Cognitive effects of long-term benzodiazepine use: a meta-analysis.

INTRODUCTION: While benzodiazepines are the most widely used psychotropic drugs, there are relatively few studies that have examined deficits in cognitive functioning after long-term use. The literature that is available is difficult to interpret due to conflicting results as well as a variety of methodological flaws.
OBJECTIVE: To systematically evaluate and integrate the available research findings to determine the effect of long-term benzodiazepine use on cognitive functioning using meta-analytical techniques.
METHODS: Thirteen research studies that employed neuropsychological tests to evaluate cognitive performance after long-term use of benzodiazepine medication met inclusion criteria. The neuropsychological tests employed in these 13 studies were each categorised as measuring one of 12 cognitive domains. Separate effect sizes were calculated for each of the 12 cognitive categories. Each study was only allowed to contribute one effect size to each cognitive category by averaging together the effect sizes from the same study if more than one type of test was used to measure a particular category. This strategy resulted in equal weight being given to each study per category, regardless of the number of tests in that category.
RESULTS: The overall mean number of patients who were benzodiazepine users was 33.5 (SD +/- 28.9) and the mean number of controls was 27.9 (SD +/- 19.6). The duration of benzodiazepine use ranged from 1 to 34 (mean 9.9) years. Long-term benzodiazepine users were consistently more impaired than controls across all cognitive categories examined, with effect sizes ranging in magnitude from -1.30 to -0.42. The mean weighted effect size was -0.74 (SD +/- 0.25). None of the effect sizes had 95% CIs that spanned zero and, therefore, all of these effects were significant and different to zero.
CONCLUSION: Moderate-to-large weighted effect sizes were found for all cognitive domains suggesting that long-term benzodiazepine users were significantly impaired, compared with controls, in all of the areas that were assessed. However, this study has several limitations, one being that it includes a relatively small number of studies. Further studies need to be conducted; ideally, well designed, controlled studies that thoroughly investigate certain areas of cognitive functioning and present data in such a way so as to be amenable to inclusion in a meta-analysis. Incorporating the information from these studies into a larger meta-analysis would allow for a more thorough and statistically sound investigation of the effects of moderator variables. The observation that long-term benzodiazepine use leads to a generalised effect on cognition has numerous implications for the informed and responsible prescription of these drugs.