A Meta-Analysis of Osteoporotic Fracture Risk with Medication Nonadherence

ObjectivesTherapy for osteoporosis reduces the risk of fracture in clinical trials; real-world adherence to therapy is suboptimal and may reduce the effectiveness of intervention. The objective was to assess the fracture risk among patients nonadherent versus adherent to therapy for osteoporosis.MethodsMedline, Embase, and CINAHL were searched for English-language publications of observational studies (January 1998–February 2009). Proceedings from two recent meetings of five relevant conferences were hand searched. Prospective and retrospective observational studies of patients with osteoporosis receiving bisphosphonates, parathyroid hormone, or selective estrogen receptor modulators denosumab were included. Studies were required to consider both fracture risk and adherence (compliance and/or persistence); any definition of adherence/fracture was acceptable. Data were analyzed using pooled comparisons of the odds and hazard ratios of fracture in noncompliance versus compliance and nonpersistence versus persistence. Sensitivity analyses were conducted to determine the effect of clinical heterogeneity on the results.ResultsTwenty-seven citations were identified, the majority of which were retrospective database analyses considering the effect of adherence to bisphosphonate therapy on fracture at any skeletal site. The absolute frequency of fracture ranged from 6% to 38% with noncompliance and from 5% to 19% with nonpersistence (104–159 weeks). Meta-analysis indicates that fracture risk increases by approximately 30% with noncompliance (odds ratio [95% confidence interval] 1.29 [1.22–1.38]; hazard ratio 1.28 [1.18–1.38]) and by 30% to 40% with nonpersistence (odds ratio 1.40 [1.29–1.52]; hazard ratio 1.32 [1.23–1.42]).ConclusionsPoor medication adherence is associated with a significantly increased risk of fracture versus optimal adherence. Improving medication adherence in patients with osteoporosis may lead to a greater reduction in fracture.