Prenatal WIC Participation Is Associated with Gestational Weight Gain: A Secondary Analysis of United States Birth Records

Abstract

Background

The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) provides breastfeeding support, nutritious supplemental foods, referrals to health care and social services, and nutrition counseling to eligible pregnant women. Evidence on maternal weight outcomes associated with prenatal WIC participation remains sparse.

Objectives

This study aims to estimate the association between prenatal WIC participation and gestational weight gain (GWG).

Methods

Birth records data from women eligible for WIC during pregnancy (defined as delivery paid with Medicaid) and of reproductive age, who gave birth to a singleton, live birth, at a hospital in the United States in 2023 were used. Reception of WIC benefits during pregnancy, maternal prepregnancy weight and height, and weight at delivery were reported in the birth records. GWG (weight at delivery minus prepregnancy weight) was standardized for gestational age by calculating gestational age-specific z-scores (GWG z-scores) from prepregnancy–BMI-class-specific references, and categorized into tertiles (low, middle, high). GWG rate (pounds gained/per week during the 2nd and 3rd trimesters) was categorized as slow, adequate, or accelerated. We used propensity scores (PS) weighting to create a balanced comparison group of WIC-eligible non–WIC-participant mothers. Using PS-weighted log binomial regression, we estimated risk ratios (RR) and 95% confidence intervals (CIs) for high/accelerated and low/slow (compared with middle/adequate GWG as reference) GWG associated with prenatal WIC participation.

Results

We analyzed 1,318,604 pregnancies. Prenatal WIC participation was associated with a small reduction in risk of low GWG z-score (RR = 0.981, 95% CI: 0.977, 0.985; P < 0.0001) and slow GWG rate (RR = 0.992, 95% CI: 0.988, 0.995; P < 0.0001). No significant association was detected for prenatal WIC participation and high GWG z-score (RR = 1.003, 95% CI: 0.999, 1.006; P = 0.18), whereas a small increase in risk of accelerated GWG rate (RR = 1.004, 95% CI: 1.002, 1.006; P < 0.0001) was observed.

Conclusions

Prenatal WIC participation provided support to modestly reduce low weight gain among WIC-eligible United States women, which may have limited clinical implications.

Introduction

The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), administered by the Food and Nutrition Service at the USDA, serves pregnant and postpartum women from low-income households who are at nutritional risk and their children aged ≤5 y. The program provides breastfeeding support, nutritious supplemental foods, referrals to health care and social services, and nutrition counseling to help mothers make healthy choices in feeding themselves and their children. WIC eligibility is based on income, that is, household income must be <185% of the United States Poverty Income Guidelines. Historically, children have been the largest category of WIC participants; WIC serves nearly half of United States infants [1]. In fiscal year 2023, close to 1.5 million infants and 3.6 million children (1–4 y old) participated in the program monthly, corresponding to 22.4% and 55.0% of WIC participants, respectively [2]. In the same year, ∼1.5 million women participated in WIC monthly (22.6% of participants); among those, half a million did so during pregnancy, or 8.1% of WIC participants [2]. Out of those who enrolled in the program during pregnancy (in 2022), 48.3% enrolled during the first trimester, 40.2% did so in the second trimester, and 11.4% enrolled in WIC in their third trimester [3].

Evidence of the health benefits from participating in WIC has accumulated over the years [4]. However, most research has focused on the infants and children served by the program, whereas research on the association between maternal health outcomes (other than birth outcomes) and prenatal WIC participation remains sparse. Specifically, little attention has been dedicated to understanding WIC program effects on maternal gestational weight gain (GWG). In the United States, almost 1 in 2 women gain excessive weight during pregnancy [5]. Excessive GWG, defined as gaining more weight than the clinical guidelines per BMI [6], has been associated with pregnancy complications [[7], [8], [9]], maternal postpartum weight retention [10] and obesity in offspring [11]. Conversely, gaining weight below guidelines has been found to be associated with small for gestational age and preterm birth [8]. A 2022 systematic review on the maternal, infant and child health outcomes associated with WIC participation included 20 studies. Of those 20 studies, only 4 of them included a maternal health outcome and just 1 study (that is, Sonchak’s 2017 study) provided direct evidence on GWG [12]. Sonchak [13] analyzed birth records data from South Carolina for births in 2004–2013 among Medicaid mothers and found that WIC participation was associated with a reduction in the likelihood of inadequate GWG (that is, gaining less weight than recommended [6]) of nearly 10% among White and Black mothers. Sonchak’s study did not address excessive GWG. More recently, Lacci-Reilly and Brunner Huber [14] examined 2018 United States birth certificates for Medicaid patients and found that WIC enrollees had a 7% higher odds of excessive GWG compared with non-WIC users. However, both of these studies determined adherence to clinical guidelines using total GWG, a metric that does not take into account duration of gestation.

