Unpublished rating scales: a major source of bias in randomised controlled trials of treatments for schizophrenia.

A recent review suggested an association between using unpublished scales in clinical trials and finding significant results. To determine whether such an association existed in schizophrenia trials. Three hundred trials were randomly selected from the Cochrane Schizophrenia Group's Register. All comparisons between treatment groups and control groups using rating scales were identified. The publication status of each scale was determined and claims of a significant treatment effect were recorded. Trials were more likely to report that a treatment was superior to control when an unpublished scale was used to make the comparison (relative risk 1.37 (95% CI 1.12-1.68)). This effect increased when a 'gold-standard' definition of treatment superiority was applied (RR 1.94 (95% CI 1.35-2.79)). In non-pharmacological trials, one-third of 'gold-standard' claims of treatment superiority would not have been made if published scales had been used. Unpublished scales are a source of bias in schizophrenia trials.