The Study to Explore Early Development (SEED) is a case–control study, implemented in 6 U.S. sites, with community-based ascertainment of children aged 2–5 years (born 2003–2006) classified as ASD, DD, or POP. Data were collected between 2007 and 2011. The original sample included 3899 children (ASD: N = 722; Possible ASD: N = 584; DD: N = 1304; POP: N = 1289). We excluded possible cases and children (N = 849) who had (1) missing infant or current anthropometric data, (2) biologically implausible values (months 0–6: weight-for-age z-scores <−5 or >5; ages 2–5 years: BMI z-scores <−4 or >5) (Centers for Disease Control and Prevention, 2016), or (3) mothers with a BMI < 14.5 kg/m². The final sample included 2466 children (ASD: N = 668; DD: N = 914; POP: N = 884).

Details of the SEED methodology were previously published (Schendel et al., 2012). ASD and DD participants were identified from special education and clinical sources providing services to children with disabilities. Children had special education or International Classification for Disease codes indicative of autism/ASD or other developmental disabilities. POP participants were selected from random samples of birth certificates within each site’s defined geographic study area. All study groups completed an in-person general developmental assessment. Children with Social Communication Questionnaire (SCQ) scores ⩾ 11 were designated as potential ASD cases regardless of initial identification method. All children with a previous ASD diagnosis or special education classification were considered potential ASD cases regardless of their SCQ scores. Final ASD case classification was based on scores from gold standard evaluation tools, the Autism Diagnostic Observation Schedule and Autism Diagnostic Interview—Revised. Some children with ASD also had cognitive delays (e.g. developmental delay or intellectual disability), but they remained in the ASD group. Children designated as potential ASD cases not meeting research criteria as an ASD case, were classified as DD or POP, depending on how they were initially identified. Caregivers completed an interview about family sociodemographics, reproductive history, and index child pregnancy. The study was approved by the Institutional Review Boards at the Centers for Disease Control and Prevention (CDC) and all study sites.

Child weights and lengths from birth to 6 months of age were ascertained from neonatal and pediatric medical records. Child current weights and heights were measured by trained staff during an in-person clinic visit. Weight was measured on a digital scale and standing height on a stadiometer with children wearing light clothing and shoes removed. Child age- and sex-specific weight and BMI percentiles and z-scores were calculated using the World Health Organization Growth Charts (birth to 24 months) (World Health Organization, 2006) and the CDC Growth Charts 2000 (>24 months) (Ogden et al., 2002). Children were classified as underweight (BMI-for-age < 5th percentile), normal-weight (BMI-for-age 5–84th percentile), overweight (BMI-for-age 85–94th percentile), or obese (BMI-for-age ⩾95th percentile).

Rapid weight gain during infancy was defined as a change in weight-for-age z-scores from birth to 6 months >0.67 standard deviation (SD) (Monteiro and Victora, 2005). This 0.67 SD represents an increase in weight on a standard growth chart that would be equal to crossing upward by at least one weight centile band (Ong and Loos, 2006).

Several birth variables were selected based on clinical significance and were assessed for their effect on infant growth during pregnancy and early childhood: (1) birth weight from medical record, (2) gestational age categorized based on infants’ estimated date of delivery (maternal report) and converted to weeks of gestation, and (3) duration of breastfeeding (never, 0–3 months, 3–6 months, >6 months) from maternal interviews. Term was defined as 37 to ⩽41 weeks gestation and post-term as >41 weeks. Level of prematurity was defined as moderate to late preterm at 32 to <37 weeks and very or extremely preterm at <32 weeks gestation (World Health Organization, 2017).

Maternal self-reported height and pre-pregnancy weight were derived from the maternal interview. Maternal pre-pregnancy BMI was calculated as weight (kg) divided by height (m) squared. Mothers were classified as underweight (BMI < 18.5 kg/m²), normal-weight (BMI 18.5–24.9 kg/m²), overweight (25.0–29.9 kg/m²), or obese (⩾30 kg/m²) (Expert Panel on the Identification, Evaluation, and Treatment of Overweight in Adults, 1998). GWG was determined by subtracting mothers’ pre-pregnancy weight from their recorded weight at delivery. Maternal GWG adequacy was determined by comparing weight gain to the IOM guidelines for weight gain during pregnancy according to pre-pregnancy BMI (underweight: 12.5–18.0 kg; normal-weight: 11.5–16.0 kg; overweight: 7.0–11.5 kg; obese: 5.0–9.0 kg) and categorized as meeting or not meeting (above or below) the weight gain recommendations (Institute of Medicine et al., 2009).

Several maternal conditions were selected based on clinical significance and assessed for the effect on infant growth during pregnancy: (1) pre-pregnancy or gestational diabetes, (2) high blood pressure (BP) before or during pregnancy and pregnancy-related high BP conditions (eclampsia, pregnancy-induced hypertension, HELLP (hemolysis, elevated liver enzymes, low platelet count) syndrome), (3) eating disorders (bulimia nervosa, anorexia nervosa, or dieting during pregnancy), (4) intrauterine growth restriction (IUGR), and (5) smoking during pregnancy. Data for pre-pregnancy and gestational diabetes, pregnancy-induced hypertension, and IUGR were abstracted from medical records; data for eclampsia, HELLP syndrome, and high BP were abstracted from medical records and maternal interviews (if data from medical records not available); data for eating disorders were abstracted from medical records, maternal interviews, and medical history forms; and data for maternal smoking were abstracted from the maternal interview.