Additional studies evaluating how participating in WIC during pregnancy may affect pregnancy weight gain when adjusting for gestational age are needed to build robust evidence on the potential program benefits for maternal health. Specifically, healthier GWG may reduce pregnancy complications [7,9], improve birth outcomes [7,9], and eventually (and synergistically) improve offspring health in the short and long terms. Intervention approaches that are cognizant of the interconnection between mothers and their offspring, like WIC, can improve (or ameliorate adverse) prenatal programming and disrupt intergenerational health disparities in the United States. Thus, we used United States birth records data and an innovative gestational-age-and-BMI-standardized GWG metric, to estimate the effect of participation in WIC during pregnancy on GWG. For clinical relevance, we also operationalized GWG in relation to adherence to current clinical recommendations [6]. Secondarily, we explored the association between prenatal WIC participation and GWG by prepregnancy maternal weight status.

Methods

This was an administrative data analysis study using natality datasets, which are based on information derived from birth certificates for all births occurring in the United States. Restricted-access birth certificate data, which includes geographical location information, was obtained from the Division of Vital Statistics at the National Center for Health and Statistics. Because this study used deidentified administrative data and did not involve any interaction with participants, the study protocol was not considered human subjects research by the Internal Review Board in our institution.

Study data and sample

The study sample included data from women and birthing people of reproductive age (15–49 y) who gave birth to a live birth in the United States in 2023 who were eligible for WIC during pregnancy, defined as having paid for delivery with Medicaid. Income eligibility limits for WIC [185% of the Federal Poverty Level (FPL)] and Medicaid can be comparable during pregnancy when states can increase the usual 138% FPL Medicaid income limit (≤213%); most states did so in 2023 [15]. Furthermore, in 2022, most WIC pregnant participants (69.2%) reported household incomes not higher than 130% of the FPL [3]. In addition, we restricted the study sample to a single year to limit the number of potential recording and programmatic changes and to reduce the number of multiple births by the same individual in the dataset.

Further criteria for inclusion in the study were singleton births occurring in a hospital in 1 of the 50 states of the United States and the District of Columbia. We excluded from the sample births with no exposure (participation in WIC unknown), outcome (maternal prepregnancy weight or delivery weight unknown), input data for outcome operationalization (BMI or gestational age unknown), or covariate (unknown) data. Births with a gestational age <22 wk or >42 wk were also excluded, based on the ability to create the main outcome variable (see Outcome data section).

Study variables

Exposure data

Starting in 2003, the United States Standard Certificate of Live Birth includes information on whether a mother was enrolled in WIC during pregnancy. Specifically, a Mother’s Worksheet for Child Birth Certificate includes a question that reads: “Did you receive WIC food for yourself because you were pregnant with this child?” Responses are coded as Yes, No, or Unknown. Observations coded as “Unknown” for this question were excluded from the study sample (see exclusion information above).

Outcome data

The maternal worksheet also includes information on maternal prepregnancy weight and height, which allowed calculation of prepregnancy BMI. A Facility Worksheet for the Certificate of Live Birth includes data on maternal weight at delivery, extracted from an admission form, and obstetric estimate of gestational age, from an obstetric admission form.

GWG was calculated as the difference between maternal weight at delivery minus maternal prepregnancy weight. Previous studies operationalized total GWG per the Institute of Medicine recommendations [11,16], which take into account maternal prepregnancy BMI but not length of gestation. Because total GWG is correlated with duration of pregnancy, we standardized GWG for gestational age by calculating gestational age-specific z-scores (GWG z-scores) from weight gain during pregnancy by maternal prepregnancy BMI class, according to previously described methods [17,18]. Specifically, GWG z-scores were calculated by comparing a woman’s weight gain to the gestational week-specific mean and SD of weight gain in the United States population, from BMI-specific GWG for gestational age charts (for gestational ages between 22 and 42 wk) [17,18]. This novel approach aimed to address the correlation between total GWG and gestational age. Using the GWG z-scores, we created tertiles within maternal prepregnancy BMI categories, to classify women with the lowest (1st tertile), middle (2nd tertile), and highest (3rd tertile) GWG z-scores.

GWG rates were also quantified as secondary outcome for clinical relevance. GWG rates were defined as pounds (lb) gained per week during the 2nd and 3rd trimesters and operationalized as follows: 1) by subtracting the mid value of the expected weight gain range during the 1st trimester (that is, 2.75 pounds [6]) from total GWG, and 2) by dividing that resulting weight gain by gestational age at delivery (obstetric estimate in completed weeks at delivery from birth certificates) minus 13 wk (that is, the length of the 1st trimester). Because trimester-specific information about weight gain was not available in the birth records, we worked under the assumption that weight gain was consistent during pregnancy to operationalize GWG rates. A categorical variable indicating adherence to clinical guidelines for GWG rate per prepregnancy BMI was created to classify women’s rate of GWG as “Slow” if it was below the recommended range, “Recommended” if within the range, or “Accelerated” if it exceeded the range [6].

