Using CPRD to describe the current management of heart failure in England and model risk trajectories to improve shared decision-making

Date of Approval: 
2018-06-01 00:00:00
Lay Summary: 
Heart failure (HF) occurs when the heart is not pumping blood around the body as well as it should. It is serious and common. It affects a growing number of people of all ages, but can be difficult to diagnose and manage, particularly in general practice. Identifying those patients who are at highest risk of emergency hospitalisation and death would help doctors and patients make better decisions around where and how rigorously their HF is managed - it would help patients to understand where they are and what they and their family need to do and how this translates into their self-care strategy. However, this "risk prediction" is hard to do well. Existing risk prediction models have various limitations that prevent their widespread use. In particular, they have been developed on patients taking part in clinical trials, who are not typical of HF patients treated in the community. We propose to use a well established research database called CPRD to improve on this.
Technical Summary: 
Heart failure (HF) is common, serious and growing in prevalence but can be difficult to diagnose and manage in primary care. Identifying those patients who are at highest risk of poor outcomes such as hospitalisation and death would help doctors and patients make better decisions around management, but it is challenging. Existing risk models have been developed on highly selected cohorts in clinical trials and have other limitations that prevent their widespread use. We propose to use CPRD to describe the management of HF and NHS resource use during two periods, ten years apart, and to model how patients' risk of death changes over time from diagnosis and from first HF admission. Methods will include time-to-event, trajectory and, if it proves feasible, multistate models. Our aim is not to produce the definitive HF risk stratification model, but instead to provide usable information on risk trajectories for shared decision-making by clinicians and patients.
Health Outcomes to be Measured: 
Death (primary outcome) - Hospitalisation (secondary outcome)
Application Number: 
18_109
Collaborators: 

Alex Bottle - Chief Investigator - Imperial College London
Alex Bottle - Corresponding Applicant - Imperial College London
Azeem Majeed - Collaborator - Imperial College London
Martin Cowie - Collaborator - King's College London
Paul Aylin - Collaborator - Imperial College London

Linkages: 
HES Accident and Emergency;HES Admitted Patient Care;HES Outpatient;ONS Death Registration Data;Patient Level Index of Multiple Deprivation