Meta-analysis: the placebo rate of abdominal pain remission in clinical trials of chronic pancreatitis.

Overview

OBJECTIVES: Treatment of the pain caused by chronic pancreatitis (CP) is not standardized. Knowledge of the response to placebo in this setting may aid the design of future trials. We aimed at investigating the placebo effect on abdominal pain remission rates in patients with CP. METHODS: MEDLINE, EMBASE, and Scopus were searched, and randomized placebo-controlled trials in CP providing data on abdominal pain remission rates in placebo arms were included. Pooled estimates of the placebo rate were calculated using random-effects logistic regression analysis. Stratum-specific rates for different patients and study-level covariates were calculated to account for heterogeneity. RESULTS: Seven randomized controlled trials (202 placebo-treated patients) met the predefined criteria. The pooled estimate of the placebo rate for pain remission was 19.9% (95% confidence interval, 9%-36%). There was a statistically significant heterogeneity among the studies (I(2) = 76%; P < 0.001). A multicenter design, a run-in period of less than 2 weeks, and absence of a washout in crossover trials were all significant sources of heterogeneity associated with higher placebo responses. CONCLUSIONS: This meta-analysis identifies for the first time the efficacy of placebo for pain in CP and variables determining it. These data provide a sound basis for designing future placebo-controlled randomized clinical trials for the treatment of pain in CP.