Statistical Analysis for Comparing Antifracture Efficacy among Antiresorptive Agents

Abstract

Objective: To develop a method that estimates and compares the antifracture efficacy [relative risk reduction (RRR) and number needed to treat (NNT)] of various antiresorptive agents when fracture trials are not available. Design, Setting and Patients: Statistical analysis using efficacy data from clinical trials of postmenopausal women with osteoporosis. Interventions: Oral alendronate 5mg, oral alfacalcidol lμg, or intranasal calcitonin 200IU daily. Methods: We used the effects on bone mineral density (BMD) in head-to-head trials together with the relationship between BMD changes and fracture risk reductions reported in published meta-analyses of antifracture trials to predict and compare RRR and NNT. Results: The predicted RRR was similar to the published values in large clinical trials, thus validating the method. With an incidence of 12% over 3 years in untreated women with pre-existing vertebral fractures (based on a large epidemiological study), the predicted NNT to prevent a new vertebral fracture was 17 for alendronate and 29 for alfacalcidol; similar results were observed for alendronate versus calcitonin. For nonvertebral fractures, the NNT was three to four times lower for alendronate, compared with alfacalcidol and calcitonin. Conclusions: There is currently insufficient evidence of antifracture efficacy from randomised trials to recommend the use of alfacalcidol or calcitonin for treating osteoporosis. Furthermore, our analyses indicate that alendronate is substantially more effective for reducing vertebral and nonvertebral fracture risk than alfacalcidol or calcitonin.