A Meta-Analysis of Prior Corticosteroid Use and Fracture Risk



The relationship between use of corticosteroids and fracture risk was estimated in a meta-analysis of data from seven cohort studies of ∼42,000 men and women. Current and past use of corticosteroids was an important predictor of fracture risk that was independent of prior fracture and BMD.

Introduction: The aims of this study were to validate that corticosteroid use is a significant risk factor for fracture in an international setting and to explore the effects of age and sex on this risk.

Materials and Methods: We studied 42,500 men and women from seven prospectively studied cohorts followed for 176,000 patient-years. The cohorts comprised the EPOS/EVOS study, CaMos, the Rotterdam Study, Dubbo Osteoporosis Epidemiology Study (DOES), and prospective cohorts at Sheffield, Rochester, and Gothenburg. The effect of ever use of corticosteroids, BMD, age, and sex on all fracture, osteoporotic fracture, and hip fracture risk alone was examined using Poisson regression in each cohort and for each sex. The results of the different studies were merged from the weighted β coefficients.

Results: Previous corticosteroid use was associated with a significantly increased risk of any fracture, osteoporotic fracture, and hip fracture when adjusted for BMD. Relative risk of any fracture ranged from 1.98 at the age of 50 years to 1.66 at the age of 85 years. For osteoporotic fracture, the range of relative risk was 2.63-1.71, and for hip fracture 4.42-2.48. The estimate of relative risk was higher at younger ages, but not significantly so. No significant difference in risk was seen between men and women. The risk was marginally and not significantly upwardly adjusted when BMD was excluded from the model. The risk was independent of prior fracture. In the three cohorts that documented current corticosteroid use, BMD was significantly reduced at the femoral neck, but fracture risk was still only partly explained by BMD.

Conclusion: We conclude that prior and current exposure to corticosteroids confers an increased risk of fracture that is of substantial importance beyond that explained by the measurement of BMD. Its identification on an international basis validates the use of this risk factor in case-finding strategies.


THE ADVERSE EFFECTS of corticosteroids on bone fragility have been appreciated for many years. A major mechanism relates to the progressive loss of bone that occurs once corticosteroids are started, but the underlying condition for which they are used may also be a factor. Irrespective of the mechanism, epidemiological data suggest that the risk of hip, forearm, and shoulder fractures is increased ∼2-fold.(1–3) The risk for vertebral fracture may be somewhat higher.(3) The largest and most recent study examined fracture risk in the general practice research data base of the United Kingdom,(4) where ∼250,000 corticosteroid users were compared with age- and sex-matched controls. A dose-dependent effect on fracture risk was noted, and at a >7.5 mg daily dose of prednisolone or its equivalent, the relative risk of vertebral fracture was 5.2, whereas between 5 and 7.5 mg daily, the relative risk was lower (2.6). The dependence of this risk on BMD is not known. Of particular interest was the observation that the increase in fracture risk had a rapid onset when starting corticosteroids and waned rapidly when they were stopped.(3) This rapid onset and offset of effect suggests that risk may in part be independent of BMD.

Corticosteroid use has been recognized as a significant risk factor in current clinical guidelines for the assessment of osteoporosis.(5–10) Under most of these guidelines, patients taking glucocorticoids should be considered for treatment if BMD falls below the threshold for osteoporosis. If, however, the use of corticosteroids is not wholly dependent on BMD, fracture risk assessment should take into account the independent risk associated with glucocorticoids. This is not well established; nor is the possible variation of risk with age and sex. The aim of this study was to examine, in an international setting, the risk of glucocorticoid use in men and women and to determine its dependence on other risk factors.


We studied 42,542 men and women drawn from seven prospective population studies. Details of each of the cohorts are published elsewhere but are summarized briefly below.

The Sheffield cohort was comprised of women ≥75 years of age selected randomly from the population of Sheffield, United Kingdom, and surrounding districts between 1993 and 1999. Approximately 35,000 women, identified from General Practitioner listings, were contacted by letter and invited to attend for the assessment of skeletal status. A total of 5873 women were willing to attend. Of these, 281 women were excluded, and the remainder was randomly allocated to treatment with the bisphosphonate, clodronate, to study its effects on fracture risk. The study is still in progress, and the material for this study was comprised of 2172 women allocated to treatment with placebo.(11) All women had baseline assessment of BMD undertaken at the femoral neck using the Hologic QDR 4500. Outcomes were assessed by six monthly home visits.