Covariate data

Maternal prepregnancy BMI was classified as underweight (<18.5 kg/m²), normal weight (18.5–<25), overweight (25–<30), or obesity class I, II or III (30–<35, 35–<40 and ≥40, respectively) [19]. Sociodemographic maternal variables (from the maternal worksheet) included age (15–19 y, 20–24 y, 25–29 y, 30–34 y, 35–39 y, 40–44 y, 45–49 y), education [no high school (HS) diploma, HS diploma/General Education Development, some college but no degree, bachelor’s degree, associate degree, master’s degree, doctorate or professional degree], United States nativity (yes, no), race/ethnicity [non-Hispanic (NH) White, NH Black, NH American Indian and Alaskan Native, NH Asian, NH Native Hawaiian and Other Pacific Islander (NHOPI), NH mixed, Hispanic]. Marital status (married, unmarried) could not be included as a covariate because >99% of the birth records in California, the state with the highest number of births in 2023, do not include that information. Relevant reproductive and parity information (from the facility worksheet) included whether the mother had a prior pregnancy (yes, no), previous preterm birth (yes, no), and previous C-section (yes, no). Other maternal data from the maternal worksheet included prepregnancy smoking (yes, no), and from the facility worksheet, prepregnancy diabetes (yes, no) and prepregnancy hypertension (yes, no). Geographic study variables (from the maternal worksheet) included state and county Federal Information Processing System (FIPS) codes (a 5-digit number that uniquely identifies a county in the United States) of maternal residence. We matched these variables to the corresponding USDA-ERS 2013 Rural-Urban Continuum Code (RUCC) [20]. This continuum characterizes metropolitan and nonmetropolitan counties based on their population size, degree of urbanization, and adjacency to a metro area. We then used RUCCs to further categorize maternal area of residence into large (code 1), small (codes 2 and 3) or nonmetro (codes 4–9) to describe the sample (for all other analyses, we used RUCCs). Temporal data included season of birth, which was defined as the quarter of the year when the birth occurred.

Statistical analysis

We used propensity score (PS) analysis to create a group of mothers who were eligible to participate (that is, used Medicaid to pay for delivery) but did not participate in WIC during pregnancy (that is, untreated group) that was comparable with the population who received WIC food during pregnancy (that is, treated group) with regard to variables that may predict WIC participation (that is, self-selection). PS are defined as a subject’s probability of treatment selection, conditional on measured covariates [21]. For calculation of PS and weights, we selected variables based on prior literature [22,23], within the limits of the data available in birth records (see covariate data above), and that were not affected by time or occur before the exposure (that is, before the index pregnancy).

We used the PSMATCH procedure in SAS 9.4 (SAS Institute Inc.) to conduct logistic regression to model the relationship between all known or hypothesized available predictors of WIC participation and program participation during pregnancy. Sociodemographic (age, education, race/ethnicity, nativity), reproductive (combined into a 3-category variable to avoid collinearity: no previous pregnancy, previous pregnancy with no prior C-section or preterm birth, previous pregnancy and prior C-section or preterm birth), health and health-related (prepregnancy diabetes, hypertension and smoking), geographical (state and RUCC), and temporal (season of birth) variables were included as potential predictors of WIC participation in this model. This procedure calculated PSs and the average treatment effect for the treated (ATT) weights. The ATT is the mean treatment effect across only the individuals that actually received treatment, as opposed to the entire population, and therefore the focus of evaluation and policy research aimed to understand the effect of a program [24]. Applying ATT weights was the suitable approach because the treatment (that is, WIC program) is not to be received by the entire population due to eligibility requirements and the need for active engagement from beneficiaries to enroll and receive benefits.

PSs were calculated using a common support region, to ensure that there is overlap in the range of PS across the treatment and the untreated group [25]. After weighting, the covariate balance between the treated and the weighted untreated samples was assessed using: 1) visual comparison of the distributions of PS and each covariate in the treated and untreated groups to assess overlap in the distributions, 2) calculation of the standardized mean difference (SMD) between the 2 exposure groups for the PS and each covariate, with SMDs of 0.25 or less considered acceptable [26], and 3) assessment of the variance ratio, that is, the ratio of the variance of the covariate in the treatment group to its variance in the untreated group; variance ratios falling between 0.5 and 2 were considered acceptable.

Frequency distributions are presented to characterize the analytic unweighted sample. To evaluate the impact of WIC prenatal participation on GWG z-scores tertiles, we used log binomial regression models to obtain the risk ratios (RR) for high and low compared with middle GWG z-scores (reference) associated with WIC participation during pregnancy, while applying the ATT weights. The same analysis was employed for GWG rate, with the classification of Recommended as the reference category. Log binomial regression, a generalized linear model for binary dependent variables, was preferred over logistic regression because 1) it produces unbiased RR and accurate confidence intervals (CIs) [27], and 2) given the GWG study outcomes are common (that is, have a population prevalence of ≥10%), the odds ratio would likely overstate the RR [28]. By using log binomial regression with a log link, we estimated RRs by exponentiation of the relevant β coefficient and corresponding 95% CIs. All models also included the same covariates used to derive the PS weights to obtain doubly-robust estimates, that is, to adjust for any potential residual group differences after weighting [26].

