Psychosis Incidence Prediction

On these pages you will find detailed predictions for new cases of psychosis in England and Wales. Predictions are based on the PsyMaptic prediction tool, a validated and peer-reviewed epidemiological prediction tool for first episode psychosis. Predicted counts of new cases and incidence rates over a given year are available at national and Local Authority District levels, overall, and stratified by broad age group and sex. More detailed predictions may be available as custom data. Full details of the current PsyMaptic psychosis incidence prediction model are provided at the bottom of this page. For previous versions, see our version log.

You may view our prediction data on geographical maps at local authority or county levels. Data is also available in tabulated form at nationalLAD and county levels. Key is provided freely under our open-use policy. More detailed predictions can also be requested.

 

 

Current PsyMaptic psychosis model: version 1.1

Rate type: incidence (new cases)

Last Updated: 1st August, 2014

Prediction model sample: 1025 incidence cases of psychotic disorder from the AESOP (South London, Nottingham and Bristol) and East London First Episode Psychosis studies (~1996-2000). See Kirkbride et al 2006 and Kirkbride et al 2008 for full details. Denominator estimated from 2001 census of Great Britain and adjusted for case ascertainment period (2 years, except Bristol: 0.75 years).

Prediction model: Negative binomial regression model fitted with age group (16-17, 18-19, 20-24… 60-64), sex, their interaction, ethnic group (10 category), Local Authority District [LAD] population density (based on usual resident population estimated in 2011 census) and 2010 Extent of Deprivation from the English Indices of Deprivation and a quadratic term for the Extent of Deprivation. Internal apparent validity assessed as well as out-of-sample  external validity. See Kirkbride et al, 2013.

Prediction model geography: LAD level (England & Wales N=348)

Forecast geography: LAD, county and national levels (England, Wales, combined)

Forecast denominator data: English and Welsh population aged 16-64 years old at LAD level, estimated using 2011 Census of Great Britain estimates stratified by age group, sex and ethnic group (as above). Source data

Forecast length: All estimated predicted counts of new cases and incidence rates are presented on a per annum basis.

Forecast predictions: Predicted counts of new cases and incidence rates are presented overall for England, Wales and England and Wales combined, and at LAD-level. All prediction estimates are stratified by broad age group (16-35 years old, 36-64 years old and 16-64 years old), sex and by selected ethnic groups (see Version 1.1 Release Notes for full details). Predicted counts are for new cases of all psychotic disorders (ICD-10 F10-39). More detailed and customised prediction reports may also be requested.

Forecast error: All forecasts include a degree of uncertainty, estimated as prediction error. This error takes into account the standard error present in both the prediction and forecast samples and allows us to estimate the likely imprecision around are predicted counts of new cases of psychosis. For all forecasts we present the 95% prediction intervals around our expected count. We would expect the observed count of cases for a given forecast to fall within the prediction intervals on 95 out of 100 times. i.e. In 95 out of 100 years, we would expect to observe the true count of new cases to lie within the prediction intervals we present, with the actual prediction being the “most likely” value this would take. We developed a method from first principles to obtain prediction intervals from negative binomial prediction models. We encourage all users of this data to take into account our prediction error when using this data for other purposes. You should also read our disclaimer. You will be asked to agree to our Terms of Use before using this data.

Complete methodology: You may read a detailed account of the development of the PsyMaptic prediction tool here