Patient sample

This is a secondary analysis of the MFMU Trial for the Prevention of Recurrent Preterm Delivery by 17-Alpha Hydroxyprogesterone Caproate. Full details are available in the original publication. Briefly, the original study was performed at 19 participating centers from 1999 through 2002. Women presenting for prenatal care were screened between 15-20 3/7 weeks’ gestation for a history of spontaneous PTB that occurred between 20-36 6/7 weeks. Eligible patients were randomized in a 2:1 fashion to weekly intramuscular injections of 250 mg of 17OHP-C or placebo, respectively, beginning at 16-20 6/7 weeks. There were 463 individual women in the original trial. For this analysis, patients with missing gestational age at delivery, maternal BMI, and/or maternal weight were excluded, leaving 443 patients for this analysis ( Figure 1 ). Reported maternal prepregnancy height and weight were obtained at the time of study enrollment and BMI was calculated. Actual weight at enrollment, total weight gain, and weight at delivery were not recorded. The MFMU study was approved by institutional review boards at each of the study sites with each subject providing written informed consent. This secondary analysis was considered exempt by the Colorado Multiple Institutional Review Board.

Patient selection from original cohort

Selection of patients for this study from initial patient cohort.

BMI , body mass index; GA , gestational age; MFMU , Maternal-Fetal Medicine Units; Wt , weight.

Heyborne. 17-Alpha hydroxyprogesterone caproate in obese women. Am J Obstet Gynecol 2015 .

Definitions and outcomes

For this analysis, we used the same primary outcome variable as the original study, namely preterm delivery <37 weeks. To investigate the interaction of maternal weight and BMI on 17OHP-C effectiveness, we used 2 standard categorical BMI definitions (3-category: <25, 25-<30, ≥30; 6-category: <18.5, 18.5-<25, 25-<30, 30-<35, 35-<40, ≥40). Weight was treated as a continuous variable. In the original data set, women with weight >300 lb were assigned a weight of 300 lb; women with these right-censored weights were excluded from further analyses using weight as a continuous variable, but remained in analyses using categorical BMI.

Statistical methods

Demographics and clinical characteristics of women randomized to 17OHP-C and placebo were summarized with mean and SE for continuous variables, and frequency and percentage for categorical measures. Differences between treatment groups were tested with 2-sample t test for continuous and χ ² for categorical measures. Demographics of women in each of 3 BMI categories were summarized similarly, and differences tested using analysis of variance for continuous and χ ² for categorical measures.

PTB rate and 95% confidence interval (CI) were calculated across BMI and weight categories according to treatment. To obtain unadjusted relative risks (RR), the probability of PTB was modeled using binomial regression, with treatment, BMI (3- and 6-category), or weight and their interactions in 3 separate models. To obtain adjusted RR, multivariable binomial regression models were estimated with an expanded set of potential covariates reported in the literature to be associated with the risk of PTB: race (black vs other), age >30 years, marital status, years of education, smoking during pregnancy, illicit drug and alcohol use during pregnancy, and >1 previous PTB. Parameters were eliminated through backwards selection separately for each model. In sensitivity analysis, models of continuous weight were estimated separately for black and non-black women. All analyses were performed in SAS 9.4 (SAS Institute Inc., Cary, NC).