Materials and Methods
Subjects for this study were recruited from among 786 women who delivered vaginally at term (38–41 weeks’ gestation) during a 1-year period (January through December 1994). Eligible women had to be free of diseases known to interfere with calcium or phosphate metabolism. Smokers and those taking aluminum-containing antacids were excluded. At enrollment, only women who did not contemplate starting oral contraception until the end of the study at 18 months post-partum were recruited. Within 3 to 5 days of delivery, eligible mothers were asked whether they wanted to breast-feed their babies. Three hundred eight who declared intent to breast-feed their infants according to a proposed, standardized schedule made up the lactation cohort. Each gave written informed consent at enrollment. They agreed to fully breast-feed their infants for 6 months, followed by a 1-month period of weaning, and to take cabergoline administered at the end of the seventh postpartum month. Cabergoline, a long-acting dopamine agonist, is used routinely for lactation suppression at our institution. Diet was unrestricted, but lactating women were allocated either no dietary supplement or a daily calcium supplement of 1 g. Calcium supplementation was restricted to lactating mothers and was assigned using a computer-generated list of random numbers. One hundred seventy-four other women who declared no intent to breast-feed were given lactation suppression with cabergoline between postpartum days 2 and 3 and formed the nonlactating cohort. Of 308 women in the lactation cohort, 274 completed the 6-month follow-up. Among those who withdrew in the first 6 months, 24 started bottle feeding between 2 and 6 months postpartum, and ten others said they wanted to take oral contraceptives after the weaning month. Of 174 women in the nonlactating cohort, 153 completed the study; the other 21 requested oral contraception after 3–5 menstrual cycles and therefore were excluded from analysis.
The women were instructed to record the date they resumed menstruation, the following menstrual bleedings, and monthly measurements of body weight. In all women, we measured bone mineral density of the radius and spine (L2–L4) at baseline (postpartum days 5–10) and at 3, 6, and 12 or 18 months postpartum. Lactating mothers also had bone mineral density measurements at 7 months postpartum, at the end of the weaning period. All bone mineral density values were adjusted for weight and height of each woman.
Bone mineral density of the radius of the nondominant arm was measured using the Osteometer DTX 100 (Osteometer A/S, Roedovre, Denmark) where the distance between the ulna and radius is less than 8 mm (automatic scanning) and the trabecular and cortical components of the radius are similar to those of the lumbar spine. The coefficient of variation was 1.2%. To evaluate bone mineral density at the lumbar spine (vertebrae L2–L4), we used a dual energy x-ray absorptiometer (Norland XR 26; Norland Corp, Fort Atkinson, WI) with 1.1% coefficient of variation.
To assess the association between radius and spine bone mineral density at the start of observation, Spearman correlation coefficient was computed, and to verify the independence of treatment group, multiple regression was used. Nonparametric analysis of variance (Kruskall-Wallis test) was used, because of the presence of nonhomogeneous variances across groups. For post hoc comparisons, the Mann-Whitney U test was used. To overcome possible differences at baseline, the relative percentage change in density with respect to baseline was computed for all times and was considered the outcome of interest. A generalized linear model was used for the analysis of repeated measurements (percent change) for radium and spine mineral density, to compare the three study groups over time. To assess the additional and independent effect of early versus late resumption of menstruation (no more than 5 versus over 5 months postpartum) on bone mineral density in breast-feeding women, with or without calcium supplementation, a new model was fitted that included time, treatment group, and time to resumption of menstruation, with the interaction of the latter two. P Stata 5.0 (StataCorp, College Station, TX) was used for computation.
Answers to all your Biology Questions
