Every year, an estimated 27,500 people with heart failure in England are hospitalised because their medication damages their kidneys. Doctors currently lack clear, personalised guidance on how often to check kidney function or how to adjust drug doses for each patient. This project aims to close that gap. The researchers have already built an algorithm that predicts how a person’s kidney function will change over time, using data from electronic health records. They will now refine that algorithm with deep-learning methods and co-design a clinical pathway with patients, GPs, and specialists. The pathway will be embedded into standard GP software, generating personalised monitoring schedules and medication thresholds for each patient. If the trial succeeds, the system could reduce preventable hospital admissions. A 5% drop in admissions would mean 1,375 fewer hospital stays per year in England alone. The team will also assess cost-effectiveness and use implementation frameworks—including the Health Inequalities Assessment Tool—to ensure the system works equitably across different populations. Early engagement with policymakers and software providers is planned to support national rollout.
View original technical description
Background Modern heart failure treatment has reduced morbidity and improved survival, but people with heart failure remain at risk of renal deterioration. This is potentially preventable through regular renal function monitoring and optimisation of drug dose/choice. Currently, healthcare professionals treating people with heart failure lack evidence-based guidance on how frequently to monitor renal function and on the best way to adjust drug dose/choice for each individual. We hypothesise that the lack of clear, personalised guidance contributes to renal deterioration and hospitalisation in people with heart failure. In preliminary work we: 1) showed there is enormous variability in care of people with heart failure and 2) created an algorithm to predict renal function change using electronic health record data. Aims and Objectives We shall develop an evidence-based system for generating guidelines for personalised renal function monitoring and treatment embedded within standard software used by primary care practitioners. We shall: Refine the accuracy of the algorithm that predicts the change in renal function in people with heart failure; Co-create with patients, primary care practitioners and specialists the clinical pathway for implementing personalised renal function monitoring and interventions; Assess the clinical effectiveness of our algorithm-guided clinical pathway, compared with current standard-of-care, in a cluster randomised controlled trial embedded within the Clinical Practice Research Datalink (CPRD); Perform health economic analysis to determine the cost-effectiveness of our algorithm-guided clinical pathway; Ensure that the patient voice is integrated throughout our studies and outputs. Methods We shall refine our algorithm using statistical and deep-learning methods and co-design algorithm-informed decision making with specialists (identifying the best algorithm and defining personalised renal function monitoring frequencies and thresholds for altering medication). We shall reach consensus using the RAND/UCLA Appropriateness Method. The algorithm will be embedded into software in collaboration with CPRD and the software provider. We shall establish patient and practitioner views on usual care using surveys, and structured and 'think aloud' interviews. This will inform workshops to co-create the implementation pathway with patients, practitioners and specialists. We shall use the behaviour change wheel, the non-adoption, abandonment, scale-up, spread, sustainability framework and the Health Inequalities Assessment Tool to support implementation. Implementation will be beta-tested in five general practices and acceptability evaluated using quantitative data and qualitative interviews. Our intervention will be tested in a randomised controlled trial delivered through CPRD. The number of people living with heart failure hospitalised due to renal deterioration/injury will be compared between 100 practices using our intervention and 100 control practices. Patient-acceptability and benefits will be evaluated using the EQ-5D-5L survey. Health economic analysis will determine the lifetime incremental costs and Quality Adjusted Life Years of implementing the renal function monitoring algorithm compared with usual care. Impact We estimate that 27,500 people with heart failure are admitted to hospital due to renal impairment caused by medication per year in England. Evan a 5% decrease in hospital admission would correspond to 1,375 admissions avoided per year. We will engage with primary care system providers and policy makers early in the project to facilitate scale-up UK wide.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know