Loading ...

TOPLINE:

Researchers developed 12 algorithms to identify individuals with psychosis using England's Mental Health Services Data Set (MHSDS), with success rates aligning with national prevalence estimates and most algorithms recognising the same core patient group.

METHODOLOGY:

• Researchers developed 12 unique algorithms using MHSDS, incorporating data elements such as mental health clusters, scores on the Health of the Nation Outcome Scales, reasons for referral, primary diagnosis, and early intervention flags, individually or in combination.
• Analysis included data from 2,378,247 patients in secondary care during 2018-2019 and from 2,220,812 patients during 2017-2018.
• Investigators validated the resulting numbers by comparing them with national estimates of psychosis prevalence.

TAKEAWAY:

• The algorithms identified 99,204-138,516 cases of psychosis for 2017-2018 and 107,545-1,46,832 cases for 2018-2019, aligning with national prevalence estimates.
• Most algorithms identified the same core group of 48,681 patients in 2018-2019.
• The findings were comparable between the durations 2017-2018 and 2018-2019.

IN PRACTICE:

"The use of these algorithms will enable population-level surveillance of outcomes among this patient population, as well as examination of their healthcare utilisation, at minimal cost," the authors wrote. "The choice of which algorithm to employ will depend on the purpose of the analysis," they added.

SOURCE:

The study was led by Claire de Oliveira, Centre for Health Economics, University of York, York, United Kingdom. It was published online on February 27 in BJPsych Open.

LIMITATIONS:

The algorithms may have misclassified some individuals as having psychosis or failed to identify true cases. The algorithms were applicable only to MHSDS version 4 and may require updates as the use of mental health clusters is phased out in the future. No single gold-standard algorithm was identified, and external validation was limited due to restricted access to primary care data and chart records.

DISCLOSURES:

The study was partly funded by the University of York's Economic and Social Research Council Transformative Research scheme and the Efficiency Research Programme of the Health Foundation. The authors reported no conflicts of interest.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.