Recent studies have suggested that use of depot medroxyprogesterone acetate (DMPA) during the reproductive years may cause or accelerate bone loss. In a cross‐sectional study, Cundy et al noted that DMPA users had a lumbar spine bone mineral density (BMD) that was 7.2% lower than that of matched controls.1 Cromer et al noted a 3.1% reduction in lumbar BMD among eight adolescents who used DMPA for 2 years,2 and Scholes et al observed an adverse relationship between DMPA use and BMD among young women.3 Several other investigations outside of the United States have associated use of DMPA with decreased bone density in the lumbar spine,4,5 as well as several regions of the hip (Ward's triangle, trochanter, femoral neck),5 and the distal forearm.6 However, no prospective study has been published, comparing adult DMPA users with women using no hormonal contraception, which controlled for demographic and behavioral factors believed to affect BMD. Thus, the independent effects of DMPA use on BMD among adult women remain unclear.
Similarly, it is not clear whether use of birth control pills during the reproductive years affects BMD. Some studies have shown that use of oral contraceptives (OC) has a beneficial effect on BMD,7–9 whereas others report no effect.10,11 Findings from these studies are difficult to interpret for several reasons. First, some studies included users of pills containing 50–100 μg of ethinyl estradiol (E2),12,13 formulations that are currently unavailable or rarely prescribed. Others combined users of different pill formulations or failed to specify the pill formulation.7–11 Studies are also limited by small sample sizes2,14 and minimal consideration of behavioral confounders.9,14 Finally, many studies on OCs did not include a control group.9,15,16
The purpose of this study was to assess the independent effect of DMPA and two different types of OC on BMD among women 18–33 years of age compared with those not using hormonal contraception over a 12‐month interval.
MATERIALS AND METHODS
All women recruited between May 16, 1996, and January 20, 1999, who had undergone a baseline bone scan as part of a larger contraceptive study, were eligible to participate. Subjects were between 18 and 33 years of age and white, black, Asian, or Hispanic. Because of the funding source (Department of Defense), all women were required to meet minimal criteria for entry into the Armed Forces (graduated high school or had GED, no felony arrests, within 36% of ideal body weight for height, and free of medical conditions or physical disabilities that would preclude satisfactory completion of military training). Women who were currently pregnant or breastfeeding, had received an injection for contraception during the past 6 months, or were taking birth control pills within the past month, or had a medical contraindication to hormonal contraception were not eligible. Subjects were recruited in person or in response to an advertisement at one of two sites: Wilford Hall Medical Center (WHMC) in San Antonio, Texas, or the Maternal and Child Health Clinic at the University of Texas Medical Branch (UTMB), Galveston, Texas. In addition, 71 women not using hormonal contraception were recruited at UTMB to serve as controls. This study was approved by the Institutional Review Boards of the Department of Defense, WHMC, and UTMB.
After obtaining informed, written consent, each subject was allowed to select the type of contraception that she would use for the duration of the study. Women who preferred injectable contraception received 150 mg of DMPA every 3 months, whereas those who chose oral contraception were randomly assigned to receive either pills containing 0.035 mg of ethinyl E2 and 1.0 mg of norethindrone or pills containing 0.030 mg of ethinyl E2 and 0.15 mg of desogestrel. Pills were referred to by the color names “green” (the desogestrel formulation) or “red” (the norethindrone formulation) and were de‐identified by a research assistant who eliminated package labeling. Randomization was carried out through the use of a random numbers table, which assigned the next eligible patient who chose to use OCs to either the “green” or the “red” formulation. Women who did not wish to use hormonal birth control were recruited to serve as controls. Controls were frequency matched on age and race/ethnicity to the entire sample of hormonal contraceptive users.
At the initial visit, height and weight were measured from which body mass index (BMI) was calculated as weight in kilograms divided by the square of the height in meters. Demographic information and medical history were recorded. Cigarette use was recorded as a dichotomous variable: yes (smoked occasionally, regularly in the past, or regularly now) or no (never smoked, smoked only once or twice in the past). Whether or not the individual engaged in weight‐bearing and/or high‐impact exercise as part of a regular exercise program was recorded as a dichotomous variable (yes/no). To determine calcium consumption, trained research personnel assisted women with recalling and recording all foods and beverages consumed during the 24 hours before the baseline study visit. Appropriate cues and prompts were given to help women with recall, and writing assistance was provided as necessary. Calcium intake was computed using Menu Mizar 3.0 for Windows (Menu Systems, Ruffs Dale, PA).
