Authors' reply

We thank Dr Vanderburg and colleagues for their comments on our paper. We used observational, non-randomised, routinely collected data to describe the rate of incident diabetes among patients treated with antipsychotics in clinical practice. These data reflect the way antipsychotics are handled by individual clinicians on the basis of their knowledge of effects and side-effects for the specific patient. Findings from analyses of our data cannot be used to infer causality between individual antipsychotics and diabetes and may be at odds with findings from randomised trials and other studies aimed at testing specific hypotheses. Our results on the individual antipsychotics describe the prevalence of diabetes among patients for whom the clinician decided to prescribe a given antipsychotic.