Because prepregnancy BMI is an important determinant of GWG [29], exploratory subgroup analyses by prepregnancy BMI were also conducted. In addition, we conducted the following sensitivity analyses by rerunning the main analyses as follow: 1) excluding births that occurred before reaching the third trimester, to reduce shorter exposure to WIC services before delivery; 2) including the covariate marital status in the PS and main analysis, which resulted in the exclusion of >99% data from California; and 3) using PS matching, instead of weighting. All analyses were conducted in SAS 9.4 (SAS Institute Inc.). Because of the large sample size, interpretation of results was based both on the magnitude of the effect estimates and on statistical significance (that is, P < 0.05).

Results

A total of 1,318,605 observations met the inclusion and exclusion criteria for the study, corresponding to 89% of the population (that is, births in 2023 for which the delivery was paid by Medicaid). Figure 1 shows a flow diagram of the study sample. Characteristics of the (unweighted) sample by WIC status are described in Table 1. A majority of participants (n = 794,998 or 60% in the unweighted sample) enrolled in WIC during pregnancy. Likely due to the large sample size, the distributions of all sample characteristics were statistically different between groups (P < 0.05). However, differences that were more evident between groups included younger age and lower education levels among WIC participants. WIC participants were also more likely to identify as Hispanic and less likely to identify as NH NHOPI. Clinical indicators also differed between groups; specifically, WIC participants were more likely to have had diabetes and hypertension before their index pregnancy. Finally, women who participated in WIC during the index pregnancy were more likely to reside in a nonmetropolitan county. Supplemental Table 1 shows the distribution of participants in the 50 states of the United States and D.C. by group, where some differences were also observed.

FIGURE 1.

Study sample. GWG, gestational weight gain.

TABLE 1.

Sample characteristics, overall and by WIC participation status¹.

Sample characteristic	Overall sample n = 1,318,605	WIC participant		P value2
		Yes 794,998 (60.3%)	No 23,607 (39.7%)
Maternal age (y)				<0.0001
15–19	100,024 (7.6%)	68,891 (8.7%)	31,133 (5.9%)
20–24	352,281 (26.7%)	223,391 (28.1%)	128,890 (24.6%)
25–29	385,074 (29.2%)	225,608 (28.4%)	159,466 (30.5%)
30–34	295,886 (22.4%)	169,303 (21.3%)	126,583 (24.2%)
35–39	146,838 (11.1%)	85,251 (10.7%)	61,587 (11.8%)
40–44	36,585 (2.8%)	21,476 (2.7%)	15,109 (2.9%)
45–49	1917 (0.1%)	1078 (0.1%)	839 (0.2%)
Maternal education				<0.0001
8th grade or less	67,448 (5.1%)	46,499 (5.8%)	20,949 (4.0%)
9th–12th grade, no diploma	201,431 (15.3%)	130,941 (16.5%)	70,490 (13.5%)
HS graduate/GED completed	569,765 (43.2%)	357,116 (44.9%)	212,649 (40.6%)
Some college credit, no degree	266,108 (20.2%)	153,681 (19.3%)	112,427 (21.5%)
Associate degree	89,893 (6.8%)	50,008 (6.3%)	39,885 (7.6%)
Bachelor’s degree	97,016 (7.4%)	46,123 (5.8%)	50,893 (9.7%)
Master’s degree	22,596 (1.7%)	8984 (1.1%)	13,612 (2.6%)
Doctorate/professional degree	4348 (0.3%)	1646 (0.2%)	2702 (0.5%)
Maternal race/Hispanic origin				<0.0001
NH White	443,538 (33.6%)	238,035 (29.9%)	205,503 (39.2%)
NH Black	279,083 (21.2%)	165,409 (20.8%)	113,674 (21.7%)
NH AIAN	14,323 (1.1%)	8533 (1.1%)	5790 (1.1%)
NH Asian	42,337 (3.2%)	23,328 (2.9%)	19,009 (3.6%)
NH NHOPI	4885 (0.4%)	2272 (0.3%)	2613 (0.5%)
NH mixed	37,553 (2.8%)	21,533 (2.7%)	16020 (3.1%)
Hispanic	496,886 (37.7%)	335,888 (42.3%)	160,998 (30.7%)
Mother born in the United States	951,207 (72.1%)	558,628 (70.3%)	392,579 (75.0%)	<0.0001
Had prepregnancy diabetes	19,453 (1.5%)	13,070 (1.6%)	6383 (1.2%)	<0.0001
Had prepregnancy hypertension	43,671 (3.3%)	27,315 (3.4%)	16,356 (3.1%)	<0.0001
Had a prior pregnancy	953,377 (72.3%)	564,147 (71.0%)	389,230 (74.3%)	<0.0001
Had a prior C-section delivery	227,187 (17.2%)	137,661 (17.3%)	89,526 (17.1%)	0.0012
Had a prior preterm birth	67,575 (5.1%)	39,325 (4.9%)	28,250 (5.4%)	<0.0001
Prepregnancy smoking	87,882 (6.7%)	52,083 (6.6%)	35,799 (6.8%)	<0.0001
Maternal area of residence3				<0.0001
Large metro	691,027 (52.4%)	399,627 (50.3%)	291,400 (55.7%)
Small metro	419,792 (31.8%)	263,582 (33.2%)	156,210 (29.8%)
Nonmetro	207,786 (15.8%)	131,789 (16.6%)	75,997 (14.5%)
Season of birth				<0.0001
January–March	293,448 (24.7%)	189,778 (23.9%)	136,065 (26.0%)
April–June	287,879 (24.3%)	190,624 (24.0%)	128,276 (24.5%)
July–September	308,342 (26.0%)	209,800 (26.4%)	132,683 (25.3%)
October–December	297,234 (25.0%)	204,796 (25.8%)	126,583 (24.2%)

Abbreviations: AIAN, American Indian and Alaskan Native; GED, General Education Development; HS, high school; NH, non-Hispanic; NHOPI, Native Hawaiian and Other Pacific Islander; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.

