Is health research undertaken where the burden of disease is greatest? Observational study of geographical inequalities in recruitment to research in England 2013-2018

• Published in: BMC medicine 2020-05, Vol.18 (1), p.133-11, Article 133
• Main Authors: Bower, Peter, Grigoroglou, Christos, Anselmi, Laura, Kontopantelis, Evangelos, Sutton, Matthew, Ashworth, Mark, Evans, Philip, Lock, Stephen, Smye, Stephen, Abel, Kathryn
• Format: Article
• Language: English
• Subjects: Alignment; Analysis; Biomedical Research - methods; Cardiovascular disease; Diabetes; Diabetes mellitus; England - epidemiology; Equity; Female; Geographical distribution; Geographical variations; Geography; History, 21st Century; Humans; Infrastructure; Male; Mathematical analysis; Measurement methods; Medical research; Mental disorders; Mental health; Middle Aged; Observational studies; Patient Selection; Patients; Populations; Primary care; Recruitment; Research activity; Risk analysis; Risk Factors; Socioeconomic Factors
• ISSN: 1741-7015; 1741-7015
• DOI: 10.1186/s12916-020-01555-4

Research is fundamental to high-quality care, but concerns have been raised about whether health research is conducted in the populations most affected by high disease prevalence. Geographical distribution of research activity is important for many reasons. Recruitment is a major barrier to research delivery, and undertaking recruitment in areas of high prevalence could be more efficient. Regional variability exists in risk factors and outcomes, so research done in healthier populations may not generalise. Much applied health research evaluates interventions, and their impact may vary by context (including geography). Finally, fairness dictates that publically funded research should be accessible to all, so that benefits of participating can be fairly distributed. We explored whether recruitment of patients to health research is aligned with disease prevalence in England. We measured disease prevalence using the Quality and Outcomes Framework in England (total long-term conditions, mental health and diabetes). We measured research activity using data from the NIHR Clinical Research Network. We presented descriptive data on geographical variation in recruitment rates. We explored associations between the recruitment rate and disease prevalence rate. We calculated the share of patient recruitment that would need to be redistributed to align recruitment with prevalence. We assessed whether associations between recruitment rate and disease prevalence varied between conditions, and over time. There was significant geographical variation in recruitment rates. When areas were ranked by disease prevalence, recruitment was not aligned with prevalence, with disproportionately low recruitment in areas with higher prevalence of total long-term and mental health conditions. At the level of 15 local networks, analyses suggested that around 12% of current recruitment activity would need to be redistributed to align with disease prevalence. Overall, alignment showed little change over time, but there was variation in the trends over time in individual conditions. Geographical variations in recruitment do not reflect the suitability of the population for research. Indicators should be developed to assess the fit between research and need, and to allow assessment of interventions among funders, researchers and patients to encourage closer alignment between research activity and burden.