M201. MODERATORS OF WEIGHT GAIN IN RANDOMIZED CONTROLLED TRIALS OF SCHIZOPHRENIA – A META-REGRESSION ANALYSIS

BACKGROUND: Weight gain is an important side effect of antipsychotics. Meta-analyses of randomized controlled trials indicate differences between the multiple antipsychotics in propensity to cause weight gain. However, antipsychotic-associated weight gain, in randomized controlled trials as well as in real life situations, might also depend on population characteristics and treatment-related factors. As a preparatory work for a systematic review and network-meta-analysis on metabolic side effects of antipsychotics (presented in another poster at this conference), we conducted a meta-regression analysis of potential moderators of weight gain. METHODS: We selected acute phase short-term, acute phase long-term as well as relapse prevention studies (all found by systematic reviews conducted by our group in the past) from our database of randomized controlled trials of antipsychotics in schizophrenia. We conducted: i) meta-regression-analyses based on pairwise meta-analysis of the mean difference (MD) in change in weight (kg) between patients exposed to antipsychotics and placebo, and ii) meta-regression-analyses based on single-arm meta-analysis of the change in weight (kg). We examined the moderators baseline weight, study duration, percentage women, and publication year. We conducted the analyses for all drugs pooled, placebo and per individual drug. For presentation of the results we focus on the drugs aripiprazole, haloperidol, quetiapine, olanzapine, and risperidone, because these are popular drugs in clinical practice, they differ in their receptor-binding profiles and multiple studies are available for these drugs. We conducted the analysis in R using the commands metacont, metagen and metareg from the package meta. RESULTS: The dataset comprises 603 randomized controlled trials with 141 584 patients that examined 40 different antipsychotics. Trial duration varied between 3 and 156 weeks (median 8 weeks). 168 studies with 49 670 patients reported on change in body weight. We found no effect of baseline weight on antipsychotic-associated weight gain, however on placebo, lower baseline weight was associated with more weight loss. Moreover, on antipsychotics, higher percentage of women was associated with less weight gain, and, on placebo, higher percentage of women was associated with more weight loss. Longer study duration was not associated with increased weight gain. On placebo, longer studies were associated with more weight loss. There was no effect of publication year. DISCUSSION: Surprisingly, we found no moderating effect of baseline weight and study duration on weight gain. However, our data suggests that men and women could have different risk of weight gain. Moreover, weight loss after switching to placebo might be higher in women and patients with less baseline weight. For interpretation, it must be noted that meta-regressions are observational evidence and thus prone to confounding. In addition, the scatter plots, presented on the poster, need to be considered to judge the robustness and magnitude of the moderated effects. Additional meta-regressions (planned to present on the poster) should address further potential moderators, such as antipsychotic dose, ethnicity, previous antipsychotic exposure and dropout rates.