Article Text

Download PDFPDF

Effect of oral contraceptives and hormone replacement therapy on bone mineral density in premenopausal and perimenopausal women: a systematic review
  1. S L Liu1,
  2. C M Lebrun2
  1. 1Queen’s University, Kingston, Ontario, Canada
  2. 2Fowler Kennedy Sport Medicine Clinic, University of Western Ontario, London, Ontario, Canada
  1. Correspondence to:
 Dr Lebrun
 Fowler Kennedy Sports Medicine Clinic, University of Western Ontario, London, Ontario N6A 3K7, Canada; clebrun{at}


Seventy five articles on the effect of oral contraceptives and other hormone replacement on bone density in premenopausal and perimenopausal women were reviewed. The evidence was appraised using the Oxford Centre for Evidence-Based Medicine levels of evidence. There is good evidence for a positive effect of oral contraceptives on bone density in perimenopausal women, and fair evidence for a positive effect in “hypothalamic” oligo/amenorrhoeic premenopausal women. There is limited evidence for a positive effect in healthy and anorexic premenopausal women. In hypothalamic oligo/amenorrhoeic women, baseline bone density has been shown to be significantly lower than that in healthy controls, therefore the decision to treat is clinically more important. The ideal formulation(s) and duration of treatment remain to be determined by further longitudinal and prospective randomised controlled trials in larger subject populations.

  • BMD, bone mineral density
  • DXA, dual energy x ray absorptiometry
  • IGF-I, insulin-like growth factor I
  • OC, oral contraceptive
  • RCT, randomised controlled trial
  • oral contraceptive
  • bone mineral density
  • hormone replacement therapy
  • oligomenorrhoea
  • amenorrhoea

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

According to Statistics Canada’s 1996–1997 National Health Population Survey, 18% of Canadian women aged 15–49 use oral contraceptives (OCs).1 In female athletes, OC use is at least as common as in the general population.2 The health benefits of OCs are contraceptive—for example, pregnancy prevention, reduced risk of ectopic pregnancy—and non-contraceptive—for example, cycle control, prevention of ovarian cancer, and reduction in dysmenorrhoea and acne.3 Whereas the pharmacological effects of both oestrogen and progesterone on bone metabolism are widely supported in the literature, the clinical effects of OC use on bone mineral density (BMD) remain unclear. Conflicting views may stem from the many confounding variables that affect BMD, including age, race, genetics, illness, smoking, weight, exercise, diet, and oestrogen status.4 The last four are especially relevant to the female athlete population, in light of the increasing prevalence of the female athlete triad. Compared with the general population, the higher levels of impact loading (in the setting of inadequate hormonal and nutritional status) may increase the female athlete’s risk of fractures and other skeletal injuries. Consequently, the female athlete faces unique concerns with respect to bone health; thus any effects of sustained OC use on BMD are of paramount importance. This review critically examines the literature to determine the effect of OCs and other forms of hormone therapy on BMD in four groups of women: healthy premenopausal, “hypothalamic” oligo/amenorrhoeic, anorexic premenopausal, and perimenopausal.


First described in the early 1990s, the female athlete triad is a clinical syndrome comprising one or more of three specific components: disordered eating, amenorrhoea, and osteoporosis.5 The World Health Organization (WHO) classifies BMD by T score—that is, the number of standard deviations below peak BMD—as follows: <−1 is normal; −1 to −2.5 is osteopenia; >−2.5 is osteoporosis.6 However, the International Society for Clinical Densitometry claims that the WHO classification should not be applied to healthy premenopausal women because it is based on studies in postmenopausal women.7 Further, recent data suggest that the female athlete triad should use osteopenia as a defining criterion rather than osteoporosis, to more accurately reflect the greater prevalence of osteopenia in the female athlete population.8

The female athlete triad is characterised by a negative energy balance, created when energy expenditure exceeds intake. This can be due to inadequate energy intake, excessive exercise, or a combination of both. A negative energy balance invariably leads to disruption of the hypothalamic-pituitary-ovarian axis, ovarian suppression, and various forms of menstrual dysfunction (including shortened luteal phase, oligomenorrhoea, and amenorrhoea). Ultimately, hypo-oestrogenism9 and the nutritional deficits contribute to the development of decreased BMD. Management of the female athlete triad is multidisciplinary, involving doctors, psychologists, and nutritionists. However, the use of OCs to treat decreased BMD found in patients with the female athlete triad is controversial.


