Strategies to improve recruitment to randomised trials Authors: Adwoa Parker, Gloria Mongelli, Camila Piccolo-Lawrance, Elizabeth Coleman, Athanasios Gkekas, Han-I Wang, Shoba Dawson, Heidi Green, Rosalind Way, Arti Rai, Hanne Bruhn, Frank Sullivan, Cherish Boxall, Markus Chan, Paul Manson, Catherine Hewitt, Caroline Fairhurst, Mike Clarke, Shaun Treweek Published: 2026 ABSTRACT AbstractBackgroundRecruiting participants to randomised controlled trials (RCTs) is challenging. Identifying effective recruitment strategies would benefit health research: poor recruitment leads to underpowered trials, reducing the reliability of findings and increasing the risk of wasted resources, ethical concerns, and trial failure. Evidence to inform recruitment strategies is increasingly generated through Studies Within A Trial (SWATs), which are methodological studies embedded within host RCTs. This is an update of a review last published in 2018.ObjectivesPrimary: to quantify the effects of strategies to improve recruitment of participants to RCTs.Secondary: to evaluate recruitment strategies' cost‐effectiveness and impact on retention, and the equity, diversity, and inclusion (EDI) characteristics of recruited participants.Search methodsWe used MEDLINE, Embase, and six other databases to identify the studies included in the review. We also sought unpublished recruitment SWATs through social media and targeted email dissemination to trial methodology networks. The latest search date was 16 February 2023.Selection criteriaWe included randomised SWATs evaluating trial recruitment strategies embedded in healthcare and non‐healthcare trials. We excluded quasi‐randomised, hypothetical, questionnaire‐only, retention‐only, or clinician incentive studies.Data collection and analysisPrimary outcome: proportion of eligible participants or centres recruited.Secondary outcomes: cost‐effectiveness, retention rates, and EDI characteristics of included participants.We conducted random‐effects meta‐analysis for strategies evaluated in at least two studies; otherwise, we synthesised results narratively. We reported effects as risk differences (RDs) with 95% confidence intervals (CIs), and assessed between‐trial heterogeneity. We used GRADE to assess the certainty of evidence for the primary outcome. We expressed cost‐effectiveness as the incremental cost per additional participant recruited in pounds sterling (GBP).Main resultsWe identified 91 eligible studies (53 new to this update), providing 94 comparisons and involving at least 176,747 participants. Eighty‐one studies involved strategies aimed at trial participants, while 10 evaluated strategies aimed at recruiters. All were healthcare studies.We found 65 recruitment strategies; 49 were evaluated in a single study. Only five strategies were supported by high‐certainty evidence according to GRADE criteria, and we focus on these strategies in the summary below.Open‐label trials versus blinded, placebo trials. Open‐label trials recruited more participants than blinded trials (RD 10%, 95% CI 8% to 12%; 3 studies, 9004 participants), corresponding to approximately 10 additional participants per 100 approached. The studies involved mostly women in the UK and Estonia. No cost or retention data were reported.Telephone reminder versus no telephone reminder. Telephone reminders to people who did not respond to an initial postal invitation boosted recruitment by 6% (95% CI 3% to 9%; 2 studies, 1450 participants), in trials with low underlying recruitment (we are less certain for trials with over 10% recruitment). The studies involved people with a mean age of 58 years in Canada and Norway. No cost or retention data were reported.Recruitment primer letter versus no letter. Pre‐recruitment letters and leaflets designed to encourage participation made little or no difference to recruitment (absolute improvement 1%, 95% CI −1% to 2%; 2 studies, 5376 participants), and were associated with increased costs compared to not sending a primer (incremental cost: GBP 2.08). The studies involved mostly older white people in the UK and Ireland.Multimedia information via a digital link/QR code plus paper participant information leaflet (PIL) versus paper PIL alone. This made little or no difference to recruitment (absolute improvement 0%, 95% CI −1% to 1%; 7 studies, 11,612 participants) and retention (absolute improvement 0%, 95% CI −2% to 3%; 5 studies, 7403 participants), and increased costs compared to not including multimedia information (incremental cost: GBP 0.78). The studies involved people in the UK.Optimised, user‐tested PIL versus standard PIL. Optimising participant information leaflets (e.g. through user‐testing the leaflet with the target population to shape its content, format, and appearance) made little or no difference to recruitment: absolute improvement was 0% (95% CI 0% to 1%; 6 studies, 27,805 participants). The studies involved people in the UK. Only one study reported EDI data; participants were mostly older women. No cost or retention data were reported.We had moderate‐certainty evidence for 13 other strategies; confidence was often reduced because the results came from single studies. Seven strategies involved changes to how potential participants received information; four involved changes to trial conduct; one targeted the recruiter or recruitment site; and one tested non‐monetary incentives.We had much less confidence in the other 47 comparisons because the studies had design flaws, were single studies, or had very uncertain results.Costs were reported in only 17 of 91 studies. Strategy impact on retention was reported in 15 studies. All but one study (99%) were from high‐income countries. The most reported demographics were age (49 studies), sex (32 studies), gender (27 studies), and education level (16 studies).Authors' conclusionsThe evidence on strategies to improve trial recruitment remains broad but lacks depth. Of 65 strategies evaluated, only five were supported by high‐certainty evidence. Open‐label trial designs and telephone reminders to non‐responders increased recruitment, while optimised participant information leaflets, recruitment primer letters, and multimedia information provided alongside a paper participant information leaflet had little or no effect. Reporting of participant characteristics was poor, limiting assessment of equity, diversity, and inclusion across most studies.Evidence is heavily skewed toward high‐income countries. Future research must prioritise evaluations in low‐to‐middle‐income settings and consistently report cost, retention, and EDI outcomes.We strongly urge the methodology research community to strengthen the evidence base by prioritising replications of existing strategies over the development and testing of new ones.FundingNational Institute for Health and Care Research (Advanced Fellowship, Adwoa Parker, reference:NIHR302256).Health Research Board, Republic of Ireland, Evidence Synthesis Ireland (grant ESI‐2021‐001)RegistrationThis review updates an earlier Cochrane review, which was first published in 2002 and subsequently updated in 2007, 2010, and 2018. Previous versions of the review and their protocols are available at:https://doi.org/10.1002/14651858.MR000013.pub2https://doi.org/10.1002/14651858.MR000013.pub3https://doi.org/10.1002/14651858.MR000013.pub4https://doi.org/10.1002/14651858.MR000013.pub5https://doi.org/10.1002/14651858.MR000013.pub6 Summary of findings Summary of findings 1Open‐label trial versus blinded trialOpen RCT versus blinded RCTPatient or population: individuals eligible for a trial Settings: any Intervention: open‐label trial Comparison: blinded, placebo trialOutcomesIllustrative effects* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect with blinded trialEffect with open‐label trialNumber recruited1806 out of 45782162 out of 4426RD 10% (8% to 12%)9004 (3 studies)⊕⊕⊕⊕ HighCost: no data reportedRetention: no data reportedEDI: all studies were conducted in two high‐income countries. Participants were almost all women; their mean age, where reported, was 78 years. There was no information on sex, ethnicity, or socioeconomic status of participants.Proportion recruitedAs measuredaRR 1.26 (1.19 to 1.32)39 per 10050 per 100 (46 to 51)Lowb10 per 10013 per 100 (12 to 13)Moderateb30 per 10038 per 100 (36 to 40)Highb50 per 10063 per 100 (60 to 66)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The effect for the open‐label trial (and its 95% CI) is based on the assumed risk in the comparison group (blinded trial) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity, and inclusion; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. Open‐label trial versus blinded trial aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. Summary of findings 2Telephone reminder versus no telephone reminderTelephone reminder versus no telephone reminderPatient or population: individuals eligible for a trial Settings: any Intervention: telephone reminder Comparison: no telephone reminderOutcomesIllustrative comparative risks* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect with no telephone reminderEffect with telephone reminderNumber recruited46 out of 71490 out of 736RD 6% (3% to 9%)1450 (2 studies)⊕⊕⊕⊕ HighcCost: no data reportedRetention: no data reportedEDI: both studies were conducted in high‐income countries. Participant mean age, where reported, was 58 years. No data reported on sex, gender, ethnicity, or socioeconomic status of participants. Both included studies had very low baseline recruitment of < 10%.Proportion recruitedAs measuredaRR 1.94 (1.25 to 3.02)6 per 10012 per 100(8 to 18)Lowb10 per 10019 per 100 (13 to 30)Moderateb30 per 10058 per 100 (38 to 91)Highb50 per 10097 per 100 (63 to 100)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect with thetelephone reminder (and its 95% CI) is based on the assumed risk in the comparison group (no reminder) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity, and inclusion; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cThe evidence for this intervention comes entirely from trials with low (< 10%) underlying recruitment. When applied to trials with higher recruitment, we would downgrade the assessment of evidence certainty to moderate due to indirectness. Telephone reminder versus no telephone reminder aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cThe evidence for this intervention comes entirely from trials with low (< 10%) underlying recruitment. When applied to trials with higher recruitment, we would downgrade the assessment of evidence certainty to moderate due to indirectness. Summary of findings 3Recruitment primer letter/leaflet versus no letter/leafletRecruitment primer letter/leaflet versus no letter/leafletPatient or population: individuals eligible for a trial Settings: any Intervention: recruitment primer letter/leaflet Comparison: no letter/leafletOutcomesIllustrative effects* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect without letter/leafletEffect with primer letter/leafletNumber recruited344 out of 3421280 out of 1955RD 1% (−1% to 2%)5376 (2 studies)⊕⊕⊕⊕ HighCost: data from 1 studyc only: adding letter/primer, ICER = GBP 208.38; incremental costs = GBP 2.08; partial incremental effectiveness = 1% (RD 0.01, 95% CI −0.01 to 0.02)Retention: no data reportedEDI: both studies were conducted in high‐income countries. Participant mean age, where reported, was 77 years; 43% male participants, 57% female; ethnicity > 99% White. No data on socioeconomic status of participants.Proportion recruitedAs measuredaRR 1.30 (0.90 to 1.17)2 per 1002 per 100 (1 to 4)Lowb10 per 10010 per 100 (6 to 18)Moderateb30 per 10030 per 100 (17 to 54)Highb50 per 10050 per 100 (28 to 90)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect for the recruitment primer letter/leaflet (and its 95% CI) is based on the assumed risk in the comparison group (no letter/leaflet) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity, and inclusion; GBP: pounds sterling; ICER: incremental cost‐effectiveness ratio; RD: risk difference; RR: risk ratio.GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cData from Arundel 2017 (falls prevention). Recruitment primer letter/leaflet versus no letter/leaflet aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cData from Arundel 2017 (falls prevention). Summary of findings 4Multimedia participant information leaflet (PIL) (provided as link/QR code) plus paper PIL versus paper PIL aloneMultimedia information (provided as link/QR code) plus paper participant information leaflet (PIL) versus paper PIL alonePatient or population: individuals eligible for a trial Settings: any Intervention: multimedia information PIL (provided as link/QR code) plus paper PIL Comparison: paper PIL aloneOutcomesIllustrative effects* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness and EDI summaryEffect with paper PIL onlyEffect with multimedia information + paper PILNumber recruited516 out of 6029451 out of 5583RD 0% (−1% to 1%)11,612 (7 studies)⊕⊕⊕⊕ HighCost: data from 1 studyc only: multimedia information plus paper PIL was more costly (GBP 0.78 higher) and slightly less effective than paper PIL alone (RD −0.01, 95% CI −0.03 to 0.01); difference not statistically significant.EDI: two studies were conducted in one high‐income country (UK). No data reported on age, sex, gender, ethnicity or socioeconomic status of participants.Proportion recruitedAs measuredaRR 0.97 (0.83 to 1.13)9 per 1008 per 100 (7 to 10)Lowb10 per 10010 per 100 (8 to 11)Moderateb30 per 10029 per 100 (25 to 34)Highb50 per 10049 per 100 (42 to 57)Number retained306 out of 3480354 out of 3923RD 0% (−2% to 3%)7403 (5 studies)⊕⊕⊕⊕ High*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect for the multimedia information participant information (provided as link/QR code) alongside paper PIL (and its 95% CI) is based on the assumed risk in the comparison group (paper PIL alone) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; GBP: pounds sterling; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cJolly 2019 (START MMI PSM‐COPD) Multimedia participant information leaflet (PIL) (provided as link/QR code) plus paper PIL versus paper PIL alone aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cJolly 2019 (START MMI PSM‐COPD) Summary of findings 5Optimised participant information leaflet (PIL) (bespoke and user‐tested) versus usual PILOptimised participant information leaflet (PIL) (bespoke and user‐tested) versus usual PILPatient or population: individuals eligible for trial Settings: any Intervention: optimised PIL (bespoke and user‐tested) Comparison: usual PILOutcomesIllustrative comparative risks* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect with usual PILEffect with optimised PIL (bespoke & user‐tested)Willingness to participate/number recruited937 out of 13,7451007 out of 14,060RD 0% (0% to 1%)27,805 (6 studies)⊕⊕⊕⊕ HighCost: no data reportedRetention: no data reportedEDI: all six studies were conducted in one high‐income country (UK). Only one study reported EDI data: participants were mostly women with a mean age of 78 years. Other studies provided no information on the sex, gender, ethnicity, or socioeconomic status of participants.Proportion willing to participate/recruitedAs measuredaRR 1.01 (0.93 to 1.1)7 per 1007 per 100(7 to 8)Lowb10 per 10010 per 100 (9 to 11)Moderateb30 per 10030 per 100 (28 to 33)Highb50 per 10051 per 100 (47 to 55)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect with theoptimised PIL (and its 95% CI) is based on the assumed risk in the comparison group (usual PIL) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity, and inclusion; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. Optimised participant information leaflet (PIL) (bespoke and user‐tested) versus usual PIL aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. Summary of findings 6Brief participant information leaflet (PIL) versus usual PILBrief participant information leaflet (PIL) versus usual PILPatient or population: individuals eligible for a trial Settings: any Intervention: brief PIL Comparison: usual PILOutcomesIllustrative comparative risks* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect with usual PILEffect with brief PILNumber recruited749 out of 2287783 out of 2346RD 0% (−1% to 2%)4633 (2 studies)⊕⊕⊕⊝ ModeratecCost: no data reportedRetention: no data reportedEDI: both studies were conducted in one high‐income country (UK). Only one study reported demographic data: participant mean age was reported as about 43 years. About 75% of participants were female, 25% male. There was no information on ethnicity or socioeconomic status of participants.Proportion recruitedAs measuredaRR 0.99 (0.93 to 1.06)33 per 10032 per 100(31 to 35)Lowb10 per 10010 per 100 (9 to 11)Moderateb30 per 10030 per 100 (28 to 32)Highb50 per 10050 per 100 (47 to 53)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect with thebrief PIL (and its 95% CI) is based on the assumed risk in the comparison group (usual PIL) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for indirectness: Chen 2011 measured entry to pre‐randomisation phase, not recruitment. Brief participant information leaflet (PIL) versus usual PIL aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for indirectness: Chen 2011 measured entry to pre‐randomisation phase, not recruitment. Summary of findings 7Participant information leaflet (PIL) developed with feedback from users versus usual PILParticipant information leaflet (PIL) developed with feedback from users versus usual PILPatient or population: individuals eligible for a trial Settings: any Intervention: PIL developed with feedback from users Comparison: usual PILOutcomesIllustrative comparative risks* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect with usual PILEffect with PIL developed with feedback from usersNumber recruited401 out of 8358441 out of 8405RD 0% (0% to 1%)16,763 (2 studies)⊕⊕⊕⊝ ModeratecCost: no data reportedRetention: no data reportedEDI: both studies were conducted in one high‐income country (UK). One study reported demographic data: participant mean age was reported as about 78 years. About 45% of participants were male. There was no information on ethnicity or socioeconomic status of participants.Proportion recruitedAs measuredaRR 1.09 (0.96 to 1.25)5 per 1005 per 100(5 to 6)Lowb10 per 10011 per 100 (10 to 13)Moderateb30 per 10033 per 100 (29 to 38)Highb50 per 10055 per 100 (48 to 63)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect with aPIL developed with feedback from users (and its 95% CI) is based on the assumed risk in the comparison group (usual PIL) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity, and inclusion; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for indirectness: Chen 2011 measured entry to pre‐randomisation phase, not recruitment. Participant information leaflet (PIL) developed with feedback from users versus usual PIL aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for indirectness: Chen 2011 measured entry to pre‐randomisation phase, not recruitment. Summary of findings 8Providing information by video versus by standard means aloneVideo information versus standard information alonePatient or population: individuals eligible for trial Settings: any Intervention: video information Comparison: standard information (mixed but not including video)OutcomesIllustrative comparative risks* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect with standard informationEffect with video informationNumber recruited237 out of 830212 out of 804RD −4% (−23% to 15%)1634 (5 studies)⊕⊝⊝⊝ Very lowc,d,eCost: data reported from one study indicated that providing information by video is both more costly (GBP 41.58 more) and less effective (−2% reduction).Retention: no data reportedEDI: all studies were conducted in high‐income countries (UK and USA). Three studies reported participants' mean age to be 34 to 70 years, with 20% to 100% female participants. Two studies reported race, one reported ethnicity. Two studies included 30% to 40% African Americans. One study included 6% Asian participants, 8% Black/African/Caribbean participants, and 6% participants from other groups, alongside White participants. All socioeconomic levels were represented; 20% to 50% were from the most disadvantaged communities.Proportion recruitedAs measuredaRR 0.92 (0.45 to 1.90)29 per 10026 per 100(13 to 55)Lowb10 per 1009 per 100 (5 to 19)Moderateb30 per 10028 per 100 (14 to 57)Highb50 per 10046 per 100 (23 to 95)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect with thevideo information (and its 95% CI) is based on the assumed risk in the comparison group (standard information) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; GBP: pounds sterling; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for study limitations: Du 2008, Du 2009, and Rodgers 2019 were at unclear risk of bias; Mattock 2020 was at high risk of bias. dWe downgraded the certainty of the evidence by one level for inconsistency. Most studies suggested little or no difference in recruitment due to the intervention. However, point estimates in Hutchison 2007, Mattock 2020, and Rodgers 2019 favoured control, while point estimates in Du 2008 and Du 2009 favoured the intervention. eWe downgraded the certainty of the evidence by one level for imprecision, with a wide confidence interval crossing the line of no effect. Providing information by video versus by standard means alone aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for study limitations: Du 2008, Du 2009, and Rodgers 2019 were at unclear risk of bias; Mattock 2020 was at high risk of bias. dWe downgraded the certainty of the evidence by one level for inconsistency. Most studies suggested little or no difference in recruitment due to the intervention. However, point estimates in Hutchison 2007, Mattock 2020, and Rodgers 2019 favoured control, while point estimates in Du 2008 and Du 2009 favoured the intervention. eWe downgraded the certainty of the evidence by one level for imprecision, with a wide confidence interval crossing the line of no effect. Summary of findings 9Monetary incentive versus no incentiveMonetary incentive versus no incentivePatient or population: individuals eligible for a trial Settings: any Intervention: monetary incentive Comparison: no incentiveOutcomesIllustrative comparative risks* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect with no incentiveEffect with monetary incentiveNumber recruited235 out of 2351522 out of 2731RD 5% (1% to 9%)5082 (12 studies)⊕⊕⊝⊝ Lowc,dCost: the estimated ICER for monetary incentives versus no incentive was GBP 3052.96, with an incremental cost of GBP 152.65 and an incremental effectiveness of 5%.Retention: no data reportedEDI: all studies were conducted in high‐income countries (Canada, the UK, and USA). In nine studies, participants had a mean age of 47 to 66 years, with 40% to 58% male. Ethnicities were mixed: 34% Black/African American and 52% White. Participants spanned all income and deprivation levels, with 27% from the most deprived areas. Urban/rural status was not reported.Proportion recruitedAs measuredaRR 1.48 (1.00 to 2.19)10 per 10015 per 100(10 to 22)Lowb10 per 10015 per 100 (10 to 22)Moderateb30 per 10044 per 100 (30 to 66)Highb50 per 10074 per 100 (50 to 100)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect with afinancial incentive (and its 95% CI) is based on the assumed risk in the comparison group (no incentive) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; GBP: pounds sterling; ICER: incremental