Neuropsychiatric indices in a high security admission sample I: estimating the prevalence