Bone mineral density of the anterior‐posterior lumbar spine (L1–L4) was determined using dual‐energy x‐ray absorptiometry at baseline and after 12 months of contraceptive use. All baseline scans were performed within 2 months of initiation of hormonal contraception. Follow‐up scans were performed on all women between 10 and 14 months after the baseline scan. In addition, follow‐up scans for hormonal contraceptive users were performed within 2 months of their 1‐year anniversary of initiating contraception. Bone mineral density measurement was performed using a single machine at each study site. Scans at UTMB were performed using a Hologic QDR 1000‐W (Hologic, Waltham, MA) bone densitometer, whereas those at WHMC were obtained with a Lunar DPX (Lunar, Madison, WI). Direct comparison of measured BMD values between the two sites was limited by the use of machines from different manufacturers. However, experts have shown strong correlations across Hologic and Lunar machines when percentage change in BMD is used in longitudinal studies as a measure of lumbar spine BMD changes.17
Short‐term precision was evaluated to examine the reproducibility of BMD outcomes. To estimate precision, 20 subjects at UTMB and 10 at WHMC were randomly selected to undergo two consecutive scans at their 12‐month visit, with an approximate interscan interval of 10–20 minutes. The coefficient of variation was computed as the root‐mean‐square averages of standard deviations (SD) of the repeated measurements.18 In vivo precision was less than 1% for the only technician at WHMC and 1.2% for the primary technician at UTMB.
Continuous variables are expressed as means ± SD. Group comparisons for these variables were conducted using analysis of variance or independent group t tests. Group differences in categorical variables were analyzed using χ2 or Fisher exact test. A two‐sided significance level of .05 was used to determine statistical significance. Separate analyses were conducted at each site on the actual BMD values (g/cm2) to estimate mean changes from baseline and conduct group comparisons using analysis of covariance (ANCOVA). Our primary outcome was change in BMD over the follow‐up interval, which was computed as the mean percent change from baseline using the formula: (follow‐up BMD − baseline BMD)/baseline BMD × 100. An ANCOVA was performed on percent change BMD, controlling for pertinent behavioral (smoking, calcium intake, weight‐bearing exercise) and demographic (race/ethnicity, age, BMI) factors related to BMD. Group mean percent changes in lumbar spine BMD, adjusted for covariates, and Bonferroni adjusted 95% confidence intervals for group comparisons were estimated using ANCOVA. Bonferroni‐adjusted P values from these models are presented for pairwise comparisons of groups.
A post‐hoc power analysis using observed mean differences and SDs demonstrated that this study had greater than 90% power to detect differences between the DMPA and pill groups, and between the norethindrone‐containing pill group and the control group. There was 71% power to detect differences between the two pill groups and between DMPA users and controls, and only 30% power to detect the difference between controls and users of desogestrel‐containing pills. Power calculations were based on the two‐sample t test for unequal n using a two‐sided significance level of .05. The estimate for the SD was the average for the two groups being compared (see Table 1).
A total of 275 women who met all inclusion criteria were enrolled in the study. Of these, 96 chose injectable contraception and were administered DMPA, whereas 179 women elected to initiate oral contraception and were randomly assigned to receive either norethindrone‐containing (n = 87) or desogestrel‐containing (n = 92) pills. Thirty‐nine percent (107 of 275) of women discontinued their hormonal method before their 12‐month visit. Of the 168 women who continued their method, 31 failed to undergo a bone scan at 12 months because of scheduling conflicts, whereas 37 additional women obtained a scan, but failed to do so within the required window (±31 days of their 12‐month anniversary date). Thus, a total of 100 users of hormonal contraception were available for analysis. Final analyses were conducted on 96 users of hormonal contraception (four women were eliminated from analysis as statistical outliers with 12‐month BMD changes greater than three SD from the sample mean). In addition, 59 of the 71 controls (83%) received a 12‐month bone scan within the required window and were included as a comparison group in the final analyses. Women who dropped out of the study did not differ from the 155 women included in the final analyses on contraceptive method (P = .71), smoking (P = .73), exercise, (P = .99), calcium intake (P = .98), race/ethnicity (P = .08), or BMI (P = .68). Those who dropped out were, on average, 1.6 years younger than those who remained in the study (P < .001) and were more likely to have been recruited at WHMC (P < .001).