Oestrogen plays a critical role in skeletal homoeostasis, with well recognised beneficial effects on bone mass, but the mechanisms by which it acts remain unclear. At the cellular level, oestrogen exerts effects on both osteoclast and osteoblast function, resulting in tonic inhibition of bone turnover and maintenance of the balance between bone resorption and formation.10 It is believed that oestrogen acts directly on bone cells in a receptor mediated manner, as suggested by oestrogen receptor expression in both osteoblasts11 and osteoclasts.12 However, oestrogen also mediates indirect actions on bone through effects on hormones, such as calcitonin and parathyroid hormone, and on cytokines and growth factors.13

Exercise also has an important effect on BMD. It has been proposed that bone is capable of sensing biomechanical strain through an internal “mechanostat”, and adjusts the level of remodelling accordingly to increase bone accretion.14 This pathway is oestrogen dependent, as oestrogen deficiency alters the set point of the mechanostat, thereby impairing detection of biomechanical strain.10 The result is an inadequate level of bone remodelling and accretion. Chronically impaired response to strain and persistent inadequate bone remodelling and accretion potentially contribute to bone loss. Therefore, in physically active hypo-oestrogenic women—that is, women with the female athlete triad—OCs may be beneficial in “resetting” the mechanostat and restoring the appropriate homoeostatic response of bone to exercise.


Study selection

The electronic databases Medline, the Cochrane database of systematic reviews (CDSR), ACP journal club, database of abstracts of reviews of effects (DARE), Cochrane central register of controlled trials (CCTR), cumulative index to nursing and allied health literature (CINAHL), and SPORTDiscus were searched to identify potentially relevant articles up until March 2005. Searches used a combination of medical subject headings and keywords (table 1).

Table 1

 Results from the electronic search strategies

There were 327 hits from Medline, 212 from CINAHL, 30 from CDSR, ACP journal club, DARE, and CCTR (combined), and 17 from SPORTDiscus. Titles and abstracts were scanned to eliminate duplicates and to assess for relevance. Additional references were found through bibliographic searches of all retrieved articles.

Studies were included if they (a) examined effects on BMD, (b) included healthy, “hypothalamic” oligo/amenorrhoeic, or anorexic premenopausal or perimenopausal women, and (c) included oestrogen and/or progesterone replacement therapy—that is, OCs or hormone replacement therapy—as a treatment.

Quality assessment and data extraction

The quality of evidence was appraised using the Oxford Centre for Evidence-Based Medicine levels of evidence,15 based on study design, including: sample size, randomisation, specific inclusion criteria, adequate follow up, and blinding (table 2).

Table 2

 Oxford Centre for Evidence-based Medicine Levels of Evidence

Articles were classified into one of four groups according to study population (healthy premenopausal, “hypothalamic” oligo/amenorrhoeic premenopausal, anorexic premenopausal, perimenopausal), then subdivided by study design (randomised controlled trial (RCT), cohort, cross sectional, case series, case report) and by effect (positive, negative, no effect). Data summarised include OC exposure (formulation, dose) and outcome (measurement of BMD).


Study selection

Seventy five studies were reviewed16–90: 11 RCTs,26–29,62,63,69,74–76,80 28 cohort,16–18,30–38,55–58,64–68,70,77,79,81–84 32 cross sectional,19–25,39–53,59–61,72,73,85–89 three case series,54,78,90 and one case report71 (table 3). Tables 4–14 give descriptions of each study. The results focus on RCTs, as they provide the strongest evidence.