cost‐effectiveness ratio; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for study limitations: Fairhurst 2022, Halpern 2021a, and Halpern 2021b were at high risk of bias. dWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 73%). Monetary incentive versus no incentive aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for study limitations: Fairhurst 2022, Halpern 2021a, and Halpern 2021b were at high risk of bias. dWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 73%). Summary of findings 10Information provided over telephone versus information provided face‐to‐faceInformation provided over telephone versus information provided face‐to‐facePatient or population: individuals eligible for a trial Settings: any Intervention: information provided over the telephone Comparison: information provided face‐to‐faceOutcomesIllustrative effects* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect with face‐to‐face informationEffect with information provided over telephoneNumber recruited100 out of 12686 out of 123RD −9% (−18% to 1%)249 participants (2 studies)⊕⊕⊕⊝ ModeratecCost: no data reportedRetention: no data reportedEDI: both studies were conducted in high‐income countries (Denmark and USA). Most participants were aged 30 to 60 years. One study reported an even male‐female split; another included only women. In one study, 55% had a master’s degree or higher, while 11% had only basic schooling. Urban/rural status was not reported.Proportion recruitedAs measuredaRR 0.90 (0.80 to 1.02)79 per 10071 per 100 (63 to 81)Lowb10 per 1009 per 100 (8 to 10)Moderateb30 per 10027 per 100 (24 to 31)Highb50 per 10045 per 100 (40 to 51)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect for information provided over the telephone (and its 95% CI) is based on the assumed risk in the comparison group (information provided face‐to‐face) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; RCT: randomised controlled trial; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for imprecision (wide CI crossing the line of no effect). Information provided over telephone versus information provided face‐to‐face aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for imprecision (wide CI crossing the line of no effect). Summary of findings 11Multimedia participant information leaflet (PIL) (viewed in person) plus paper PIL versus paper PIL aloneMultimedia participant information leaflet (PIL) (viewed in person) plus paper PIL versus paper PIL alonePatient or population: individuals eligible for a trial Settings: any Intervention: multimedia information (MMI) PIL (viewed in person) plus paper PIL Comparison: paper PIL onlyOutcomesIllustrative effects* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness and EDI summaryEffect with paper PIL onlyEffect with multimedia information + paper PILNumber recruited7 out of 913 out of 14RD 17% (−89% to 123%)23 (3 studies)⊕⊝⊝⊝ Very lowc,dCost: no data reportedEDI: all studies were conducted in a high‐income country (UK), and involved non‐white male and female children aged between six and 13 years, living mainly in disadvantaged areas.Proportion recruitedAs measuredaRR 1.02 (0.72 to 1.44)78 per 10079 per 100 (56 to 100)Lowb10 per 10010 per 100 (7 to 14)Moderateb30 per 10031 per 100 (22 to 43)Highb50 per 10051 per 100 (36 to 72)Number retained5 out of 911 out of 14RD 28% (−79% to 136%)23 (3 studies)⊕⊝⊝⊝Very lowc,e*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect for the multimedia information participant information leaflet (PIL) (viewed in person) plus paper PIL (and its 95% CI) is based on the assumed risk in the comparison group (paper PIL only) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by two levels for imprecision (extremely wide CI crossing the line of no effect; RD = 0). dWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 64%). eWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 57%). Multimedia participant information leaflet (PIL) (viewed in person) plus paper PIL versus paper PIL alone aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by two levels for imprecision (extremely wide CI crossing the line of no effect; RD = 0). dWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 64%). eWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 57%). Summary of findings 12Multimedia participant information leaflet (PIL) (viewed in person) versus paper PILMultimedia information participant information leaflet (PIL) (viewed in person) versus paper PILPatient or population: individuals eligible for a trial Settings: any Intervention: multimedia information PIL (viewed in person) Comparison: paper PILOutcomesIllustrative effects* (95% CI)Relative effect (95% CI)No.of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness and EDI summaryEffect with paper PILEffect with multimedia information PILNumber recruited489 out of 736486 out of 692RD 4% (−1% to 8%)1428 (3 studies)⊕⊕⊕⊝ ModeratecCost: no data reportedEDI: all studies were conducted in a high‐income country (UK). Mean age range: 6 to 9 years. Female participants made up 39.3% to 76.2%. Two studies (BALANCE and FORCE) reported ethnicity. BALANCE: 42.9% White, 47.6% non‐White, 4.8% missing. FORCE: 80.2% White, 10.1% Asian, 5% Black, 2% mixed, 2.2% other, 0.5% not stated. Median deprivation scores (3.5 to 4.5) indicated moderate to substantial deprivation. Urban/rural status was not reported.Proportion recruitedAs measuredaRR 1.05 (0.98 to 1.13)66 per 10070 per 100 (65 to 75)Lowb10 per 10011 per 100 (10 to 11)Moderateb30 per 10032 per 100 (29 to 34)Highb50 per 10053 per 100 (49 to 57)Number retained484 out of 736483 out of 692RD 22% (−51% to 96%)1428 (3 studies)⊕⊝⊝⊝Very lowd,e*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect for the multimedia information participant information leaflet (viewed in person) (and its 95% CI) is based on the assumed risk in the comparison group (paper PIL) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for imprecision. dWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 64%). eWe downgraded the certainty of the evidence by two levels for imprecision (extremely wide CI crossing the line of no effect; RD = 0). Multimedia participant information leaflet (PIL) (viewed in person) versus paper PIL aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for imprecision. dWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 64%). eWe downgraded the certainty of the evidence by two levels for imprecision (extremely wide CI crossing the line of no effect; RD = 0). Summary of findings 13Multimedia information (provided as link/QR code) versus paper participant information leaflet (PIL)Multimedia information (provided as link/QR code) versus paper participant information leaflet (PIL)Patient or population: individuals eligible for a trial Settings: any Intervention: multimedia information (provided as link/QR code) Comparison: paper PILOutcomesIllustrative effects* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness and EDI summaryEffect with paper PILEffect with multimedia informationNumber recruited52 out of 11270 out of 117RD 13% (1% to 26%)229(2 studies)⊕⊕⊕⊝ ModeratecCost: no data reportedEDI: all studies were conducted in a high‐income country (UK). Participants’ mean age was 6 years and median age 12 years. Between 62.5% and 76.2% of participants were female. Median deprivation scores ranged from 3.5 to 4.5, indicating moderate to substantial deprivation. One study reported ethnicity: 42.9% White, 47.6% non‐White, and 4.8% missing. Urban/rural status was not reported.Proportion recruitedAs measuredaRR 1.29 (1.01 to 1.65)46 per 10060 per 100 (46 to 76)Lowb10 per 10013 per 100 (10 to 17)Moderateb30 per 10039 per 100 (30 to 50)Highb50 per 10065 per 100 (51 to 83)Number retained47 out of 11264 out of 117RD 13% (0% to 26%)229 (2 studies)⊕⊕⊕⊝Moderatec*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect for the multimedia information (provided as link/QR code) (and its 95% CI) is based on the assumed risk in the comparison group (paper participant information leaflet (PIL) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; RCT: randomised controlled trial; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for imprecision (wide CIs). Multimedia information (provided as link/QR code) versus paper participant information leaflet (PIL) aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for imprecision (wide CIs). Summary of findings 14Monetary incentives (change in value)Monetary incentives (change in value)Patient or population: individuals eligible for a trial Settings: any Intervention: higher‐value monetary incentives Comparison: lower‐value monetary incentivesOutcomesIllustrative effects* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness, retention, and EDI summaryEffect with lower incentive valueEffect with higher incentive valueNumber recruited180 out of 429196 out of 435RD 3% (−13% to 19%)864 (2 studies)⊕⊝⊝⊝ Very lowc, d,eCost: the estimated ICER of changing financial incentives was GBP 8214.83, with an incremental effectiveness of 3%.Retention: no data reportedEDI: both studies were conducted in a high‐income country (USA). The result applies to middle‐aged males and females with a range of ethnicities and socioeconomic statuses living in the USA. Urban/rural status was not reported.Proportion recruitedAs measuredaRR 1.08 (0.74 to 1.57)42 per 10045 per 100 (31 to 66)Lowb10 per 10011 per 100 (7 to 16)Moderateb30 per 10032 per 100 (22 to 47)Highb50 per 10054 per 100 (37 to 79)*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect for the higher‐value financial incentives (and its 95% CI) is based on the assumed risk in the comparison group (lower‐value financial incentives) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; GBP: pounds sterling; RD: risk difference; RR: risk ratioGRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for study limitations: both studies were at an overall high risk of bias. dWe downgraded the certainty of the evidence by one level for imprecision (very wide CI which crossed the line of no effect; RD = 0). eWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 83%). Monetary incentives (change in value) aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for study limitations: both studies were at an overall high risk of bias. dWe downgraded the certainty of the evidence by one level for imprecision (very wide CI which crossed the line of no effect; RD = 0). eWe downgraded the certainty of the evidence by one level for inconsistency (substantial heterogeneity; I2 = 83%). Summary of findings 15Non‐monetary incentive (branded pen versus no pen)Non‐monetary incentive (branded pen versus no pen)Patient or population: individuals eligible for a trial Settings: any Intervention: branded pen Comparison: no penOutcomesIllustrative effects* (95% CI)Relative effect (95% CI)No. of participants (studies)Certainty of the evidence (GRADE)Cost‐effectiveness and EDI summaryEffect with no penEffect with penNumber recruited238 out of 5140188 out of 4500RD 0% (−1% to 1%)9640 (3 studies)⊕⊕⊕⊝ ModeratecCost: trial branded pens resulted in a small increase in cost (incremental cost = GBP 0.45) with no difference in recruitment.EDI: one study reported age (mean 46.1, range 18 to 70) and sex (83% female). Most participants were British (87.1%), with small numbers from other White (5.3%), Asian (3.7%), mixed (1.2%), Black (0.6%), and Arab (0.6%) groups. No Gypsy/Traveller or African participants. Conducted in the UK, across urban and rural areas.Proportion recruitedAs measuredaRR 0.91 (0.74 to 1.11)5 per 1004 per 100 (4 to 6)Lowb10 per 1009 per 100 (7 to 11)Moderateb30 per 10027 per 100 (22 to 33)Highb50 per 10046 per 100 (37 to 56)Number retained49 out of 124227 out of 620RD 0% (−2% to 2%)1862 (1 study)⊕⊕⊕⊝Moderated*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the footnotes. The effect for the branded pen (and its 95% confidence interval) is based on the assumed risk in the comparison group (no pen) and the relative effect of the intervention (and its 95% CI).CI: confidence interval; EDI: equity, diversity and inclusion; GBP: pounds sterling; RD: risk difference; RR: risk ratioGRADE Working Group levels of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for study limitations: the risk of bias was high for Fairhurst 2022 and was unclear for White 2023. dWe downgraded the certainty of the evidence by one level for imprecision (single study). Non‐monetary incentive (branded pen versus no pen) aThis is the baseline recruitment measured in the included studies. bWe selected the low, moderate, and high illustrative recruitment levels of 10%, 30%, and 50% based on our prior experience with trial recruitment. cWe downgraded the certainty of the evidence by one level for study limitations: the risk of bias was high for Fairhurst 2022 and was unclear for White 2023. dWe downgraded the certainty of the evidence by one level for imprecision (single study). Background All trials need to recruit participants, but this is often a challenge. Poor recruitment can lead to an underpowered study, which may report clinically relevant effects to be statistically non‐significant. A non‐significant finding increases the risk that an effective intervention will be abandoned before its true value is established, or that there will be a delay in demonstrating this value while more trials or meta‐analyses are done. Underpowered trials also raise an ethical problem: researchers have exposed participants to an intervention with uncertain benefit but may still be unable to determine whether the intervention does more good than harm on completion of the trial. Poor recruitment can also lead to trial extension, higher costs, or trial abandonment. Investigations of recruitment differ in their estimates of the proportion of studies that achieve their recruitment targets, but it is likely that around half meet their target [1, 2, 3, 4, 5]. For example, McDonald and colleagues found that only 38 (31%) of 114 trials achieved their original recruitment target and 65 (53%) were extended [3]. A more recent replication of this work found that the number of trials meeting recruitment targets had increased to around 63% (245/388), but 32% (79/245) of these trials required an extension to their recruitment period to meet the target, meaning approximately 43% of trials recruit to target without an extension [1]. An additional 22% (86/388) of trials recruited to within 80% of their final recruitment target, with 36% (31/86) of these requiring an extension to their recruitment period [1]. Poor recruitment can have significant public health implications [6]. Evidence from the dexamethasone arm of the COVID‐19 RECOVERY trial suggests that delays in trial completion – such as those caused by slow recruitment – led to avoidable deaths in the UK [7]. Faster recruitment could lead to earlier trial completion, and if the intervention proves effective, earlier access could benefit more people. Had RECOVERY recruited 50% of eligible participants rather than the typical 15%, nearly 3000 additional lives could have been saved through earlier access to dexamethasone, which was shown to be an effective treatment. The economic implications may also be substantial: increasing recruitment from 15% to 50% at all sites could have delivered an incremental net benefit of 13.9 million pounds sterling (GBP) to the UK from a healthcare system perspective [8]. These findings highlight the vital importance of efficient trial processes – not only for saving lives but also for maximising the health and economic value of research. This systematic review updates our previous reviews [9, 10, 11]. In addition to updating the literature search, we adjusted the eligibility criteria by excluding quasi‐randomised and hypothetical trials (i.e. where people are asked to imagine whether they would take part in a trial that is not really happening, rather than being invited to a real trial). We also included three new outcomes: cost‐effectiveness of recruitment strategies; impact of recruitment strategies on participant retention; and equity, diversity, and inclusion (EDI) characteristics of the participants. Description of the methods being investigated Strategies to improve participant recruitment aim to increase the number of people who consent and enrol in a trial. These interventions can be implemented at three levels: the participant, the recruiter, or the overall trial management. Methods include changes to how potential participants are contacted or invited to join, strategies directed at recruiters (such as training or prompts), modifications to trial design or management (e.g. altering consent procedures), and changes to the information or materials used to present trial details. Objectives Primary: to quantify the effects of strategies to improve recruitment of participants to RCTs. Secondary: to evaluate recruitment strategies' cost‐effectiveness and impact on retention, and the equity, diversity, and inclusion (EDI) characteristics of recruited participants. Methods We followed guidance in the Methodological Expectations for Cochrane Intervention Reviews (MECIR) manual when undertaking the review [12], and PRISMA 2020 for reporting the review [13]. Differences between protocol and review Since the 2018 update, we amended the protocol in several ways to enhance methodological rigour and address emerging priorities in trial methodology research. We narrowed the inclusion criteria to include only randomised controlled trials. We excluded hypothetical and quasi‐randomised trials to ensure the meta‐analysis reflects high‐certainty, real‐world evidence. We also introduced three new secondary outcomes to provide a more comprehensive evaluation of recruitment strategies: retention rates – to assess downstream effects on trial completion; cost‐effectiveness – to provide practical resource implications; and equality, diversity, and inclusion (EDI) characteristics – to identify representation gaps. Criteria for considering studies for this review Types of studies We included randomised Studies Within A Trial (SWATs), including individually randomised and cluster‐randomised designs, that evaluated strategies to improve participant recruitment to randomised trials. Types of data We included any intervention that aimed to improve participant recruitment to a randomised trial, except incentives or disincentives for clinicians to recruit participants, which is covered by another Cochrane review [14]. We did not impose any restrictions on the number or types of participants included or the study setting. The interventions being studied could be directed at potential participants (e.g. patients being randomised to a trial), collaborators (e.g. clinicians recruiting patients for a trial), or others (e.g. research ethics committees). Examples of eligible interventions include: letters introducing the trial signed by influential people; alternative methods of providing information about the trial to potential participants; additional training for collaborators; monetary incentives for participants; telephone follow‐up after expressions of interest; and modifications to the trial design (e.g. preference designs). Types of methods We included randomised Studies Within A Trial (SWATs) of strategies to improve recruitment to randomised trials. These were not limited to healthcare trials since strategies that work in other fields (e.g. education, housing) could be applicable to healthcare settings. We excluded quasi‐randomised SWATs and recruitment strategies evaluated within quasi‐randomised host trials because these designs are at high risk of bias and contribute low‐certainty evidence. We also excluded studies conducted in hypothetical trial settings because participant decisions in hypothetical scenarios may not reflect real‐world decisions about trial participation. Additionally, given the growing body of evidence from evaluations embedded in real trials, we no longer consider their inclusion justified. We excluded studies focused on questionnaire response rates, trial retention, or incentives and disincentives for clinicians to recruit participants because these topics are addressed in separate Cochrane methodology reviews [14, 15, 16, 17, 18]. Types of outcome measures Primary outcomes Proportion of eligible individuals or centres recruited (i.e. randomised to the trial). Secondary outcomes We added the following outcomes for this update. Previous versions of this review did not include any secondary outcomes. Cost‐effectiveness of recruitment strategies, measured using the incremental cost‐effectiveness ratio (ICER), defined as the incremental cost per additional participant recruited. We defined incremental cost as the additional cost of one strategy compared with another, and the ICER as the additional cost per additional participant recruited.Impact of the recruitment strategy on retention rates. We defined retention as the proportion of participants randomised to each SWAT arm who were retained in the host trial (e.g. attended follow‐up visits or completed questionnaires). Where available, we used retention at the host trial's primary outcome time point; otherwise, we used the first reported retention measure.Participant equity, diversity, and inclusion (EDI) characteristics, as reported by trialists, to inform equity‐related assessments of the populations to whom the evidence applies. Cost‐effectiveness of recruitment strategies, measured using the incremental cost‐effectiveness ratio (ICER), defined as the incremental cost per additional participant recruited. We defined incremental cost as the additional cost of one strategy compared with another, and the ICER as the additional cost per additional participant recruited. Cost‐effectiveness of recruitment strategies, measured using the incremental cost‐effectiveness ratio (ICER), defined as the incremental cost per additional participant recruited. We defined incremental cost as the additional cost of one strategy compared with another, and the ICER as the additional cost per additional participant recruited. Impact of the recruitment strategy on retention rates. We defined retention as the proportion of participants randomised to each SWAT arm who were retained in the host trial (e.g. attended follow‐up visits or completed questionnaires). Where available, we used retention at the host trial's primary outcome time point; otherwise, we used the first reported retention measure. Impact of the recruitment strategy on retention rates. We defined retention as the proportion of participants randomised to each SWAT arm who were retained in the host trial (e.g. attended follow‐up visits or completed questionnaires). Where available, we used retention at the host trial's primary outcome time point; otherwise, we used the first reported retention measure. Participant equity, diversity, and inclusion (EDI) characteristics, as reported by trialists, to inform equity‐related assessments of the populations to whom the evidence applies. Participant equity, diversity, and inclusion (EDI) characteristics, as reported by trialists, to inform equity‐related assessments of the populations to whom the evidence applies. Search methods for identification of studies Electronic searches To update the review, we re‐screened the 68 previously included studies and searched the following electronic databases, with no language restrictions, for eligible studies published from 1 January 2015 onwards. We searched the following databases on 16 February 2023: MEDLINE and MEDLINE In‐Process (Ovid);Embase (Ovid);Science Citation Index and Social