Subjects

Three hundred six full-term healthy infants (163 males and 143 females) participated in the study. The subjects were recruited at Roosevelt Hospital in midtown Manhattan, NY. The study was approved by the St Luke’s–Roosevelt Hospital Institutional Review Board (Institutional Review Board #06-016; approval date January 2007). A written consent was obtained from a parent before participation.

Race was determined by self-identification of each subject from the following 4 categories: (1) Asian, (2) non-Hispanic black (African American), (3) non-Hispanic white, and (4) Hispanic. Subjects were asked to select the category into which each parent and grandparent fell. When all categories were the same, the subject was identified by that category. When multiple categories were selected, the individual was classified as “other.” In the current sample, the maternal race/ethnic distribution was as follows: 143 white, 37 African American, 70 Hispanics, 47 Asian, and 9 other.

Study procedures

Subjects were recruited while an inpatient at Roosevelt Hospital from the maternity floors. Inclusion criteria for the study included newborns that were healthy with no known birth defects, congenital abnormality or an admission to the neonatal intensive care unit, a singleton birth, and full term (longer than 37 weeks). Mothers diagnosed with gestational diabetes, hypertension, or preeclampsia were excluded.

Anthropometric measurements and body composition assessment were conducted prior to infant discharge between 1 and 2 days after birth. Measurements were not taken during the first 24 hours (day 0) because pilot data from 8 infants collected in our laboratory suggest that there may be an initial weight loss during this period. For these 8 infants, the mean (SDs) weights measured during days 0, 1, and 2 were 3286 g (680 g), 3163 g (669 g), and 3136 g (682 g), respectively. Using a repeated-measures analysis of variance, the mean weights at days 1 and 2 were significantly less than the mean weight at day 0 ( P < .0001); however, the mean weights at days 1 and 2 were not significantly different from each other ( P = .2679).

Our sample size was not sufficient to include women who had a prepregnancy underweight BMI or who gained below what was recommended for weight during pregnancy. Twenty-two mothers had an underweight prepregnancy BMI; of these, 8 gained below the recommended weight during pregnancy, 9 gained an appropriate amount, and 5 gained an excessive amount. For a prepregnancy BMI of normal weight, overweight, and obese, 70 (21%), 2 (2%), and 10 (16%), respectively, gained below the recommended weight during pregnancy.

Presented in Table 1 are the IOM 2009 GWG recommendations. Based on their prepregnancy BMI, women were classified as gaining an appropriate or excessive amount of weight. When a woman gained within the recommended range for their prepregnancy BMI, she was classified as appropriate. When she gained greater than the recommended amount of weight, she was classified as gaining an excessive amount of weight. Gestational weight gain was calculated using inpatient chart–extracted prepregnancy weight when possible or self-reported prepregnancy weight and the self-reported highest weight measured during pregnancy. Good agreement has been reported between maternal recall of prepregnancy weight and medical records for prepregnancy weight such that recall of maternal prepregnancy weight is considered a satisfactory substitute when chart extraction is incomplete.

Air displacement plethysmography (Pea Pod)

An infant board (Shorr Productions, Olney, MD) was used to measure body length to the nearest 0.1 cm. Two measurements of body length were taken for each infant and an average of the measurements was used. The Pea Pod body composition system (Life Measurement Instruments, Concord, CA) was used to measure body volume. The Pea Pod was calibrated once daily prior to beginning testing. A calibration cylinder with a known volume was used to calibrate the chamber and a 5000 g weight was used to calibrate the scale.

Testing procedures have been described in detail elsewhere. Infants were undressed and wore a standard tight-fitting hat (Allentown Scientific Associates, Inc, Allentown, PA) to minimize air trapped in the hair for body volume and body weight measurements. The infant was placed naked on the scale, and a body weight was obtained to the nearest 0.0001 kg. Any irremovable items such as the umbilical clamp and identification bands were tared for the body weight and body volume measurements. After body weight was measured, the infant was placed inside the Pea Pod wearing a wig cap, and a body volume measurement was performed. Assessment of the body volume required approximately 2 minutes. Body density was then converted to percent fat (%fat) using sex-specific equations by Fomon et al.

Prior studies have shown the Pea Pod is a valid tool to measure %fat when compared with the gold standard 4-compartment model (4C) and deuterium, in which no differences were found for %fat between the Pea Pod and the 4C model (16.9 ± 6.5% and 16.3 ± 7.2%, respectively) or between the Pea Pod and deuterium dilution (20.32 ± 6.87% and 20.39 ± 6.68%, respectively). Ellis et al explored the contribution of the variation in hydration, protein, and bone mineral fractions on the differences between %fat from the Pea Pod vs a 4C model. The bone mineral fraction explained the greatest variation (16%); the protein and hydration fractions explained an additional 0.2% and 0.1%, respectively. The mean age of the infants in these prior validation studies was approximately 8 weeks (range, 0.4–23.0 weeks), which is older than the current study population.

Statistical analysis

Analysis of covariance was used to investigate the main effects of GWG category and prepregnancy BMI category and their interaction on infant body composition (percentage body fat [FM] and fat-free mass [FFM]). Percentage body fat is defined as the percentage of body mass comprised as fat mass (ie, the relative amount of fat mass taking body mass into consideration [100 ★ (FM (grams)/body mass (grams)]). For all models, the covariates were infant sex, infant age (days), gestational age, maternal race/ethnicity, and maternal age. Infant weight was not considered as a covariate for the analyses of the effects of GWG category and prepregnancy BMI category on FM and FFM because infant weight itself may be affected by these 2 variables. That is, if the analysis were to adjust for infant weight, it could artificially remove the effects of the GWG and prepregnancy BMI on FM or FFM.

Covariates should never be influenced by the factors being studied. By definition, %fat is the proportion of the infant’s weight that is fat; therefore, %fat was also analyzed to determine the effects of GWG and prepregnancy BMI on the composition of the infant’s weight. Significant main effects and interactions were investigated further using pair-wise comparisons. Fisher’s protected least significant difference was used to test differences for multiple comparisons. For each model, a residual analysis was performed to determine whether the distribution of the residuals were consistent with the assumptions required to perform the analysis of covariance. Data were analyzed using SPSS (version 17; SPSS Inc, Chicago, IL). Statistical significance was set at P < .05.