (−0.009 SDs; 95% CI, −0.016 to −0.001) (Table 3). They were also more likely to be AGA (0.9% points; 95% CI, 0.5 to 1.3) and less likely to be SGA (−0.4% points; 95% CI, −0.7 to −0.1) and LGA (−0.5% points; 95% CI, −0.8 to −0.2) or have LBW (−0.2% points; 95% CI, −0.4 to −0.004). We were unable to rule out the null hypothesis that there was no association with preterm birth, very LBW, length of admission at birth, or readmissions in the first year of life. When excluding mothers whose pregnancy included October 2009, results were similar to the primary analysis, except that differences for gestational age and birth weight were no longer statistically significant at P < .05 (eTable 2 in the Supplement).

There were racial/ethnic differences in the estimated associations of the revised WIC food package with maternal health (Table 4). Black women, Hispanic women, Asian women, and women of other races had reduced gestational diabetes, while white women did not. The increase in less-than-recommended GWG was less pronounced among Hispanic, Asian, and other women compared with white women, and the reduction in more-than-recommended GWG was less pronounced among Hispanic and Asian women.

For infant health, reduced birth weight was less pronounced among Hispanic and Asian women compared with white women, and birth weight increased among other women. Hispanic and Asian women were also more likely to have higher gestational age at delivery and reduced likelihood of preterm birth than white women in response to the revised WIC food package. Asian women had lower likelihood of having an AGA infant, while black, Asian, and other women had increased likelihood of having an LGA infant. Black women were more likely to have a very LBW infant compared with white women.

In this study, we estimated the associations of the 2009 revised WIC food package with perinatal outcomes. We used quasi-experimental DID analysis in a large diverse sample of California births. The revised WIC food package was intended to improve maternal nutrition by providing more whole grains, fruits, vegetables, and low-fat milk. We found improvements in a number of outcomes, including maternal preeclampsia and weight gain, gestational age, and appropriate birth weight for gestational age.

While previous studies have found beneficial associations of the revised WIC food package with household food expenditures and maternal dietary quality,^12-17 this is the first study to our knowledge to examine downstream health benefits during the perinatal period. In several cases, the magnitude of the findings represents a clinically meaningful change, which is likely to have a substantial impact at the population level.²⁹ For example, we found a reduction in preeclampsia of 0.6% points, which represents a 17% reduction from a base of 3.5%. Nutritional factors such as low fiber and high lipid intake have been implicated as risk factors for preeclampsia, perhaps because they trigger inflammatory responses and endothelial dysfunction.³⁰ However, most of these studies have involved correlational analyses that are unable to account for confounding by unobserved characteristics that may differ for women who eat different amounts of these foods. Our study provides quasi-experimental evidence suggesting this association may be causal and that policies to improve maternal nutrition may be an effective strategy to prevent preeclampsia.

Similarly, we found reductions in maternal GWG among WIC recipients after the implementation of the revised food package. More than 40% of women in this sample gained more weight than recommended, and the revised food package resulted in a reduction of 3.2% points, or 7.8% compared with the control group. This could be explained by the substitution of low-fat milk for whole milk or the provision of whole-grain instead of white bread, which has a higher glycemic index.³¹ Because excessive GWG has been correlated with increased obesity and chronic disease later in life for mothers and children^32,33 and because low-income children are at high risk of severe obesity,³⁴ future studies should examine whether the revised WIC food package had effects on long-term health.

For infant health outcomes, effect sizes were more modest perhaps because these represent outcomes further downstream from the primary intervention. Given the sample size, it is unlikely that analyses were underpowered. Nevertheless, while these may not be clinically meaningful changes for an individual patient, some represent meaningful changes in disease distribution at the population level.²⁹ For example, we found that the revised food package was associated with longer gestation (roughly 1% of an SD). While there was also evidence of reduced birth weight, we found accompanying reductions in SGA, LGA, and LBW babies—about 4% to 5% lower than baseline levels—suggesting that the reduction in birth weight might be explained by fewer LGA and more AGA infants rather than an increase in SGA infants. Alternatively, it may be due to in utero fetal selection, in that high-risk fetuses may be more likely to be born owing to better nutrition.³⁵ Moreover, the change in birth weight represented reduction in the z score of 0.009, which may not be meaningful in a sample this large. As expected, the reduction in LBW (0.2% points) was smaller than the reduction that was documented among WIC recipients when the program was first introduced (1.4% points)⁸ perhaps because the package revisions represent only an incremental change over and above the availability of the package itself. We were unable to rule out the null hypothesis that there was no association with infant hospitalizations, possibly because of the rarity of this outcome or because it is driven by factors other than nutrition. Nevertheless, these findings suggest improved maternal nutrition translated into improvements in newborn outcomes.