Sciences Citation Index (Web of Science). MEDLINE and MEDLINE In‐Process (Ovid); MEDLINE and MEDLINE In‐Process (Ovid); Embase (Ovid); Embase (Ovid); Science Citation Index and Social Sciences Citation Index (Web of Science). Science Citation Index and Social Sciences Citation Index (Web of Science). We added the following databases for this update: Applied Social Sciences Index and Abstracts (ASSIA) (searched to 31 December 2023);CINAHL Plus (Cumulative Index to Nursing and Allied Health Literature; EBSCO) (searched to 31 December 2023);ERIC (EBSCO) (searched to 31 December 2023);Online Resource for Research in Clinical triAls (ORRCA) (searched to 31 December 2019, when it was last updated);PsycINFO (EBSCO) (searched to February Week 2 2023). Applied Social Sciences Index and Abstracts (ASSIA) (searched to 31 December 2023); Applied Social Sciences Index and Abstracts (ASSIA) (searched to 31 December 2023); CINAHL Plus (Cumulative Index to Nursing and Allied Health Literature; EBSCO) (searched to 31 December 2023); CINAHL Plus (Cumulative Index to Nursing and Allied Health Literature; EBSCO) (searched to 31 December 2023); ERIC (EBSCO) (searched to 31 December 2023); ERIC (EBSCO) (searched to 31 December 2023); Online Resource for Research in Clinical triAls (ORRCA) (searched to 31 December 2019, when it was last updated); Online Resource for Research in Clinical triAls (ORRCA) (searched to 31 December 2019, when it was last updated); PsycINFO (EBSCO) (searched to February Week 2 2023). PsycINFO (EBSCO) (searched to February Week 2 2023). To ensure global coverage, we also consulted regional sources such as LILACS (Latin American and Caribbean Health Sciences Literature). We did not search the Cochrane Methodology Review (CMR) Group Specialised Register as it has not been updated since July 2012. However, we included eligible papers identified from the CMR Group Specialised Register for the previous review (last search date 11 February 2015) in this review update. Supplementary material 1 details the full search strategies for all databases. Searching other resources To identify unpublished and ongoing recruitment SWATs [19], we issued a call for information via social media and email. On 23 March 2023, we posted on X (formerly Twitter), inviting trial methodology researchers who had conducted randomised recruitment SWATs to share their findings. On 27 March 2023, we circulated a similar request through the Trial Forge SWAT Network, the UK Trial Managers’ Network, the UK Clinical Research Collaboration (CRC) Registered Trials Units Network, the Medical Research Council and National Institute for Health and Care Research (MRC‐NIHR) Trial Methodology Partnership, and the Health Research Board Trial Methodology Network in Ireland. We implemented a study‐integrity audit to identify retractions, expressions of concern, or corrigenda. We checked the digital object identifiers of all eligible studies against the Crossref/CrossMark API and the Retraction Watch database, and used this information to ensure that the results reflected the most up‐to‐date, corrected versions of each included study. Data collection and analysis We prepared a revised protocol for this updated review, including it as Supplementary material 6 to make it available alongside this review in the Cochrane Library. Selection of studies We downloaded the search results to Endnote reference management software for deduplication and then uploaded the records to Covidence to enable screening and data extraction [20]. Working independently, two review authors screened the titles and abstracts of all records retrieved from searches of the electronic databases. We resolved any disagreements by involving a third author. We retrieved the full texts of studies that appeared to meet the inclusion criteria. Two review authors independently assessed all potentially eligible studies to determine if they met the inclusion criteria. We resolved any disagreements through discussion or by involving a third author. Authors Adwoa Parker, Shaun Treweek, Frank Sullivan, Elizabeth Colement, Mike Clarke, Caroline Fairhurst, and Catherine Hewitt were lead authors or co‐authors of a number of studies included in this review. They were not involved in study eligibility decisions, data extraction, risk of bias assessment, or GRADE assessments for any included studies they authored. Data extraction and management Working independently, two review authors extracted data from each included study using a data extraction form designed for the purpose. We resolved any differences in data extraction through discussion. We extracted data on: recruitment strategy evaluated;country in which the study was carried out;population;study setting;host trial;randomisation method;number and proportion of participants in the SWAT intervention and comparator groups;participant characteristics, including age, sex, gender, ethnicity, socioeconomic status, and geographic location, using the PRO EDI tool [21];costs;proportion of participants or clusters retained. recruitment strategy evaluated; recruitment strategy evaluated; country in which the study was carried out; country in which the study was carried out; population; population; study setting; study setting; host trial; host trial; randomisation method; randomisation method; number and proportion of participants in the SWAT intervention and comparator groups; number and proportion of participants in the SWAT intervention and comparator groups; participant characteristics, including age, sex, gender, ethnicity, socioeconomic status, and geographic location, using the PRO EDI tool [21]; participant characteristics, including age, sex, gender, ethnicity, socioeconomic status, and geographic location, using the PRO EDI tool [21]; costs; costs; proportion of participants or clusters retained. proportion of participants or clusters retained. For the cost‐effectiveness analysis, we extracted cost data from newly included studies, where available. We also extracted cost data from previously included studies identified in a systematic review of economic evaluations conducted alongside SWATs [22]. For each study, we recorded the costs of the intervention and comparator strategies, and any reported incremental costs. Where incremental costs were not reported, we estimated them by attributing relevant unit costs to each strategy and calculating the difference. Assessment of risk of bias in included studies We assessed risk of bias in the included studies using the original Cochrane risk of bias tool [23], for consistency with previous versions of the review [11]. We adapted the assessment to reflect the methodological nature of the interventions evaluated. Working independently, two review authors performed a blinded risk of bias assessment of each study before comparing assessments. We resolved any discrepancies through discussion. We assessed random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other potential sources of bias. We judged each domain as low, high, or unclear risk of bias. We considered a study to be at low overall risk of bias if all domains were judged to be at low risk of bias, at high overall risk of bias if one or more domains were judged to be at high risk of bias, and otherwise at unclear risk of bias. We did not exclude studies judged to be at high overall risk of bias from the meta‐analyses (Supplementary material 4). However, because we have lower confidence in these studies' evidence, we do not discuss them individually in the Results or Discussion, except where their data contribute to a meta‐analysis and can be considered alongside evidence from studies at lower risk. We suggest that readers should not draw conclusions about intervention effects solely from the results of studies at high overall risk of bias. Rather, these studies may help identify interventions that warrant further evaluation using more rigorous methods. Methodological quality data for all included studies are presented in the risk of bias graph (Figure). Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies Measures of the effect of the methods We calculated risk differences (RDs) with 95% confidence intervals (CIs) to estimate the effects of recruitment strategies on recruitment. In accordance with guidance in the Cochrane Handbook for Systematic Reviews of Interventions, we estimated RDs for single‐study comparisons using fixed‐effect models with Mantel‐Haenszel weighting [24]. For comparisons involving multiple studies, we conducted random‐effects meta‐analyses using inverse‐variance weighting [24]. In the summary of findings tables, we report the effect of recruitment strategies using both risk ratios (RRs) and risk differences (RDs) with 95% CIs, to provide a relative effect measure alongside the absolute impact of the strategies. Unit of analysis issues For recruitment SWATs that randomised individuals and clusters, the unit of analysis was the participant. We included cluster‐randomised trials in meta‐analyses only when sufficient data were available to adjust for clustering (e.g. intra‐cluster correlation coefficients or cluster‐adjusted estimates). When cluster‐adjusted risk differences (RDs) or risk ratios (RRs) were not available, we used odds ratios (ORs) as the summary effect measure. Dealing with missing data We contacted study authors to request missing data needed for the risk of bias assessment, the number of participants randomised to each group, and the number of participants recruited. Assessment of heterogeneity We sought statistical evidence of heterogeneity of results using the Chi2 test for heterogeneity, and we quantified the degree of heterogeneity using the I2 statistic [25]. The I² is independent of the number of studies and reflects variation based on the treatment effect. Where we detected substantial heterogeneity, we informally investigated possible explanations and summarised the data using a random‐effects meta‐analysis. Assessment of reporting biases We assessed the potential for reporting bias for the primary outcome using a funnel plot when at least 10 studies were available for a meta‐analysis. Data synthesis We used the Online Resource for Recruitment research in Clinical triAls (ORRCA) project's taxonomy of recruitment interventions to classify the interventions included in this review [26]. We split one ORRCA category ('Recruitment information needs') into two to distinguish interventions aimed at the consent process from those aimed at more general participant information. Our seven categories are as follows. Category A: design: changes to the general design of the trial made specifically to increase recruitment.Category B: pre‐trial planning: work done before the trial starts (possibly in a separate study) to explicitly make it more likely that recruitment will be successful.Category C: trial conduct changes: initiatives implemented once the trial has started, including better ways of identifying participants, changes to how data are collected, changes to the type of data collected, and tailoring recruitment to different types of participants.Category D: modifications to the consent process: changes to the personnel obtaining consent, the timing of consent, what sort of consent information is presented and how it is presented.Category E: modifications to the information given to potential participants about the trial: including who provides it, when, where, what sort of information is presented, and how the information is presented.Category F: interventions aimed at the recruiter or recruitment site: any interventions aimed at the recruiter or recruitment site staff, such as changes to training.Category G: incentives: monetary and other incentives for participants (but not staff, which is covered in a separate review [14]). Category A: design: changes to the general design of the trial made specifically to increase recruitment. Category A: design: changes to the general design of the trial made specifically to increase recruitment. Category B: pre‐trial planning: work done before the trial starts (possibly in a separate study) to explicitly make it more likely that recruitment will be successful. Category B: pre‐trial planning: work done before the trial starts (possibly in a separate study) to explicitly make it more likely that recruitment will be successful. Category C: trial conduct changes: initiatives implemented once the trial has started, including better ways of identifying participants, changes to how data are collected, changes to the type of data collected, and tailoring recruitment to different types of participants. Category C: trial conduct changes: initiatives implemented once the trial has started, including better ways of identifying participants, changes to how data are collected, changes to the type of data collected, and tailoring recruitment to different types of participants. Category D: modifications to the consent process: changes to the personnel obtaining consent, the timing of consent, what sort of consent information is presented and how it is presented. Category D: modifications to the consent process: changes to the personnel obtaining consent, the timing of consent, what sort of consent information is presented and how it is presented. Category E: modifications to the information given to potential participants about the trial: including who provides it, when, where, what sort of information is presented, and how the information is presented. Category E: modifications to the information given to potential participants about the trial: including who provides it, when, where, what sort of information is presented, and how the information is presented. Category F: interventions aimed at the recruiter or recruitment site: any interventions aimed at the recruiter or recruitment site staff, such as changes to training. Category F: interventions aimed at the recruiter or recruitment site: any interventions aimed at the recruiter or recruitment site staff, such as changes to training. Category G: incentives: monetary and other incentives for participants (but not staff, which is covered in a separate review [14]). Category G: incentives: monetary and other incentives for participants (but not staff, which is covered in a separate review [14]). We conducted random‐effects meta‐analyses when at least two studies evaluated the same recruitment strategy. Unlike previous versions of this review, which used fixed‐effect models in the absence of substantial heterogeneity, we used random‐effects models throughout because we anticipated substantial heterogeneity across most comparisons. In accordance with updated guidance in the Cochrane Handbook for Systematic Reviews of Interventions [27], we estimated between‐study variance using the restricted maximum likelihood (REML) method. We calculated confidence intervals for the summary effect using the Wald‐type method by default. However, when the estimated between‐study variance was greater than zero and more than two studies contributed to a comparison, we used the Hartung‐Knapp‐Sidik‐Jonkman (HKSJ) method. To synthesise cost‐effectiveness evidence across studies within each comparison, we reported the incremental cost‐effectiveness ratio (ICER), defined as the additional/incremental cost of the intervention strategy compared with the control strategy per additional participant recruited. We measured incremental effectiveness using the RD, defined as the number of additional participants recruited, and defined incremental cost as the difference in cost between the intervention and control strategies. When all studies in a comparison reported cost data, the RD used in the cost‐effectiveness analysis matched the RD estimated from the pooled effectiveness results. When only a subset of studies reported cost data, we re‐estimated the RD using data from those studies only. For single‐study comparisons, we estimated RDs using fixed‐effect meta‐analysis with Mantel‐Haenszel weighting [27]. For comparisons involving multiple studies, we conducted random‐effects meta‐analyses using inverse‐variance weighting [27]. For comparisons with cost data from multiple studies, incremental costs were weighted according to the same weights applied in the estimation of the RD. This approach enabled us to estimate the overall incremental cost of each recruitment strategy. We were unable to synthesise incremental costs using single‐arm meta‐analysis methods because the included SWATs did not report cost variability data. We calculated ICERs only when both incremental cost and incremental effectiveness data were available for a given strategy. All ICERs are reported in pounds sterling (GBP) at 2023 price levels. For studies reporting cost data in GBP at price levels other than 2023, we adjusted costs to 2023 values using the UK Consumer Price Index (CPI) [28]. For studies reporting cost data in other currencies and price years, we first adjusted costs to 2023 values using the CPI for the relevant country [29], and then converted the figures to GBP using the UK integrated tariff exchange rates [30]. Subgroup analysis and investigation of heterogeneity We planned to explore the following factors in subgroup analyses for our primary outcome (proportion of eligible participants recruited). We assumed sufficient studies would be identified for these analyses and that these factors represented plausible explanations for any observed heterogeneity. Type of allocation concealment used to evaluate recruitment strategies (adequate versus inadequate or unclear).Setting of the study recruiting participants (e.g. primary versus secondary care; healthcare versus non‐healthcare settings).Type of health condition in healthcare studies (e.g. cancer versus lifestyle change).Design of the study recruiting participants (e.g. open‐label versus blinded studies, trials with placebo arms versus those without).Target group (e.g. ethics committees, clinicians, patients). Type of allocation concealment used to evaluate recruitment strategies (adequate versus inadequate or unclear). Type of allocation concealment used to evaluate recruitment strategies (adequate versus inadequate or unclear). Setting of the study recruiting participants (e.g. primary versus secondary care; healthcare versus non‐healthcare settings). Setting of the study recruiting participants (e.g. primary versus secondary care; healthcare versus non‐healthcare settings). Type of health condition in healthcare studies (e.g. cancer versus lifestyle change). Type of health condition in healthcare studies (e.g. cancer versus lifestyle change). Design of the study recruiting participants (e.g. open‐label versus blinded studies, trials with placebo arms versus those without). Design of the study recruiting participants (e.g. open‐label versus blinded studies, trials with placebo arms versus those without). Target group (e.g. ethics committees, clinicians, patients). Target group (e.g. ethics committees, clinicians, patients). However, there were too few studies to permit subgroup analysis except for one strategy (monetary incentive versus no incentive – detailed in the Effects of methods). Sensitivity analysis We did not perform sensitivity analyses because the limited number of replications for most strategies precluded a meaningful assessment of the robustness of the results. If the evidence base for specific recruitment strategies expands in future review updates, we will consider the feasibility of conducting sensitivity analyses. Certainty of the evidence assessment We used the GRADE framework to assess the certainty of the evidence for our primary outcome (proportion of eligible individuals or centres recruited), and where data were available, our secondary outcomes (cost‐effectiveness, retention, participant EDI characteristics) [31]. We present GRADE assessments for meta‐analysed comparisons in the corresponding summary of findings tables [32]. These tables compare specific recruitment strategies with standard practice or alternative interventions and report the primary and secondary outcomes. As all included studies were randomised SWATs, we started each GRADE assessment at high certainty and considered downgrading the evidence for risk of bias, inconsistency, indirectness, imprecision, and publication bias [31, 32]. For all comparisons, we assessed: risk of bias, downgrading by one level when studies were judged to have an overall high or unclear risk of bias. This differs from previous versions of the review, in which we automatically downgraded studies at high overall risk of bias by two levels;inconsistency, downgrading when there was substantial unexplained heterogeneity across studies;indirectness, downgrading when there were key differences across study interventions or methods;imprecision, downgrading when confidence intervals were wide or crossed the line of no effect (risk difference = 0);publication bias, downgrading when we detected evidence of reporting bias. risk of bias, downgrading by one level when studies were judged to have an overall high or unclear risk of bias. This differs from previous versions of the review, in which we automatically downgraded studies at high overall risk of bias by two levels; risk of bias, downgrading by one level when studies were judged to have an overall high or unclear risk of bias. This differs from previous versions of the review, in which we automatically downgraded studies at high overall risk of bias by two levels; inconsistency, downgrading when there was substantial unexplained heterogeneity across studies; inconsistency, downgrading when there was substantial unexplained heterogeneity across studies; indirectness, downgrading when there were key differences across study interventions or methods; indirectness, downgrading when there were key differences across study interventions or methods; imprecision, downgrading when confidence intervals were wide or crossed the line of no effect (risk difference = 0); imprecision, downgrading when confidence intervals were wide or crossed the line of no effect (risk difference = 0); publication bias, downgrading when we detected evidence of reporting bias. publication bias, downgrading when we detected evidence of reporting bias. For single‐study comparisons, we: did not assess inconsistency as it is inapplicable;downgraded by one level for imprecision due to sparse data and by an additional level when confidence intervals were wide and crossed the line of no effect. did not assess inconsistency as it is inapplicable; did not assess inconsistency as it is inapplicable; downgraded by one level for imprecision due to sparse data and by an additional level when confidence intervals were wide and crossed the line of no effect. downgraded by one level for imprecision due to sparse data and by an additional level when confidence intervals were wide and crossed the line of no effect. We used the GRADE assessments to inform the narrative summaries of intervention effects presented in the Results, in accordance with GRADE guidance [33]. Results Description of studies Results of the search We conducted the literature search on 15 and 16 February 2023, identifying 21,127 records. Additionally, we screened the 68 studies included in the previous version of the review, as well as 148 studies identified via reference checking and handsearching (Supplementary material 7). After deduplication, we screened 21,072 titles and abstracts, excluded 20,876 records, and retrieved 196 full‐text articles for detailed screening. A total of 91 studies (53 new to this update), providing 94 comparisons and reported across 78 articles, were eligible for inclusion (see the study flow diagram in Figure). PRISMA study flow diagram *The number of included studies does not match the number of eligible records because the reporting differs across the 78 included records, ultimately yielding a total of 94 comparisons. Four records reported a total of 17 unique SWATs not published elsewhere, resulting in a sub‐total of 91 SWATs. Three of these were 3‐arm SWATs, each contributing one additional comparison. Adding these three 3‐arm comparisons results in a total of 94 comparisons. Included studies Studies came from 10 countries; there were also four multinational studies, one conducted in Australia and New Zealand, one conducted in the UK and Ireland, one involving the UK, Netherlands, Belgium, Poland, and Greece, and one involving 19 countries. The UK and USA dominated, with 46 and 23 studies, respectively. The next largest contributor was Australia with six studies. The full breakdown is given in Table. 