Pre-Pregnancy Body Mass Index, Gestational Weight Gain, and Other Maternal Characteristics in Relation to Infant Birth Weight

Ihunnaya O. Frederick; Michelle A. Williams; Anne E. Sales; Diane P. Martin; Marcia Killien

Disclosures

Matern Child Health J. 2008;12(5):557-567. 

In This Article

Methods

Study Design and Population

The data for this analysis were collected as part of the Omega Study, an ongoing, prospective cohort study of the risk factors of preeclampsia and gestational diabetes mellitus among pregnant women receiving prenatal care from obstetrics clinics affiliated with Swedish Medical Center, Seattle, and Tacoma General Hospital, Tacoma, WA, from December 1996 to October 2004. Eligible participants were pregnant women who planned to deliver at Swedish Medical Center or Tacoma General Hospital, were ≤20 weeks gestation at time of enrollment, 18 years or older, and English-speaking. Women were referred to the study during their first obstetrics clinic visit by nursing staff at participating clinics. Study staff followed up with a telephone call, provided details of study protocol, answered questions, and set up appointments to complete the informed consent process. Each participant signed an informed consent form. The protocol for the study was approved by the institutional review boards of Swedish Medical Center, Seattle, Tacoma General Hospital, Tacoma, WA, and the University of Washington, Seattle, WA.

Analytical Population

During the period of the study, 3,899 eligible participants were approached and 3,000 women (77%) were enrolled in the study. Reasons for non-participation included: (1) not having time to participate in the interview; (2) having no interest in the goals of the study; and (3) missed appointments. Of the enrolled, 41 women experienced spontaneous abortion, 19 women underwent voluntary termination of pregnancy, and 10 women delivered stillborn infants. After excluding those women as well as women with missing or incomplete pregnancy outcome information (n = 154), and those who delivered twins or triplets (n = 106), a total of 2,670 women delivering singleton infants remained for this analysis.

Data Collection

A structured questionnaire was used by trained research interviewers in in-person interviews to obtain maternal socio-demographic and anthropometric characteristics, i.e. age, weight, height, education, as well as family medical history information. Interviews were conducted between 16-20 weeks gestation. Confirmatory data on pre-pregnancy weight and height, labor and delivery information, as well as baby's health at delivery was obtained from maternal and infant medical records review.

Identification of Exposure

The main exposure variable, pre-pregnancy BMI, was calculated using self-reported weight (in pounds) at 3 months before conception and converted into kilograms, divided by maternal height in meters squared (kg/m2). Women were grouped into four categories according to pre-pregnancy BMI: <19.8 kg/m2 (lean); 19.8-26.0 kg/m2 (average); 26.1-29.0 kg/m2 (overweight), and ≥29.0 kg/m2 (obese). These categories were used by the IOM to develop gestational weight gain guidelines for optimal pregnancy outcome .[14] We used the average pre-pregnancy BMI group as the referent group in categorical analysis because it was considered representative of normal pre-pregnancy weight.[14]

Identification of Covariates

Gestational weight gain was calculated as the difference in weight (in kilograms) between last recorded maternal weight within 4 weeks of delivery, and self-reported maternal weight at three months before conception. The IOM recommended gestational weight gain for categories of pre-pregnancy BMI are 12.5-18.0 kg (28-40 lbs) (lean); 11.5-16.0 kg (25-35 lbs) (average); 7.0-11.5 kg (15-25 lbs) (overweight); and ≥7.0 kg (≥15 lbs) (obese).[14] Compliance with IOM guidelines was categorized into three groups: below, within, and above guideline. In addition, gestational weight gain was categorized into two groups: ≤median and >median when it was considered necessary to collapse categories to increase cell numbers in analysis. Rate of gestational weight gain was calculated as total gestational weight gain divided by completed weeks of gestation. Gestational age was determined using self-reported date of last menstrual period (LMP) and confirmed by earliest ultrasound, when available, or by physician's best LMP estimate. Preterm birth was defined as gestational age <37 completed weeks at delivery. Other covariates were maternal age, race/ethnicity, parity, education, marital status, smoking status, gestational weight gain, infant gender, preterm birth, gestational diabetes, and preeclampsia.