The Rotterdam study, begun in 1990, is a prospective cohort study that aimed to examine and follow-up all residents ≥55 years of age living in Ommoord, a district of Rotterdam.(12) By 1993, 7983 residents had been included (response rate, 78%). BMD was assessed at the femoral neck by DXA using a Lunar DPX-L.(13) Fracture follow-up was achieved through an automated link with General Practitioner computer systems and hospital admission data. Fracture data were collected and validated by two independent research physicians. For this analysis, validated fracture follow-up was available for 7774 participants (3065 men), with an average follow-up time of 6 years. Femoral neck BMD was measured in 5778 individuals (2431 men).

The Gothenburg study was comprised of a randomly drawn population cohort of ∼7000 women 50-70 years of age who were followed for up to 8 years.(14) Seventy percent of those contacted agreed to participate. Assessment included a standardized questionnaire that recorded information on risk factors for osteoporosis. Clinical fractures were identified prospectively through the radiology departments servicing the region. BMD was assessed at baseline at the distal forearm using the Osteometer DTX-200.

The Dubbo Osteoporosis Epidemiology Study (DOES) is a population-based study with multiple assessments of skeletal status in men and women ≥60 years of age from Dubbo, Australia.(15) Participation in the study was 56% of the population. Baseline measurements included BMD at the femoral neck, assessed using DXA (GE-Lunar, DPX, and Prodigy). Fractures were identified through radiologists reports from the two centers servicing the region.

The Canadian Multicenter Osteoporosis study (CaMos) is an ongoing prospective age-stratified cohort. The study is documenting the incidence of fractures and risk factors in a random sample of 9424 men and women ≥25 years of age selected by telephone listings. The sampling frame is from nine study centers in seven provinces.(16) Characterization of individuals was by interview. BMD was measured by DXA at the hip with Hologic QDR in seven centers and the Lunar DPX Alpha in two centers. Machines were cross-calibrated using the same European Spine Phantom. BMD was also measured at the lumbar spine, an ultrasound scan was taken at the heel, and a lateral thoracic and lumbar spine X-ray was taken in individuals ≥50 years of age.

The Rochester cohort was recruited from two random population samples stratified by decade of age: one of women who were subsequently followed for up to 20 years(17) and another sample of women and men who were followed for 8 years.(18) BMD of the right femoral neck was measured by dual photon absorptiometry in the first cohort (cross-calibrated to DXA) and by DXA (Hologic QDR 2000) in the second group. Fractures were ascertained by periodic interview combined with review of the in-patient and out-patient medical records of all local care providers.

The EVOS (European Vertebral Osteoporosis Study) was comprised of age- and sex-stratified random samples from 36 centers in 19 European countries.(19) Equal numbers of men and women were drawn in each center within six 5-year age bands (50-54, 55-59, 60-64, 65-69, 70-74, and 75-79 years). A baseline radiograph for vertebral fracture prevalence was undertaken in 15,570 men and women. BMD was measured in 3461 men and women from 13 centers by DXA at the femoral neck using Pencil Beam machines that were cross-calibrated using the European Spine Phantom. This sample provided the framework for the EPOS study (European Prospective Osteoporosis Study), where repeated assessment was undertaken in 29 of the centers.(20,21)

Baseline and outcome variables

Ever use of oral steroids was used to characterize corticosteroid exposure, because the questionnaires in most cohorts did not distinguish between ever and current use of corticosteroids. The duration of use was not analyzed. For the CaMos study, patients who had ever taken corticosteroids for more than 1 month were identified, and at Rochester, patients who had ever taken corticosteroids for more than 6 months. In three cohorts, documentation of current use was available. In Rotterdam, the distinction was made between current use (n = 159) and noncurrent use (n = 7624). In the DUBBO cohort, the distinction was between never use (n = 2068), past but not current use (n = 25), and current use (n = 58). For Sheffield, the distinction was made between never (n = 1963), ever (n = 137), and current use (n = 64). BMD measurements were available in 83% of these individuals.

BMD was assessed by multiple techniques. For the purposes of this analysis, we used BMD assessed at the femoral neck by DXA, with the exception of the Gothenburg cohort, where BMD was assessed by DXA at the distal forearm. Details are provided in Table 1.