1Countries where the included studies took placeCountryNumber of studiesUK46USA26aAustralia6Canada3Multinational3 (one involving the UK and Ireland, one involving Australia and New Zealand, and one involving 19 countries)Estonia2France2Denmark1Norway1Sweden1Republic of Ireland1aThere were 23 included studies from the USA; however, three of these studies are three‐arm SWATs (two of which were reported in a single published paper). Countries where the included studies took place aThere were 23 included studies from the USA; however, three of these studies are three‐arm SWATs (two of which were reported in a single published paper). Eighty‐one studies involved strategy interventions aimed at trial participants, and 10 studies evaluated strategy interventions aimed at those recruiting participants (Supplementary material 8). At least 176,747 individuals were involved in 89 studies; it was not clear how many participants were recruited in two studies (Supplementary material 9). The figure of 176,747 includes individuals who were successfully recruited as well as those who were approached about recruitment but declined or did not respond. We describe the included studies in Supplementary material 2. The full data package for the review is available as Supplementary material 5. Equity, diversity, and inclusion characteristics of included participants (PRO‐EDI) Nearly all studies (99%) were conducted in Global North, high‐income countries [34, 35], with only one study (1%) including both high‐ and middle‐income settings. Just over half of the studies (53%) reported participants’ age. Fewer reported other characteristics: 35% reported sex, 29% reported gender, 14% reported ethnicity, and 13% reported race. Socioeconomic status was reported in 17% of studies using education, 15% using a deprivation index, and another 15% using other measures. Overall, few studies provided information on participant diversity beyond age. Table outlines the EDI characteristics of the included participants. See Supplementary material 11 for a detailed report of the EDI data presented in the SWATs. 2Reported EDI characteristics of participants in included studiesEquity, diversity, and inclusion (EDI) characteristicsNumber (%) of studies reporting this characteristic (of 91 studies)Location (country/countries of data collection and site co‐ordination)High‐income country (Global North) only90 (99)Mix of high‐ and middle‐income countries*1 (1)Age49 (54)Sex32 (35)Gender27 (30)Ethnicity13 (14)Race12 (13)Socioeconomic status: deprivation index14 (15)Socioeconomic status: education16 (18)Socioeconomic status: other14 (15)*Monaghan 2007 was undertaken in Australia, Canada, Czech Republic, Estonia, France, Germany, Hungary, India, Ireland, Italy, Lithuania, Malaysia, the Netherlands, New Zealand, the Philippines, Poland, Russia, Slovakia, and the United Kingdom. Reported EDI characteristics of participants in included studies *Monaghan 2007 was undertaken in Australia, Canada, Czech Republic, Estonia, France, Germany, Hungary, India, Ireland, Italy, Lithuania, Malaysia, the Netherlands, New Zealand, the Philippines, Poland, Russia, Slovakia, and the United Kingdom. Cost‐effectiveness We identified and extracted costing data from 17 of 91 studies (19%). Of these, 11 studies (64%) explicitly reported incremental costs including costing data, and eight (47%) explicitly reported direct estimates of incremental cost‐effectiveness ratios (ICERs) including costing data. Overall, cost‐effectiveness results, presented in terms of the ICER, were reported for nine of the 65 strategies (14%). The corresponding ICERs are presented in Table 7a, as part of Supplementary material 12, and detailed cost‐effectiveness data, along with the underlying calculations, are also available in Supplementary material 12. Impact of recruitment strategies on retention rates Thirteen of the 91 studies (14%) reported the impact of recruitment strategies on participant retention rates. Of these, four (31%) reported data at the host trial’s primary outcome time point, and nine (69%) reported data at other time points. Analyses of the impact of recruitment strategies on retention rates are available in Supplementary material 13. We followed the approach used by Gillies and colleagues in their 2021 Cochrane review on strategies to improve retention rates [18]: specifically, we used the retention rate at the primary outcome time point for the host trial, where reported. If these data were not reported, we used the retention rate for the first time point reported. Excluded studies As depicted in the flow diagram (Figure), we excluded 118 studies in this review update. This included 31 previously included studies that no longer met the eligibility criteria (28 studies were hypothetical and three were quasi‐randomised). Reasons for exclusions are listed in the flow diagram (Figure), and all studies excluded at the full‐text screening stage are documented in Supplementary material 3, with the reasons for exclusion. Risk of bias in included studies We rated 52 studies to have an overall low risk of bias, 25 to have an unclear risk, and 15 studies to have a high risk. For details, see the individual risk of bias tables in Supplementary material 2, as well as Figure and Figure for summaries of the risk of bias assessment. Risk of bias summary: review authors' judgements about each risk of bias item for each included study Effects of methods In Table, we present an overview of the included studies organised according to the seven intervention categories described in the Data synthesis section. We assigned studies to these categories using the available information and the authors’ stated focus. We acknowledge that both the distinctions between categories and the intervention descriptions leave room for subjective interpretation. In Table, we present an overview of the synthesis, the included studies' characteristics, and outcomes. 3Intervention categoriesStudyHost trial interventionType of participantsA–Design. This includes changes to the general design of the trial made specifically to increase recruitment. N = 7Avenell 2004Drug: vitamin D tabletPatients (adults): attending a fracture clinic or orthopaedic wardCooper 1997Drug/surgery: medical management or transcervical resection of the endometriumPatients (adults): first‐time attendees at a gynaecological clinicFowell 2006Drug: anti‐emetics only if symptomaticPatients (adults): cancer inpatients receiving palliative careHemminki 2004Drug: HRTPatients (adults): postmenopausal women considering HRTLitchfield 2005Device: alternative delivery systems (NovoPen and Innovo) for insulinPatients (probably adults): people with type 1 diabetesPaul 2011Drug: adjuvant treatmentPatients (probably adults): with colorectal cancerVeerus 2016Drug: postmenopausal hormone therapyPatients (adults): postmenopausal womenB–Pre‐trial planning. This includes work done before the trial starts (possibly in a separate study) that explicitly aims to increase recruitment success. N = 1Garvelink 2015Training: training interprofessional home‐care teams in shared decision‐makingPatients (adults): elderly patients receiving home careC–Trial conduct changes. This includes initiatives implemented once the trial has started, such as better ways of identifying participants, changes to how data are collected, changes to the type of data collected and tailoring recruitment to different types of participant. N = 6Bracken 2019Drug: testosterone for type 2 diabetesPatients (adults): obese or overweight men with prediabetes or type 2 diabetes and low testosteroneFree 2010Behaviour: mobile phone‐based smoking cessationHealthy volunteers (adults): smokersFree 2011Behaviour: mobile phone‐based smoking cessationHealthy volunteers (adults): smokersNystuen 2004Therapy: psychologist intervention for issues linked to psychological problems or musculoskeletal painPatients (adults): on sick leave receiving benefitsTreweek 2012Drug: antibiotic prescribingHealth professionals (adults): family doctorsWong 2013Screening: colorectal cancer screeningHealthy volunteers (adults): eligible for colorectal cancer screeningD–Modification to the consent form or process. This includes changes to personnel helping with consent, when consent is taken, what sort of consent information is presented and how it is presented. N = 5 (2 of which were excluded from the results)Abd‐Elsayed 2012 [136]*Drug or blood storage trialsPatients (adults): eligible for 1 of 3 trials, all of whom had substantial illness requiring major surgery (cardiac)Coyne 2003Drug: variousPatients (adults): eligible for cancer trialLipstein 2021 [137]*Drug: methotrexate for paediatric Crohn’s disease patientsPatients (children and young adults): paediatric Crohn’s disease patients.Trevena 2006Screening: colorectal cancerHealthy volunteers (adults): eligible for colorectal screeningWadland 1990Lifestyle: smoking cessationHealthy volunteers (adults): smokersE–Modification to the information given to potential participants about the trial. This includes who provides it, when, where, what sort of information is presented, and how the information is presented. N = 52 (4 of which were excluded from the results)Arundel 2017Therapy: podiatry intervention for falls preventionPatients (adults): older patients who have attended routine podiatry services or with a fear of fallingBergenmar 2014Drug: variousPatients (probably adults): eligible for cancer trialsBobb 2016Drug: oral antiseptic for preventing hospital‐acquired pneumoniaPatients (adults): emergency department patients expected to be admitted to hospitalBrierley 2012Therapy: cognitive behavioural therapyPatients (adults): depressionBrown 2015Lifestyle: lifestyle intervention for pregnant women with gestational diabetes or prediabetesPatients (adults): pregnant women with gestational diabetes or pre‐diabetesChen 2011UnclearPatients (probably adults): unclear what typeChow 2020Lifestyle: self‐management intervention to treat heart diseasePatients (adults): childhood cancer survivors at risk of heart disease or heart failureCockayne 2017 (START PIL REFORM)aDevice: orthosisPatients (adults): podiatryCrane 2016Lifestyle: weight gain preventionHealthy volunteers (adults): normal and overweight young adultsDear 2011 [138]*Information: access to cancer trials sitePatients (adults): have cancerDu 2008Cancer trials (unspecified)Patients (adults): lung cancerDu 2009Cancer trials (unspecified)Patients (adults): women with breast cancerDwyer 2022Therapy: cognitive occupation‐based programmePatients (adults): people with multiple sclerosis and cognitive difficultyFelicitas‐Perkins 2017 [139]*Unspecified: cancer trialsPatients (adults): Filipino patients eligible for an ongoing oncology trialFord 2004Screening: prostate, lung and colorectal cancer screeningHealthy volunteers (adults): men eligible for prostate, lung and colorectal cancer screeningFoss 2016VaccinationHealthy volunteers (adults): pregnant womenFracasso 2013Cancer trials (unspecified)Patients (adults): cancer (various)Hughes‐Morley 2016Therapy: training for mental health professionalsPatients (adults): adults with severe mental illnessHutchison 2007Cancer trials (unspecified)Patients (probably adults): cancer (various)Ives 2001Unclear but probably drugPatients (adults): people with HIVJolly 2019 (START MMI PSM‐COPD)bTherapy: telephone self‐management coachingPatients (adults): primary care patients with chronic obstructive pulmonary diseaseKamen 2018Cancer trials (unspecified)Patients (adults): oncology patients eligible for an ongoing oncology trialKendrick 2001Injury prevention trialHealthy volunteers (adults and children): familiesKimmick 2005Cancer trials (various)Patients (adults): cancer (various)Knapp 2020 (START PIS CASPER)aBehaviour: behavioural activation for sub‐threshold depressionHealthy volunteers (adults): older adults with sub‐threshold depressionKnapp 2023e (Thermic‐3)cTherapy: warm versus cold blood during open‐heart surgeryPatients (children): children undergoing open‐heart surgeryKnapp 2023a (BALANCE)cTherapy: customised movies viewed on a consolePatients (children): children with lazy eyeKnapp 2023b (BAMP)cTherapy: orthodontic treatmentPatients (children): children with an underbiteKnapp 2023c (CHAMP‐UK)cTherapy: eye drop therapyPatients (children): children with short‐sightednessKnapp 2023d (FORCE)cDevice: splint versus soft bandage immobilisationPatients (children): children with a torus fractureKnapp 2023f (UKALL)cTherapy: alternative chemotherapy dosingPatients (children and young adults): patients with lymphocytic cancerLarkey 2002Various targeting cardiovascular disease, cancer and osteoporosisHealthy volunteers: (adults) womenMan 2015 (START PIS Healthlines Depression)aTherapy: telephone support and self‐managementPatients (adults): cardiovascularMadurasinghe 2021 (START PIS HeLP Diabetes)aBehaviour: web‐based self‐management programmePatients (adults): people with type 2 diabetesMadurasinghe 2023c (START MMI HI‐Light)bTherapy: home interventions and light therapy for vitiligoPatients (children and adults): people with vitiligoMadurasinghe 2023d (START MMI seAFOod)bTherapy/drug: fish oil and aspirin for polyp preventionHealthy volunteers (adults): individuals at high risk of colon cancerMadurasinghe 2023a (START MMI GHT2000a)bBehaviour: web‐delivered exercise self‐referral toolPatients (adults): inactive adults with high normal blood pressure or hypertensionMadurasinghe 2023b (START MMI GHT2000b)bBehaviour: exercise programme referralPatients (adults): inactive adults with high normal blood pressure or hypertensionMattock 2020Behaviour: positive parenting interventionHealthy volunteers (children): parents of toddlers exhibiting behavioural problemsMcCaffery 2019Therapy: occupation therapy for falls preventionHealthy volunteers (adults): elderly people who had fallen or had a fear of fallingMiller 2021aDrug: vitamin D for falls preventionHealthy volunteers (adults): older people at risk of falling, with low vitamin DParis 2015Clinical trials (various)Patients (adults): adults being screened for an ongoing trialParker 2018 (START PIS ECLS)aScreening: a new cancer screening testHealthy volunteers (adults): smokers undergoing lung cancer screeningPaul 2014Screening: colorectal cancerHealthy volunteers (adults): colorectal cancer screeningRogers 2019Drug: comparing different treatments for goutPatients (adults): chronic gout patients with cardiovascular risk factorsSaif 2020Drug: Alzheimer’s disease treatment for Alzheimer's preventionHealthy volunteers (adults): adults at risk of developing Alzheimer’s diseaseSheridan 2020 (START PIS ISDR)aScreening: personalised risk‐based screening for diabetic retinopathyPatients (children and adults): diabetic patients attending diabetic retinopathy screeningSkinner 2019 [140]*Behaviour: educational videoPatients (adults): oncology patientsSimes 1986Unclear: cancerPatients (adults): cancerSundaresan 2017Therapy: alternative timings for radiotherapy after a radical prostatectomyPatients (adults): high‐risk prostate cancer patientsWhite 2023Therapy: medical appointments with a specially‐trained GPPatients (adults): patients with persistent, medically unexplained, physical symptomsWitham 2018 [141]*Drug: vitamin K supplementationHealthy volunteers (adults): older people at risk of fallsF–Interventions aimed at the recruiter or recruitment site. This includes anything aimed at the recruiter or recruitment site staff, such as changes to training. N = 9 (2 of which were excluded from the results)Agni 2022Surgery: high‐dose dual‐antibiotic cement versus single‐antibiotic cement during partial hip replacementPatients (adults): older adults with intracapsular hip fracture requiring acute surgeryEthier 2017 [142]*Drug: comparing on‐demand versus continuous use medication for gastric refluxPatients (adults): people with gastric reflux requiring protein pump inhibitorsIsaksson 2025Drug: antidepressant medication versus placebo for stroke recoveryPatients (adults): patients who have recently had a strokeJefferson 2018Therapy: cast treatment versus surgical fixationPatients (adults): patients with a wrist fractureLee 2017Therapy: pain educationStaff at primary care clinics (sites are target, not patients)Liénard 2006Drug: breast cancer treatmentStaff at breast cancer treatment centres (sites are target, not patients)Maxwell 2017Drug: restarting versus avoiding antiplatelet drugs following a strokePatients (adults): patients who had experienced a stroke due to intracerebral haemorrhageMonaghan 2007Unclear: diabetes managementStaff at clinical sites recruiting to a diabetes and vascular disease treatment trial (sites are target, not patients)Tilley 2012 [143]*Drug: Parkinson's diseaseNeurologists, primary care doctors and internists (adults)Tilley 2021Drug: pharmacological treatments for various conditions depending on host trial, namely atrial fibrillation, blood and marrow transplants, colon cancer prevention, and Parkinson’sPatients (adults): patients eligible for various host trials covering atrial fibrillation, blood and marrow transplants, colon cancer prevention, and Parkinson’sG–Incentives. Financial and other incentives for participants. N = 12Fairhurst 2022Lifestyle: yoga programmePatients (adults): patients with multiple chronic conditionsFree 2010Lifestyle: mobile phone‐based smoking cessationHealthy volunteers (adults): smokersGiguere 2015Behaviour: online educationHealthy volunteers (adults): family physiciansHalpern 2021a ($200)dBehaviour: varenicline and behavioural activation for smoking cessationPatients (adults): depressed smokersHalpern 2021b ($100)dDevice: gamification intervention with a social support partner to increase patient mobility after hospital dischargePatients (adults): hospital in‐patientsJennings 2015aeDrug: NSAIDPatients (adults): arthritisJennings 2015beDrug: hyperuricaemiaPatients (adults): symptomatic hyperuricaemiaJennings 2015ceDrug: hypertensionPatients (adults): hypertensionJennings 2015deDrug: hypertensionPatients (adults): hypertensionJennings 2015eeDrug: diuretic therapyPatients (adults): metabolic syndromeWhite 2023Therapy: medical appointments with a specially‐trained GPPatients (adults): patients with persistent, medically unexplained, physical symptomsWhiteside 2019Therapy: occupation therapy for falls preventionHealthy volunteers (adults): elderly people who had fallen or had a fear of fallingGP: general practice; HIV: human immunodeficiency virus; HRT: hormone replacement therapy; MMI: multimedia information; NSAID: non‐steroidal anti‐inflammatory drugs.*Studies excluded from the results because we judged them to be at high overall risk of bias, and they tested a comparison not evaluated in another study. aCockayne 2017 (START PIL REFORM); Knapp 2020 (START PIS CASPER); Madurasinghe 2021 (START PIS HeLP Diabetes); Man 2015 (START PIS Healthlines Depression); Parker 2018 (START PIS ECLS); and Sheridan 2020 (START PIS ISDR) were all from the MRC START PIS programme, a series of embedded trials evaluating the impact of optimised participant information sheets within multiple host trials. bJolly 2019 (START MMI PSM‐COPD); Madurasinghe 2023a (START MMI GHT2000a); Madurasinghe 2023b (START MMI GHT2000b); Madurasinghe 2023c (START MMI HI‐Light); and Madurasinghe 2023d (START MMI seAFOod) were all from the MRC START MMI programme, a series of embedded trials evaluating the impact of MMI on trial recruitment within multiple host trials. cKnapp 2023a (BALANCE); Knapp 2023b (BAMP); Knapp 2023c (CHAMP‐UK); Knapp 2023d (FORCE); Knapp 2023e (Thermic‐3); and Knapp 2023f (UKALL) were all from the TRECA (TRials Engagement in Children and Adolescents) study. This was a series of co‐ordinated trials evaluating the impact of modified participant information sheets on trial recruitment within multiple paediatric host trials. dHalpern 2021a ($200) and Halpern 2021b ($100) are from a single paper that describes two separate embedded 3‐arm recruitment trials. e Jennings 2015a, Jennings 2015b, Jennings 2015c, Jennings 2015d, and Jennings 2015e are from a single paper that describes five embedded recruitment trials. Intervention categories GP: general practice; HIV: human immunodeficiency virus; HRT: hormone replacement therapy; MMI: multimedia information; NSAID: non‐steroidal anti‐inflammatory drugs. *Studies excluded from the results because we judged them to be at high overall risk of bias, and they tested a comparison not evaluated in another study. aCockayne 2017 (START PIL REFORM); Knapp 2020 (START PIS CASPER); Madurasinghe 2021 (START PIS HeLP Diabetes); Man 2015 (START PIS Healthlines Depression); Parker 2018 (START PIS ECLS); and Sheridan 2020 (START PIS ISDR) were all from the MRC START PIS programme, a series of embedded trials evaluating the impact of optimised participant information sheets within multiple host trials. bJolly 2019 (START MMI PSM‐COPD); Madurasinghe 2023a (START MMI GHT2000a); Madurasinghe 2023b (START MMI GHT2000b); Madurasinghe 2023c (START MMI HI‐Light); and Madurasinghe 2023d (START MMI seAFOod) were all from the MRC START MMI programme, a series of embedded trials evaluating the impact of MMI on trial recruitment within multiple host trials. cKnapp 2023a (BALANCE); Knapp 2023b (BAMP); Knapp 2023c (CHAMP‐UK); Knapp 2023d (FORCE); Knapp 2023e (Thermic‐3); and Knapp 2023f (UKALL) were all from the TRECA (TRials Engagement in Children and Adolescents) study. This was a series of co‐ordinated trials evaluating the impact of modified participant information sheets on trial recruitment within multiple paediatric host trials. dHalpern 2021a ($200) and Halpern 2021b ($100) are from a single paper that describes two separate embedded 3‐arm recruitment trials. e Jennings 2015a, Jennings 2015b, Jennings 2015c, Jennings 2015d, and Jennings 2015e are from a single paper that describes five embedded recruitment trials. 