Abstract
Introduction: The UK and global life expectancy is increasing, but life years lived in ill health is also increasing. Disease burden, and health and social care service use is highest in older age. Prevention, treatment and management of conditions of older age (e.g. frailty and multi-morbidity) is a research priority. Efficient trials need to better recruit and retain older participants to produce robust and generalisable evidence for our aging population. Synthesised qualitative and quantitative evidence regarding trial retention does not generally include the oldest and frail in society, and recommendations likely do not resonate with this population who have different needs and barriers to research participation. Here we present a systematic review and meta-analysis of participant and trial characteristic impacting retention rates in physical rehabilitation trials in community dwelling, frail, older people. Method: Medline, Embase, CENRTAL, CINAHL and PEDro were searched from 2010-2025. RCTs of physical rehabilitation intervention in a frail community dwelling population were included. Two reviewers independently screened studies before extracting aggregated study data. Results: Following automated deduplication, 2123 titles/abstracts were screened, 340 with full text screen, 36 studies were included. Pooled retention rates were 82%, with no between trial arm impact. Meta-regression analysis indicates significant retention effect of trial duration, mode of data collection, and geographical region. Average trial participant age and frailty severity, type of primary outcome, and number of trial sites were not significantly associated with retention rates. Other retention strategies were very sparsely reported. Conclusion: Retention rates observed are very similar to those reported in another frailty trials review, and lower than retention in trials generally (89%). Simple trial design change may significantly improve retention for this population. The aggregate nature of age and frailty severity data likely impacted analysis. Individual participant data analysis is required to further explore these associations.