Exploring the use of modern matching algorithms as a key component of the public health toolkit.

Background: The use of data for group comparisons is integral for public health evidence-based practice. Modern matching algorithms, which cover a range of matching and weighting methods, can be applied to problems of estimating of causal effects in non-randomised studies, of generalising and transporting estimates from one sample to a wider population, and of exploring health inequalities by seeking unconfounded descriptive comparisons. However, these methods are likely underused in public health. Research question: Can modern matching algorithms, applied to the analysis of health and social care data, provide new prevalence estimates of public health conditions or behaviours, and describe health inequalities through unconfounded descriptive comparisons? Aim: To advance the application of modern matching techniques in public health and to explore how to improve the understanding and trustworthiness of evidence produced using these techniques. Specific objectives: (1) to assess how matching methods are used in public health, what matching methods have been used, and for what challenges (2) to undertake two practical applications of matching analyses relevant to local public health priorities (3) to develop a guide to support public health professionals and the public with the understanding of matching. Methods: Four specific studies are proposed. Study area 1 will be a scoping review of the use and presentation of matching for public health application. Study 2 will be a target trial emulation study looking at the impact of additional mental health practitioner roles on antidepressant prescribing in general practice. Study 3 will be a tapered matching study using the CUREd+ research database to determine disparities in breast cancer survival within South Yorkshire. Study area 4 will be the development of a guide on the use and presentation of matching for a public and public health and policy audience which will use qualitative methodologies to understand barriers and enablers to using, and trusting, matching methods. Timeline for delivery: This will be a 36-month fellowship full-time, commencing October 2025. Study area 1 will start immediately. Study area 2 has a protocol already developed so will start following the study area 1. Study area 3 will run alongside study area 1 in the first instance to confirm all database requirements and associated procedures are in place to begin analysis in 2026. Study 4 will run throughout the full project as all three other study areas will feed into this co-produced guide. Impact and dissemination: The evidence produced throughout this fellowship will support public health professionals by advancing the application of modern matching methods for addressing public health knowledge needs and making these approaches understandable and trustworthy. It will contribute to understanding health inequalities which may inform public health policy making and targeted interventions or services. A range of dissemination methods will be used including a written guide, an innovative infographic co-produced with the public, oral presentations to public health professionals, conference presentations, and academic papers through a range of collaborations.