4Overview of synthesis and included studiesStudyCountryHost trial areaStudy designInterventionsOutcomesAnalysisA–Design. This includes changes to the general design of the trial made specifically to increase recruitment (N = 7). Note: EDI data reporting was very limited for studies in this category; interpret results with caution regarding generalisability to under‐served groups.Avenell 2004UKFracture preventionRandomised trialInvestigated the effect of different trial designsProportion recruited to host trialMACooper 1997UKManagement strategies for heavy menstrual bleedingRandomised embedded trialInvestigated the effect of different trial designsProportion recruited to host trialMAFowell 2006UKPalliative careCluster‐randomised cross‐over SWATInvestigated the effect of different trial designsProportion recruited to host trialNarrative summaryHemminki 2004EstoniaPostmenopausal hormone therapyRandomised SWATInvestigated the effect of different design methodsProportion recruited to host trialMALitchfield 2005UKDelivery systems for insulinCluster‐randomised SWATInvestigated electronic data captureNumber of participants recruited to the trialMAPaul 2011UKAdjuvant treatment for colorectal cancerRandomised SWATInvestigated the effect of the randomisation time pointProportion recruited to host trialNarrative summaryVeerus 2016EstoniaPostmenopausal hormone therapyRandomised SWATInvestigated the effect of participant being unblinded to the host trial interventionProportion recruited to trialMAB–Pre‐trial planning. This includes work done before the trial starts (possibly in a separate study) that explicitly aims to increase recruitment success (N = 1). Note: EDI data were not reported for the study in this category; interpret results with caution regarding generalisability to under‐served groups.Garvelink 2015CanadaFrailtyCluster‐randomised cross‐over trialInvestigated pre‐engagement with potential trial sites during proposal developmentProportion recruited to host trialMAC–Trial conduct changes. This includes initiatives implemented once the trial has started, such as better ways of identifying participants, changes to how data are collected, changes to the type of data collected and tailoring recruitment to different types of participant (N = 6). Note: EDI data reporting was very limited for studies in this category; interpret results with caution regarding generalisability to under‐served groups.Bracken 2019AustraliaTestosterone for type 2 diabetesRandomised SWATInvestigated telephone call reminder versus SMS reminderProportion recruited to host trial, cost‐effectivenessMAFree 2010UKSmoking cessationRandomised SWATInvestigated whether including GBP 5 with invitation or sending SMS messages to potential participants increased recruitmentProportion recruited to host trialMAFree 2011 Smoking cessationRandomised SWATInvestigated effect of mentioning scarcity on recruitmentProportion recruited to host trialMANystuen 2004NorwayGetting people to return to workRandomised SWATInvestigated the effect of different telephone remindersProportion recruited to host trialMATreweek 2012UKAntibiotic prescribing by GPsRandomised SWATInvestigated use of different modes of invitation to take part in trialProportion recruited to host trialMAWong 2013CanadaColorectal cancerRandomised SWATInvestigated use of telephone reminders to non‐respondersProportion recruited to host trialMAD–Modification to the consent form or process. This includes changes to personnel helping with consent, when consent is taken, what sort of consent information is presented, and how it is presented (N = 5, 2 of which were excluded from the results). Note: EDI data reporting was very limited for studies in this category; interpret results with caution regarding generalisability to under‐served groups.Abd‐Elsayed 2012*USACardiac surgeryRandomised SWATInvestigated the use of different consent form presentations Intervention A: consent documents on heavy‐ weight cream‐coloured paper (20‐pound) and a blue folder Comparator: consent documents as photocopies stapled togetherProportion recruited to host trialNot presentedCoyne 2003USACancerCluster‐randomised SWATInvestigated the effect of different consent methodsProportion recruited to host trialNarrative summaryLipstein 2021*USAPaediatric Crohn’s diseaseCluster‐randomised cross‐over trialInvestigated enhanced pre‐consent discussion with aid of posterProportion recruited to host trialNot presentedTrevena 2006AustraliaColorectal cancerRandomised SWATInvestigated the effect of different trial information methodsProportion recruited to host trialMAWadland 1990USASmoking cessationRandomised SWATIntervention: consent form read out by researcherProportion recruited to host trialMAE–Modification to the information given to potential participants about the trial. This includes who provides it, when, where, what sort of information is presented, and how the information is presented (N = 52, 4 of which were excluded from the results). Note: EDI data reporting was very limited for studies in this category; interpret results with caution regarding generalisability to under‐served groups.Arundel 2017UK and IrelandFalls preventionRandomised SWATInvestigated a recruitment primer leafletProportion recruited to host trial, cost‐effectivenessMABergenmar 2014SwedenCancerRandomised SWATInvestigated use of audio recording to improve communication about the trialProportion recruited to host trialMABobb 2016USAEmergency department trialsRandomised SWATTelemedicine‐enabled research informed consent versus standard face‐to‐face consentProportion recruited to host trialMABrierley 2012UKDepressionRandomised SWATInvestigated effect of length of the participant information leaflet on recruitmentProportion recruited to host trialMABrown 2015USALifestyle intervention for gestational diabetes or pre‐diabetesRandomised SWATInvestigated a targeted invitation letter for minority groupsProportion recruited to host trialMAChen 2011UKClinical area unclearRandomised SWATInvestigated different versions of the participant information leaflet (PIL)Proportion recruited to pre‐randomisation phase of host trialMAChow 2020USASelf‐management for coronary heart disease following paediatric cancer treatment3‐arm randomised SWATInvestigated shortened introductory packets with simplified consentProportion recruited to host trialMACockayne 2017 (START PIL REFORM) aUKFalls preventionRandomised SWATInvestigated two different versions of the participant information leaflet (PIL): optimised PIL (bespoke and user‐tested) or a template‐developed PILProportion recruited to host trialMACrane 2016USABehavioural intervention for weight gain preventionRandomised SWATInvestigated gender‐targeted recruitment postcardsProportion recruited to host trialMADear 2011*AustraliaCancer (unspecified)Cluster‐randomised cross‐over trialInvestigated whether information provided through a website improved recruitmentSelf‐reported (by participant) recruitment to a trialNot presentedDu 2008USALung cancerRandomised SWATInvestigated the effect of different methods of providing information about the trialProportion recruited to host trialMADu 2009USABreast cancerRandomised SWATIntervention: 18‐minute video. The video presents an overview of phase I, II and III clinical trials and the importance of cancer clinical research to society. The video addresses common concerns regarding clinical trials and cancer treatment from the patient's perspective, such as side effects, expected risks and benefits, eligibility criteria, the enrolment process, and treatment costs.Enrolment in therapeutic trialsMADwyer 2022Republic of IrelandCognitive rehabilitation intervention for multiple sclerosisRandomised SWATInvestigated patient‐designed‐and‐informed PILProportion recruited to host trial, proportion retained at 6 months.MAFelicitas‐Perkins 2017*USACancer (unspecified)Randomised SWATInvestigated culturally‐tailored educational DVD about clinical trialsProportion recruited to host trialNot presentedFord 2004USACancerRandomised SWATInvestigated the effect of different trial information and consent methodsProportion recruited to host trialMAFoss 2016DenmarkVaccinationRandomised SWATInvestigated the effect of different trial information and consent methodsProportion recruited to host trialMAFracasso 2013USACancerRandomised SWATInvestigated coaching as a way of improving recruitmentProportion recruited to host trialNarrative summaryHughes‐Morley 2016USACare planning training for community mental health teamsCross‐factorial, cluster‐randomised SWATInvestigated leaflet advertising patient and public involvementProportion recruited to host trialMAHutchison 2007UKCancer trials (unspecified)Randomised SWATInvestigated providing trial information by videoProportion recruited to host trialMAIves 2001UKHIVRandomised SWATInvestigated the effect of different trial information methodsProportion recruited to host trialMAJolly 2019 (START MMI PSM‐COPD)bUKCOPDCluster‐randomised cross‐over trialInvestigated multimedia trial informationProportion recruited to host trial, proportion retained at 6 and 12 months, intervention costsMAKamen 2018USAOncologyRandomised SWATInvestigated multimedia trial informationProportion recruited to host trialMAKendrick 2001UKInjury prevention, recruiting family unitsCluster‐randomised cross‐over trialInvestigated the effect of different trial information methodsProportion recruited to host trial, proportion retained at 6 and 12 months, intervention costsMAKimmick 2005USACancerCluster‐randomised cross‐over trialInvestigated the effect of different trial information methodsProportion recruited to host trialNarrative summaryKnapp 2020 (START PIS CASPER)bUKBehavioural activation for depressionRandomised SWATInvestigated optimised participant information leaflet (bespoke and user‐tested)Proportion recruited to host trial, proportion retained.MAKnapp 2023e (Thermic‐3)cUKPaediatric congenital heart defects3‐arm randomised SWATInvestigated multimedia trial informationProportion recruited to host trial, proportion retained at 3 months.MAKnapp 2023a (BALANCE)cUKPaediatric lazy eye3‐arm randomised SWATInvestigated multimedia trial informationProportion recruited to host trial, proportion retained at 16 weeks.MAKnapp 2023b (BAMP)cUKPaediatric underbite3‐arm randomised SWATInvestigated multimedia trial informationProportion recruited to host trial, proportion retained at 18 monthsMAKnapp 2023c (CHAMP‐UK)cUKPaediatric short‐sightedness3‐arm randomised SWATInvestigated multimedia trial informationProportion recruited to host trial, proportion retained at 6 monthsMAKnapp 2023d (FORCE)cUKPaediatric wrist fractureCluster‐randomised cross‐over trialInvestigated multimedia trial informationProportion recruited to host trial, proportion retained at 6 weeksMAKnapp 2023f (UKALL)cUKPaediatric and adolescent leukaemia or lymphoma3‐arm randomised SWATInvestigated multimedia trial informationProportion recruited to host trial, proportion retained at 3 monthsMALarkey 2002USAUnclearCluster‐randomised cross‐over trialInvestigated the effect of different methods of training lay advocates for trialsProportion recruited to host trialNarrative summaryMan 2015 (START PIS Healthlines Depression)aUKExtra support for depressionRandomised SWATInvestigated an alternative way of presenting patient information materialsProportion recruited to host trialMAMadurasinghe 2021 (START PIS HeLP Diabetes)aUKType 2 diabetes self‐managementRandomised SWATInvestigated optimised participant information leaflet (bespoke and user‐tested)Proportion recruited to host trialMAMadurasinghe 2023c (START MMI HI‐Light)bUKVitiligoRandomised coordinated SWAT (across 5 host trials)Investigated multimedia trial informationProportion recruited to host trial, proportion retained at 9 monthsMAMadurasinghe 2023d (START MMI seAFOod)bUKBowel cancer preventionRandomised coordinated SWAT (across 5 host trials)Investigated multimedia trial informationProportion recruited to host trial, proportion retained at 12 monthsMAMadurasinghe 2023a (START MMI GHT2000a)bUKExercise for high blood pressureRandomised coordinated SWAT (across 5 host trials)Investigated multimedia trial informationProportion recruited to host trialMAMadurasinghe 2023b (START MMI GHT2000b)bUKExercise for high blood pressureRandomised coordinated SWAT (across 5 host trials)Investigated multimedia trial informationProportion recruited to host trialMAMattock 2020UKPaediatric behavioural difficultiesRandomised SWATInvestigated providing information by videoProportion recruited to host trialMAMcCaffery 2019UKFalls preventionRandomised SWATInvestigated a handwritten personalised invitation letterProportion recruited to host trial, proportion retained at 3 monthsMAMiller 2021aUSAFalls prevention3‐arm randomised SWATInvestigated different wording of risk of underlying conditionProportion recruited to host trialMAParis 2015FranceMultiple host trials in different clinical areas, including heart surgery, dementia, diabetes, and sleep disordersRandomised SWATInvestigated simplified PILsProportion recruited to host trialMAParker 2018 (START PIS ECLS)aUKLung cancer diagnosisRandomised SWATInvestigated optimised participant information brochure (bespoke and user‐tested)Proportion recruited to host trialMAPaul 2014AustraliaColorectal cancerRandomised SWATInvestigated pre‐recruitment primer letterProportion recruited to host trialMARogers 2019UKArthritisRandomised SWATInvestigated DVD with invitation packProportion recruited to host trial, cost‐effectiveness.MASaif 2020USAAlzheimer's diseaseRandomised SWATInvestigated web‐based courseProportion recruited to host trialMASheridan 2020 (START PIS ISDR)aUKDiabetic retinopathy screeningCluster‐randomised cross‐over trialInvestigated optimised participant information leaflet (bespoke and user‐tested)Proportion recruited to host trialMASkinner 2019*AustraliaCancerRandomised SWATInvestigated the effect of different consent methodsProportion recruited to host trialMASimes 1986USACancer (unspecified)Randomised SWATInvestigated in‐office video viewingProportion recruited to host trialNot presentedSundaresan 2017Australia and New ZealandRadiotherapy for prostate cancerRandomised SWATInvestigated decision aidProportion recruited to host trialMAWhite 2023UKCommunity‐based clinic for patients with persistent, medically unexplained, physical symptoms2x2 factorial randomised SWATInvestigated brief PIL and trial‐branded penProportion recruited to host trial, cost‐effectiveness.MAWitham 2018*UKFalls preventionRandomised SWATInvestigated information sheet with different wordingProportion recruited to host trialNot presentedF–Interventions aimed at the recruiter or recruitment site. This includes anything aimed at the recruiter or recruitment site staff, such as changes to training (N = 9, 2 of which were excluded from the results). Note: EDI data were not reported for studies in this category; interpret results with caution regarding generalisability to under‐served groups.Agni 2022UKPartial hip replacement following hip fracture2x2 factorial cluster‐randomised SWATInvestigated additional support and/or digital nudges for Trainee Principal Investigators (TPI)Total number of patients randomised, from each site, in their first 6 months of trial recruitmentNarrative summaryEthier 2017*UK, Netherlands, Belgium, Poland and GreeceGastric refluxCluster‐randomised cross‐over trialInvestigated software for automatically flagging eligible patients for a trial during GP appointmentsProportion recruited to host trial (only reported for 2 out of 5 countries)Not presentedIsaksson 2025SwedenStroke recoveryCrossover cluster‐randomised SWATInvestigated teleconference with recruiting centres and commitment contractsNumber of included subjects before and after interventionNarrative summaryJefferson 2018UKCast versus surgery for wrist fractureCluster‐randomised cross‐over trialInvestigated remote site initiation visitsProportion recruited to host trial, proportion retained at each follow‐up time point including 12 months (host trial primary endpoint), intervention costsMALee 2017AustraliaRecruiting GP practices to low back pain trialCluster‐randomised cross‐over trialInvestigated the use of a teaser campaign to increase recruitment of clinical centresProportion of clinics recruited to host trialMALiénard 2006FranceBreast cancerCluster‐randomised cross‐over trialInvestigated the effect of organising visits by the trial co‐ordination team to centres participating in a multicentre trialProportion recruited to host trialNarrative summaryMaxwell 2017UKAntiplatelet drugs following strokeStepped‐wedge cluster randomised SWATInvestigated a recruitment co‐ordinator interventionNumber of patients randomised into host trial per month per siteNarrative summaryMonaghan 200719 countriesDiabetesCluster‐randomised cross‐over trialInvestigated the effect of different levels of communication between the trial co‐ordination team and participating sitesProportion recruited to host trialNarrative summaryTilley 2012*USAParkinson’s DiseaseCluster‐randomised cross‐over trialInvestigated effect of a recruitment coordinatorProportion recruited to host trialNot presentedTilley 2021USAVarious: atrial fibrillation, blood and marrow transplants, colon cancer prevention, Parkinson’sCluster‐randomised cross‐over trialInvestigated trust‐based educational programmeProportion recruited to host trialMAG–Incentives. Financial and other incentives for participants (N = 12). Note: EDI data reporting was very limited for studies in this category; interpret results with caution regarding generalisability to under‐served groups.Fairhurst 2022UKYoga for older adults with multimorbidities2x2 factorial cluster‐randomised SWATInvestigated trial‐branded pen and unconditional GBP 5 incentive enclosed in trial recruitment packProportion recruited to host trial, cost‐effectivenessMAFree 2010UKSmoking cessationRandomised SWATInvestigated whether including GBP 5 with invitation or sending SMS messages to potential participants increased recruitmentProportion recruited to trialMAGiguere 2015CanadaRecruiting family physicians to a medical education trialRandomised SWATInvestigated CAD 300 monetary incentiveProportion recruited to host trialMAHalpern 2021a ($200)dUSASmoking cessation3‐arm randomised SWATInvestigated higher versus lower monetary incentive versus nothingProportion recruited to host trialMAHalpern 2021b ($100)dUSAAmbulation3‐arm randomised SWATInvestigated higher versus lower monetary incentive versus nothingProportion recruited to host trialMAJennings 2015aeUKArthritisRandomised coordinated SWAT (across 5 host trials)Investigated effect of GBP 100 financial incentive on recruitmentProportion recruited to host trial, cost‐effectivenessMAJennings 2015beUKHyperuricemiaRandomised coordinated SWAT (across 5 host trials)Investigated effect of GBP 100 financial incentive on recruitmentProportion recruited to host trial, cost‐effectivenessMAJennings 2015ceUKHypertensionRandomised coordinated SWAT (across 5 host trials)Investigated effect of GBP 100 financial incentive on recruitmentProportion recruited to host trial, cost‐effectivenessMAJennings 2015deUKHypertensionRandomised coordinated SWAT (across 5 host trials)Investigated effect of GBP 100 financial incentive on recruitmentProportion recruited to host trial, cost‐effectivenessMAJennings 2015eeUKHypertensionRandomised coordinated SWAT (across 5 host trials)Investigated effect of GBP 100 financial incentive on recruitmentProportion recruited to host trial, cost‐effectivenessMAWhite 2023UKCommunity‐based clinic for patients with persistent, medically unexplained, physical symptoms2x2 factorial randomised SWATInvestigated brief PIL and trial‐branded penProportion recruited to host trial, cost‐effectiveness.MAWhiteside 2019UKFalls preventionRandomised SWATInvestigated branded pen with invitation packProportion recruited to host trial, costs, proportion retained at 3 months.MACAD: Canadian dollars; EDI: equity, diversity, and inclusion; GBP: pounds sterling; GP: general practice; HIV: human immunodeficiency virus; HRT: hormone replacement therapy; MA: meta‐analysis; MMI: multimedia information; NSAID: non‐steroidal anti‐inflammatory drugs; PIL: participant information leaflet; SWAT: Study Within A Trial*Studies excluded from the results sections because we judged them to be at overall high risk of bias studies, and they tested a comparison not evaluated in another study. aCockayne 2017 (START PIL REFORM); Knapp 2020 (START PIS CASPER); Madurasinghe 2021 (START PIS HeLP Diabetes); Man 2015 (START PIS Healthlines Depression); Parker 2018 (START PIS ECLS); and Sheridan 2020 (START PIS ISDR) were all from the MRC START PIL programme, a series of embedded trials evaluating the impact of optimised participant information leaflets (bespoke & user‐tested) within multiple host trials. bJolly 2019 (START MMI PSM‐COPD); Madurasinghe 2023a (START MMI GHT2000a); Madurasinghe 2023b (START MMI GHT2000b); Madurasinghe 2023c (START MMI HI‐Light); and Madurasinghe 2023d (START MMI seAFOod) were all from the MRC START MMI programme, a series of embedded trials evaluating the impact of MMI on trial recruitment within multiple host trials. cKnapp 2023a (BALANCE); Knapp 2023b (BAMP); Knapp 2023c (CHAMP‐UK); Knapp 2023d (FORCE); Knapp 2023e (Thermic‐3); and Knapp 2023f (UKALL) were all from the TRECA (TRials Engagement in Children and Adolescents) study. This was a series of co‐ordinated trials evaluating the impact of modified participant information sheets on trial recruitment within multiple paediatric host trials. dHalpern 2021a ($200) and Halpern 2021b ($100) are from a single paper that describes two separate embedded 3‐arm recruitment trials. e Jennings 2015a, Jennings 2015b, Jennings 2015c, Jennings 2015d, and Jennings 2015e are from a single paper that describes five embedded recruitment trials. Overview of synthesis and included studies CAD: Canadian dollars; EDI: equity, diversity, and inclusion; GBP: pounds sterling; GP: general practice; HIV: human immunodeficiency virus; HRT: hormone replacement therapy; MA: meta‐analysis; MMI: multimedia information; NSAID: non‐steroidal anti‐inflammatory drugs; PIL: participant information leaflet; SWAT: Study Within A Trial *Studies excluded from the results sections because we judged them to be at overall high risk of bias studies, and they tested a comparison not evaluated in another study. aCockayne 2017 (START PIL REFORM); Knapp 2020 (START PIS CASPER); Madurasinghe 2021 (START PIS HeLP Diabetes); Man 2015 (START PIS Healthlines Depression); Parker 2018 (START PIS ECLS); and Sheridan 2020 (START PIS ISDR) were all from the MRC START PIL programme, a series of embedded trials evaluating the impact of optimised participant information leaflets (bespoke & user‐tested) within multiple host trials. bJolly 2019 (START MMI PSM‐COPD); Madurasinghe 2023a (START MMI GHT2000a); Madurasinghe 2023b (START MMI GHT2000b); Madurasinghe 2023c (START MMI HI‐Light); and Madurasinghe 2023d (START MMI seAFOod) were all from the MRC START MMI programme, a series of embedded trials evaluating the impact of MMI on trial recruitment within multiple host trials. cKnapp 2023a (BALANCE); Knapp 2023b (BAMP); Knapp 2023c (CHAMP‐UK); Knapp 2023d (FORCE); Knapp 2023e (Thermic‐3); and Knapp 2023f (UKALL) were all from the TRECA (TRials Engagement in Children and Adolescents) study. This was a series of co‐ordinated trials evaluating the impact of modified participant information sheets on trial recruitment within multiple paediatric host trials. dHalpern 2021a ($200) and Halpern 2021b ($100) are from a single paper that describes two separate embedded 3‐arm recruitment trials. e Jennings 2015a, Jennings 2015b, Jennings 2015c, Jennings 2015d, and Jennings 2015e are from a single paper that describes five embedded recruitment trials. We report the results of studies rated as being at low or unclear risk of bias here. The full list of strategies (irrespective of risk of bias) is given in Supplementary material 8; participant numbers per study are presented in Supplementary material 9; and the data and analyses are in Supplementary material 10. We produced summary of findings (SoF) tables for all strategies investigated by at least two studies, for a total of 15 SoF tables (Table; Table; Table; Table; Table; Table; Table; Table; Table; Table; Table; Table; Table; Table; Table). Category