¹Unweighted frequencies and percentages are presented.

²Chi-square test.

³Defined based on the Rural-Urban Continuum Codes developed by the USDA (2013).

Table 2 presents the distribution of maternal weight status and the GWG variables in the overall (unweighted) sample and by WIC participation. The proportion of participants who had overweight or obesity was 64%. On the basis of clinical recommendations for GWG rate, most participants (61%) were categorized as having had an accelerated GWG rate. Once again, all variable distributions were statistically different between groups, although the magnitude of the differences seemed small in most cases. However, WIC participants seemed more likely to have some level of obesity than non-WIC participants did. Although statistically significant, observed differences in GWG-for-GA z-scores and GWG rate (lb/wk) between groups were negligible.

TABLE 2.

Distribution of maternal weight, and gestational weight outcomes in the overall sample and by WIC participation status¹.

Outcome	Overall sample n = 1,318,605	WIC participant		P value
		Yes 794,998 (60.3%)	No 23,607 (39.7%)
Weight status, per BMI				<0.00012
Underweight	44,385 (3.4%)	26,104 (3.3%)	18,281 (3.5%)
Normal weight	432,090 (32.8%)	245,994 (30.9%)	186,096 (35.5%)
Overweight	362,518 (27.5%)	217,875 (27.4%)	144,643 (27.6%)
Obesity class 1	248,503 (18.8%)	154,865 (19.5%)	93,638 (17.9%)
Obesity class 2	131,342 (10.0%)	84,138 (10.6%)	47,204 (9.0%)
Obesity class 3	99,767 (7.6%)	66,022 (8.3%)	33,745 (6.4%)
GWG-per-GA z-score, mean (SD)	−0.3 (1.1)	−0.3 (1.1)	−0.3 (1.1)	<0.00013
GWG z-score tertiles				<0.00012
Lowest GWG z-scores	444,596 (33.7%)	269,625 (33.9%)	174,971 (33.4%)
Middle GWG z-scores	435,415 (33.0%)	263,006 (33.1%)	172,409 (32.9%)
Highest GWG z-scores	438,594 (33.3%)	262,367 (33.0%)	176,227 (33.7%)
GWG rate in lb/wk, median (IQR)4	0.9 (0.6–1.4)	0.9 (0.6–1.4)	1.0 (0.6–1.4)	<0.00015
GWG rate per IOM guidelines				<0.00012
Slow	337,615 (25.6%)	204,737 (25.8%)	132,878 (25.4%)
Within recommendations	176,625 (13.4%)	106,912 (13.4%)	69,713 (13.3%)
Accelerated	804,365 (61.0%)	483,349 (60.8%)	321,016 (61.3%)

Abbreviations: GA, gestational age; GWG, gestational weight gain; IOM, Institute of Medicine; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.

¹Unweighted frequencies and percentages are presented, unless otherwise specified.

²Chi-square test.

³ANOVA F test.

⁴During the 2nd and 3rd trimesters.

⁵Wilcoxon nonparametric test.

Calculation of PS and corresponding ATT weights resulted in the exclusion of a single observation that fell outside the common support region. After applying weights, the SMD of the logit of the PS between the treated and control groups was reduced from 0.27 (in the region sample) to −0.002 (in the weighted sample), reflecting a percent reduction of 99.1, and the variance ratio in the weighted sample was 0.98; all of these indicators were considered acceptable. Balance between groups for each covariate distribution was also assessed. Supplemental Table 2 shows balance diagnostic statistics for the logit of the PS, and for all the covariates included in the PS analysis (as dummy variables). SMDs and variance ratios for all the covariates in the weighted sample fell within acceptable values (values ranged between −0.01 and 0.01, and between 0.95 and 1.08, respectively), indicating that balance was achieved. As shown in Supplemental Table 3, the distributions of the covariates included in the PS analysis were balanced by WIC group status when PS weights were applied.

Table 3 shows the results from ATT-weighted log binomial regression models for the main analyses. Participation in WIC during pregnancy was associated with a 1.9% lower risk of low GWG z-score (P < 0.0001); no significant association was detected for WIC prenatal participation and high GWG z-score (P = 0.18). When GWG was operationalized based on clinical recommendations, significant, albeit very small, associations were observed. Prenatal WIC participation was associated with a <0.5% lower risk of slow GWG rate (P < 0.0001), and with a <0.5% increase in risk of accelerated GWG rate (P < 0.0001).

TABLE 3.

