Population Health Screening Benefits and Harms: Evidence Synthesis methods to support policy-making

Research question What are the most accurate evidence-synthesis methods (for indirect evidence, use of routine data, and international standardisation) to evaluate screening programmes for national policy-makers? Background Policy-makers and governments make decisions about which screening programmes to implement based on synthesis of research evidence. This synthesis and subsequent decisions could be improved through understanding when indirect evidence is appropriate, integration of routine screening data into the reviews, and fostering international collaboration to reduce research waste and improve review quality. Aims The overall aim is to develop a Centre of Evidence Synthesis for Screening Policy, leading international efforts to optimise evidence synthesis for policy-makers, enabling them to maximise benefits of screening and minimise harms and inequity. Evaluate and delineate circumstances under which indirect evidence chains can accurately augment or replace direct evidence in screening evaluations Describe how routine data (e.g. on ethnicity and socioeconomic status) can be used to bridge evidence gaps and assess generalisability in screening evaluations Develop a list of requirements to adapt an evidence review of screening to be applicable across countries. As an exemplar of using indirect evidence chains incorporating routine data, generalisable across multiple countries, evaluate the benefits and harms of incorporating artificial intelligence in breast cancer screening, A cross-cutting theme across all aims is to evaluate the impact by ethnicity and socioeconomic status, to facilitate a reduction in inequity from screening programmes. Methods Policy-makers guidelines for evidence synthesis, and previous reviews (n=1000), will be systematically extracted from websites of 22 national and 2 international policy-making organisations. Previous evidence syntheses from all disease-specific screening programmes with both direct and indirect evidence will be identified and analysed to ascertain drivers of inconsistency in results between approaches. Factors affecting generalisability of direct and indirect trial evidence to practice will be identified using mixed-effect models of routine cancer data from 28 countries. Simulation studies will explore the impact of drivers/factors identified in this study and the literature on accuracy of conclusions. New methodological guidance for international evidence synthesis will be developed by extracting findings from this research, published primary quantitative research, and other methodological guidance, building on the US Preventive Services Task Force analytic framework for evidence synthesis in screening, (best fit framework approach). International consensus for core standards across 22 screening committees will be developed (Delphi methods), and applied to three reviews for UK, Canadian, and Irish policy-makers. A best-practice exemplar review of incorporating artificial intelligence into the breast cancer screening test will be produced, using routine data from 13 million women to supplement the indirect evidence pathway and incorporating all methodological findings. Anticipated impact and dissemination Changes to methodological guidance for screening evidence reviews in the UK, Canada and Ireland, resulting in more evidence-based decisions about all future screening programmes. An international collaboration for multi-national joint evidence reviews in screening, reducing research waste, improving review quality, and producing reviews for countries who do not currently have access. Self-sustaining Centre of Evidence Synthesis for Screening Policy, to improve quality of UK based primary research.