Daniela De Angelis (Theme Lead) and Robert Goudie
This theme aims to develop, apply and disseminate principled analytical approaches to substantive problems in population health protection and health improvement to inform decision-making. Our work is motivated by multi-disciplinary collaboration around national and international public health priorities, including infectious disease elimination, pandemic preparedness, acute hospital care and disease prevention. More specifically, our research spans the following five areas.
Data integration in disease burden models
Disease burden can often be only quantified through the integration of multiple sources of evidence, typically through complex Bayesian models. Building and assessing these evidence synthesis models is challenging. Motivated by the estimation of bloodborne
virus and respiratory infection burden and to inform monitoring and treatment of hospital patients, we are extending and improving model building and assessment methods, to enable their systematic, practical and routine use for any Bayesian evidence synthesis.
Acute hospital efficiency, resourcing and effectiveness
Hospitals in the UK face many pressures, with increasing demand for services making sustaining and improving the quality of care that patients receive challenging. Innovations to make routine care more efficient and resilient are required. We are using both nationally representative day-by-day data and minute-by-minute data from selected hospitals to develop and evaluate such innovations through principled statistical approaches.
From our contributions to the COVID-19 pandemic response, lessons have been learnt on the requirements for effective pandemic preparedness. Our work focuses on developing tools to monitor disease transmission, particularly over the course of a long-lasting epidemic characterised by distinct phases, defined, for example, by policy changes or viral evolution.
Evaluation of non-randomised interventions
There is an increasing demand for approaches to evaluating non-randomised interventions. We are working on methods for the assessment of such population health interventions in two contexts: retrospectively, to evaluate interventions already implemented; and prospectively, to assess the long-term impacts of interventions or scenarios that might modify disease risk.
Optimal design of disease surveillance
Infectious disease surveillance is typically based on sentinel, convenience sampling. This lack of design leads to selection biases. Designing surveillance to answer specific questions is a natural, and crucial improvement over “making the best” of the observational data available. We are developing designs to improve disease surveillance, driven by the need for estimation of respiratory disease burden and detection of undiagnosed blood-borne virus infections as transmission elimination is approached.