Maternal Obesity, Gestational Weight Gain, and Asthma in Offspring

Kristen J. Polinski, MSPH; Jihong Liu, ScD; Nansi S. Boghossian, PhD, MPH; Alexander C. McLain, PhD


Prev Chronic Dis. 2017;14(12):E109 

In This Article


The Early Childhood Longitudinal Study–Birth Cohort (ECLS–B) is a longitudinal study conducted by the National Center for Education Statistics (NCES) and is designed to collect information on childhood health, development, care, and education. A nationally representative sample of children born in the United States in 2001 was selected from birth certificates, and children were followed through kindergarten entry as part of the ECLS–B.[8] We used birth certificate data as well as data from assessments that took place when children were aged 9 months, 2 years, and 4 years. The weighted longitudinal response rates, which take into account response for all rounds of data collection, were reported by NCES as follows: 74.1% at 9-month collection, 69.0% at 2-year collection, and 63.1% at 4-year assessment.[9] The Arnold School of Public Health is licensed to use the ECLS–B restricted-use data.

Primary Outcome

The presence of childhood asthma was assessed at each wave of data collection. In each interviewer-administered assessment, a guardian of the child was asked, "Since your child turned (x) years of age, has a doctor, nurse or other medical professional ever told you that your child has asthma?"

Primary Maternal Exposures

In the 9-month assessment, a trained interviewer asked the mother, "How much did you weigh just before you became pregnant with [child]?" The mother was also asked to provide information on her height. These 2 questions were used to obtain data for calculation of the mother's pre-pregnancy BMI (measured as weight in kilograms divided by height in meters squared [kg/m2). Maternal BMI was classified as underweight (<18.5 kg/m2), normal weight (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2) and obese (≥30.0 kg/m2).[10]

We obtained data on total gestational weight gain from the birth certificate, and if data on gestational weight gain were missing in birth certificates (19% of certificates), we used self-reported values. We used 4 measures of total gestational weight gain. First, we treated data on gestational weight gain as continuous data. Second, we determined a measure of weight gain adequacy based on recommendations of the Institute of Medicine (IOM) by using methods described previously.[11,12] Briefly, we calculated a ratio of the observed total gestational weight gain to the expected gestational weight gain for each mother; we then categorized the ratio according to the percentage of IOM weight gain recommendations met (ie, inadequate, adequate. and excessive) based on ranges of the mother's pre-pregnancy BMI.[12] We determined expected gestational weight gain by using the following equation: Expected gestational weight gain = recommended first trimester total weight gain + [(gestational age at weight measure at or before delivery − 13 weeks) × recommended rate of gain in second and third trimesters],[11,12] where recommended total first-trimester weight gain was assumed to be 2.0 kg for underweight or normal-weight mothers, 1.0 kg for overweight mothers, and 0.5 kg for obese mothers. Recommended rate of weight gain in second and third trimesters was based on the assumption that underweight, normal-weight, overweight, and obese women gain weight at the rate of 0.51 kg, 0.42 kg, 0.28 kg, and 0.22 kg per week, respectively.[13] Third, we calculated the weekly rate of gestational weight gain in the second and third trimesters by dividing the estimated total weight gain in the second and third trimesters (ie, total gestational weight gain minus estimated weight gain in the first trimester per IOM guidelines) by the number of weeks in the second and third trimesters (ie, gestational age at delivery minus 13). Fourth, we divided data on gestational weight gain into 6 categories (<5 kg, 5–9 kg, 10–15 kg, 16–19 kg, 20–24 kg, and ≥25 kg) to enable comparisons between our data and data from other studies.[14] We considered less than 5 kg of weight gain to be an extreme low and 25 kg be an extreme high, and we designated 10 to 15 kg as the reference level.


All 8 covariates were one-time measurements taken from either the birth certificate or the 9-month assessment. In addition to sociodemographic variables (mother's race, maternal age, child's sex), we selected, on the basis of research, the following covariates: child's birth weight, parity, gestational age, participation in the Special Supplemental Nutrition Program for Women, Infants and Children (WIC) in the previous 12 months, and smoking during pregnancy.[14–20]

Statistical Analyses

Per NCES data-use guidelines, we rounded all sample sizes to the nearest 50; thus, a sample size of 100 reflects a sample size that ranges from 75 to 124. Before any exclusions, the sample size was 10,700. After excluding twins, children with a birth weight of 500 g or less, and children whose gestational age was less than 28 weeks or more than 45 weeks, the sample was reduced to 8,550. We then removed children missing data on ECLS–B weights, which removed those who did not have a preschool assessment, reducing the sample to 6,900. Next, we removed children who had missing data on other covariates (n = 450), leaving an analytic sample of 6,450; of these 450 children, 100 were missing data on gestational weight gain and 250 were missing data on pre-pregnancy BMI.

All analyses were adjusted for the clustered sample design and weighted to account for oversampling of particular groups and attrition.[8] We tabulated baseline data on descriptive characteristics for the total analytic sample by using weighted percentages, means, and 95% confidence intervals (CIs), with inference provided by weighted χ 2 tests (categorical variables) or t tests (continuous variables). With repeated asthma measurements, logistic regression models via generalized estimating equations with exchangeable correlation structure produced odds ratios (ORs) and 95% CIs for an asthma diagnosis by age 4 years. We ran 4 models, one for each measure of gestational weight gain. Each model included pre-pregnancy BMI and adjusted for all 8 covariates. We tested an interaction between each measure of gestational weight gain with pre-pregnancy BMI and child's age at asthma diagnosis. No interactions were significant at the .10 level, and thus no interaction terms were included in any model.

As a sensitivity analysis, we applied multiple imputation methods to impute missing birth certificate data for gestational weight gain and BMI by using self-reported data and other covariates. Ten imputed data sets were analyzed and combined by using standard multiple imputation techniques.[21] Briefly, we first imputed 10 data sets for continuous gestational weight gain by using PROC MI. Next, we used PROC GENMOD, using the same 8 covariates described above, to analyze each of the 10 imputed data sets. Then, PROC MIANALYZE was used to combine results for inference. All analyses were implemented in SAS version 9.3 (SAS Institute Inc).