There were no significant differences between the four contraceptive groups in their characteristics (Table 2), although women who selected an oral method of birth control were significantly less likely to report smoking cigarettes than those who used DMPA or nonhormonal methods (P = .02).
Using the actual BMD values (g/cm2), separate ANCOVAs were performed on BMD data obtained at UTMB and WHMC because of the use of DXA machines from different manufacturers. Both sites revealed similar group patterns in BMD change after 1 year (Table 3). Women using DMPA, on average, experienced a loss in bone density. Using data from UTMB only, the results of the ANCOVA revealed that the loss in BMD among DMPA users was significantly greater than that observed among controls (P = .03). Women using OCs experienced slight gains or no change in BMD at both sites. Bone mineral density changes among users of desogestrel‐containing pills compared with users of norethindrone‐containing pills did not differ at UTMB (P = .39) or WHMC (P = .95).
Initial analyses were conducted to evaluate potential interaction effects in our ANCOVA models. Separate models including all main effects and the interactions (method × site), (method × smoking status), and (method × age) were tested. None of the interaction terms were significant (P = .94, P = .11, and P = .94, respectively). Analysis of covariance was performed on percent change BMD, with method, race/ethnicity, site, smoking, and exercise status included as fixed factors, and calcium intake, BMI, and age included as continuous covariates. All covariates remained in the model regardless of their P value. The model evidenced a statistically significant effect only for contraceptive method [F(3,142) = 11.43, P < .001]. Users of DMPA experienced an average loss of 2.7% in BMD over the 12‐month interval compared with controls who sustained a 0.37% mean loss (P = .01, Table 4). On average, users of both types of birth control pills demonstrated a gain, with users of norethindrone‐containing pills experiencing a 2.33% gain, and users of desogestrel‐containing pills demonstrating a 0.33% gain in BMD. Table 5 presents 95% confidence intervals for pairwise differences in percent change in BMD between the contraceptive groups. The gain among users of pills was significantly different from controls for users of norethindrone‐containing pills (P = .01), but not for users of desogestrel‐containing pills (P = .99). Mean changes in BMD significantly differed between DMPA users and women who used either type of birth control pill (both P < .002).
We observed that use of DMPA for 12 months has an adverse effect on BMD, compared with OCs or nonhormonal methods. On average, women who used DMPA experienced approximately a 2.7% loss in BMD compared with a 0.37% loss in those not using hormonal contraception and small gains among OC users of 0.33% and 2.33%, respectively, for users of desogestrel‐ and norethindrone‐containing pills. The mechanism of the effect of DMPA on BMD is unknown. However, studies have shown that DMPA users have significantly lower serum E2 levels than users of nonhormonal contraception.15,19 In the absence of adequate levels of estrogen, (postmenopausal, anorexia nervosa, GnRH therapy), bone resorption outstrips formation, and bone mass decreases.20 Thus, it seems logical that the bone loss associated with DMPA use is caused by hypoestrogenism and a subsequent increase in bone turnover. Alternatively, the decrease in BMD could be related to the exogenous glucocorticoid‐like effects of DMPA.21
This study compares the effects of two different types of OCs on BMD. Previously, it was not possible to compare the effects of the type of pill on BMD because earlier studies did not specify the type, merged different types, or included only one pill formulation. In contrast, we randomized OC users to a norethindrone‐containing or desogestrel‐containing pill. A small gain in BMD was noted among users of both types of pills, which was significantly different from controls for norethindrone pills. We did not detect a difference between users of desogestrel pills and controls, which may have been because of insufficient power. In addition, we did not detect a difference between these two pills in their effect on BMD. The lack of a significant difference between the two pill groups in this study suggests that no firm conclusions should be drawn regarding a potential difference between these two types of pills until more data are available.
