Development of small population prevalence model for heart failure

Date of Approval
Application Number
Technical Summary

Heart failure is a common, long term condition of enormous public health importance, which causes significant morbidity and mortality and contributes significantly to NHS health care costs. In this project we will be developing a prevalence model for heart failure in the English population,[1] using CPRD as a new data source,. Read codes for doctor diagnosed heart failure will be used to extract all definite and possible cases of heart failure from the CPRD cohort and the prevalence at the end of each year and cumulative prevalence will be estimated. In addition to diagnostic Read codes entered in primary care electronic health records (EHRs) we will use other data to identify likely or possible cases of heart failure, including Hospital Episode Statistics (HES) ICD-10 diagnostic data linked to CPRD, and other clinical data including test, prescribing and symptom data. This model will then enable us to identify patients in addition to GP-diagnosed cases who possibly have heart failure but do not have a GP diagnosis. Logistic regression models will then be fitted to estimate risk factor odds ratios, which will then be converted using a well-established method to small population prevalence estimates with confidence intervals using corresponding local data on risk factors.

Health Outcomes to be Measured

Heart failure


Michael Soljak - Chief Investigator - Imperial College London
Michael Soljak - Corresponding Applicant - Imperial College London
Azeem Majeed - Collaborator - Imperial College London
Mahsa Mazidi - Collaborator - Imperial College London
Martin Cowie - Collaborator - King's College London (KCL)
Roger Newson - Collaborator - Imperial College London


HES Admitted Patient Care;Patient Level Index of Multiple Deprivation;Practice Level Index of Multiple Deprivation