Association between WIC participation and gestational weight gain (GWG) outcomes in the overall sample (n = 1,318,604).

Exposure1	Risk ratios of GWG outcome (95% CI)2
	Low GWG z-scores3	P value	High GWG z-scores3	P value
WIC participation	0.981 (0.977, 0.985)	<0.0001	1.003 (0.999, 1.006)	0.1802
	Slow GWG rate4	P value	Accelerated GWG rate4	P value
WIC participation	0.992 (0.988, 0.995)	<0.0001	1.004 (1.002, 1.006)	<0.0001

Abbreviations: ATT, Average Treatment Effects on the Treated; CI, confidence interval; IOM, Institute of Medicine; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.

¹Exposure reference level was WIC eligible (that is, delivery paid with Medicaid) but no participation during pregnancy.

²Risk ratios were obtained using a log binomial regression model with a log link and by weighting the sample using ATT weights of propensity scores (PS). All models were further adjusted by all the covariates used in the calculation of the PSs.

³Low GWG z-scores refer to the 1st tertile; high GWG z-scores refer to the 3rd tertile of the distribution specific to the maternal prepregnancy BMI category. Reference category was the middle (2nd) tertile.

⁴Slow and accelerated GWG rates are based on the IOM recommendations (IOM, 2009). Reference category was the recommended GWG rate.

Balance diagnostics within prepregnancy maternal weight subgroups indicated that covariate balance was maintained, that is, all SMDs and variance ratios were acceptable, with the only exception of a variance ratio of 0.44 for the dummy variable for the state of North Dakota in the underweight subgroup (data not shown). Results from the subgroup analysis by prepregnancy maternal weight category are shown in Table 4. The association between prenatal WIC participation and low GWG z-score was stronger (more protective) among participants in the underweight and normal weight groups, of similar magnitude among participants with prepregnancy overweight or obesity class 1, and weaker and no longer significant for participants with prepregnancy obesity class 2 or 3. Associations between WIC participation and high GWG z-score were not statistically significant, as in the overall sample, with the exception of a 1% increased risk for participants in the normal weight category and a slight lower risk (<1%) among those in the overweight group. RR patterns were similar when the GWG outcome was operationalized based on adherence to clinical guidelines (that is, GWG rate), except that no protective association for accelerated GWG rate was observed for any maternal weight category.

TABLE 4.

Association between WIC¹ participation and gestational weight gain (GWG) outcomes by maternal prepregnancy weight status.

	Risk ratios of GWG outcome (95% CI)2
Weight status3	Low GWG z-scores4	P value	High GWG z-scores4	P value
Underweight	0.949 (0.930, 0.968)	<0.0001	0.989 (0.969, 1.009)	0.2802
Normal weight	0.974 (0.968, 0.980)	<0.0001	1.013 (1.006, 1.020)	0.0001
Overweight	0.981 (0.974, 0.987)	<0.0001	0.992 (0.985, 0.999)	0.0276
Obesity class 1	0.988 (0.980, 0.997)	0.0059	1.005 (0.996, 1.013)	0.3096
Obesity class 2	1.001 (0.989, 1.013)	0.9099	1.005 (0.994, 1.018)	0.3718
Obesity class 3	0.996 (0.983, 1.010)	0.6060	1.005 (0.991, 1.019)	0.5249
Weight status3	Slow GWG rate5	P value	Accelerated GWG rate5	P value
Underweight	0.952 (0.936, 0.970)	<0.0001	0.991 (0.975, 1.007)	0.2590
Normal weight	0.981 (0.976, 0.987)	<0.0001	1.004 (1.001, 1.007)	0.0079
Overweight	0.987 (0.979, 0.996)	0.0025	1.003 (1.000, 1.006)	0.0366
Obesity class 1	0.997 (0.988, 1.006)	0.4834	1.004 (1.001, 1.008)	0.0139
Obesity class 2	0.996 (0.987, 1.006)	0.4703	1.001 (0.996, 1.007)	0.6400
Obesity class 3	0.994 (0.985, 1.003)	0.1565	0.998 (0.991, 1.005)	0.5638

Abbreviations: ATT, average treatment effects on the treated; CI, confidence interval; GWG, gestational weight gain; IOM, Institute of Medicine; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.

¹Exposure is WIC participation (compared with no WIC participation) during pregnancy within the specified weight status group.

²Risk ratios were obtained using a log binomial regression model with a log link and by weighting the sample using ATT weights of propensity scores (PS). All models were further adjusted by all the covariates used in the calculation of the PSs.

³Defined based on BMI (kg/m²) using standard cutoffs (Weir and Jan, 2024).

⁴Low GWG z-scores refer to the 1st tertile; high GWG z-scores refer to the 3rd tertile of the distribution specific to the maternal prepregnancy BMI category. Reference category was the middle (2nd) tertile.

⁵Slow and accelerated GWG rates are based on the IOM recommendations (IOM, 2009); recommended GWG rate was the reference category.

Finally, findings from the 3 sensitivity analyses, each of which was done using a reduced sample size, were similar to the main analysis (that is, all the RRs were around 1 and in the same direction as the main analyses), indicating that our findings are robust to some changes in the dataset (Supplemental Table 4).

