Booster vaccines halved the risk of COVID-19 hospitalisation and death among adults aged 50 and over in England, according to new research using the OpenSAFELY platform developed by the Bennett Institute for Applied Data Science at the University of Oxford.
The study, published in Vaccine on 18 February and led by the universities of Bristol and Oxford, analysed linked GP and hospital records for more than 3.4 million adults who received an autumn 2022 booster, comparing them with the same number of eligible but unboosted people matched by age, vaccination history, clinical vulnerability and region.
Over nearly a year of follow-up, boosted individuals had substantially lower risks of COVID-19 hospitalisation (3.78 versus 6.81 per 1,000) and death (0.29 versus 0.61 per 1,000). Protection was strongest in the first 70 days and declined over time. Moderna and Pfizer-BioNTech boosters performed similarly.
William Hulme, statistical epidemiologist at Oxford's Bennett Institute for Applied Data Science, is among the study's authors, contributing to its design, methodology, data curation and funding.
The Bennett Institute created and maintains OpenSAFELY, a secure analytics platform that holds linked data on 24 million people registered at NHS general practices – roughly 45% of England's population. Rather than extracting patient data for external analysis, approved researchers write analytic code that runs inside NHS infrastructure; the patient-level data never leaves the NHS. All code for this study is openly shared on GitHub, allowing other researchers to review and reuse it.
This combination of scale, security and transparency allowed the Bristol and Oxford team to track hospitalisations and deaths across millions of people over nearly a year – exactly the kind of large-scale, rapid analysis the platform was designed to enable.
Dr Paul Madley-Dowd, research fellow in medical statistics and health data science at the University of Bristol and the study's corresponding author, said the findings reinforced the importance of booster vaccination for people over 50.
To test the robustness of their approach, the researchers also examined fracture rates – an outcome unlikely to be affected by vaccination. They found only a small difference between groups, which supports the validity of the study's main conclusions while acknowledging that some unmeasured factors may still influence the results.
The study was funded by NHS England, the Wellcome Trust, the MRC, the NIHR, the NIHR Bristol Biomedical Research Centre, and the Bennett Institute for Applied Data Science.
Reference: Madley-Dowd, P. et al. (2026) 'Effectiveness of bivalent BA.1 mRNA booster vaccines during the autumn 2022 COVID-19 booster programme in adults aged 50+ in England: observational matched cohort study using OpenSAFELY', Vaccine. DOI: 10.1016/j.vaccine.2026.128276