Interventions for reducing benzodiazepine use in older people: meta-analysis of randomised controlled trials. BJP, 204, . The authors regretfully report that errors were made in some of the data analyses in the above review, as follows.

1. 1. A unit-of-analysis error was made through combining both treatment and control conditions when studies included two or more comparator groups or multiple time points.

2. 2. Awithdrawal study by Habraken et al (1997) was inappropriately included in the meta-analysis. This should not have been included as the authors used a control condition in which participants were required to keep using benzodiazepines rather than stop using them. All other withdrawal studies used comparator conditions in which participants were required to stop using these drugs with an active or non-active control intervention.

3. 3. There were a few small errors in the calculation of the odds ratio and/or standard error for Pit et al (2007) and Avorn et al (1992).

Consequently, all raw data have now been re-checked, all errors corrected and all analyses re-calculated. After re-analysis, the pattern of results and conclusions remain unchanged from those originally reported in the review, with the minor exception of the following.

1. 1. There is no longer any evidence of heterogeneity in effect sizes for withdrawal with psychotherapy at 0.5-3 months. Thus, the conclusion that this type of intervention may not always be effective (v. control conditions) in individual settings no longer stands.

2. 2. Although type of intervention just failed to reach significance in the univariate meta-regression of prescribing interventions, it remained significant in the subgroup analysis. Consequently, the conclusion that the odds of not using benzodiazepines were higher for multifaceted prescribing interventions than for single-faceted ones remains unchanged.

Annotations to the published review and supplementary tables detailing these changes are presented as a data supplement to this correction.