Discussion

Using records from Medicaid-paid births in 2023 at the national level, this study showed that WIC participation during pregnancy was associated with a small reduction in the risk of low gestational-age and BMI-adjusted GWG (1.9%), and an even smaller risk reduction was detected for slow GWG rate (<0.5%). Prenatal WIC participation was not associated with risk of high gestational-age and BMI-adjusted GWG, but a minor and likely inconsequential increased risk of accelerated GWG rate (<0.5%) was detected. These results were robust to further sample restrictions and to applying a different analytic approach.

A previous study reported a 10% reduced risk of inadequate GWG among White and Black WIC mothers in Medicaid in South Carolina for births in 2004–2013 [13]. However, that study measured adherence of total GWG to clinical recommendations per prepregnancy BMI, without consideration of the length of gestation. In our study, associations between WIC participation and low GWG and slow GWG rate, although statistically significant, were of small magnitude and might lack clinical significance. Thus, the discrepancy on the magnitude of the association in Sonchak’s study and ours suggests that accounting for gestational age (in addition to prepregnancy BMI) considerably attenuates the association between participating in WIC during pregnancy and reduced risk of gaining low gestational weight. This points to the importance of adjusting GWG to account for differences in gestational duration in perinatal research, as it has been suggested by others [30]. This becomes particularly relevant for WIC research because prenatal participation in the program has been shown to reduce preterm deliveries [4], which means that WIC prenatal participants may have longer pregnancies on average and therefore more opportunity to gain gestational weight.

More recently, another study also using 2018 United States birth certificates for Medicaid patients reported that WIC users had a 7% increased odds of exceeding recommendations of weight gain, compared with non-WIC users [14]. Direct comparability with that study is not possible because it also used the total GWG metric, as opposed to the GWG rate that adjusts for gestational age, and the authors reported an odds ratio, which overestimates an RR (what we estimated) when the outcome is not rare [28]. Using a GWG metric that took into account gestational age, we found no association with high GWG z-scores and only a minimal and likely inconsequential increase in risk (<0.5%) of accelerated GWG rate among WIC participants (compared with nonparticipants) in our study. Reaching statistical significance for such a close-to-the-null estimate was likely due to the extremely large size of the dataset and consequent reduction in standard errors. Thus, the implications of this finding for maternal and offspring clinical outcomes are likely trivial [31]. Nevertheless, once again, the inconsistency between study findings highlights the need to adjust GWG metrics for length of gestation in perinatal research. Such an approach would provide more accurate estimates of association and could reduce potential harmful research finding implications.

Besides pointing out the importance of adjusting for gestational age when studying GWG, our study findings also suggest that the WIC benefits and services that pregnant participants received were not enough to help prevent high or accelerated GWG. Among those WIC benefits, the most popular is probably the food package, which includes a cash value benefit (CVB) as a fixed dollar amount to spend specifically on fruits and vegetables. In fiscal year 2023, the CVB for pregnant WIC participants was $44/mo, an amount 4 times higher than the one provided before the COVID-19 pandemic. The WIC food package is intended to support access to nutritious foods in resource-limited households, and in synergy with WIC nutrition education, promote healthy eating for women and children in low-income households. Although still limited, the current evidence indicates that dietary patterns during pregnancy that are higher in fruits and vegetables, nuts, legumes, and fish, and lower in added sugar, and red and processed meat are associated with lower risk of excessive GWG [32]. Findings from a study with a diverse, urban population of postpartum women at a tertiary care center can shed some light. In that study, Cheu et al. [33] found that food insecurity, characterized by limited or uncertain access to nutritionally adequate and safe food, was associated with lower median total GWG, but not with excessive GWG. Overall, the evidence may suggest that the food package and CVB partially addresses limited caloric intake among some pregnant WIC participants, but it may not have been enough to substantially displace the consumption of cheaper foods that tend to be high in fat, sodium, and sugar, which may have resulted in high weight gain. However, some changes to the WIC prenatal food package implemented in 2024 (that is, after the study births), such as including fish and other nuts (besides peanuts), may better support healthy weight gain among pregnant WIC participants.

In addition to the food packages, pregnant WIC participants have access to health and immunization screenings, referrals, and nutrition counseling. However, WIC participants reported having inconsistent discussions with healthcare providers about GWG and no knowledge about different gestational weight goals based on prepregnancy weight status [34]. Strengthening linkages between WIC and the health care sector may be a good strategy to better promote healthy weight gain during pregnancy among women with limited resources. A pilot study in Maryland tested an integrated WIC and obstetric service model to prevent postpartum weight retention (PPWR) among 53 African-American women with obesity [35]. This small pilot intervention successfully decreased PPWR (P < 0.05), and although GWG was not the main outcome, the study findings suggested that this model might also be effective in reducing excessive GWG, which affects 1 in 2 United States women [5]. Pregnant women who received integrated WIC and obstetric care services gained on average ∼10 lb less than their counterparts in the control group (P = 0.11) [35]. Current efforts to connect primary care and Medicaid touchpoints with potential WIC beneficiaries are underway, although these are mostly focused on enrollment [36,37]. Our findings and those from the pilot study in Maryland suggest that expanding this linkage to provision of coordinated prenatal care could be a more effective approach to reduce high or accelerated weight gain during pregnancy. Further, qualitative studies elucidating the impact of weight stigma on quality of prenatal care for patients with obesity highlight the importance of incorporating patient-centered, emotionally supportive care into such integrated approaches [38,39]. Supporting pregnant WIC participants to gain healthy weight during pregnancy could have much broader public health and equity implications, because it could also reduce adverse birth outcomes [8], increase breastfeeding rates [40], and reduce child obesity [41].