Data were analyzed using the SAS System for Windows (Version 9.3; SAS Institute, Cary, NC). Bivariate analyses assessed differences in demographic/anthropometric characteristics and maternal medical conditions by case status using analysis of variance (ANOVA), Kruskal–Wallis, independent t-tests, or Wilcoxon tests for continuous variables and chi-square/Fisher’s exact tests for categorical variables. A generalized linear model (PROC GENMOD) assessed the association between maternal pre-pregnancy BMI, GWG (met, above, or below IOM recommendation), rapid weight gain, and case status on child BMI status (overweight/obese vs underweight/normal-weight) at ages 2–5 years. The interaction between predictors and case status was tested for significance in all models. The influence of demographic, maternal, and birth variables on outcomes was tested by adding these variables as covariates to the models. Maternal and child demographic covariates included maternal education, race/ethnicity, poverty status, child age, and sex. Poverty status was determined using the U.S. Census Bureau 2015 Poverty Thresholds (United States Census Bureau, 2015). Maternal covariates included diabetes, high BP conditions, eating disorders, IUGR, and smoking status. Birth covariates included birth weight, prematurity status, and breastfeeding duration. Analyses related to the association between maternal GWG and child weight status excluded mothers with preterm infants (N = 414) to avoid misclassification of IOM status due to a shortened gestational period. P-values < 0.05 were considered statistically significant.

At the clinic visit, children were, on average, 59 months of age across groups (Table 1). A significantly higher proportion of children with ASD were male (82%). Children with ASD showed the highest frequency of rapid weight gain (44.0%) during the first 6 months of life when compared with children in the DD (35.9%) and POP (33.3%) groups. Children with ASD showed higher overweight/obesity rates than children in the POP group (P < 0.001), but not the DD group (P = 0.247). A significantly smaller proportion of mothers in the ASD (P < 0.001) and DD (P < 0.001) groups than mothers in the POP group attended college; a higher proportion of mothers in the ASD group lived below the poverty level (P < 0.001); and a higher proportion of mothers in the POP group were White (P < 0.001). Mothers in the ASD and DD groups showed higher pre-pregnancy overweight/obesity rates than mothers of the POP group. There was a significant relationship between GWG adequacy and case status: the proportion of mothers not meeting the recommendations for GWG adequacy in the ASD and DD groups was higher (65%) than in the POP group (59%). Mothers in the ASD and DD groups showed higher prevalence of gestational diabetes, hypertension, and smoking during pregnancy and a greater proportion of them breastfed for <3 months compared to mothers in the POP group. Fewer mothers in the POP group delivered preterm babies (14%) compared to mothers in the ASD (20.8%) and DD (26.9%) groups, respectively.

There was a significant main effect of maternal pre-pregnancy BMI on all children’s weight status (P < 0.001), but no significant interaction between children’s case status and mothers’ pre-pregnancy BMI (P = 0.25). This finding indicates that the association between maternal pre-pregnancy BMI and child weight status was consistent across groups. After adjusting for case status, mothers who were overweight/obese prior to pregnancy were 2.38 times more likely to have an overweight/obese child at the time of the clinic visit as compared to mothers who were underweight/normal-weight pre-pregnancy (P < 0.001; Table 2). Of the mothers who were overweight/obese pre-pregnancy, 33% had an overweight/obese child while 16.8% of the mothers who were underweight/normal-weight pre-pregnancy had an overweight/obese child (chi-square = 82.3; P < 0.001). The main effect of maternal pre-pregnancy BMI on child weight status remained statistically significant in adjusted analyses. After controlling for all covariates, mothers with pre-pregnancy overweight/obesity were two times more likely to have an overweight/obese child when compared to mothers who were normal-weight pre-pregnancy (95% confidence interval (CI): 1.57–2.53; P < 0.001).

When controlling for (1) case status and demographic covariates and (2) case status, demographic, and maternal covariates, mothers whose GWG fell below the IOM recommendations had (1) 34% (95% CI: 0.47–0.92) and (2) 31% (95% CI: 0.49–0.96) lower odds of having an overweight/obese child when compared to mothers who exceeded the recommendations, respectively. In addition, when controlling for case status, mothers who exceeded the GWG recommendations were 1.48 (95% CI: 1.17–1.87) times more likely to have an overweight/obese child as compared to mothers who met the weight gain recommendations (P = 0.001). This association remained statistically significant after adding demographic, maternal, and birth covariates (birth weight and breastfeeding) separately and simultaneously to the model (P = 0.033; Table 3). Sensitivity analyses showed a significant association between birth weight and adherence to GWG recommendations (P < 0.001). Infants whose mothers exceeded the recommendations had significantly higher birth weights (3302.2 ± 768.4 g) compared to infants whose mothers met (3203.5 ± 662.7 g) or gained less than the recommended amount of weight during pregnancy (2916.4 ± 824.3 g; P < 0.001). There also was a significant association between breastfeeding and adherence to GWG recommendations (P < 0.001). Infants whose mothers gained less than the recommended amount of weight during pregnancy were significantly less likely to have been breastfed (21.1%) than those whose mothers met (11.5%) or exceeded (12.2%) the GWG recommendations (P < 0.001).

Among children with ASD, those with rapid weight gain (i.e. change in weight-for-age z-scores from birth to 6 months >0.67 SD) showed a significantly higher frequency of overweight/obesity at ages 2–5 years compared to those without (37.6% vs 19.1%; P < 0.001). When adjusting the model for covariates, the association between rapid weight gain and childhood overweight/obesity among those with ASD remained significant (Table 4). Children with ASD and rapid weight gain during infancy had 3.47 times greater odds of being overweight/obese during early childhood than children with ASD who did not show rapid weight gain after controlling for all covariates (95% CI: 1.85–6.51; P < 0.001). The variable responsible for this increase in the odds ratio was the addition of birth covariates (infant birth weight) to the model, indicating that low birth weight infants were at greater risk for rapid weight gain.