Background The treatment of polymyalgia rheumatica (PMR) requires long-term corticosteroid therapy, which is known to induce bone mineral density loss and predispose patients to fracture. Therefore, we hypothesise that patients with PMR are more likely to suffer from osteoporosis (OP), as defined by a T-score threshold of ≤-2.5.

Objectives To ascertain whether steroid use in PMR patients is associated with an increased propensity to OP, and whether other factors increase the propensity.

Methods Study subjects were recruited from patients attending for DEXA scan at a scanner in the North West of England from 2004–2014. Patients with PMR were identified from this cohort for a case-control study: cases were defined as OP on DEXA scan, and controls were patients with T-score >-2.5; the exposure was steroid use (current, previous or ever). Odds ratios within the case-control study were estimated using Stata v13.1, which was used for all statistical analysis. A logistic model was fitted with OP as the dependent variable and age, gender, co-morbidities, medication predisposing to OP and risk factors for OP as the explanatory variables. Following univariate analysis, variables with p<0.2 were fitted into a multivariate logistic model, which was stratified by OP overall, and OP of either the femoral neck or lumbar spine.

Results 649 patients with PMR were identified. There were 108 patients with OP (cases), of whom 95 patients (87.96%) were female, and 541 controls, of whom 388 patients (71.72%) were female (p=4.11E-04). Mean age in cases was 74.59 years (SD 8.78) and 70.18 years (SD 8.49) in controls (p=1.23E-06).

“Current” and “ever” use of steroids were associated with reduced odds of OP: OR 0.62 (95% CI 0.26–1.10, p=0.07) and OR 0.16 (95% CI 0.04–0.58, p=3E-04), respectively. There was no association with “previous” steroid use.

Following multivariate logistic regression, any history of steroid use (OR_adj 0.14 [95% CI 0.04–0.44]), age >70 years (OR_adj 2.37 [95% CI 1.51–3.72]) and female gender (OR_adj 3.05 [95% CI 1.64–5.66]) were significantly associated with OP overall. This model retained significance in lumbar spine OP: steroid use OR_adj 0.24 (95% CI 0.08–0.78), age >70 years OR_adj 1.72 (95% CI 1.08–2.75) and female gender OR_adj 2.83 (95% CI 1.46–5.48). Steroid use was not associated with femoral neck OP, whilst age>70 years (OR_adj 7.01 [95%CI 2.93–16.79]) and female gender (OR_adj 2.68 [95% CI 1.11–6.49]) retained their significant association.

Conclusions This study demonstrates that steroid use is protective of OP in PMR patients, despite the increased risk of OP with chronic steroid use. This may be because clinicians are more diligent in prescribing bone protection in these patients, due to awareness of the increased iatrogenic risk of OP that long-term corticosteroid use predisposes to.

Disclosure of Interest None declared