Because of the difficulty in randomizing women to a particular contraceptive method, we did allow all subjects to select whether they would use injectable, oral, or a nonhormonal method. In the absence of a randomized design, we carefully inspected the data for evidence of bias (because of self‐selection) in the composition of the contraceptive groups. Specifically, we examined group‐based differences in age, race/ethnicity, weight‐bearing exercise habits, calcium intake, cigarette smoking, and BMI. These analyses were conducted univariately using appropriate statistical tests, and multivariably by conducting analyses of variance that included the relevant interaction terms. Although only smoking status significantly differed by method, we included all behavioral and demographic factors as covariates in our final analysis. Our inclusion of multiple covariates and our choice to apply a Bonferroni correction to evaluate pairwise comparisons between contraceptive groups represents a conservative strategy toward detecting method‐related effects on BMD. The contraceptive‐related differences we observed in BMD remained significant after controlling for behavioral and demographic correlates such as smoking and age, which have been influential in the broader literature on bone density.
Calcium intake data demonstrated that regardless of the type of contraception they used, few women ingested an adequate amount of calcium. In fact, the daily mean calcium intake among all women was 565 mg (SD = 379). Only 7% of women 18–24 years old ingested the recommended daily amount (1200 mg) for women aged 11–24 years, whereas only 12% of women 25–33 years of age met recommendations for their age group (1000 mg/day).22 This is particularly disconcerting considering that most women in our study had healthy habits — 61% reported exercising three times per week or more, and none were obese. Thus, it appears that women of reproductive age may not ingest adequate amounts of calcium, even if they engage in other healthy behaviors. This is especially of concern if their contraception places them at risk of bone loss.
This study has limitations that bear mentioning. Ideally, all women would have been scanned on a single bone densitometer. We minimized this shortcoming by first inspecting the pattern of data within each site as differences between measured BMD and as percent change BMD over the study interval. Only upon observation of consistent patterns at UTMB and WHMC did we merge the data from the two sites and report overall percent change in BMD. We also note that most women in this study were white, and all were within 36% of their ideal body weight and had obtained a high school degree. As most women in the United States do not fit this profile, our findings are not readily applicable to the general population. Furthermore, a 39% method discontinuation rate was observed within the 12‐month study period. This discontinuation rate is similar, and in many cases, lower, than that found in other published studies. For example, among users of various OC formulations, 12‐month discontinuation rates have ranged from nearly 36–66%,23,24 and among users of injectable contraception, 1‐year discontinuation rates have ranged from 48–77%.24–28 Acknowledging that high discontinuation rates temper the conclusions that can be drawn in contraceptive studies, we are careful to apply the findings of this research only to women who continued their contraceptive method for a 1‐year period. Finally, we collected data over 12 months only. Additional studies are needed to determine the effects of these methods on BMD compared with controls over longer durations of use.
1. Cundy T, Cornish J, Roberts H, Elder H, Reid IR. Spinal bone density in women using depot medroxyprogesterone contraception. Obstet Gynecol 1998;92:569–73.
2. Cromer BA, McArdle Blair J, Mahan JD, Zibners L, Naumovski Z. A prospective comparison of bone density in adolescent girls receiving depot medroxyprogesterone acetate (Depo-Provera), levonorgestrel (Norplant), or oral contraceptives. J Pediatr 1996;129:671–6.
3. Scholes D, Lacroix AZ, Ott SM, Ichikawa LE, Barlow WE. Bone mineral density in women using depot medroxyprogesterone acetate for contraception. Obstet Gynecol 1999;93:233–8.
4. Paiva LC, Pinto-Neto AM, Faundes A. Bone density among long-term users of medroxyprogesterone acetate as a contraceptive. Contraception 1998;58:351–5.
5. Tang OS, Tang G, Yip P, Li B, Fan S. Long-term depot-medroxyprogesterone acetate and bone mineral density. Contraception 1999;59:25–9.
6. Bahamondes L, Perrotti M, Castro S, Faundes D, Petta C, Bedone A. Forearm bone density in users of Depo-Provera as a contraceptive method. Fertil Steril 1999;71:849–52.
7. Lindsay R, Tohme J, Kanders B. The effect of oral contraceptive use on vertebral bone mass in pre- and post-menopausal women. Contraception 1986;34:333–40.
