The present study represents a secondary analysis of data collected from May 2000 through January 2003 on a cohort (N = 433) of adolescent girls who enrolled in a 2-year study of hormonal contraception and bone mineral density. Detail study methods have been previously reported (12
). The study population consisted of post-menarcheal girls, age 12 – 18 years, attending one of four urban adolescent health clinics in a large metropolitan area. Adolescent girls requesting contraception, and selecting either DMPA or OC, were eligible to participate. In addition, adolescent girls who planned to receive no hormonal contraception were eligible for enrollment as control subjects. One major aim of the parent study was to evaluate whether estrogen supplementation in subjects receiving DMPA resulted in decreased BMD losses. As such, DMPA subjects were randomized to receive either monthly injections of 5mg estradiol cypionate or placebo. The current analyses include only those subjects selecting DMPA for contraception (N = 97), both those randomized to receive estradiol cypionate or placebo.
Exclusion criteria for study participation included pregnancy or DMPA use within the preceding 6 months; OC use within the preceding 3 months; alcohol or drug dependence; medical condition (e.g. renal disease) or medication use (e.g. corticosteroids) know to be associated with the outcomes of interest; contraindication to estrogen use; weight exceeding 250 lbs (upper limit for DEXA scanner); and need for confidential contraceptive care. Subjects younger than 18 years gave written assent for participation, and written informed consent was obtained from a parent or legal guardian. Subjects aged 18 years provided their own written informed consent for participation. The study protocol was approved by the institutional review board of the participating institutions.
At baseline, 6, 12, 18, months, clinical and behavioral information was obtained from each study participant. Height and weight were measured by using the same stadiometer (Easy Glide Bearing stature board) and Mettler-Toledo scale. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters. Gynecologic age was calculated as the number of years since menarche. Physical activity was assessed with a survey that asked each participant to classify herself as inactive, normal, or active.
Since Le and colleagues found that early weight gain, defined as more than 5% weight gain at 6 months of DMPA, was most predictive of continued weight gain in adult women (11
), we dichotomized our DMPA users into those who did or did not experience a > 5% increase in body weight at 6 months of DMPA use. Girls who experienced more than a 5% weight gain at 6 months were classified as “early weight gainers” and girls who experienced less than or equal to a
5% weight gain at 6 months were classified as “non-early weight gainers.” In our study, group comparisons at baseline were completed using chi-square tests or Fisher's exact tests, as appropriate, for categorical variables and t-tests or Wilcoxon rank sum tests, as appropriate, for continuous variables. One-way Analysis of Variance (ANOVA) modeling was used to compare groups at each of the times points (baseline, 6 months, 12 months, 18 months) on BMI. Using BMI ≥85th and <95th sex- and age-specific percentile from the 2000 CDC growth charts as the definition of overweight and BMI ≥95th sex- and age-specific percentile from the 2000 CDC growth charts as the definition of obese, we calculated the percentage of both early and non-early weight gainers who were overweight or obese at each time point.
Repeated measures ANCOVA modeling was then used to explore the association between early weight gain and the percentage change in BMI after 12 and 18 months of DMPA use. In the repeated measures ANCOVA modeling, we incorporated a factor for group and time, while adjusting for possible baseline characteristics (age, gynecologic age, race, ethnicity, BMI, menstrual regularity, duration of menses, menstrual cycle length, sexual activity, previous pregnancy, physical activity, and randomization to estradiol or placebo). Results with p-value less than or equal to an alpha-level of 0.05 were considered statistically significant. Data analyses were conducted with SAS statistical software, version 9.1 (SAS Institute Inc., Cary, NC).