Authors' reply

To clarify, we did mean to state that earlier onset psychoses could represent individuals with stronger risk factors for the disorder. We agree that there is likely to be some heterogeneity in the outcomes, even if the loading of putative causal factors is similar for any set of given individuals. However, there is evidence that cases of childhood-onset psychosis spectrum disorders in general, when carefully defined, tend to be more severe and more homogeneous, with stronger family histories of schizophrenia spectrum disorders than adult-onset cases. ^1–3 Therefore, there may still be much to be learned about causality, even if the assumption of homogeneous clinical outcomes does not hold strictly true. Moreover, we believe the high rate of false-positive reports is likely to be due, at least in most instances, to clinicians initially wrongly attributing perceptual disturbance in children to an underlying psychotic illness. In most cases, it is likely that voice experiences and other potentially psychotic phenomena may result from processes that may be conceptualised as more psychologically driven, such as dissociation. Such experiences are commonly reported in community samples of children and adolescents, who are likely to share few, if any, of the risk factors associated with the development of early-onset schizophrenia spectrum disorders. ⁴

We would concur with Dr Bhavsar that the use of confidence intervals (which were indeed based on Poisson standard errors) in this situation may not have been strictly necessary. However, they do attempt to communicate some of the uncertainty regarding the estimates of incidence. We also agree that in this study most of this uncertainly will be due to the degree of completeness of case ascertainment using the surveillance design, rather than variability in the disease process.