Table 3

 Summary of articles reviewed

Table 4

 Healthy premenopausal women: positive effect of oral contraceptives on bone mineral density

Table 5

 Healthy premenopausal women: no effect of oral contraceptives on bone mineral density

Table 6

 Healthy premenopausal women: negative effect of oral contraceptives on bone mineral density

Table 7

 Oligo/amenorrhoeic premenopausal women: positive effect of oral contraceptives on bone mineral density

Table 8

 Oligo/amenorrhoeic premenopausal women: no effect of oral contraceptives on bone mineral density

Table 9

 Oligo/amenorrhoeic premenopausal women: negative effect of oral contraceptives on bone mineral density

Table 10

 Anorexic premenopausal women: positive effect of oral contraceptives on bone mineral density

Table 11

 Anorexic premenopausal women: no effect of oral contraceptives on bone mineral density

Table 12

 Anorexic premenopausal women: negative effect of oral contraceptives on bone mineral density

Table 13

 Perimenopausal women: positive effect of oral contraceptives on bone mineral density

Table 14

 Perimenopausal women studies: no effect of oral contraceptives on bone mineral density

Data extraction

Healthy premenopausal women

Forty six studies in healthy premenopausal women were reviewed. Ten (three cohort,16–18 seven cross sectional19–25) showed a positive effect, 29 (four RCTs,26–29 nine cohort,30–38 15 cross sectional,39–53 one case series 54) showed no effect, and seven (four cohort,55–58 three cross sectional 59–61) showed a negative effect. All of the RCTs showed no effect on BMD, as measured by either dual energy x ray absorptiometry (DXA)26,27,29 or quantitative computed tomography.28 However, three of the four RCTs also showed a positive effect on bone turnover, as shown by decreased urinary concentrations of the bone resorption markers pyridinoline, deoxypyridinoline,27,29 and cross linked N-telopeptides.28 Further, the RCTs were comparison studies evaluating the effects of different doses/formulations of OCs, but two did not include a control group,26,28 and two used self selected control groups choosing not to receive contraception,27,29 which may have affected the validity of the results. No RCT showed a negative effect. But notably, two cohort studies56,57 and one cross sectional study59 examined the combination of exercise and OCs on BMD. As previously discussed, exercise is believed to have a positive effect on BMD, according to Frost’s mechanostat theory.14 However, Burr et al56 showed that either exercise or OCs alone was associated with a suppression of the normal increase in femoral neck BMD in women 18–31 years old, but the combination of exercise and OCs together had a less suppressive effect than either alone. Similarly, Weaver et al57 suggested that exercise in combination with OCs compromised attainment of peak spinal BMD. Hartard et al59 reported that women with long term exercise and short term OC use had the highest lumbar spine and femoral neck BMD, whereas women with long term exercise and long term OC use had comparable BMD values to women with short term exercise and either long or short term OC use, suggesting that OCs offset the beneficial effects of exercise on BMD.

Oligo/amenorrhoeic premenopausal women

Ten studies on oligo/amenorrhoeic premenopausal women were reviewed. Menstrual irregularities were classified as “hypothalamic” oligo/amenorrhoea—that is, functional menstrual irregularity—or that occurring in the absence of an organic cause (except for two cohort studies which included subjects with primary ovarian failure,66 and from a variety of unspecified causes65). Although these conditions often occur in athletic females, as previously discussed, it is the energy deficit, rather than the activity itself, that leads to the menstrual dysfunction. In the reproductive literature, eumenorrhoea is defined as cycles with intervals of 25–34 days, whereas oligomenorrhoea typically refers to menstrual cycles longer than 35 days. The term amenorrhoea (secondary) connotes a persistent absence of menstrual cycles, commonly for three or more months after the establishment of regular menses. However, confusion often arises when comparing studies, because of the inconsistency of definitions, particularly in earlier research.