E strategies (i.e. studies investigating modifications to the information given to potential participants about the trial) dominated. We detected substantial heterogeneity in four comparisons. Whilst we planned to undertake subgroup analyses to explore plausible explanations for heterogeneity, there were too few studies to allow subgroup analyses for three of these comparisons. We conducted a subgroup analysis for the comparison of monetary incentives to no incentives, exploring the effect of the type of allocation concealment used, which is presented as Analysis 63.2 and described below. Category A: design Overview Seven studies (9625 participants) evaluated five trial design comparisons. High‐certainty evidence from three pooled studies shows that an open‐label trial design significantly increases recruitment compared to a blinded, placebo‐controlled design (RD 10%). Conversely, single‐study evidence for other strategies, including patient preference designs and internet‐based data collection, is of low certainty or remains insufficient to facilitate GRADE assessments due to small sample sizes and wide confidence intervals. None of the included studies reported cost or retention data for any of these strategies. We describe results for the individual comparisons below. Open‐label versus blinded, placebo‐controlled design An open‐label design compared to a blinded, placebo‐controlled design results in a large increase in recruitment: RD = 10% (95% CI 8% to 12%); GRADE: high; Analysis 1.1; Summary of findings table 1. This is based on three studies (9004 participants): Avenell 2004 [36] (fracture prevention), RoB: low; Hemminki 2004 [37] (postmenopausal hormone therapy), RoB: low; and Veerus 2016 [38] (postmenopausal hormone therapy), RoB: unclear. This result and its GRADE assessment apply mainly to women (although one study did involve some men) in the UK and Estonia. Applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. None of the studies reported cost or retention data. Patient preference randomisation versus conventional randomised allocation A patient preference design increased total participation but may have resulted in little or no difference to recruitment to the randomised trial: RD = −4% (reduced recruitment) (95% CI −15% to 7%); GRADE: low (−2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 2.1. This is based on single‐study evidence (273 participants): Cooper 1997 [39] (management strategies for heavy menstrual bleeding), RoB: low. This result and its GRADE assessment apply mainly to higher‐income women in the UK (Scotland); applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Cooper 1997 did not report cost or retention data. Electronic data capture versus paper‐based data capture Internet‐based, electronic data collection compared to paper‐based data capture may reduce recruitment: RD = −13% (95% CI −24% to −3%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 3.1. This is based on single‐study evidence (28 practices and 80 participants involved): Litchfield 2005 [40] (delivery systems for insulin), RoB: unclear. This result and GRADE assessment apply only to people living in the UK. Because participant characteristics were insufficiently reported, applicability to other populations is unclear. For other populations, certainty would be downgraded for indirectness. Litchfield 2005 did not report cost or retention data. Cluster‐randomised design versus Zelen design Compared to a Zelen design, a cluster‐randomised design may increase recruitment (6/24 recruited in the cluster arm versus 0/29 in the Zelen arm). This is based on single‐study evidence involving mostly older people receiving end of life care in the UK (Wales; two cluster sites and 53 participants): Fowell 2006 [41] (palliative care), RoB: low; comparison 4A, no forest plot available, Supplementary material 10. We were unable to generate a forest plot or complete a GRADE assessment for this comparison due to lack of data. Fowell 2006 did not report cost or retention data. Two‐stage randomisation to choose duration of treatment versus single randomisation A single, upfront randomisation to three or six months of treatment may increase recruitment compared to delayed, two‐stage randomisation to decide whether a second 3‐month treatment should be given (5.21 versus 4.09 participants per year, per centre). Data on numbers recruited were not available for one arm. This is based on single‐study evidence involving people living in the UK (215 participants): Paul 2011 [42] (adjuvant treatment for colorectal cancer), RoB: low; comparison 5A, no forest plot available, Supplementary material 10. We were unable to generate a forest plot or complete a GRADE assessment for this comparison due to lack of data. Paul 2011 did not report cost or retention data. Category B: pre‐trial planning Overview There was only one study evaluating a single comparison assessing the effect of pre‐trial planning on improving recruitment. The study involved 33 healthcare organisations. Pre‐engagement with potential trial sites during proposal development versus no pre‐engagement Pre‐engagement with potential trial sites during proposal development may increase recruitment to the randomised trial: RD = 50% (95% CI 18% to 81%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 6.1. This is based on single‐study evidence: Garvelink 2015 [43] (frailty), RoB: unclear. This result and its GRADE assessment apply to directors of health and social services centres in Canada; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Garvelink 2015 did not report cost or retention data. Category C: trial conduct changes Overview Six studies (6592 participants) evaluated five conduct‐related strategies. High‐certainty evidence from two pooled studies shows that telephone reminders to non‐responders increase recruitment (RD 6%), although this specifically applies to trials with low underlying recruitment. Conversely, single‐study evidence of moderate certainty suggests that SMS messages mentioning trial scarcity (RD 3%) or including participant quotes (RD 4%) probably increase recruitment. Single‐study evidence also indicates that email and postal invitations perform similarly. While single‐study evidence suggests telephone reminders probably improve recruitment more than SMS, they involve higher costs (GBP 3.64 incremental cost per participant) with statistically non‐significant effectiveness. None of the studies reported retention data. We describe results for the individual comparisons below. Telephone reminder versus no telephone reminder Using a telephone reminder to contact non‐responders to a postal invitation increases recruitment: RD = 6% (95% CI 3% to 9%); GRADE: high; Analysis 7.1; Table. This is based on two studies (1450 participants): Nystuen 2004 [44] (getting people to return to work), RoB: low; Wong 2013 [45] (colorectal cancer), RoB: low. Note: the evidence for this intervention comes entirely from trials with low (< 10%) underlying recruitment. When applied to trials with higher recruitment, we would downgrade the evidence to moderate certainty for indirectness. In addition, this result and its GRADE assessment apply to people with a mean age of 58 years living in Canada and Norway; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be further downgraded for indirectness. Neither study reported cost or retention data. SMS reminder mentioning scarcity versus SMS reminder with no mention Mentioning the scarcity of trial places in SMS messages probably increased recruitment: RD = 3% (95% CI 1% to 6%); GRADE: moderate (−1 level: imprecision – single study); Analysis 8.1. This is based on single‐study evidence (1862 participants): Free 2011 [46] (smoking cessation), RoB: low. This result and its GRADE assessment apply to younger people living in the UK; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Free 2011 did not report cost or retention data. SMS messages containing quotes from existing participants versus no messages Giving quotes from previous participants in SMS messages probably increased recruitment: RD = 4% (95% CI 2% to 6%); GRADE: moderate (−1 level: imprecision – single study); Analysis 9.1. This is based on single‐study evidence (811 participants): Free 2010 [47] (smoking cessation), RoB: low. This result and its GRADE assessment apply to younger people living in the UK; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Free 2010 did not report cost or retention data. Email invitation versus postal invitation Using email invitations probably made little or no difference to recruitment compared to postal invitations: RD = 1% (95% CI −3% to 4%); GRADE: moderate (−1 level: imprecision – single study); Analysis 10.1. This is based on single‐study evidence (1760 participants): Treweek 2012 [48] (antibiotic prescribing by GPs), RoB: low. This result and its GRADE assessment apply to general practitioners living in the UK (Scotland); applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Treweek 2012 did not report cost or retention data. Telephone reminders versus SMS reminders Using telephone reminders probably increases recruitment compared to SMS reminders. RD = 5% (95% CI −1% to 11%); GRADE: moderate (−1 level: imprecision – single study); Analysis 11.1. This is based on single‐study evidence (709 participants): Bracken 2019 [49] (testosterone for type 2 diabetes), RoB: low. This result and its GRADE assessment apply to middle‐aged men living in Australia; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Telephone reminders, versus SMS reminders, have an estimated ICER of GBP 72.72, with incremental costs of GBP 3.64 (consisting of inflation‐adjusted implementation, i.e. landline charges per call for telephone reminders versus bulk prices for SMS reminders and staff time costs) and an incremental effectiveness of 5%. However, the effectiveness of this strategy is not statistically significant. Bracken 2019 did not report retention data. Category D: modification to the consent process Overview Five studies evaluated five comparisons assessing the effect of modifying the consent process. We present only three studies (482 participants) here as two were at high risk of bias. Single‐study evidence of low certainty suggests that opt‐out consent may improve recruitment (RD 19%), based on a study of older, affluent participants. Single‐study evidence for other strategies, such as having a researcher read consent details or using easy‐to‐read forms, remains very uncertain or insufficient to undertake GRADE assessments. None of the included studies reported cost or retention data. We describe results for the individual comparisons below. Opt‐out consent versus opt‐in consent Opt‐out consent may improve recruitment: RD = 19% (95% CI 3% to 35%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 12.1. This is based on single‐study evidence (152 participants): Trevena 2006 [50] (colorectal cancer), RoB: unclear. This result and its GRADE apply mainly to better‐off males and females above the age of 50 living in Australia; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Trevena 2006 did not report cost or retention data. Researcher reading out consent versus participant reading consent It is very uncertain whether a researcher reading out the consent details affects recruitment: RD = 6% (95% CI −13% to 25%); GRADE: very low (−1 level: study limitations – unclear RoB; −2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 13.1. This is based on single‐study evidence (104 participants): Wadland 1990 [51] (smoking cessation), RoB: unclear. This result and its GRADE assessment apply to people living in the USA; applicability to other populations is unclear because the authors provided no further participant characteristics. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. Wadland 1990 did not report cost or retention data. Easy‐to‐read versus standard consent form An easy‐to‐read consent form may make little or no difference to recruitment compared to standard consent: RD = 3% (P = 0.32); Comparison 14D, Supplementary material 10 (no forest plot available). The authors of this cluster‐randomised trial did not present centre‐level recruitment data, or provide an intracluster correlation coefficient. However, they did consider intracluster correlation in their analysis and found that recruitment did not differ significantly between the two trial groups. This is based on single‐study evidence (226 participants): Coyne 2003 [52] (cancer), RoB: unclear. We were unable to complete a GRADE assessment for this comparison due to lack of data. This result applies mostly to well‐educated white women above the age of 50 years living in the USA. Coyne 2003 did not report cost or retention data. Category E: modification to the information given to potential participants about the trial Overview Fifty‐two studies evaluated 37 comparisons, with 48 studies included, involving at least 109,968 participants. It was unclear how many participants were involved in two studies, although at least 126 centres were involved. We judged four studies (7.7%) to be at high risk of bias and do not present them here. Single‐study evidence of moderate certainty shows that enclosing a trial‐relevant questionnaire probably increases recruitment (RD 5%). Moderate‐certainty evidence from two pooled studies suggests that providing multimedia information via links or QR codes probably leads to a large increase in recruitment and retention (RD 13%). Conversely, high‐certainty evidence from pooled studies demonstrates that optimised participant information leaflets (PILs)(bespoke and user‐tested) and recruitment primer letters make little to no difference. Moderate‐certainty evidence from pooled studies also suggests that brief or user‐feedback‐optimised PILs do not significantly improve recruitment. Some strategies may reduce recruitment: moderate‐certainty evidence from two pooled studies indicates that telephone‐based information is less effective than face‐to‐face information (RD −9%). Single‐study evidence of low certainty suggests that handwritten personalisation or ethnicity‐specific risk framing may slightly decrease participation. Evidence for other strategies, such as video‐based information or ethnic‐matched recruiters, remains very uncertain or insufficient to facilitate GRADE assessments. Cost data show that while additional leaflets are low‐cost, video‐based information costs more for lower effectiveness. Most studies lacked cost or retention data. The results for the 33 individual comparisons are described below. Brief participant information leaflet (PIL) versus full PIL Using a brief patient information leaflet (PIL) probably makes little or no difference to recruitment compared to a full PIL: RD = 0% (95% CI −1% to 2%); GRADE: moderate (−1 level: indirectness, Chen 2011 actually measured entry to pre‐randomisation phase); Analysis 17.1; Table. This is based on two studies (4633 participants): Chen 2011 [53] (clinical area unclear), RoB: low; Brierley 2012 [54] (depression), RoB: low. This result and its GRADE assessment apply to people, especially women, around 40 years old living in the UK. For other populations, certainty would be downgraded for indirectness. Neither study reported cost or retention data. Study‐related questionnaire plus trial invitation versus trial invitation Enclosing a questionnaire covering issues relevant to the trial with the invitation probably increases recruitment: RD = 5% (95% CI 2% to 8%); GRADE: moderate (−1 level: imprecision – single study); Analysis 18.1. This is based on single‐study evidence (2393 participants): Kendrick 2001 [55] (injury prevention, recruiting family units), RoB: low. This result and its GRADE assessment apply to people living in the UK; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Kendrick 2001 did not report cost or retention data. PIL developed with feedback from users versus usual PIL Optimised PILs developed with feedback from potential participants or stakeholders, but without formal iterative user‐testing, probably make little or no difference to recruitment: RD = 0% (95% CI 0% to 1%); GRADE: moderate (−1 level: indirectness, Chen 2011 actually measured entry to pre‐randomisation phase); Analysis 19.1; Table. This is based on two studies (16,763 participants): Chen 2011 (clinical area unclear), RoB: low; Cockayne 2017 (START PIL REFORM) [56] (falls prevention), RoB: low. This result and its GRADE assessment apply to older people living in the UK; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. None of the studies reported cost or retention data. Enhanced recruitment package plus recruitment at churches versus standard recruitment package Recruitment at a church and other enhancements, such as having African American recruiters, may result in little to no difference in recruitment: RD = 1% (95% CI 0% to 2%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 20.1. This is based on single‐study evidence (6246 participants): Ford 2004 [57] (cancer), RoB: unclear. This result and its GRADE assessment apply to African American men living in the USA; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Ford 2004 did not report cost or retention data. Enhanced versus standard recruitment package An enhanced recruitment package including more contact may make little or no difference in recruitment: RD = 0% (95% CI −1% to 0%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 21.1. This is based on single‐study evidence (6376 participants): Ford 2004 (cancer), RoB: unclear. This result and its GRADE assessment apply to African American men living in the USA; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Ford 2004 did not report cost or retention data. Enhanced recruitment package plus baseline data over telephone versus standard recruitment package An enhanced recruitment package including more contact by telephone may make little or no difference in recruitment: RD = 0% (95% CI −1% to 1%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 22.1. This is based on single‐study evidence (6372 participants): Ford 2004 (cancer), RoB: unclear. This result and its GRADE assessment apply to African American men living in the USA; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Ford 2004 did not report cost or retention data. Audio record of information given about trial versus no audio record It is very uncertain whether providing an audio record of the discussion about the trial affects recruitment: RD = −3% (reduced recruitment) (95% CI −19% to 13%); GRADE: very low (−1 level: study limitations – unclear RoB; −2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 23.1 This is based on single‐study evidence (130 participants): Bergenmar 2014 [58] (cancer), RoB: unclear. This result and its GRADE assessment apply to middle‐aged, mostly female participants with a broad range of educational backgrounds living in Sweden. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. Bergenmar 2014 did not report cost or retention data. Clinical trial booklet plus standard information versus standard information It is very uncertain whether providing a clinical trial booklet together with standard information affects recruitment: RD = 20% (95% CI −5% to 46%); GRADE: very low (−1 level: study limitations – unclear RoB; −2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 24.1. This is based on single‐study evidence (31 participants): Ives 2001 [59] (HIV), RoB: unclear. This result and its GRADE assessment apply to males in their thirties with a good education living in an urban population in the UK. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. Ives 2001 did not report cost or retention data. Total information disclosure versus standard disclosure It is very uncertain whether providing total information disclosure rather than leaving it to recruiters as to what to reveal affects recruitment: RD = 11% (95% CI −6% to 28%); GRADE: very low (−1 level: study limitations – unclear RoB; −2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 25.1. This is based on single‐study evidence (57 participants): Simes 1986 [60] (cancer), RoB: unclear. This result and its GRADE assessment apply to middle‐aged white women whose first language is English, with a high‐school or lower education, who live in Australia. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. Simes 1986 did not report cost or retention data. Bespoke user‐tested PIL versus usual PIL Optimised PILs (bespoke and user‐tested) tailored for a specific trial and refined through formal user testing make little or no difference to recruitment: RD = 0% (95% CI 0% to 1%); GRADE: high; Analysis 27.1; Table. This is based on six studies (27,805 participants): Cockayne 2017 (START PIL REFORM) (falls prevention), RoB: low; Knapp 2020 (START PIS CASPER) [61] (behavioural activation for depression), RoB: low; Madurasinghe 2021 (START PIS HeLP Diabetes) [62] (type 2 diabetes self‐management), RoB: low; Man 2015 (START PIS Healthlines Depression) [63] (extra support for depression), RoB: low; Parker 2018 (START PIS ECLS) [64] (lung cancer diagnosis), RoB: unclear; Sheridan 2020 (START PIS ISDR) [65] (diabetic retinopathy screening), RoB: high. This result and its GRADE assessment apply to people living in the UK; applicability to other populations is unclear because the authors of most of the studies provided no further participant characteristics. Only one out of the six studies reported EDI data, and participants were mostly older women. For other populations, certainty would be downgraded for indirectness. None of the studies reported cost or retention data for this comparison. Recruitment primer letter versus no letter Sending a recruitment primer letter/leaflet makes little or no difference to recruitment: RD = 1% (95% CI −1% to 2%); GRADE: high; Analysis 28.1; Table. This is based on two studies (5376 participants): Arundel 2017 [66] (falls prevention), RoB: low; Paul 2014 [67] (colorectal cancer), RoB: low. This result and its GRADE assessment apply mainly to older white people living in the UK and Ireland; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. We extracted cost‐effectiveness data from Arundel 2017. Adding a recruitment leaflet has an estimated ICER of GBP 208.38, with incremental costs of GBP 2.08 (consisting of inflation‐adjusted printing, packing, supply, and labelling costs) and a partial incremental effectiveness of 1%. However, the partial effectiveness of this strategy is not statistically significant (RD 0.01, 95% CI −0.01 to 0.02). Neither study reported retention data. Information provided over telephone versus information provided face‐to‐face Providing information over the telephone rather than face‐to‐face probably reduces recruitment: RD = −9% (reduced recruitment) (95% CI −18% to 1%); GRADE: moderate (−1 level: imprecision – wide CI crossing the line of no effect); Analysis 29.1; Table. This is based on two studies (249 participants): Bobb 2016 [68] (emergency department trials), RoB: low; Foss 2016 [69] (vaccination), RoB: low. This result and its GRADE assessment apply to people, especially women, between the ages of 30 and 60 who have a high degree of education and are living in the USA or Denmark. For other populations, certainty would be downgraded for indirectness. Neither study reported cost or retention data for this comparison. Providing information by video versus standard information It is very uncertain whether providing trial information by video affects recruitment compared to standard information: RD = −4% (reduced recruitment) (95% CI −23% to 15%); GRADE: very low (−1 level: study limitations – unclear and high RoB; −1 level: inconsistency; −1 level: imprecision – wide CI crossing the line of no effect); Analysis 30.1; Table. This is based on five studies (1634 participants): Du 2008 [70] (lung cancer), RoB: unclear; Du 2009 [71] (breast cancer), RoB: unclear; Hutchison 2007 [72] (cancer), RoB: low; Mattock 2020 [73] (paediatric behavioural difficulties), RoB: high; Rogers 2019 [74] (arthritis), RoB: unclear. This result and its GRADE assessment apply to older, mainly ethnic majority males and females from a broad range of socioeconomic backgrounds living in high‐income countries. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. We extracted cost‐effectiveness data from Rogers 2019. Providing information by video is dominated by standard information, as it is both more costly (with incremental costs of GBP 41.58 consisting of inflation‐adjusted video production, DVD manufacture, packaging, and postage costs) and less effective (with a partial incremental effectiveness of −2%). However, the partial effectiveness of this strategy is not statistically significant (RD: from −0.02, 95% CI −0.07 to 0.04). None of the studies reported retention data for this comparison. Invitation letter emphasising ethnicity‐specific disease risk versus standard letter An invitation letter that emphases ethnicity‐specific disease risk may reduce recruitment slightly compared to a standard invitation leaflet: RD = −3% (reduced recruitment) (95% CI −12% to 6%); GRADE: low (−2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 31.1. This is based on single‐study evidence (445 participants): Brown 2015 [75] (lifestyle intervention for gestational diabetes or pre‐diabetes), RoB: low. This result and its GRADE assessment apply to women over the age of 30 from a range of ethnic groups who live in the USA. For other populations, certainty would be downgraded for indirectness. Brown 2015 did not report cost or retention data. Handwritten personalised invitation letter versus standard letter A handwritten, personalised invitation leaflet may reduce recruitment compared to a standard invitation leaflet: RD = −4% (reduced recruitment) (95% CI −8% to 0%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 32.1. Additionally, a handwritten personalised invitation leaflet may reduce retention compared to a standard invitation leaflet: RD = −4% (reduced retention) (95% CI −8% to 0%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 32.2. These findings are based on single‐study evidence (317 participants): McCaffery 2019 [76] (falls prevention), RoB: unclear. This result and its GRADE assessment apply to people living in the UK; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. McCaffery 2019 did not report cost data. Different website wording, including infographic, about the risk of underlying condition versus standard wording Different website wording, including an infographic, about the risk of the underlying condition may make little or no difference to recruitment, compared to standard wording, but the evidence is very uncertain: RD = 0% (95% CI −0.01 to 0.01); GRADE: very low (−2 levels: study limitations – both included studies are high RoB; −1 level: indirectness – ‘visitors’ to the website may not be unique); Analysis 33.1. These findings are based on single‐study evidence involving three arms, published in a single paper (2605 participants): Miller 2021a [77] (falls prevention), RoB: high; and Miller 2021b [78] (falls prevention), RoB: high. This result and its GRADE assessment apply to people aged 70 years and over living in the USA; applicability to other populations is unclear because the authors provided no further participant characteristics. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. Neither study reported cost or retention data. Shorter recruitment package versus standard recruitment package It is very uncertain whether a shorter recruitment package with standard consent affects recruitment: RD = 2% (95% CI −13% to 17%); GRADE: very low (−1 level: study limitations – unclear RoB; −2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 34.1. This is based on single‐study evidence (170 participants): Chow 2020 [79] (self‐management for coronary heart disease following paediatric cancer treatment), RoB: unclear. This result and its GRADE assessment apply to people living in the USA; applicability to other populations is unclear because the authors provided no further participant characteristics. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. Chow 2020 did not report cost or retention data. Shorter recruitment package with simplified consent versus standard recruitment package It is very uncertain whether a shorter recruitment pack with simplified consent affects recruitment: RD = 2% (95% CI −13% to 17%); GRADE: very low (−1 level: study limitations – unclear RoB; −2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 35.1. This is based on single‐study evidence (175 participants): Chow 2020 (self‐management for coronary heart disease following paediatric cancer treatment), RoB: unclear. This result and its GRADE assessment apply to people living in the USA; applicability to other populations is unclear because the authors provided no further participant characteristics. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. Chow 2020 did not report cost or retention data. Leaflet advertising patient and public involvement (PPI) in trial plus standard trial invitation versus standard trial invitation only Including a leaflet advertising patient and public involvement (PPI) in a trial alongside the standard trial invitation probably makes little or no difference to recruitment: RD = −1% (reduced recruitment) (95% CI −2% to 0%); GRADE: moderate (−1 level: imprecision – single study); Analysis 36.1. This is based on single‐study evidence (8182 participants): Hughes‐Morley 2016 [80] (care planning training for community mental health teams), RoB: low. This result and its GRADE assessment apply to middle‐aged adults, predominantly female and relatively socioeconomically disadvantaged, living in the UK. For other populations, certainty would be downgraded for indirectness. Hughes‐Morley 2016 did not report cost or retention data. Gender‐targeted recruitment postcard versus generic recruitment postcard Gender‐targeted recruitment postcards may have little or no impact on recruitment compared to generic recruitment postcards: RD = 0% (95% CI −0% to 0%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 37.1. This is based on single‐study evidence (30,000 participants): Crane 2016 [81] (behavioural intervention for weight gain prevention), RoB: unclear. This result and its GRADE assessment apply mainly to females who live in the USA. For other populations, certainty would be downgraded for indirectness. Crane 2016 did not report cost or retention data. Researcher‐optimised PIL versus standard PIL An optimised PIL, modified by the research team to improve readability without direct user input or feedback, may reduce recruitment: RD = −9% (reduced recruitment) (95% CI −17% to 0%); GRADE: low (−2 levels: imprecision – single study; wide CI including the line of no effect); Analysis 38.1. This is based on single‐study evidence (480 participants): Paris 2015 [82] (multiple host trials in different clinical areas, including heart surgery, dementia, diabetes, and sleep disorders), RoB: low. This result and its GRADE assessment apply to middle‐aged people with a broad range of socioeconomic backgrounds who live in France. For other populations, certainty would be downgraded for indirectness. Paris 2015 did not report cost or retention data. Brief PIL plus standard PIL versus standard PIL Providing a brief PIL alongside the standard PIL may have little or no effect on recruitment: RD = 0% (95% CI −1% to 1%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 40.1. This is based on single‐study evidence (6946 participants): White 2023 [83] (community‐based clinic for people with persistent, medically unexplained, physical symptoms), RoB: unclear. This result and its GRADE assessment apply mainly to white British females around the age of 46 with a good education living in the UK. For other populations, certainty would be downgraded for indirectness. We also extracted cost‐effectiveness data from White 2023. Providing a brief PIL alongside the standard PIL, versus standard PIL alone, has an undefined ICER, as the incremental effectiveness (the denominator) is zero. This strategy presents an incremental cost of GBP 0.57 consisting of the inflation‐adjusted cost of adding a brief PIL. Overall, adding a brief PIL alongside the standard PIL results in a small increase in cost with no difference in recruitment. White 2023 did not report retention data. Decision aid about RCTs versus blank notebook A decision aid about RCTs may result in a large increase in trial recruitment compared to a blank notebook: RD = 12% (95% CI −1% to 24%); GRADE: low (−2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 41.1. This is based on single‐study evidence (129 participants): Sundaresan 2017 [84] (radiotherapy for prostate cancer), RoB: low. This result and its GRADE assessment apply to older males with a broad range of countries of birth and levels of education living in Australia and New Zealand. For other populations, certainty would be downgraded for indirectness. Sundaresan 2017 did not report cost or retention data. Interactive online digital educational tool versus time‐matched webinar An interactive online digital educational tool may make little or no difference to recruitment compared to a time‐matched webinar: RD = 1% (95% CI −1% to 3%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 42.1. This is based on single‐study evidence (351 participants): Saif 2020 [85] (Alzheimer's disease), RoB: unclear. This result and its GRADE assessment apply to older people, especially females, from a range of ethnicities (though predominantly white) with varying amounts of education living in the USA. For other populations, certainty would be downgraded for indirectness. Saif 2020 did not report cost or retention data. PPI‐designed and ‐informed PIL versus standard researcher‐designed PIL It is very uncertain whether a PPI‐designed and ‐informed PIL affects recruitment compared to a standard researcher‐designed PIL: RD = −2% (reduced recruitment) (95% CI −15% to 12%); GRADE: very low (−1 level: study limitations – unclear RoB; −2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 43.1. These findings are based on single‐study evidence (207 participants): Dwyer 2022 [86] (cognitive rehabilitation intervention for multiple sclerosis), RoB: unclear. This result and its GRADE assessment apply to people living in Ireland; applicability to other populations is unclear because the authors provided no further participant characteristics. For other groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. Additionally, it is very uncertain whether a PPI‐designed and ‐informed PIL affects retention compared to a standard researcher‐designed PIL: RD = 4% (reduced retention) (95% CI −9% to 18%); GRADE: very low (−1 level: study limitations – unclear RoB; −2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 43.2. Dwyer 2022 did not report cost data. Multimedia education about clinical trials (DVD plus booklet) versus standard educational booklet only Providing multimedia education about clinical trials via a DVD and booklet may increase recruitment: RD = 4% (95% CI −5% to 14%); GRADE: low (−2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 45.1. This is based on single‐study evidence (418 participants): Kamen 2018 [87] (oncology), RoB: low. This result and its GRADE assessment apply to older people, especially females, who are predominantly white, with a range of incomes and levels of education, living in the USA. For other populations, certainty would be downgraded for indirectness. Kamen 2018 did not report cost or retention data. Multimedia information (viewed in person) plus paper PIL versus paper PIL alone It is very uncertain whether providing multimedia information (viewed in person) alongside a paper PIL increases recruitment: RD = 17% (95% CI −89% to 123%); GRADE: very low (−1 level: inconsistency; −2 levels: imprecision – extremely wide CI crossing the line of no effect); Analysis 47.1; Table. Additionally, it is very uncertain whether providing multimedia information (viewed in person) alongside a paper PIL affects retention compared to a paper PIL only: RD = 28% (95% CI −79% to 136%); GRADE: very low (−1 level: inconsistency; −2 levels: imprecision – extremely wide CI crossing the line of no effect); Analysis 47.2. These findings are based on three studies (23 participants): Knapp 2023a (BALANCE) [88] (paediatric lazy eye), RoB: low; Knapp 2023b (BAMP) [89] paediatric underbite), RoB: low; Knapp 2023f (UKALL) [90] paediatric and adolescent leukaemia or lymphoma), RoB: low. This result and its GRADE assessment apply to non‐white male and female children aged between six and 13 years, living in the UK, mainly in disadvantaged areas. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. None of the studies reported cost data for this comparison. Multimedia information (viewed in person) PIL versus paper PIL Providing a multimedia information PIL (viewed in person) probably increases recruitment slightly compared to a paper PIL: RD = 4% (95% CI −1% to 8%); GRADE: moderate (−1 level: imprecision); Analysis 48.1; Table. These findings are based on three studies (1428 participants): Knapp 2023a (BALANCE) paediatric lazy eye), RoB: low; Knapp 2023b (BAMP) paediatric underbite), RoB: low; Knapp 2023d (FORCE) [91] paediatric wrist fracture), RoB: low. This result and its GRADE assessment apply to male and female children aged between six and 12 years, of mixed ethnicities, living in the UK, mainly in disadvantaged areas. For other populations, certainty would be downgraded for indirectness. Additionally, it is very uncertain whether providing multimedia information (viewed in person) affects retention compared to a paper PIL: RD = 22% (95% CI −51% to 96%); GRADE: very low (−2 levels: imprecision – extremely wide CI crossing the line of no effect; −1 level: inconsistency); Analysis 48.2. None of the studies reported cost data for this comparison. Multimedia information (provided as link/QR code) plus paper PIL versus paper PIL Providing multimedia information in the form of a link or QR code alongside a paper PIL makes little or no difference to recruitment: RD = 0% (95% CI −1% to 1%); GRADE: high; Analysis 49.1; Table. This is based on seven studies (11,612 participants): Jolly 2019 (START MMI PSM‐COPD) [92] (COPD), RoB: low; Knapp 2023e (Thermic‐3) [93] (paediatric congenital heart defects), RoB: low; Knapp 2023c (CHAMP‐UK) [94] (paediatric short‐sightedness), RoB: low; Madurasinghe 2023a (START MMI GHT2000a) [95] (exercise for high blood pressure), RoB: low; Madurasinghe 2023b (START MMI GHT2000b) [96] (exercise for high blood pressure), RoB: low; Madurasinghe 2023c (START MMI HI‐Light) [97] (vitiligo), RoB: low; Madurasinghe 2023d (START MMI seAFOod) [98] (bowel cancer prevention), RoB: low. This result and its GRADE assessment apply to people living in the UK, some of whom were male and female children between the ages of nine months and nine years with mixed ethnicities; however, reporting in most studies was very limited. For other populations, certainty would be downgraded for indirectness. We also extracted cost‐effectiveness data from Jolly 2019 (START MMI PSM‐COPD). Multimedia information plus paper PIL is dominated by paper PIL only, as it is both more costly (with incremental costs of GBP 0.78 consisting of the intervention inflation‐adjusted unit costs) and less effective (with a partial incremental effectiveness of −1%). However, the partial effectiveness of this strategy is not statistically significant (RD −0.01, 95% CI −0.03 to 0.01). Additionally, providing multimedia information in the form of a link or QR code alongside a paper PIL makes little or no difference to the likelihood of being retained compared to a paper PIL only: RD = 0% (95% CI −2% to 3%); GRADE: high; Analysis 49.2. This is based on five studies (7403 participants): Jolly 2019 (START MMI PSM‐COPD) (COPD), RoB: low; Knapp 2023e (Thermic‐3) (paediatric congenital heart defects), RoB: low; Knapp 2023c (CHAMP‐UK) (paediatric short‐sightedness), RoB: low; Madurasinghe 2023c (START MMI HI‐Light) (vitiligo), RoB: low; Madurasinghe 2023d (START MMI seAFOod) (bowel cancer prevention), RoB: low. Multimedia information (provided as link/QR code) versus paper PIL Providing multimedia information in the form of a link or QR code probably results in a large increase in recruitment compared to a paper PIL: RD = 13% (95% CI 1% to 26%); GRADE: moderate (−1 level: imprecision – wide CI); Analysis 50.1; Table. This is based on two studies (229 participants): Knapp 2023e (Thermic‐3) (paediatric congenital heart defects), RoB: low; Knapp 2023c (CHAMP‐UK) (paediatric short‐sightedness), RoB: low. This result and its GRADE assessment apply to studies involving children with a range of ethnicities, of both sexes, aged between approximately eight months and nine years, living in the UK, mainly in disadvantaged areas. In one study (Knapp 2023c (CHAMP‐UK), parents were predominantly female, highly educated, and employed. For other populations, certainty would be downgraded for indirectness. Additionally, providing multimedia information in the form of a link or QR code probably results in a large increase in retention (assessed at three months) compared to a paper PIL only. RD = 13% (95% CI 0% to 26%); GRADE: moderate (−1 level: imprecision – wide CI); Analysis 50.2. This is based on two studies: Knapp 2023e (Thermic‐3) (paediatric congenital heart defects), RoB: low; Knapp 2023c (CHAMP‐UK) (paediatric short‐sightedness), RoB: low. None of the studies reported cost data for this comparison. Educational material providing additional information about a trial versus standard information alone An educational package providing additional information about a trial did not significantly increase recruitment compared to standard information alone (31% of participants aged over 65 in both intervention and control groups in year 2, P = 0.83), comparison 51E, no forest plot available, Supplementary material 10. Although the authors of this cluster‐randomised trial did not present centre‐level recruitment data, or provide an intracluster correlation coefficient, they did consider intracluster correlation in their analysis. This is based on single‐study evidence (126 centres): Kimmick 2005 [99] (cancer), RoB: unclear. We were unable to complete a GRADE assessment for this comparison due to lack of data. This study provided minimal information, and we have therefore not presented the characteristics of included participants. Kimmick 2005 did not report cost or retention data. Trained recruiters from a similar ethnic background to study population already taking part in a trial as lay advocates Trained recruiters from a similar ethnic background to the trial population acting as lay advocates may increase recruitment compared to untrained recruiters (8/28 trained Hispanic recruiters recruited one or more women versus 0/26 untrained Hispanic women and 2/42 untrained Anglo controls). The authors of this cluster‐randomised trial did not report an analysis that corrected for the clustering or provide an intracluster correlation coefficient. Larkey 2002 [100] reported data at the recruiter aggregate level on whether a recruiter did or did not recruit anyone to the trial. This is based on single‐study evidence (number of participants unclear): Larkey 2002 (unclear), RoB: unclear, comparison 52E, no forest plot available, Supplementary material 10. We were unable to complete a GRADE assessment for this comparison due to lack of data. This study provided minimal information, and we have therefore not presented the characteristics of included participants. Larkey 2002 did not report cost or retention data. Coach to support recruitment of minority participants versus no coach Using a coach to promote trust among minority participants did not significantly increase trial recruitment compared to usual care. The authors report the number of participants in each arm who went on to be recruited in a trial (16 participants in the intervention group and 13 participants in the control group, P = 0.35). However, we were unable to determine the proportion of participants who were recruited in a trial in each study arm as they did not report the numbers of participants who were randomised to receive coaching versus usual care. This is based on single‐study evidence (75 participants): Fracasso 2013 [101] (cancer), RoB unclear, comparison 53E, no forest plot available, Supplementary material 10. We were unable to complete a GRADE assessment for this comparison due to lack of data. This study provided limited information, and we have therefore not presented the characteristics of included participants. Fracasso 2013 did not report cost or retention data. Category F: interventions aimed at the recruiter or recruitment site Overview Ten studies evaluated 10 comparisons targeting recruiters or sites. We judged two studies to be at high risk of bias and do not present them here. The remaining eight studies involved at least 5045 participants. It was unclear how many participants were involved in four studies, although 316 clinical sites were involved between them. Single‐study evidence of low certainty shows that remote site initiation visits may increase recruitment (RD 2%) and are more cost‐effective than face‐to‐face visits, saving approximately GBP 42.29 per site. Conversely, single‐study evidence suggests that postcard teaser campaigns (moderate certainty) and educational programmes for ethnic minority recruitment (low certainty) appear to have little or no impact. Other strategies, including recruiter training and site teleconferences, showed mixed results in individual studies but lacked sufficient data to undertake GRADE assessments. Data on cost and retention were rarely reported. We describe the results for the eight individual comparisons below. Teaser campaign using postcards versus no teaser Using a postcard teaser campaign probably makes little or no difference to recruitment: RD = 0% (95% CI −4% to 5%); GRADE: moderate (−1 level: imprecision – single study); Analysis 54.1 This is based on single‐study evidence (670 participants): Lee 2017 [102] (recruiting GP practices to low back pain trial), RoB: low. This result and its GRADE assessment apply to practices and clinics in Australia; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. Lee 2017 did not report cost or retention data. Educational programme to tailor recruitment strategies for minority recruitment versus usual recruitment Using an educational programme to tailor recruitment for ethnic minority recruitment may have little or no impact on recruitment: RD = 0% (95% CI −4% to 3%); GRADE: low (−1 level: study limitations – unclear RoB; −1 level: imprecision – single study); Analysis 55.1. This is based on single‐study evidence (2107 participants): Tilley 2021 [103] (various host trials: atrial fibrillation, blood and marrow transplants, colon cancer prevention, Parkinson’s), RoB: unclear. This result and its GRADE assessment apply to Latinx white people, especially those who speak English, who live in the USA. For other populations, certainty