Category: Cancer Rehabilitation; Health Services Research; Cross-Cutting
Research Objective: Successful rehabilitation research on exercise and physical activity (PA) interventions depends on effectively recruiting and retaining participants. This meta-analysis and meta-regression examined dropout and recruitment efficiency (RE) in exercise and PA studies in the breast cancer (BC) population.
Data Sources: Pubmed and Cumulative Index to Nursing and Allied Health Literature databases.
Study Selection: Randomized and non-randomized studies were abstract screened for eligibility by trained study team members. Eligible studies must: 1) be written in English, 2) implement an exercise or PA intervention involving human adults, 3) report recruitment outcomes, 4) peer-reviewed, and 5) have the BC population constitute >50% of the sample.
Data Extraction: Dropout, RE, participant characteristics, study characteristics, research design, and intervention factors were extracted. RE was defined as the number of participants enrolled divided by the number of potential participants contacted.
Data Synthesis: Twenty-seven out of 2836 identified studies were eligible for meta-analysis. Random-effects meta-analyses calculated point-estimates for dropout (d=.16, 95% CI=[.11, .22]; p<.001) and RE (d=.30, 95% CI=[.24, .36]; p<.001). The aggregate z-score for participant variables (e.g., age, education) explained significant dropout variance (b=-0.14; p<.05). The aggregate z-score for study variables (e.g., compensating participants) was significant in explaining RE heterogeneity (b=-0.09; p<.05) and research design approached significance (b=-0.06; p<.1).
Conclusions: BC population exercise and PA studies experienced 16% dropout and 30% RE. Dropout decreased with higher levels of participant variables, and RE decreased with greater study funding and number of sites as well as with higher quality research designs. Given these findings, investigators can use evidence-based approaches for estimating and optimizing dropout and RE, such as considering the patient population, incentives, and use of a high quality research design.
Authors Disclosures: None.
Jeffrey Hoover– Graduate Student/Graduate Research Assistant, University of Kansas
Aqeel Alenazi– Lecturer, Prince Sattam Bin Abdulaziz University, Kansas City, Missouri
Caio Vinicius Messias Sarmento– Graduate student, University of Kansas Medical Center, Kansas City, Kansas
Shaima Alothman– NA, NA, riyadh, Ar Riyad
Mohammed Alshehri– Research assistance, KUMC, Lenexa, Kansas
Abdalghani Yahya– PhD. Candidate and an assistant professor, University of Kansas Medical Center/University of St. Augustine, Kansas City, Kansas
Jason Rucker– Clinical Assistant Professor, University of Kansas Medical Center, Kansas City, Kansas
Bader Alqahtani– Assistant Professor, Prince Sattam Bin Abdulaziz University, Pittsburgh, Pennsylvania
Patricia Kluding– Professor and Chair, University of Kansas Medical Center, Kansas City, Kansas