In this study, we also explored WIC and GWG associations by BMI subgroups, providing granular findings by obesity severity that could be helpful for targeted WIC programming efforts. An interesting pattern observed was the stronger protective effect of prenatal WIC participation on risk of low GWG (or slow GWG rate) among women at higher risk, that is, those in the underweight or normal weight category before pregnancy [5]. The very small increased risk of high GWG or accelerated GWG rate observed for most BMI categories, although likely inconsequential given its insignificant magnitude, underlines that in the United States pregnant women with limited resources remain at risk of gaining too much weight during pregnancy, regardless of their prepregnancy weight status.

This study has several limitations that should be considered when interpreting its results. First, the analysis relied on birth certificate data, which was not collected for research purposes and lacked some relevant covariate information. For instance, information on when during pregnancy women enrolled in WIC, WIC agency where they received services, and participation in other government assistance programs was not available. Gestational weight trajectory may differ depending on the gestational age at which participants begin receiving WIC benefits. Furthermore, WIC implementation may not only differ at the state and county level (which we were able to adjust for), but also at the WIC agency level where there may be variability in program implementation. In addition, differences in data reporting by states prevented us from including marital status in the main analysis. Being married was associated with lower WIC uptake among pregnant individuals with low income in a national survey [22]. We addressed this by conducting a sensitivity analysis adjusting for marital status (but excluding data from California due missing marital status information); such analysis resulted in similar associations as those observed in the main analysis. Second, the secondary outcome of GWG rate was quantified under the assumption that weight gain occurred consistently throughout the second and third trimesters and using an assumed first trimester weight gain based on previous literature, which may not have been reflective of the weight gain trajectory for all pregnancies. Also, both GWG outcomes were likely based on self-reported prepregnancy weight, which may have introduced recall bias and misclassification of BMI, potentially leading to misclassification of GWG also. However, although pregnant women tend to underestimate their prepregnancy weight [42], self-reported weight resulted in 91% correct BMI classifications for pregnant women in a retrospective cohort study conducted in Oregon and Washington [43]. Third, despite confounder balance resulting from the use of PS weighting, the potential for residual confounding bias in estimates cannot be ruled out. In an attempt to reduce potential residual group differences after weighting, we adjusted for all the covariates used to calculate the PS weights in the models to generate “doubly-robust” estimates [26]. This study also has several strengths. The sample was drawn from the national population of live births during a predefined period, which favors generalizability of findings among those eligible for WIC during pregnancy (that is, Medicaid participants). Related to this, the very large sample size allowed us to conduct stratified analysis by BMI categories, including specification of the severity of obesity (that is, obesity class). In addition, we used a recommended method to characterize GWG to standardize values not only for BMI categories but also for gestational age, resulting in a GWG measure that was uncorrelated from gestational age. This novel approach has been proposed as a best practice for studying GWG in perinatal research [30]. Also, this approach includes GWG charts for gestational ages between 22 and 42 wk, allowing the inclusion of most preterm births in the analysis. Finally, results of the sensitivity analyses we conducted reflected the robustness of the study findings.

Altogether, findings from this study at the national level suggested a small benefit from prenatal WIC participation on reducing low GWG and none or a likely inconsequential minimum increased risk of gaining too much gestational weight among WIC participants. Subgroup differences by maternal weight category were evident and could assist implementation of targeted GWG programming by WIC. Additional studies to investigate WIC program efficacy to support healthy GWG with a mechanistic approach (for example, assessing the role of prenatal WIC CVB redemption on dietary intake during pregnancy) are warranted.

Author contributions

SLM is the Principal Investigator of the study, designed the research, analyzed the data, wrote the paper, and had primary responsibility for final content; CDF codeveloped the analysis plan and reviewed and edited the manuscript for intellectual content; and both authors: read and approved the final manuscript.

Data availability

Data described in the manuscript will not be made available because data sharing is restricted by a data use agreement with the National Center for Health Statistics. The code book for the public use dataset (with no location variables) is publicly and freely available without restriction at https://ftp.cdc.gov/pub/health_statistics/nchs/dataset_documentation/DVS/natality/UserGuide2023.pdf. The analytic code will be made available upon reasonable request.

Funding

This article was supported by the RIDGE Partnership, with funding from the USDA Economic Research Service and Food and Nutrition Service. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the USDA. The source of support had no involvement in study design, analysis, interpretation of data, or writing of the report.

Conflict of interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: SLM reports financial support was provided by USDA Economic Research Service and Food and Nutrition Service. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: