Psychosis risk prediction tool features in BJGP podcast

A project the Bristol BRC is collaborating with University College London on has been featured on a British Journal of General Practice (BJGP) podcast.

The research team behind this project has developed and validated a risk prediction model for primary care providers (such as GPs) to assess whether their patients are at risk of developing psychosis.

The team is currently investigating whether the tool (P-risk) could be implemented in GP surgeries.

Dr Sarah Sullivan, senior research fellow in quantitative primary care at the University of Bristol, spoke to Nada Khan, associate editor at the journal.

Dr Sullivan said:

“Psychosis often has poor outcomes for people and most of the risk factors for developing it are either almost impossible or very difficult to change. Getting people into specialist care more quickly is one thing that we know is strongly associated with better outcomes and GPs play an important role in this.

“Many of the early warning signs for psychosis are non-specific. This makes picking them up difficult, especially as GPs may not have had the chance to build up experience in diagnosing the condition. We thought a data-driven approach might help solve some of these problems.

“In our training data P-risk correctly identified when someone was at risk of developing psychosis 77 per cent of the time. When applied to external validation data, this changed slightly to 78 per cent. This means P-risk works as well as other risk algorithms GPs use every day, such as QDiabetes.”

“We’re now going to run P-risk on GP computers for six months in parts of the country. From that we’ll collect non-identifiable summary data on how the tool ran. We’ll also interview GPs, patients and secondary care mental health clinicians on their views and opinions of P-risk.”