Of the 10 studies of OC and other hormone replacement in this population, seven (two RCTs,62,63 five cohort64–68) showed a positive effect, two (one RCT,69 one cohort70) showed no effect, and one case report71 showed a negative effect on BMD. In all studies that compared baseline BMDs with that of healthy controls or age matched reference values, baseline BMDs were significantly lower in the oligo/amenorrhoeic subjects.65–71 Hergenroeder et al62 showed an increase in total body and lumbar spine BMD with OCs, compared with medroxyprogesterone or placebo. Although well designed, this was a small study with only five subjects per treatment group, followed over a 12 month time span. In a somewhat larger study (18–24 subjects per group), Castelo-Branco et al63 examined the effects of two doses (20 or 30 μg) of ethinyl oestradiol-containing OCs on lumbar spine BMD. Both doses increased BMD, whereas the BMD of the control group decreased.63 Conversely, Gibson69 showed that lumbar spine and hip BMD did not significantly change with OCs, calcium carbonate, or control. This trial was conducted over 18 months; however, data from only nine months were reported because of a high dropout rate.69 Further, the OC treated group in this study did show a non-significant increase in BMD after nine months.69 No RCT showed a negative effect of OC treatment on BMD.

Anorexic premenopausal women

Eight studies on premenopausal women with anorexia nervosa were reviewed. Subjects were defined as having anorexia nervosa by either the Diagnostic and statistical manual of mental disorders (DSM)-IIIR or DSM-IV criteria (except for two studies,73,79 in which the criteria used were not explicitly stated). Two cross sectional studies72,73 showed a positive effect, five studies (three RCTs,74–76 one cohort,77 one case series78) showed no effect, and one cohort study79 showed a negative effect. Klibanski et al74 found no overall change in lumbar spine BMD from baseline in either the oestrogen treated or control group. However, the effect of oestrogen on BMD was greatest in patients with the lowest initial body weight, and diminished with increasing patient weight.74 Control patients with a baseline body weight <70% of ideal experienced a significant decrease in BMD, whereas oestrogen treated patients with baseline body weight <70% of ideal did not experience any significant change in BMD, suggesting that, in anorexic women, oestrogen may have a body weight dependent effect on BMD.74 Gordon et al75 showed no effect of either dehydroepiandrosterone or OCs on total hip BMD in anorexic women. In both groups, non-significant increases in lumbar BMD and significantly decreased N-telopeptide concentrations were reported.75 Grinspoon et al76 examined the effect of OCs, recombinant human insulin-like growth factor I (IGF-I), OCs plus IGF-I, or placebo plus IGF-I on BMD at several skeletal sites. No effect of OCs on BMD was detected at any site by factorial analysis, but by four group analysis it was found that, despite being ineffective alone, OCs may augment the effects of IGF-I on BMD in anorexic women.76 No RCT showed a negative effect of OC treatment on BMD.

Perimenopausal women

Eleven studies on perimenopausal women were reviewed. Eight (one RCT,80 four cohort,81–84 three cross sectional85–87) supported a positive effect, whereas three (two cross sectional,88,89 one case series90) showed no effect. Volpe et al80 showed a non-significant increase in spinal BMD in the OC treated group compared with a significant decrease in BMD in the control group. No study showed a negative effect of OC treatment on BMD.


This review critically examines current literature to determine the effect of OCs (and other hormone treatment) on BMD in four groups: healthy premenopausal, “hypothalamic” oligo/amenorrhoeic premenopausal, anorexic premenopausal, and perimenopausal women. Because of the number and diversity of the studies, it was not possible to perform a formal meta-analysis of the results. However, the type of evidence, based on study type and including subject numbers, is summarised below.

There is good evidence supporting a positive effect of OCs on BMD in perimenopausal women. Of 11 studies found, eight (with a combined total of 5854 subjects) showed a positive effect, including one RCT (with 17 subjects). Three studies (of 885 women) did not find any effect. No study showed a negative effect.

There is also fair evidence supporting a positive effect of OCs on BMD in oligo/amenorrhoeic premenopausal women. Of 10 studies, seven (with a total of 379 subjects) showed a positive effect, including two RCTs in a total of 88 women. Although another RCT of 34 women reported no effect, there was still a non-significant trend towards increased BMD in the OC group in this study. In addition, a RCT of 45 women examining the effect of OCs on bone metabolism showed decreased markers of bone resorption in the OC treated group, compared with placebo, supporting a beneficial effect of OCs in this group91 (table 15). Only one case report showed a negative effect.

