Demographic characteristics of women who completed the 6-month follow-up are shown in Table 1. At baseline, there were no differences among the three groups in age, height, body weight, parity, and body mass index (BMI), with all subjects aged between 27 and 36 years.
Wanting to use oral contraception was the reason for most withdrawals in the lactation cohort. Between 12 and 18 months, 22 lactating women who were taking no calcium supplement withdrew from the study, of whom 18 decided to take oral contraceptives. The respective numbers of lactating women taking calcium were 18 and 16, leaving 113 women taking no calcium and 121 with calcium in the lactating cohort available for the final 18-month evaluation.
Strong correlation between spine and radius measurements was found at the start, with a Spearman correlation coefficient of 0.7197 (P . There was a significant progressive decline in bone mineral density in the lactation cohort during the first 6 months post-partum. A significant recovery of bone mass was found thereafter, levels that at 18 months were higher than baseline levels. In the nonlactating cohort, bone mineral density of the spine increased progressively and at 18 months postpartum was 1.9% higher than baseline levels. A similar pattern was found for bone mineral density of the radius, with an increase up to 1.1% by 18 months postpartum. When groups were compared over time, we found a significant decrease in bone loss in between lactating women receiving no calcium and lactating women with calcium, and between those and controls, for radius and spine mineral density, as well as a significant decrease over time. However, calcium supplementation did not allow us to separate breast-feeding mothers for the spine measurement. No interaction was observed between time and group. Twelve months after breast-feeding cessation, no significant differences in bone mineral density of radius or spine were found between groups.
To assess the impact of the return of menstruation on bone mineral density, lactating women were further subclassified according to whether menses resumed within 5 months postpartum or later. Results are shown in Table 3. Among lactating women taking no calcium, 77 of 135 had menstrual flow within 5 months, compared with 80 of 139 taking calcium. There were significant differences in bone mineral density at the levels of radius and spine in both groups between women who had menstruated early and those who had not. Such differences were evident at 6–7 months postpartum, but did not persist at 18 months. All nonlactating controls resumed menses within 35–55 days postpartum, thus they were not categorized by timing of menstruation.
Among breast-feeding women, calcium supplementation and amenorrhea duration independently predicted changes in radius mineral density over time, but no interaction was observed between them. To better illustrate the phenomenon, all pairwise comparisons among the four subgroups of subjects, which were significant at 1% level, were done. A decreasing loss of radius mineral density was observed from the no-calcium, late-menstruation subgroup to the with-calcium, late-menstruation subgroup, then to the no-calcium, early menstruation subgroup and with-calcium, early menstruation subgroup, in that order. In spine measurements for the same subgroups, amenorrhea duration was the only independent predictor of percentage changes in spine mineral density over time. A significant qualitative interaction was present between calcium supplementation and time to menses resumption. Pairwise comparisons of all subgroups showed significant differences at 1% level for all of them, except between the no-calcium, early menstruation subgroup and the with-calcium, early-menstruation subgroup, where no differences were found. As with the radius, decreasing bone loss was observed across subgroups.
A more rapid weight loss was seen in controls compared with lactating women at 6 months (mean weight change, -4.3 ± 2.8 kg compared with -2.8 ± 2.5 kg, respectively, P .01). At 18 months postpartum, mean values were comparable among all groups.