AbstractAimTo quantify recruitment, retention and differential retention rates and associated trial, participant and intervention characteristics in randomised controlled trials (RCTs) evaluating the effect of exercise therapy in people with multimorbidity.Data sourcesMEDLINE, EMBASE, CINAHL and CENTRAL from 1990 to April 20, 2020.Study selectionRCTs including people with multimorbidity comparing exercise therapy with a non-exposed comparator group reporting at least one of the following outcomes: physical function, health-related quality of life, depression symptoms, or anxiety symptoms.Data extraction and synthesisRecruitment rates (proportion of people randomised/proportion of people eligible), retention rates (proportion of people providing the outcomes of interest/proportion randomised) and differential retention rates (difference in proportion of people providing the outcomes in the intervention group and comparator group) were calculated. Meta-analysis using a random-effects model was used to estimate pooled proportions. Methodological quality was assessed using Cochrane ´Risk of Bias tool 2.0´ for individual studies, and the GRADE approach was used to assess the overall quality of the evidence.ResultsTwenty-three RCTs with 3363 people were included. The pooled prevalence for recruitment rate was 75% (95%CI 66 to 84%). The pooled prevalence for retention rate was 90% (95%CI 86 to 94%) at the end of the intervention (12 weeks; interquartile range (IQR) (12 to 12)). Meta-regression analyses showed that increasing age and including a higher proportion of people with hypertension was associated with lower retention rates. Retention rates did not differ between the intervention and comparator groups. The overall quality of the evidence was deemed very low.ConclusionThree in four eligible people with multimorbidity were randomised to RCTs using exercise therapy, of which nine out of 10 provided end of treatment outcomes with no difference seen between the intervention and comparison groups. However, the results must be interpreted with caution due to large differences between the included studies.Trial registrationClinicalTrials.govCRD42020161329. Registered on 28 April 2020.