Table 15

 Biochemical evidence: positive effect of oral contraceptives on bone metabolism

There is limited evidence supporting a positive effect of OCs on BMD in anorexic premenopausal women. Of eight studies, two cross sectional ones of 483 women found a positive effect. Five studies (with 247 total subjects) showed no effect. However, it appears that body weight at initiation of OC treatment may play a role in determining the effect of OCs on BMD.74 Thus, calculation of body weight, as a percentage of ideal, may be an important step in deciding whether to treat anorexic patients with OCs. This evidence may not be helpful in deciding treatment for women with the female athlete triad though, as anorexics are quite distinct in their hormonal condition and state of activity. Sundgot-Borgen & Torstveit92 reported that a higher percentage of Norwegian elite athletes met the criteria for subclinical eating disorders—that is, athletic amenorrhoea or “eating disorders not otherwise specified”—than for clinical eating disorders (anorexia or bulimia nervosa). Women with clinical eating disorders are more sedentary than women with the female athlete triad syndrome, and oestrogen deficiency appears to play less of a role, and IGF-I deficiency more of a role, in decreased BMD in women with clinical eating disorders than in those with the syndrome.76

There is limited evidence supporting a positive effect of OCs on BMD in healthy premenopausal women. Of 46 studies, 29 showed no effect, including all of the RCTs. However, one RCT29 showed a non-significant trend towards increased BMD, and three RCTs27–29 showed decreased concentrations of bone resorption markers in the OC group. Likewise, one RCT93 and two cohort studies94,95 examining the effect of OCs on bone metabolism also suggested similar beneficial results (table 15). A total of seven studies (cohort and cross sectional) of 1361 women suggested a negative effect of OCs on BMD. This is somewhat worrisome, and a variety of potential explanations were given.

Interestingly, there are also data from three studies showing that a combination of exercise and OC use in healthy premenopausal women may have a negative effect on BMD. Postulated reasons for the negative interaction between exercise and OC use are: inadequate bone mineralisation because of nutritional calcium deficiency,56,57 suppression of endogenous pituitary releasing hormone, oestrogen, and progesterone peaks with resultant alteration of the bone mechanostat,59 and the differential effects of different progestins on BMD.59

According to the Oxford Centre for Evidence-Based Medicine levels of evidence,15 the strongest level of evidence (1a) is derived from a systematic review with homogeneity of RCTs. The next best level (1b) is from individual RCTs, with evidence from other study designs carrying less weight. In this review, focus was placed on the RCTs, with supporting evidence from other study types. All of the RCTs included had methodological limitations. In three of the RCTs, subjects were asked whether they desired contraception or not. Those that desired contraception were randomised to one of several treatment groups, and those who did not choose contraception served as the controls, necessitating the concern of self selection bias.27,29,63 Three other studies compared the effect of different types/doses of OCs on BMD, but did not include a non-treatment control group for comparison.26,28,75 Five studies had non-treatment control groups62,69,74,76,80 but only one was placebo controlled.62 Only one study was double blinded,28 but two other studies were single blinded.27,62 Reported reasons for not including a placebo control and for not blinding subjects were: the expected bone loss if a placebo control was used,75 and the expected withdrawal bleeding in subjects who were initially amenorrhoeic taking OCs.76 The duration of the RCTs ranged from nine months76 to three years.31,80 The follow up rate was good, being <80% in only two studies.28,69

The cohort studies included in this review were generally of good quality. In all of them, BMD was measured in the same way in both the OC exposed and non-exposed groups, and confounding variables were identified and accounted for. Further, the groups were similar,17,18,30–32,34,35,55–57,64,67,70,79,81–84 and the follow up rate was >80%16,18,30–32,35,64,67,70,77,79,81,83,84 in most of the studies. However, in several of the studies, follow up was <80%,17,33,34,55–57,65,66,82 and the groups differed in factors potentially contributing to selection bias.33,65,66,77

Many of the studies reviewed were cross sectional.19–25,45–53,59–61,72,73,85–89 In addition, three case series54,78,90 and one case report71 were also reviewed. Evidence from these types of study is weaker, as confounding variables are less likely to have been controlled for, and the results may be more subject to selection and recall bias. Cross sectional studies and case reports are not specifically classified under the Oxford Centre for Evidence-Based Medicine levels of evidence; however, it was felt that they could provide useful evidence that should be included in this review.

