Are Antipsychotics Effective for the Treatment of Anorexia Nervosa?

Abstract

To assess the utility of antipsychotics for weight gain and improvement of illness-related psychopathology in patients with anorexia nervosa. PubMed, the Cochrane Library databases, and PsycINFO citations from the inception of the databases until March 27, 2012, were searched without language restrictions using the following keywords: randomized, random, randomly, and anorexia nervosa. In addition, we hand-searched for additional studies eligible for inclusion in this meta-analysis and contacted authors for unpublished data. Included in this study were randomized placebo- or usual care-controlled trials of antipsychotics in patients with anorexia nervosa. Two independent evaluators extracted data. The primary outcome of interest was body weight, expressed as the standardized mean difference (SMD) between the 2 groups in baseline to endpoint change of body mass index (BMI), endpoint BMI, or daily weight change. SMD, risk ratio (RR), and number needed to harm (NNH) ± 95% confidence interval (CI) were calculated. Across 8 studies (mean duration = 9.6 weeks; range, 7-12 weeks), 221 patients (mean age = 22.5 years, 219 [99.1%] females) with anorexia nervosa were randomly assigned to olanzapine (n = 54), quetiapine (n = 15), risperidone (n = 18), pimozide (n = 8), sulpiride (n = 9), placebo (n = 99), or usual care (n = 18). Both individually (P = .11 to P = .47) and pooled together (SMD = 0.27, 95% CI, -0.01 to 0.56; P = .06, I2 = 0%; 7 studies, n = 195), weight/BMI effects were not significantly different between antipsychotics and placebo/usual care. Moreover, pooled antipsychotics and placebo/usual care did not differ regarding scores on questionnaires related to anorexia nervosa (P = .32, 5 studies, n = 114), body shape (P = .91, 4 studies, n = 100), depressive symptoms (P = .08, 4 studies, n = 103), and anxiety (P = .53, 4 studies, n = 121). Individually, quetiapine (1 study, n = 33) outperformed usual care regarding eating disorder attitudes (P = .01) and anxiety (P = .02). While rates of dropout due to any reason (P = .83, I2 = 0%) and due to adverse events (P = .54, I2 = 5%) were similar in both groups, drowsiness/sedation occurred significantly more often with antipsychotics than placebo/usual care (RR = 3.69, 95% CI, 1.37-9.95; I2 = 67%, P = .01; NNH = 2, P = .001; 5 studies, n = 129), but most other adverse effects were only sparsely reported. Although limited by small samples, this meta-analysis failed to demonstrate antipsychotic efficacy for body weight and related outcomes in females with anorexia nervosa.