Data

North Carolina’s Pregnancy Nutrition Surveillance System is compiled by linking data from the North Carolina Special Supplemental Nutrition Program for Women, Infants, and Children (NC WIC) with birth certificates and fetal death certificates. With the exception of the coded NC WIC participant identification number, data from the North Carolina’s Pregnancy Nutrition Surveillance System used for this study were deidentified and obtained through a data use agreement with the North Carolina Department of Health and Human Services. This agreement and the study protocol were approved by the Duke University Institutional Review Board.

Study population

Women who were enrolled in the NC WIC program for >1 pregnancy between 1996 and 2004 were identified by the coded NC WIC participant identification that was provided in the dataset. Maternal demographic and clinical characteristics, including parity, were collected when women were admitted to the NC WIC program during pregnancy and at the postpartum recertification visit when the mother renews her participation in NC WIC. In an attempt to capture only the first and second live births for each woman (and thus exclude women who may have had an intervening pregnancy during which they were not a participant in the NC WIC program), only women with sequential pregnancies for which parity was recorded as “0” and “1” were included (n = 47,903). Subjects who had a multifetal gestation during either their first or second pregnancy (n = 1247) or a gestational age at delivery recorded as <34 weeks for either their first or second pregnancy (n = 2520) were excluded. There were also 16,189 women with either a missing height, a missing prepregnancy weight for either pregnancy, and/or a missing GWG for either pregnancy who were excluded.

To eliminate potential coding errors, only those women with a recorded height between 48-72 inches, recorded prepregnancy weights between 80-600 lb, and calculated prepregnancy BMI of ≥15 kg/m ² were included in the analysis. This resulted in a final analysis sample of 27,771 women.

Measures

In 2009, the IOM released updated recommendations for GWG based on prepregnancy BMI: <18.5 kg/m ² : 28-40 lbs; 18.5-24.9 kg/m ² : 25-35 lbs; 25-29.9 kg/m ² : 15-25 lbs; ≥30 kg/m ² : 11-20 lbs. Prepregnancy BMI was calculated with the use of self-reported prepregnancy weight recorded at the NC WIC prenatal visit for each pregnancy and a height measured by WIC office staff. GWG for each pregnancy was obtained from the birth certificate. Based on prepregnancy BMI, subjects were then categorized as having “inadequate” GWG if they gained less than the IOM recommendations, “appropriate” GWG if they gained within the IOM recommendations, and “excessive” GWG if they gained more than the IOM recommendations.

We examined several other candidate predictors and potential confounders of the association between GWG in the first and second pregnancies: age at first delivery (<21 or ≥21 years), race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, and other), education (less than high school or high school or higher), smoking during the second pregnancy (yes/no), breastfeeding duration after the first pregnancy (<24 or ≥24 weeks), and difference in maternal age between the 2 deliveries (<2 or ≥2 years). Maternal age, race/ethnicity, education, and smoking were variables that were obtained from the birth certificate. The “other” category for race included those women who were coded as >1 racial category. Although variables for smoking in both pregnancies were available, only the variable for smoking in the second pregnancy was used because it was correlated highly with smoking in the first pregnancy (r = 0.62). The age variable was dichotomized at 21 years to examine how a first pregnancy that occurs in the teenage/adolescent years affects GWG. Breastfeeding duration was recorded at WIC postnatal visits and was defined as “any breastfeeding,” as opposed to “exclusive breastfeeding.”

Statistical analysis

GWG category in the first pregnancy served as the primary predictor variable for the primary outcome of GWG category in the second pregnancy. To evaluate for potential colinearity among covariates, correlation coefficients were calculated and demonstrated that none were highly correlated. All variables were modeled with reference cell coding. Excessive and inadequate GWG were compared with appropriate GWG as the reference for both the primary predictor and outcome variables. A polytomous logistic regression model was used to test the association of excessive or inadequate GWG in the first pregnancy with excessive or inadequate GWG in the second pregnancy, while being controlled for the described covariates. To examine whether there were any differences in GWG patterns by BMI, a stratified analysis was done according to the woman’s BMI before her first pregnancy. Separate polytomous logistic regression models that tested the association of GWG category in the first pregnancy with GWG category in the second pregnancy, controlling for the described covariates, were calculated for women in each of 4 prepregnancy BMI categories: underweight, normal weight, overweight, and obese. The polytomous models are summarized by odds ratios with corresponding 95% CIs. All statistical tests used a 2-sided α of .05. Analysis was performed with SAS software (version 9.2; SAS Institute Inc, Cary, NC).