Abstract
CONTEXT: Three recent nested case-control studies conducted in automated databases suggest that users of 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (statins) have a risk of hip and other osteoporotic fractures half that of non-users of any lipid-lowering drug. However, this comparison may be biased by unmeasured factors associated with treated hyperlipidemias. OBJECTIVE: To compare the risk of hip fracture among users of statins and other lipid-lowering agents, which is less susceptible to bias than the comparisons performed in the previous studies. DESIGN AND SETTING: Retrospective cohort study conducted in the Tennessee Medicaid program between 1 January 1989 through 31 December 1998. SUBJECTS: New users of all lipid-lowering drugs and randomly selected non-user controls who at baseline were at least 50 years of age and did not have life threatening illness, nursing home residence, or diagnosed dementia or osteoporosis. There were 12506 persons with new use of statins, 4798 with new use of other lipid lowering drugs, and 17280 non-user controls. MAIN OUTCOME MEASURE: Fracture of the proximal femur (hip), excluding pathological fractures or those resulting from severe trauma. RESULTS: During 66690 person years of follow up, there were 186 hip fractures (2.8 per 1000). Relative to non-users, the adjusted incidence rate ratios (95% confidence interval) were 0.62 (0.45 to 0.85) for statin users and 0.44 (0.26 to 0.95) for other lipid-lowering drugs. When compared directly with the other drugs, the adjusted incidence rate ratio for statins was 1.42 (0.83-2.43). CONCLUSION: These data provide evidence that the previously observed protective effect of statins may be explained by unmeasured confounding factors.