Context

Inhaled corticosteroids (ICS) are recommended for the management of moderate to severe chronic obstructive pulmonary disease (COPD).1 2 Previous investigations on the adverse effects of ICS on bone health in individuals with COPD have yielded conflicting results. Some trials3 have found that ICS can worsen bone mineral density (BMD) but not fracture risk, whereas other studies4 5 have found no such effect. Meta-analyses have also yielded contradictory findings.6 7 Understanding the strength and characteristics of this association is important as fractures can have serious consequences in patients with COPD.

Findings

Sixteen RCTs with a total of 17 513 participants were analysed, while seven observational studies with a total of 68 937 individuals were included. For the RCTs, 180 events occurred in the ICS group compared with 141 events in the comparator group (overall OR for fracture with ICS 1.27, 95% CI 1.01 to 1.58). There was no statistically different risk in fractures comparing the ICS/LABA group to LABA alone or ICS alone versus placebo. Sensitivity analysis including studies with longer-term follow-up (156 weeks' duration) showed no increased risk of fractures with ICS. For the observational studies, meta-analytic estimates showed increased odds of fracture comparing current or ever ICS users to non-users (OR 1.21, 95% CI 1.12 to 1.32). A dose–response meta-regression showed that with each 500 µg increase in beclomethasone dose equivalent of ICS, there was an increased fracture risk (OR 1.09, 95% CI 1.06 to 1.12).

Commentary

Although previous meta-analyses of similar topics were inconclusive, this study seems to confirm that the use of ICS in COPD is associated with an increased fracture risk. This analysis adds to previous publications by including large and small observational trials and RCTs from published and unpublished reports. Despite these methods, the effect size, albeit statistically significant, is of unclear clinical significance.

There are important considerations when attempting to understand the results of these analyses. First, as the authors point out, one very large study [Towards a Revolution in COPD Health5 (TORCH)] accounts for a substantial proportion of the sample size that ultimately contributes to the RCT analysis. The proportion of individuals experiencing the outcome in TORCH (4.2%) is much higher than nearly all other studies included in the analysis, thus potentially driving the observed association. Second, unpublished data have not undergone rigorous peer review, and thus may lead to ascertainment bias. Finally, a wide range of ICS were incorporated into the observational meta-analysis, including triamcinolone which is not approved by the Food and Drug Administration for COPD, whereas only studies assessing the effects of budesonide and fluticasone were included in the RCT analysis. Interestingly, triamcinolone has a clearer association with BMD loss,3 potentially explaining the more substantial association seen in the observational studies.

Ultimately, the potential harm associated with the use of ICS must be balanced with the potential clinical benefit in exacerbation reduction, forced expiratory volume decline and mortality. While bone loss in patients with COPD can have serious ramifications, the magnitude of benefit and risk must be individualised to each patient.