People with psychotic disorders like schizophrenia die 15-years sooner than the general population, mostly due to physical
conditions like diabetes and heart disease. Early signs of these physical conditions are already detectable in young people
when they first develop psychosis. This is an important problem because over half the NHS budget for treating psychotic
disorders is used for treating physical conditions.
In the general population, computer-based tools are routinely used to determine who is at highest risk of future conditions
like heart disease, so steps can be taken to prevent them from developing. However, existing tools are inaccurate when
used in young people with psychotic disorders.
Therefore, using data from 1,000 young people with psychotic disorders, we developed a tool specifically for this group:
The Psychosis Metabolic Risk Calculator (PsyMetRiC). It aims to help clinicians determine who is at greatest risk for future physical health problems.
To bring PsyMetRiC toward routine use, it now needs testing and refining in a larger sample. We will use the CPRD and
QResearch resources, which include data from up to 40million people. This will allow us to: accurately test PsyMetRiC; refine
PsyMetRiC by including additional information to make predictions; and, predict longer-term conditions like diabetes. These steps will improve PsyMetRiC’s accuracy and usefulness.
In future, PsyMetRiC will undergo a clinical trial where it will help decide which treatments are offered to who.
This work will pave the way toward preventing physical conditions in people with psychotic disorders, thus improving
quality and length of life.
Young people with psychotic disorders are at substantial cardiometabolic risk, and physical comorbidity is the leading cause of a 15-year reduced life expectancy in people with psychotic disorders. There is a clear need for improved strategies to tackle this comorbidity. While cardiometabolic risk prediction algorithms are used routinely in the general population to encourage informed, personalised treatment decisions for primary prevention of adverse cardiometabolic outcomes, general-population-based algorithms are unsuitable for young people with psychosis.
In my NIHR Doctoral Research Fellowship, using data from three UK Psychosis Early Intervention Services, I developed and
externally validated a cardiometabolic risk prediction algorithm tailored for young people with psychosis: the Psychosis
Metabolic Risk Calculator (PsyMetRiC). PsyMetRiC predicts up-to six-year risk of developing metabolic syndrome in young
people with psychosis from routinely collected data, and was accompanied by an online data visualization tool
(http://psymetric.shinyapps.io/psymetric).
PsyMetRiC now needs testing and refining in larger, population-representative samples, before it can be taken forward
toward routine clinical use. The CPRD and QResearch databases represent an unrivalled opportunity for this work. These datasets have been used previously to develop and validate general population-based cardiometabolic risk prediction algorithms (e.g. QRISK3).
Therefore, first we will use the QResearch sample to make improvements and refinements to the PsyMetRiC algorithm, such as including additional predictors and featuring more distal clinically-relevant outcomes such as 10-year risk of type 2 diabetes and cardiovascular disease.
We aim to use CPRD as a 'model validation' sample to:
1) Test the existing PsyMetRiC algorithm in a large population-representative sample, using logistic regression.
2) Test the refined PsyMetRiC algorithms, using logistic regression and/or cox proportional hazards regression.
This work will pave the way for a package of complex intervention development including PsyMetRiC, to improve the
physical health of young people with psychosis.
1) Stage 1: Metabolic syndrome computed either from biological information as per the International Diabetes Federation (IDF)-consensus definition, or from general practice records or hospital episode statistics (ICD-10 code E88.81), at the latest assessment up to six-years after baseline. The outcome will be measured up to six-years later to match the original PsyMetRiC study;
2) Stage 2: A composite continuous Z-score outcome consisting of fasting plasma glucose (FPG), high-density lipoprotein cholesterol (HDL), triglycerides (TG), systolic blood pressure (SBP), body mass index (BMI), measured at the latest assessment five-years after baseline. The outcome will be measured at five-years to best match up with existing shorter-term risk prediction study timeframes;
3) Stage 3a: type 2 diabetes computed either from biological information (FPG or random plasma glucose or glycated haemoglobin) or recorded diagnosis from general practice records or hospital episode statistics (ICD-10 code E11*) up to 15 years after baseline assessment, focused on 10-year risk. Across all measures, the earliest assessment meeting the outcome criteria will be used as the outcome date;
4) Stage 3b: cardiovascular disease from recorded diagnosis (general practice records or hospital episode statistics or linked mortality records) up to 15 years after baseline assessment, focused on 10-year risk. We will include ICD-10 codes: G45 (transient ischaemic attack and related syndromes), I20 (angina pectoris), I21 (acute myocardial infarction), I22 (subsequent myocardial infarction), I23 (complications after myocardial infarction), I24 (other acute ischaemic heart disease), I25 (chronic ischaemic heart disease), I63 (cerebral infarction), and I64 (stroke not specified as haemorrhage or infarction). Across all measures, the earliest assessment meeting the outcome criteria will be used as the outcome date.
Benjamin Perry - Chief Investigator - University of Cambridge
Benjamin Perry - Corresponding Applicant - University of Cambridge
Emanuele Osimo - Collaborator - University of Cambridge
Graham Murray - Collaborator - University of Cambridge
Jan Stochl - Collaborator - University of Cambridge
Peter Jones - Collaborator - University of Cambridge
Rachel Upthegrove - Collaborator - University of Birmingham
Simon Griffin - Collaborator - University of Cambridge
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Townsend Index