Date of Approval:
Heart failure is a serious and common disease with a high risk of needing admission to hospital as an emergency when symptoms worsen. If we can better identify which types of patients are hospitalised as an emergency, it will help with their clinical management by their GP and healthcare team. To do this, we plan to combine GP consultation records from the Clinical Practice Research Datalink (CPRD) with databases of hospital admissions and death certificates for participating practices who have given their consent for their data to be used in research. We will build statistical models to predict which patients are hospitalised or die within a year of being diagnosed with heart failure, taking into account factors such as their age, gender, medications and other medical problems; most research so far in this area has focused on hospital admission data, which lack some key information that CPRD captures.
Our objectives are to identify the main predictors of emergency (re)hospitalisation in patients with heart failure and their relative importance. In this observational study with a retrospective cohort design, we will first identify in the linked CPRD-HES-ONS database each patient's first NHS contact for HF (either practice consultation or hospital admission) and take this as the diagnosis date and start of follow-up. All patients in the database up to March 2014 will be considered, taking into account patient and practice data quality flags, database follow-up, eligibility and coverage periods. Predictors will include demographics, comorbidities, frailty, physiology and medications. After descriptive analysis and an assessment of missing data, statistical methods will include survival analysis in a competing risks framework and multiple logistic regression, with forced entry of predictors where possible. We will explore the need for adjustment for clustering within practices.
Alex Bottle - Chief Investigator - Imperial College London
Azeem Majeed - Collaborator - Imperial College London
Martin Cowie - Collaborator - King's College London
Paul Aylin - Collaborator - Imperial College London