Table Table 1.. Details of Cohorts Studied
original image

Fracture ascertainment was undertaken by self-report (Sheffield, EVOS/EPOS) and/or verified from hospital or central databases (Gothenburg, CaMos, DOES, Sheffield, EVOS/EPOS, Rochester, Rotterdam). The EPOS study also included sequential systematic radiography to define incident vertebral deformities, but these were not included for this analysis. Information on all clinical fractures was used for this report. In addition, fractures considered to be caused by osteoporosis were analyzed, and hip fracture alone was considered separately. An osteoporotic fracture was one considered to be caused by osteoporosis by the investigator. For the EVOS/EPOS study, osteoporotic fractures comprised hip, forearm, humeral, or spine fractures. For the CaMos Study, they comprised fractures of the spine, pelvis, ribs, distal forearm, forearm, and hip. In the other cohorts (Sheffield, Rotterdam, Rochester, Gothenburg, DOES), fractures at sites considered to be characteristic for osteoporosis(22) were extracted.

Statistical methods

The risk of fracture was estimated by Poisson regression applied to each cohort and each sex separately. Covariates included time since start of follow-up, current age, use of corticosteroids, current age × corticosteroids, and BMD. We additionally excluded BMD from the model, and in further analyses, included a history of previous fragility fracture and rheumatoid arthritis. The β-coefficients for each sex and each cohort is a linear function of age: βk + βk + 1.age. The estimated value of the β-coefficients and their variance was determined for each age from the age of 50-85 years. The results of each cohort and the two sexes were weighted according to the variance and merged to determine the weighted mean and SD. The risk ratio of those ever treated with corticosteroids versus those not treated was equal to emean.


The total sample studied was 42,542 men and women followed for 176,286 person years. During this time, there were 3682 fractures; 2867 fractures were thought to be related to osteoporosis, including 583 hip fractures. Details by cohort are given in Table 2. BMD measurements were available in 72% of individuals.

Table Table 2.. Details of Patients Studied and Fracture Outcomes
original image

The exposure to the ever use of corticosteroids increased almost linearly with age, from 3.0% at the age of 30, to 3.7% at the age of 50 years, and up to 5.2% at the age of 80 years.

The ever use of corticosteroids was associated with a significantly increased risk of any fracture at all ages compared with those with no history of corticosteroid use (Table 3). This increase in relative risk was not explained by differences in BMD. For example, at the age of 50 years, the relative risk was 1.98 compared with an individual never treated with corticosteroids but with the same BMD, and was 1.99 without BMD in the model. The relative risk ranged from 1.98 at the age of 50 years to 1.66 at the age of 85 years, and the increase in relative risk was most marked at ages younger than 65 years. There was, however, no statistical difference in relative risk by age or between men and women.

Table Table 3.. Risk Ratio of Any Fracture and 95% CIs Associated With Ever Use of Corticosteroids According to Age and Adjusted for BMD
original image

For osteoporotic fractures, risk ratios were higher than those for all fractures combined (Table 3). As in the case for all fractures, relative risk was higher at younger ages, but not significantly so, nor was there a significant difference in relative risk between men and women. There was a small increase in relative risk by the exclusion of BMD from the model, but the quantitative effect was small: relative risk at 50 years was 2.54 and 2.63, respectively.

The highest gradients of risk were observed for hip fracture (Table 3). The risk ratios ranged between 2.13 and 4.42, depending on age. As in the case of osteoporotic fractures, the relative risk was higher in the younger ages, but not significantly so. Also, there was no significant difference between men and women. When BMD was excluded from the model, the risk ratio was lower up to the age of 75 years (Fig. 1).

Figure FIG. 1..

Risk of hip fracture associated with ever use of corticosteroids according to age, with and without BMD.

A summary of the relative risks over all ages is also given in Table 3. Computed from the entire database, the risk relative to the population was calculated from these relative risks and the prevalence of prior exposure to glucocorticoids. Because exposure to corticosteroids in the population was low, downward adjustment of the relative risks was small.

The exclusion of the Gothenburg cohort, in whom BMD was assessed at the forearm, made no difference to the overall conclusions (data not shown).

Current use of corticosteroids was documented in cohorts from Rotterdam, Dubbo, and Sheffield. BMD at the femoral neck was lower in current users of corticosteroids, but the effect was small (Table 4).

Table Table 4.. BMD at the Femoral Neck According to Use of Corticosteroids
original image

In a further model, we examined age, corticosteroid use, and prior fracture. Exposure to corticosteroids was associated with a significantly increased risk of any fracture, an osteoporotic fracture, and a hip fracture (Table 5). Prior fracture was also associated with an independent risk. There was no significant difference in risk between men and women. Rheumatoid arthritis, documented in three cohorts when current corticosteroid use was recorded (CaMoS, DOES, Sheffield), was given as a reason for corticosteroid use in 14%. In a further model, there was an independent fracture risk of corticosteroid use adjusted for rheumatoid arthritis for any fracture (RR = 1.68; 95% CI = 1.40-2.01), for osteoporotic fracture (1.80; 1.47-2.20), and for hip fracture (2.30; 1.50-3.55). Conversely, rheumatoid arthritis was associated with a significant risk of any fracture (1.45; 1.16-1.80), osteoporotic fracture (1.56; 1.20-2.02), and hip fracture (1.95; 1.11-3.42). The risk persisted after adjustment for corticosteroid use in the case of any fracture (1.38; 1.11-1.72) and osteoporotic fracture (1.46; 1.12-1.90), but was of borderline significance for hip fracture (1.76; 0.97-3.19; p = 0.06).