A review by Kuohung et al96 evaluated 13 studies examining the effect of low dose OCs—that is, 20–40 μg ethinyl oestradiol—on BMD in women of all ages, including postmenopausal women. Their results suggested that there was fair evidence supporting a favourable effect of OC use on BMD.96 However, in premenopausal and perimenopausal women, there have been mixed results. Previous reviews have attributed these divergent results to differences in study design,4,97,98 inadequate sample sizes,4,97 and heterogeneity in study populations, because of the many confounders affecting BMD,2,4 such as genetics (race), lifestyle (smoking, alcohol, nutrition, exercise), and hormonal (menstrual history, age at menarche, parity, breast feeding) factors. There was a wide diversity in study populations examined among the papers reviewed, but we attempted to define more homogeneous populations by classifying studies into four groups according to health, menstrual status, and reproductive age (premenopausal or perimenopausal). However, an important distinction between reproductive age and skeletal age should be noted. As the average age of menopause ranges from 40 to 58 years,99 a woman classified as “premenopausal” can be anywhere from age 40 and below, and thus may be either skeletally immature or mature. Recker et al16 found that women do not reach skeletal maturity, as reflected by peak bone mass, until around 30 years of age. As skeletal maturity was not an inclusion criterion in any of the studies reviewed, it is unclear whether the subjects had attained peak bone mass or not. This heterogeneity in skeletal maturity may be partly responsible for the variability in results, especially in the cohort and cross sectional studies in healthy premenopausal women, where the evidence seemed to be split between positive effect and no effect. Interestingly, an RCT conducted in skeletally immature cynomolgus monkeys showed that OC treatment actually inhibited net bone accretion and/or growth by reducing bone metabolism,100 whereas no RCT in humans has yet shown a negative effect of OCs on BMD. Thus there is the potential that the effect of OC treatment on BMD may be, in part, dependent on skeletal (rather than reproductive) maturity.

What is already known on this topic

  • To date, there have been mixed results (either positive or no effect) in studies examining the effect of oral contraceptives and other hormone therapy on bone density in healthy premenopausal and perimenopausal women

  • Previous reviews have not taken into account health or menstrual status

Other factors affecting the results include the method and anatomical site of BMD measurement. Among the reviewed studies, there were seven different methods used: 125I photon absorptiometry,19,39 single photon absorptiometry,16,21,30,57,64,85,88x ray/computed tomography,81 quantitative computed tomography,28,41,74 dual photon absorptiometry,16,20,21,30,40,42,51,54,64,82,88,90 single x ray absorptiometry,47,49,50 and DXA.17,18,22–27,29,31–38,43–46,48,52,55–63,65,68,89 There were six different anatomical sites of BMD measurement: lumbar spine,16,17,20–25,27–31,33,35,38,40,42,44,45,48,51,54,57–64,72,74–81,83,86–90 hip (femoral neck, trochanter, Ward’s triangle),22–25,32,33,35,38,40,42,44,45,48,51,56–62,66,67,69–72,75–77,84,86,87 hand,81 heel,37 radius,16,19,21,23,30,45,47,49–54,76,82,85,88,89 and total body.16,23–25,33,35,45,46,48,52,57,62,68,72,73,75,76 This is important because the type of bone varies between anatomical site—for example, vertebral bodies are primarily trabecular, whereas the femur is predominantly cortical,62—and each method allows more accurate measurement of different types of bone—for example, DXA for trabecular, single photon absorptiometry for cortical.96 Furthermore, trabecular bone is more active than cortical; thus the effects of oestrogen may be more readily apparent in trabecular bone.4 Variations in location and method of BMD measurement may also account for previous discordant findings.