We were also unable to rule out the null hypothesis that there was no association of the revised WIC food package with gestational diabetes and preterm birth. Of note, these were the 2 outcomes for which the parallel trends assumption was not met, so the results for these outcomes should be interpreted cautiously. Prior work suggests that maternal nutrition is an important determinant of gestational diabetes,³⁶ but it may be that the revised food package did not reduce the consumption of other products like sugar-sweetened beverages that may be more important culprits than refined grains, or that the origins of gestational diabetes lie in the preconception period. Similarly, prior work suggests that maternal psychosocial stressors and biomedical factors are important determinants of preterm birth.³⁷ While prior work has found that WIC receipt is associated with reduced preterm birth,³⁸ it may be that revisions to the food package did not provide any additional benefit.

There were inconsistent racial/ethnic differences in the association of the revised food package with maternal and infant health. Black women, Hispanic women, Asian women, and women of other races were less likely to develop gestational diabetes, although this result should be interpreted cautiously because this outcome failed the parallel trends test. Hispanic and Asian women were less likely than white women to experience reduced GWG. These women were also more likely to have heavier infants of greater gestational age. The burden of adverse birth outcomes and infant mortality is born disproportionately by black women,^39,40 and our study suggests that differences in the associations of the revised WIC food package with outcomes do not explain these findings. Future studies should attempt to further illuminate and address these inequities. For example, prior work has found that WIC counselors offer disparate counseling to white and black women²² and that Hispanic women are more likely than black women to reside in close proximity to WIC-approved stores,⁴¹ concerning for structural racism embedded in the clinical and community contexts.^42,43

This study has several limitations. First, linked birth certificate and hospital discharge data were only available through 2012; future studies should attempt to replicate these analyses when more recent data become available from state agencies. Second, our study was conducted in California, and results may not generalize to other states; while the WIC food package is intended to be standard nationwide, the foods available in local stores may differ and the characteristics of pregnant women in California may differ from those in other states. Also, we did not have information on the timing of WIC enrollment during pregnancy, as this information is not provided on birth certificates. Finally, DID analysis relies on the assumption that there were no other exposures that may differentially influence outcomes between the treatment and control groups over the course of the study period. The Great Recession of 2008 to 2009 may represent such a confounder. While we were not able to test this counterfactual scenario, prior work has found that adverse health effects of the recession accrued disproportionately to individuals of lower socioeconomic status,⁴⁴ which is the opposite of our findings and therefore unlikely to explain our results.

In conclusion, we found that revisions to improve the nutritional content of the WIC food package in October 2009 were associated with improvements in maternal and infant health at the population level, with inconsistent differences by race/ethnicity. More than a quarter of pregnant and postpartum women in the United States receive WIC, making it among the largest safety net programs for vulnerable families. These findings should be incorporated into future cost-benefit analyses of WIC.⁴⁵ The results suggest that social policy may be an important lever to reduce health disparities, and it strengthens the evidence for improving maternal nutrition during pregnancy.

Corresponding Author: Rita Hamad, MD, PhD, University of California, San Francisco, 995 Potrero Ave, Bldg 80, Ward 83, San Francisco, CA 94110 (rita.hamad@ucsf.edu).

Accepted for Publication: March 15, 2019.

Published Online: July 1, 2019. doi:10.1001/jamapediatrics.2019.1706

Author Contributions: Dr Hamad had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Hamad, Jelliffe-Pawlowski.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Hamad, Collin, Jelliffe-Pawlowski.

Critical revision of the manuscript for important intellectual content: Collin, Baer, Jelliffe-Pawlowski.

Statistical analysis: Collin.

Obtained funding: Jelliffe-Pawlowski.

Administrative, technical, or material support: Collin, Baer, Jelliffe-Pawlowski.

Supervision: Hamad, Jelliffe-Pawlowski.

Conflict of Interest Disclosures: None reported.

Funding/Support: Funding for this work was provided by the National Institutes of Health (grant K08-HL132106), the University of California San Francisco National Center of Excellence in Women’s Health, and the California Preterm Birth Initiative at the University of California San Francisco.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.