Hormonal contraceptives have only small, reversible effects on bone density, according to a multicenter study conducted in Africa, Asia and Latin America.1 Compared with women who do not use hormonal methods, women who use combined oral contraceptives experience a significant increase in bone density, while those who use the injectable depot medroxyprogesterone acetate (DMPA) or the levonorgestrel implant experience a significant decrease. These changes disappear after the first 2-3 years of current use and appear to be clinically insignificant.
A total of 2,545 women aged 30-34 were enrolled between 1994 and 1997 at family planning clinics in Bangladesh, Brazil, China, Egypt, Mexico, Thailand and Zimbabwe. During interviews, each woman provided information on her social and demographic characteristics, obstetric and contraceptive history, dietary habits, and height and weight. Seventy-one women were excluded because they were currently pregnant or lactating (or had been within the prior six months), had undergone hysterectomy or oophorectomy, or reported diseases or drug intake that could influence calcium metabolism.
The remaining 2,474 women were categorized as users of hormonal contraceptives if they had used hormonal contraceptives for at least two years over their lifetime and as never-users if their lifetime exposure amounted to no more than six months. Study participants who had used more than one hormonal contraceptive were assigned to the method most recently used for at least two years. Thus, 33% were classified as pill users, 14% as DMPA users and 25% as implant users, while 28% were considered never-users.
To determine bone density, the radius was measured near the wrist and the ulna was measured at midshaft. The average bone density at both sites differed significantly by country, with women in Zimbabwe having the highest values at both the radius and the ulna and women in Bangladesh having the lowest values. As a group, Asian centers (those in Ban-gladesh, China and Thailand) had the lowest bone density readings of any region.
When the data were adjusted for study center, an analysis of covariance showed that bone density at both the radius and the ulna were associated with body mass index, age, total months of lactation, total months since last lactation and the occupation of the woman's partner. Parity and coffee consumption were related to bone density at the radius only. Bone density at both sites was associated with hormonal method use: Values were highest for women who did not use hormonal contraceptives, followed by those who relied on the pill, those who used the implant and those who relied on the injectable.
Most of these associations persisted after adjustment for the other variables in the analysis. Bone density values for women who relied on the pill (both all users and exclusive users) were not significantly different from those for never-users. Women who used DMPA had lower bone density; only the difference at the radius was significant for all users, while the differences at both the radius and the ulna were significant for exclusive users. Women whose only hormonal method had been the levonorgestrel implant had significantly lower bone density at the ulna. The decrease in bone density for women who used DMPA or levonorgestrel was approximately 0.01 g/cm2, a relatively small decline given that a decrease of one standard deviation below the measurements in never-users equals approximately 0.05 g/cm2.
Further analysis of bone density for exclusive users and never-users revealed a pattern of significant change for all three methods during short-term current use. Women who had been using the pill for only 2-3 years had higher values at both bone sites than did never-users, while women who had been relying on either DMPA or the levonorgestrel implant for a similar period had significantly lower values than never-users at both bone sites. No other differences in bone density were found for any of the contraceptive methods for longer durations of current or past use, which suggests that such changes are reversible over time.
According to the researchers, the study's limitations include its cross-sectional design, the limited age range of the participants and the lack of bone density measurements at the femoral neck and spine. They note that the substantial variations across countries remain unexplained because several potentially important variables were not explored, such as calcium intake, dietary habits in childhood and exercise. The women displayed minimal variation in smoking and alcohol consumption, so the potential influence of these variables on bone density could not be determined.
The investigators point out that the small decreases in bone density found in this study among women using the implant and the injectable are not considered abnormal and fall short of the definition of low bone mass--and well short of that for osteoporosis. They conclude that the changes in bone density observed soon after the beginning of hormonal contraceptive use appear to be temporary and clinically insignificant.--L. Ninger
1. Petitti DB et al. for the WHO Study of Hormonal Contraception and Bone Health, Steroid hormone contraception and bone density: a cross-sectional study in an international population, Obstetrics and Gynecology, 2000, 95(5):736-744.