The type, dose, and formulation of OC used also differed between the studies reviewed. In two studies, mestranol was used,19,32 whereas in the rest, various doses of ethinyl oestradiol were used (10 μg,65 20 μg,38,54,55,58,63,65,75,77,83,84,90 30 μg,17,26,29,31,36,37,45,63,65,66,68,71,77,81,82,89 35 μg,17,26,36,48,62,65,74,76,77 ⩽50 μg,47,56,57,81 50–100 μg,19,45,64,65,78 or unknown/unspecified doses,16,18,20–25,30,32–34,49–51,59,60,70,72,73,79,80,85–88) and in combination with six different progestins or other hormones (levonorgestrel,28,38,48,56,68,75,77,81 norgestrel,20,77,78 norgestimate,77 norethindrone,17,36,48,62,76,77 gestodene,27,29,82 desogestrel17,26,31,36,54,55,63,66,71,83,84,88 cyproterone acetate,26,64 or drospirenone29,37). A study on postmenopausal women examining the effect of oestrogen dose on bone loss has suggested a dose-response effect: at <15 μg ethinyl oestradiol, net bone loss occurs, and at >25 μg ethinyl oestradiol, net bone gain occurs, but between 15 and 25 μg ethinyl oestradiol, neither bone gain nor loss occurs.101 If this dose-response effect holds true in premenopausal and perimenopausal women, the doses used in some of the studies may have been insufficient to show any effect on BMD. In addition, different progestins vary in their effects on bone.97,102,103 For example, one study showed that a portion of norethindrone is converted into ethinyl oestradiol in the body, resulting in potential bone-sparing properties.104

What this study adds

  • This study reviews the evidence in premenopausal and perimenopausal women, including all study types (randomised controlled trials, as well as all other types)

  • The studies are stratified according to health, menstrual status, and reproductive age, in order to more clearly define effects of oral contraceptives and other hormone therapy on bone mineral density in each group

The definition of OC exposure also differed greatly in the cohort and cross sectional studies. Some used the “non-user” v “user” distinction,18,32–34,41,45,51–53,55–57,64–67,72,73,77,79,81–84 some used “ever” v “never”,20–23,40,43,44,47,60,88 and others further subdivided “ever” users into “current” and “past” users.16,25,30,50 Still others used specific time periods to define OC users—for example, >2 months,39 ⩾6 months and still at the age of 22,35 ⩾2 years,49 ⩾4 years,70 never/2–9 years/>10 years,85 or >3 years if <22 years old or >50% of the time after menarche if >22 years old61—yet these time periods seemed arbitrary, as no reasons for their selection were given. Cobb et al24 have suggested the concept of “cumulative oestrogen exposure” as a quantitative method of defining OC exposure, derived by multiplying the oestrogen dose per month by the total number of months that OCs were used. Use of this quantitative method in the future may make comparison between studies easier.

Clearly, a number of confounding variables influence the effect of OCs on BMD, which may contribute to the divergent results in the literature.


There is good evidence for a positive effect of OCs on BMD in perimenopausal women, and fair evidence in “hypothalamic” oligo/amenorrhoeic premenopausal women. However, there is limited evidence in anorexic and healthy premenopausal women for any positive effect. Further RCTs should be carried out to confirm these results. Ideally, any future studies would also take into account skeletal maturity, as well as reproductive maturity. In addition, studies of women with menstrual dysfunction should use consistent definitions of eumenorrhoea, oligomenorrhoea, and amenorrhoea.

Of significance to the female athlete is the combined effect of OCs and exercise on BMD, but to date there is a lack of evidence in this area. Ultimately, the decision to prescribe OCs to support BMD in the female athlete should be made on an individual basis, taking into account lifestyle and hormonal factors. Current literature does not show any evidence of a negative effect of OC use on BMD in women. OC use may have a favourable effect on BMD, especially in premenopausal women with athletic oligo/amenorrhoea. In these women, baseline BMD has been shown to be significantly lower than that in healthy controls; therefore the decision to treat is clinically more important. Hence, in oligo/amenorrhoeic athletes, the best therapeutic option to support BMD in those desiring contraception, or in those athletes in whom other conservative measures have not resulted in return of normal ovulatory menses in a reasonable amount of time, may be OCs. The “ideal” formulation(s) and duration of treatment remain to be determined by further longitudinal and prospective RCTs.



  • Competing interests: none declared