Table Table 5.. Independent Risk Ratio (RR) of Ever Use of Corticosteroids and Prior Fracture According to Type of Fracture and Gender
original image


The principal finding of this study, which was undertaken in large and internationally drawn population-based cohorts, is that prior corticosteroid use confers a substantial increase in fracture risk, as has been shown in a large United Kingdom study based in general practice.(4) This study additionally shows that this risk is largely independent of BMD or a prior fragility fracture. A strength of this study is that the estimate of risk is derived from several studies in an international setting from randomly selected population cohorts. As expected, the risk was higher for osteoporotic fractures than for all fractures, and higher still for hip fracture alone. We did not find any significant differences in the increase in fracture risk between men and women, nor were we able to find a significant difference in risk with age, but the estimated dependence of age is substantial (see Fig. 1) as reported for BMD.(23) Much larger samples would be required to verify this.

The mechanism for the BMD-independent increase in risk could not be determined from this study, but could be caused, at least in part, by the nature of the underlying diseases for which corticosteroids were prescribed. In the cohorts in which this could be examined, rheumatoid arthritis was associated with an independent risk of fracture that persisted when adjusted for corticosteroid use. There was, however, an adverse effect of corticosteroid treatment, even when adjusted for rheumatoid arthritis. Adverse effects of corticosteroids on muscle strength and metabolism may also have increased the liability of falling and impaired protective responses to falling, thereby increasing fracture risk. A further possibility is the effects of these agents on skeletal architecture, which seems to differ from the effects of gonadal deficiency at sites of cancellous bone.(24,25) It is also suggested that glucocorticoids affect osteocyte viability,(26) which might induce alterations in the material properties of bone.

Irrespective of the mechanism, these data indicate that the risk of all fractures is substantially greater in corticosteroid-induced osteoporosis than in postmenopausal osteoporosis for the same level of BMD. These findings, derived from an international setting, have important practical implications for intervention thresholds. Health economic analyses suggest that intervention is cost-effective when treatment is targeted to women with a T-score of −2.5 SD at the femoral neck.(27) Because exposure to corticosteroids confers a risk over and above that provided by BMD, intervention thresholds for BMD can be less stringent, at a T-score of approximately −1.5 SD, and still yield the same cost-effectiveness. Such thinking has now been incorporated into practice guidelines.(8,28)

The prevalence of corticosteroid use is relatively low, and the impact of treating such patients with bone active agents on fracture burden in the general community will not be great. Nevertheless, the strength of the association is high, and the impact on individuals taking corticosteroids will be high.

This study has some limitations that should be mentioned. The greatest problem is the construct of the question concerning corticosteroid use and the documentation on characterization of fracture events. These differed substantially between cohorts. The effect of this heterogeneity is likely to weaken rather than strengthen the association that we found. In addition, the majority of cohorts did not document the current use of oral corticosteroids. Because BMD may recover somewhat when steroids are stopped, this is likely to explain the modest adjustment of the risks by including BMD in the models. As before, this factor might substantially underestimate the risk ratio associated with current corticosteriod use. These limitations indicate the need for further prospective data to more accurately characterize the risk of current use.

We conclude that ever use of corticosteroids confers a substantial risk for future fractures and that this risk is largely independent of BMD. The consistency of the association in an international setting provides the rationale for the use of this risk factor in case-finding strategies. Moreover, patients identified can be targeted for treatment at lower BMD thresholds than individuals of the same age with osteoporosis caused by gonadal deficiency.


This study was supported in part by the NHS R&D National Coordinating Centre for Health Technology Assessment (NCCHTA) acting on behalf of the NHS Executive, United Kingdom. The views and opinions expressed in this paper do not necessarily reflect those of the NCCHTA. We are grateful to Dr Center for help with the DOES Study. We are also grateful to Lilly, Hologic, Roche, IGEA, the Alliance for Better Bone Health, Novartis, and the International Osteoporosis Foundation for their unrestricted support of this work.