8. Kleerekoper M, Brienza RS, Schultz LR, Johnson CC. Oral contraceptive use may protect against low bone mass. Arch Intern Med 1991;151:1971–6.
9. Recker RR, Davies KM, Hinders SM, Heaney RP, Stegman MR, Kimmel DB. Bone gain in young adult women. JAMA 1992;268:2403–8.
10. Mazess RB, Barden HS. Bone density in premenopausal women: Effects of age, dietary intake, physical activity, smoking, and birth-control pills. Am J Clin Nutr 1991;53:132–42.
11. Hreshchyshyn MM, Hopkins A, Zylstra S, Anbar M. Associations of parity, breast-feeding, and birth control pills with lumbar spine and femoral neck bone densities. Am J Obstet Gynecol 1988;159:318–22.
12. Goldsmith NF, Johnston JO. Bone mineral: Effects of oral contraceptives, pregnancy, and lactation. J Bone Joint Surg Am 1975;57:657–8.
13. Lloyd T, Buchanan JR, Ursino GR, Myers C, Woodward G, Halbert DR. Long-term oral contraceptive use does not affect trabecular bone density. Am J Obstet Gynecol 1989;160:402–4.
14. Ricci S, Mango D, Manna P, Tibollo FG, Serra GB, Granata Q. The use of low-dose oral contraceptives for one year does not affect vertebral bone density. In: Christiansen C, Overgaard K, eds. Osteoporosis 1990. Copenhagen: Osteopress, 1990:1188–9.
15. Naessen T, Olsson SE, Gudmundson J. Differential effects on bone density of progestogen-only methods for contraception in premenopausal women. Contraception 1995;52:35–9.
16. Castelo-Branco C, Marti′nez de Osaba MJ, Pons F, Vanrell JA. Effects on bone mass of two oral contraceptives containing ethinyl-estradiol and cyproterone acetate or desogestrel: Results of a 2-year follow-up. Eur J Contracept Reprod Health Care 1998;3:79–84.
17. Pocock NA, Noakes KA, Griffiths M, Bhalerao N, Sambrook PN, Eisman JA, et al. A comparison of longitudinal measurements in the spine and proximal femur using Lunar and Hologic instruments. J Bone Miner Res 1997;12:2113–8.
18. Gluer CC, Blake G, Lu Y, Blunt BA, Jergas M, Genant HK. Accurate assessment of precision errors: How to measure the reproducibility of bone densitometry techniques. Osteoporosis Int 1995;5:262–70.
19. Taneepanichskul S, Intaraprasert S, Theppisai U, Chaturachinda K. Bone mineral density in long-term depot medroxyprogesterone acetate acceptors. Contraception 1997;56:1–3.
20. Raisz LG. Physiology and pathophysiology of bone remodeling. Clin Chem 1999;45:1353–8.
21. Aedo AR, Landgren BM, Diczfalusy E. Studies on ovarian and adrenal steroids at different phases of the menstrual cycle. III. Steroid and lutropin levels before and after the administration of a single contraceptive dose of depot medroxyprogesterone acetate (DMPA). Contraception 1981;24:117–35.
23. Kirkman RJ, Bromham DR, O'Connor TP, Sahota JE. Prospective multicentre study comparing levonorgestrel implants with a combined contraceptive pill: Final results. Br J Fam Plann 1999;25:36–40.
24. Rosser M, Stone J, Graves W, Hatcher R. Depo-provera: An excellent contraceptive for those who continue to use it. Prim Care Update Ob Gyn 1998;5:172.
25. Hubacher D, Goco N, Gonzalez B, Taylor D. Factors affecting continuation rates of DMPA. Contraception 1999;60:345–51.
26. Davidson AR, Kalmuss D, Cushman LF, Rombero D, Heartwell S, Rulin M. Injectable contraceptive discontinuation and subsequent unintended pregnancy among low-income women. Am J Public Health 1997;87:1532–4.
27. Sangi-Haghpeykar H, Poindexter AN, Bateman L, Ditmore JR. Experiences of injectable contraceptive users in an urban setting. Obstet Gynecol 1996;88:227–33.
© 2001 by The American College of Obstetricians and Gynecologists.
28. Westfall JM, Main DS, Barnard L. Continuation rates among injectable contraceptive users. Fam Plann Perspect 1996;28:275–7.