Psychotropic medications and the risk of fracture: a meta-analysis.

Background: Older adults throughout the developed world are at significant risk of osteoporotic fractures. Many studies have examined the relationship between the use of psychotropic medications and the risk of fractures, but these studies have reported conflicting results.

Purpose: To resolve discrepancies, we carried out a meta-analysis to assess the risk of fractures among users of several classes of psychotropic drugs.

Data Sources: We retrieved studies published in any language by systematically searching MEDLINE, LILACS, EMBASE and ISI Proceedings databases and by manually examining the bibliographies of the articles retrieved electronically as well as those of recent reviews.

Study Selection: We included 98 cohort and case-control studies, published in 46 different articles, that reported relative risk (RR) estimates and confidence intervals (CIs) or sufficient data to calculate these values.

Data Synthesis: Study-specific RRs were weighted by the inverse of their variance to obtain fixed- and random effects pooled estimates. The random effects RR of fractures was 1.34 (95% CI 1.24, 1.45) for benzodiazepines (23 studies), 1.60 (95% CI 1.38, 1.86) for antidepressants (16 studies), 1.54 (95% CI 1.24, 1.93) for non-barbiturate antiepileptic drugs (13 studies), 2.17 (95% CI 1.35, 3.50) for barbiturate antiepileptic drugs (five studies), 1.59 (1.27, 1.98) for antipsychotics (12 studies), 1.15 (95% CI 0.94, 1.39) for hypnotics (13 studies) and 1.38 (95% CI 1.15, 1.66) for opioids (six studies). For non-specified psychotropic drugs (10 studies), the pooled RR was 1.48 (95% CI 1.41, 1.59).

Limitations: Main concerns were the potential for residual confounding and for publication bias.

Conclusion: Globally, the increase in the risk of fractures among psychotropic drug users is moderate. Further research is needed, especially to examine high-risk populations and newer medications. Future studies should be prospective and emphasise control of confounding bias.