Background It is widely believed that use of glucocorticoids (GC) increase risk of fracture through reduction in bone quality as well as reduction in bone mineral density (BMD). Data supporting this hypothesis have been provided by a small clinical trial (van Staa). Use of steroids is also included in the FRAX™ risk assessment tool. However to date, observational studies have not demonstrated this effect of steroid on bone in larger cohorts.

Objectives To determine the propensity for fracture in patients currently exposed to glucocorticoids compared to those referred for risk factors that are not steroid related, after adjustment for bone density.

Methods We used data from patients referred for first BMD estimation by Dual X-Ray absorptiometry (DEXA) in the North West of England. Patients details recorded at time of scan included age at scan, height weight and the presence of risk factors for scanning including the presence of fragility fractures. Drug history, BMD in the spine (mean of L1-L4 vertebrae) and the femoral neck are also recorded. Two risk sets were identified, those on current steroids and those referred with no prior exposure to steroids. The prevalence of fractures in each group was compared using a chi-squared test. A logistic model was the fitted to adjust for possible confounders including age at scan, gender, differences in body mass index (BMI) and BMD. The analysis was done for both the femoral neck and the lumbar spine.

Results 20210 subjects were included in the analysis, mean age was 63.2 years (SD 13.0 years), 16.531 (81.7%) were female. Patients who were on steroids were not significantly older than the comparator group 63.1 years vs 62.8 years (p=0.21). The prevalence of fractures was surprisingly significantly lower in the steroid group (13% vs 21.7% p<0.001). The logistic model adjusting for age gave an odds ratio of 0.55 (95%CI 0.50,0.60), when adjusting for age and gender the odds were unchanged OR 0.54 (95%CI 0.50,0.59). When lumbar spine BMD was included in the model the odds were 0.55 (95%CI 0.50,0.60). Hip BMD also did not alter the findings when included together with age at scan and gender OR 0.50 95%CI 0.45,0.55. Adjusting for BMI also did not alter this observation (Or 0.54 95%CI 0.79,0.58 for lumbar spine and 0.53 95%CI 0.48,58 for the femoral neck).

Conclusions The above would indicate that steroids are not associated with excessive fracture risk after adjusting for confounders. Therefore we would conclude that that current corticosteroid use leads to less fractures at the same bone mass level. These results may be subject to unmeasured confounding and need to be interpreted with caution, but for now they do not lend support to the detrimental effect of steroids on bone. Further analysis on the indications for scanning in the comparison and steroid cohort will be performed.

References

1. Van Staa et al, Arthritis and Rheumatism 2003:48;3224-3229.

Disclosure of Interest None declared