Identification of Outcome

Infant birth weight was measured in grams, and abstracted from medical records. For categorical analysis, infant birth weight was grouped according to clinically relevant categories: LBW (<2,500 g), normal birth weight (2,500-3,999 g), and macrosomia (≥4,000 g), and according to the World Health Organization recommendation.[22]

Statistical Analysis

Frequency distributions of independent variables were calculated according to the three groups of infant birth weight. Bivariate associations were assessed using scatter plots, correlation coefficients, and simple linear regression. We further explored suggestion of a nonlinear relationship between infant birth weight and pre-pregnancy BMI (from scatter plot) by adding higher order-terms of pre-pregnancy BMI to regression models to find the best fit for the data. We evaluated compliance with the IOM gestational weight gain guidelines by assessing the proportion of women gaining below, within and above guidelines for each category of pre-pregnancy BMI. For the obese group, we assumed it reasonable to use the IOM recommended upper limit of weight gain for overweight women (11.3 kg) as have investigators in previous studies.[15,21]

Multivariate regression modeling procedures were used to assess associations between infant birth weight and pre-pregnancy BMI and other covariates. We considered the following covariates as potential confounders and controlled for them in the final model: maternal age, race, education, marital status, parity, infant gender, smoking status, preterm birth, gestational diabetes, and preeclampsia, based on prior study results.[5,6,7,13,15,16,17,18] We assessed confounding by entering potential confounders into a regression model one at a time with the dependent variable [25] and evaluated the estimated beta coefficient. In situations where potential confounders may be specified as continuous, dichotomous, or grouped-linear, we included in the final model, whenever appropriate, the most parsimonious variable specification that achieved the greatest control of confounding. The final regression model included those covariates that were significantly associated with infant birth weight, increased coefficient by 10% or more, as well as covariates of a priori subject-matter knowledge.[23] We considered gestational weight gain an intermediary variable between pre-pregnancy BMI and infant birth weight.[23,24] In addition, gestational weight gain was considered a potential effect modifier of the association between pre-pregnancy BMI and infant birth weight.[15,21]

We used generalized linear modeling (GLM) estimation to calculate relative risk (RR) and 95% confidence intervals (CI) of LBW and macrosomia, respectively across categories of pre-pregnancy BMI.[21] We assessed linear trend in RRs by assigning a score of 1, 2, 3, 4 respectively to the categories of pre-pregnancy BMI and treating it as a continuous variable in a regression model.[25] Some investigators have noted independent effect, as well as the combined effect of pre-pregnancy BMI and gestational weight gain on pregnancy outcomes.[15,16,17,18,21] We explored the possibility of interaction of pre-pregnancy BMI with gestational weight gain in relation to the risks of LBW and macrosomia, referencing normal birth weight. We explored possible effect modification by gestational weight gain in relation to the risk of LBW or macrosomia by constructing eight-level combined categories of pre-pregnancy BMI (four groups) and gestational weight gain (two groups: ≤median and >median weight gain). We calculated relative risks for LBW and macrosomia twice using different referent groups. The first referent group consisted of women of average pre-pregnancy BMI with >median gestational weight gain (19.8-26.0 kg/m2 and >15.9 kg), and the second included women of average pre-pregnancy BMI with ≤median gestational weight gain (19.8-26.0 kg/m2 and ≤15.9 kg).

Finally, we used ordered logistic regression procedures to assess the extent to which pre-pregnancy BMI is associated with the three clinically relevant categories of birth weight (LBW, normal birth weight, and macrosomia) and calculated predicted probabilities of delivering a LBW, normal birth weight or macrosomic infant given categories of pre-pregnancy BMI.[26] This estimation procedure is well suited for analysis involving an outcome variable with more than two categories with ordinal grouping (i.e. the three categories of infant birth weight).[27] Stata 9.0 statistical package (Stata Corporation, College Station, Texas) was used for all analyses. All P-values are two-tailed and statistical significance was calculated at the α = 0.05 level.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.

processing....