Association of fracture risk with benzodiazepine among adults in South Korea.

This study examined the association between fracture and benzodiazepine (BZD) prescription in Korean adults using case-crossover (CCO) and self-controlled case-series (SCCS) designs, which have the advantage to control confounding bias, such as individual characteristics. Patients with fracture were defined as patients who visited the emergency room and orthopedics department with the ICD-10 diagnosis code for fracture. Fractures due to motor vehicle accidents and stroke were excluded. Whereas the CCO design presented odds ratio (OR) using a conditional logistic regression model, SCCS design showed incidence rate ratio (IRR) using a conditional Poisson regression model. The concomitant drugs that can affect the fracture were adjusted. Sensitivity analysis and subgroup analysis by age (elderly vs. nonelderly), action mechanism (short-acting vs. long-acting), and prescription duration (short-term user vs. long-term user) were conducted. The adjusted OR (AOR) for control period I (prior to 90 days from case) was 1.39 (95% CI=1.25-1.54) for all BZD prescriptions. The adjusted ORs for other control periods showed similar trends. The adjusted IRRs (AIRR) during the first 4 weeks, 4-8 weeks, 8-12 weeks, and 12-16 weeks from new BZD use were 1.46 (95% CI=1.28-1.66), 1.23 (95% CI=1.01-1.49), 1.09 (95% CI=0.86-1.37), and 1.38 (95% CI=1.07-1.77), respectively. Regardless of age group, action mechanism, or prescription duration, fracture risk was higher during case period than control. The risk for fracture was higher in both elderly and non-elderly people with BZD prescription than in those without BZD prescription. Careful monitoring for people who start BZD treatment and further research in the non-elderly is required.