To the Editor.
—The conclusion from the meta-analysis by Dr Mayo-Smith and colleagues1 that benzodiazepines are more effective than placebo in preventing delirium among patients withdrawing from alcohol is not convincingly supported by the data. First, chlordiazepoxide was the only benzodiazepine studied for its effect on delerium in the 4 placebo-controlled trials included in the meta-analysis. Although it is tempting to extrapolate from chlordiazepoxide to other benzodiazepines, the differences noted between different drugs in this class with respect to other aspects of alcohol withdrawal therapy (eg, oversedation and breakthrough of withdrawal symptoms) caution that an assumption of equivalence for delirium control may need empirical validation.Second, the appropriateness of pooling the 4 studies for meta-analysis seems questionable in view of the studies' disparate findings. Three of the studies essentially suggested equivalence between placebo and chlordiazepoxide with respect to the proportion of patients with delirium (combined: 3/56 [placebo] vs 2/69