would be downgraded for indirectness. Tilley 2021 did not report cost or retention data. Remote versus face‐to‐face site initiation visits Using remote site initiation visits probably increases recruitment compared to face‐to‐face site initiation visits: RD = 2% (95% CI −5% to 10%); GRADE: moderate (−1 level: imprecision – single study); Analysis 56.1. This is based on single‐study evidence (688 participants): Jefferson 2018 [104] (cast versus surgery for wrist fracture), RoB: low. This result and its GRADE apply to the UK; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. We also extracted cost‐effectiveness data from Jefferson 2018. Remote visits dominate face‐to‐face site initiation visits, as they are both less costly (with inflation‐adjusted cost savings of GBP 42.29, taking into consideration travel costs and the average cost of setting up a site) and more effective (with an incremental effectiveness of 6%). However, the effectiveness of this strategy is not statistically significant. Additionally, remote site initiation visits may make little or no difference to retention compared to face‐to‐face site initiation visits: RD = 1% (95% CI −6% to 9%); GRADE: low (−2 levels: imprecision – single study; wide CI crossing the line of no effect); Analysis 56.2. Recruitment coordinator plus training versus usual recruitment Maxwell 2017 [105] compared having a recruitment coordinator conducting teleconferences with site teams to enhance trial recruitment using software for identifying eligible participants (with a second teleconference after six months) to standard recruitment. A stepped wedge approach was used to evaluate this intervention; raw figures at each step were not provided, so it was not possible to calculate a risk difference. The authors found an adjusted rate ratio of 1.06 (95% CI 0.55 to 2.03, P = 0.87), suggesting that the intervention may make little to no difference to recruitment. This is based on single‐study evidence (72 hospital sites): Maxwell 2017 (antiplatelet drugs following stroke), RoB: low, comparison 57F, no forest plot available, Supplementary material 10. We were unable to complete a GRADE assessment for this comparison due to lack of data. This study provided limited information, and we have therefore not presented the characteristics of included participants. Maxwell 2017 did not report cost or retention data. Teleconference and commitment contracts with site teams versus no teleconference Isaksson 2025 [106] compared using teleconference and commitment contracts with site teams to enhance recruitment versus standard recruitment. A stepped wedge approach was used to evaluate this intervention; raw figures at each step were not provided, so it was not possible to calculate a risk difference. The authors found that the intervention increased recruitment from 1.9 participants/centre/60 days to 2.1 participants/centre/60 days – a 10% increase. This is based on single‐study evidence (20 tertiary care centres): Isaksson 2025 (stroke recovery), RoB: unclear, comparison 58F, no forest plot available, Supplementary material 10. We were unable to complete a GRADE assessment for this comparison due to lack of data. This study provided limited information, and we have therefore not presented the characteristics of included participants. Isaksson 2025 did not report cost or retention data. On‐site initiation visit versus no initiation visit For this on‐site initiation visit strategy, the authors did not present the proportion of eligible participants recruited, only the number recruited: visited sites recruited 302 participants while those not receiving visits recruited 271. This is based on single‐study evidence (573 participants): Liénard 2006 [107] (breast cancer), RoB: low, comparison 59F, no forest plot available, Supplementary material 10. We were unable to complete a GRADE assessment for this comparison due to lack of data. This study provided limited information, and we have therefore not presented the characteristics of included participants. Liénard 2006 did not report cost or retention data. Additional communication from central trial coordinator to sites versus standard communication Monaghan 2007 [108] investigated different levels of communication between the trial coordination team and participating sites, comparing the effect of additional communication strategies such as tailored feedback on recruitment. The median total number of participants in the additional communication group was 37.5, compared to 37.0 in the standard communication group. Intervention centres achieved half their recruitment targets in 4.4 months, compared to 5.8 months for control centres. This is based on single‐study evidence (167 clinical sites): Monaghan 2007 (diabetes), RoB: low, comparison 60F, no forest plot available, Supplementary material 10. We were unable to complete a GRADE assessment for this comparison due to lack of data. This study provided limited information, and we have therefore not presented the characteristics of included participants. Monaghan 2007 did not report cost or retention data. Recruiter training with ongoing support and/or personalised nudge versus standard training and support Agni 2022 [109] investigated the effect of recruiter training with ongoing support and/or theory‐informed personalised nudge to the recruiter compared to standard recruiter support and training. The authors did not collect the number of eligible participants approached in the study. Thus, it was not possible to calculate a risk difference for this intervention. The total number of participants recruited in the enhanced Trainee Principal Investigator (TPI)‐only arm was 279; 147 in the digital nudge‐only arm; 410 for the arm which received both interventions; and 379 in the control arm. The authors concluded that the enhanced TPI education intervention improved recruitment (incident rate ratio (IRR) 1.23, 95% CI 1.09 to 1.40, P = 0.001) – but that the digital nudge did not improve recruitment (IRR 0.89, 95% CI 0.79 to 1.01, P = 0.07). This is based on single‐study evidence (1215 participants): Agni 2022 (partial hip replacement following hip fracture), RoB: low, comparison 61F, no forest plot available, Supplementary material 10. We were unable to complete a GRADE assessment for this comparison due to lack of data. This study provided limited information, and we have therefore not presented the characteristics of included participants. Agni 2022 did not report cost or retention data. Category G: incentives Overview Twelve studies involving 4934 participants evaluated three incentive types. This included one publication involving five Studies Within A Trial (SWATs) of the same strategy, and another involving two SWATs testing three different comparisons each. Pooled evidence of low certainty indicates that monetary incentives increase recruitment (RD = 4%), particularly at higher values. The estimated cost‐effectiveness ratio is GBP 3052.96 per additional recruit. It remains very uncertain whether higher monetary amounts are more effective than lower ones, as this single‐study evidence (across two trials) is statistically non‐significant. Moderate‐certainty evidence from three pooled studies shows that trial‐branded pens make little or no difference to recruitment or retention, representing a small cost (GBP 0.45) with no gain in effectiveness. Retention data were rarely reported. We describe the results for the 12 studies below. Monetary incentive versus no incentive Monetary incentives offered to potential participants may improve recruitment compared to no incentive: RD = 5% (95% CI 1% to 9%); GRADE: low (−1 level: study limitations high RoB; −1 level: inconsistency); Analysis 63.1; Table. This is based on 12 studies (5082 participants), including five trials within a single published paper, and two 3‐arm trials within a single published paper: Fairhurst 2022 [110] (yoga for older adults with multimorbidity), RoB: high; Free 2010 (smoking cessation), RoB: low; Giguere 2015 [111] (recruiting family physicians to a medical education trial), RoB: low; Jennings 2015a [112]; Jennings 2015b [113]; Jennings 2015c [114]; Jennings 2015d [115]; Jennings 2015e [116] (primary care, older people, mainly hypertension), RoB: low; Halpern 2021a ($200) [117] and Halpern 2021a ($500) [118] (smoking cessation), RoB: high; Halpern 2021b ($100) [119] and Halpern 2021b ($300) [120] (ambulation), RoB: high. Note: the evidence mainly comes from high incentive amounts. The comparisons are shown split by incentive value, grouped into incentives up to GBP 5 (two studies), around GBP 100 (six studies), and over GBP 100 (three studies, one providing two comparisons). This result and its GRADE assessment apply to people of both sexes aged from their thirties to mid‐sixties, probably mostly white ethnicity (but good Black/African American involvement in two studies), and a reasonable spread of socioeconomic statuses living in Canada, the UK, and the USA. For other populations, certainty would be downgraded for indirectness. As this was the only comparison with 10 or more studies available, we created a funnel plot, which showed a fairly symmetrical distribution, suggesting that publication bias is unlikely (Figure). The symmetry indicates that small and large studies report similar effects, and studies with negative or non‐significant results do not appear to be systematically missing. Funnel plot for participant recruitment for the comparison of monetary incentives versus no monetary incentives (category G strategy) A subgroup analysis to investigate variations in allocation concealment as a potential source of heterogeneity separated the studies into two groups: those with adequate allocation concealment (eight studies) and those with inadequate or unclear allocation concealment (two 3‐arm studies); Analysis 63.2. There was no statistical evidence of a difference in the recruitment effect between studies with adequate allocation concealment (RD = 3%, 95% CI 0% to 5%) and those with inadequate or unclear concealment (RD = 10%, 95% CI −2% to 22%). Although the point estimate was larger in studies with inadequate or unclear concealment, the test for subgroup differences was not statistically significant (Chi² = 1.41, df = 1, P = 0.24, I² = 29%). We extracted cost‐effectiveness data from all 10 studies included in this comparison. Monetary incentives, versus no incentives, have an estimated ICER of GBP 3052.96, with incremental costs of GBP 152.65 (consisting of the weighted‐average and inflation‐adjusted amount of the monetary incentive which varied from GBP 5 to USD 500 (approximately GBP 340) across the included studies) and an incremental effectiveness of 5%. The effectiveness of this strategy is also statistically significant. Nevertheless, due to the substantial heterogeneity in the size of the monetary incentives offered, we conducted sensitivity analyses by different levels of monetary incentive (low: up to GBP 5; medium: around GBP 100; high: above GBP 100) to assess the robustness of these findings; the results can be found in Table 7a in Supplementary material 12. None of the studies reported retention data for this comparison. Higher versus lower monetary incentive It is very uncertain whether offering a higher value of monetary incentive affects recruitment compared to a lower monetary incentive: RD = 3% (95% CI −13% to 19%); GRADE: very low (−2 levels: study limitations – high RoB; −1 level: inconsistency; −1 level: imprecision – wide CI crossing the line of no effect); Analysis 64.1; Table. This is based on two studies, which are both trials within a single published article (864 participants): Halpern 2021a ($200) versus Halpern 2021a ($500) (smoking cessation), RoB: high; Halpern 2021b ($100) versus Halpern 2021b ($300) (ambulation), RoB: high. This result and its GRADE assessment apply to middle‐aged males and females, with a range of ethnicities and socioeconomic statuses, living in the USA. For other specific groups, an additional GRADE downgrade for indirectness would be warranted, although the certainty of the evidence is already very low. We extracted cost‐effectiveness data from both studies included in this comparison. Changing monetary incentives has an estimated ICER of GBP 8214.83, with incremental costs of GBP 246.44 (consisting of the weighted‐average and inflation‐adjusted differences in monetary incentives between the intervention and the comparison groups evaluated from the two included studies) and an incremental effectiveness of 3%. However, the effectiveness of this strategy is not statistically significant. None of the studies reported retention data for this comparison. Non‐monetary incentive (branded pen versus no pen) Offering a non‐monetary incentive in the form of a trial‐branded pen alongside the trial invitation probably has little or no effect on recruitment: RD = 0% (95% CI −1% to 1%); GRADE: moderate (−1 level: study limitations – high and unclear RoB); Analysis 65.1; Table. This is based on three studies (9640 participants): Fairhurst 2022 (yoga for elderly patients with multimorbidity), RoB: high; White 2023 (community‐based clinic for patients with persistent, medically unexplained, physical symptoms), RoB: unclear; Whiteside 2019 [121] (falls prevention), RoB: low. This result and its GRADE assessment apply to people in the UK, probably in both urban and rural settings; applicability to other populations is unclear because the authors provided no further participant characteristics. For other populations, certainty would be downgraded for indirectness. We extracted cost‐effectiveness data from all three studies. Trial branded pens, versus no pen, have an undefined ICER, as the difference in effectiveness (the denominator) is zero. This strategy presents an incremental cost of GBP 0.45, consisting of the inflation‐adjusted branded pen costs. Overall, trial‐branded pens result in a small increase in cost with no difference in recruitment. Only one of the three included studies reported retention results (Whiteside 2019), which showed that offering a non‐monetary incentive in the form of a trial‐branded pen alongside the trial invitation probably has little or no effect on retention (at three months): RD = 0% (95% CI −2% to 2%); GRADE: moderate (−1 level: imprecision – single study); Analysis 65.2. This is based on one study: Whiteside 2019 (falls prevention), RoB: low. Discussion Summary of main results Researchers looking to the literature for evidence‐informed recruitment strategies will find an expanding body of studies that continue to offer plenty of variety but little depth. Despite the large number of SWATs, the evidence remains fragmented and methodologically inconsistent; much of it relies on single‐study evidence, limiting opportunities for meta‐analysis and preventing the generation of high‐certainty findings. As a result, the literature continues to offer more questions than answers. We assessed five strategies as having high‐certainty evidence using the GRADE framework, only two of which improve recruitment (using an open‐label trial design compared to a blinded design and using a telephone reminder to contact non‐responders to postal invitations). Both have challenges associated with their use. Blinding is used to support internal validity, which means switching to an open‐label trial as a recruitment intervention is an unlikely proposition in most cases. Contacting non‐responders is widely considered to be cold‐calling, and ethics committees often rule against it. The three other strategies with high‐certainty evidence – optimised participant information leaflet (PIL; bespoke and user‐tested), sending a recruitment primer letter to participants before they are formally invited to a trial, and providing multimedia information via a link or QR code alongside a paper PIL – all make little or no difference to recruitment. We assessed 13 strategies as having moderate‐certainty evidence according to GRADE criteria, mostly embedded within single, well‐conducted studies. We assessed the evidence for all other comparisons as being of low or very low certainty, or as at high risk of bias if insufficient data were available to apply the GRADE criteria. There is only one evaluation of an intervention used pre‐trial to support recruitment (category B), rated as low‐certainty evidence, and no evaluations of a consent‐related intervention (category D) with a GRADE rating above low‐certainty evidence. Little or no effect for optimised PILs as a recruitment strategy Our review provides high‐certainty evidence that optimised PILs (bespoke and user‐tested) do not improve trial recruitment rates. While enhancing clarity is valuable, the effect on the initial decision to enrol is minimal, suggesting several underlying mechanisms. The lack of recruitment efficacy may stem from potential participants not fully engaging with the PIL. The leaflet may not be the critical component in their decision: Dwyer 2022, for instance, found that 89.3% of participants in their SWAT reported deciding to consent before receiving the PIL, despite acknowledging its formal role. This suggests factors like rapport with the clinical team or general attitudes toward research often dominate the enrolment decision, making the content of PILs largely inconsequential to recruitment success. Alternatively, the benefits of PIL optimisation may lie elsewhere, primarily in improved participant understanding, and the quality of informed consent. Evidence supports the idea that user‐testing enhances comprehension, though this effect does not translate into better recruitment [122]. Therefore, while PIL optimisation is an ethical and regulatory necessity, trial teams should not rely on it as a strategy to boost participant recruitment. This is consistent with principles of shared decision‐making and the function of decision aids [123, 124]: these tools improve knowledge and reduce post‐trial regret, but their core purpose is simply to clarify the decision for individuals, ensuring participation is the right choice for them. Cost‐effectiveness Our findings highlight a persistent and significant gap in the evidence on the cost‐effectiveness of recruitment strategies, primarily due to limited reporting of cost data. Of the 91 SWATs included in this review, only 17 reported cost outcomes, which allowed for cost‐effectiveness estimates in just nine out of the 65 comparisons. In six of these nine comparisons, the analysis was based on a single study – either because only one study was available for the comparison, or because other studies in the same comparison did not report cost data. Such evidence gaps result in uncertain and limited cost‐effectiveness findings. The cost‐effectiveness analysis showed wide variation across recruitment strategies, with ICERs ranging from GBP 72.72 (telephone versus SMS reminders) to GBP 8214.83 (monetary incentive value change) per additional participant recruited. These estimates are uncertain due to few contributing studies and wide confidence intervals of incremental effectiveness. Monetary incentives were the only strategy with a statistically significant incremental effect (ICER: GBP 3052.96). However, the cost‐effectiveness results remain uncertain, as the sensitivity analyses showed large variation in ICER values (from GBP 229.33 for an incentive of GBP 5 to GBP 4279.82 for an incentive of GBP 100). Additionally, none of the subgroup incremental effects were statistically significant, which limits the generalisability of these findings. The ICER uncertainty is further compounded by inconsistent cost data reporting across studies; for example, some omitted detailed breakdowns, and others did not account for staff time, which collectively hindered accurate comparisons. To improve this, clear guidance is needed for costing SWAT recruitment strategies. We support the PROMETHEUS programme’s call for streamlined economic evaluations alongside all SWATs to strengthen the evidence base and inform policy [125]. Finally, the absence of variability measures (e.g. standard deviations) surrounding the reported costs, and consequently, the estimated incremental costs, introduces persistent uncertainty into the cost‐effectiveness results presented in our review. Impact of recruitment on retention rates Retention outcomes were reported in 15 SWATs. Only four reported retention at the host trial’s primary outcome time point, while the remaining 11 reported retention at another time point. Overall, the evidence base is limited and highly uncertain, with most comparisons relying on single studies and wide confidence intervals around the estimated effects. Future recruitment SWATs should aim to report retention at the primary outcome time point of the host trial to align with core retention outcome standards and improve comparability. We recognise that waiting until the primary time point may in some cases lead to an unacceptable delay in publishing the recruitment results, and that retention at other time points may be the best that can be included. Regardless, even when the primary aim is recruitment, researchers should routinely aim to collect and report retention data, given the potential for unintended effects on participant follow‐up and engagement. Multimedia information provided to potential participants, particularly when delivered via link or QR code, shows some potential for improving retention and warrants further investigation through replication. Equity, diversity, and inclusion (EDI) The evidence identified in this review was heavily concentrated in high‐income countries in the Global North. Overall, 99% of included studies were conducted in these settings, with 76% coming from the UK or USA. Only one study included participants from middle‐income countries (India, Indonesia, Malaysia, and the Philippines). Consequently, none of the five recruitment strategies identified with high‐certainty evidence have been robustly tested in the Global South or low‐income settings. This stark geographic imbalance significantly limits the transferability of our findings to these contexts, where recruitment barriers, healthcare infrastructures, and cultural norms may differ substantially. A key limitation of the included trials is the unclear and inconsistent reporting of participant sex and gender. While we recorded whether trial teams explicitly used the terms 'sex', 'gender', or both, it is difficult to determine if these terms were used accurately or interchangeably in some articles. Only six of the included studies explicitly reported both sex and gender, which suggests that these specific teams at least recognised the need to distinguish between the two. However, most studies were ambiguous on this point. Our data extraction process sometimes involved changing the recorded variable (e.g. correcting 'sex' to 'gender' or vice versa) depending on whether the primary authors referred to social/cultural roles (men and women) or biological characteristics (males and females), based on the definitions provided in our review guidance from PRO‐EDI [21]. This inconsistency limits the applicability of the evidence and prevents meaningful subgroup analyses based on biological or social differences, ultimately hindering conclusions about whether the recruitment strategies tested are effective for specific population groups. Future recruitment SWATs should say more about who was included in the evaluation and, as a minimum, report the six PRO‐EDI mandatory characteristics of age, sex, gender, ethnicity, socioeconomic status, and location [21]. Trial teams may also wish to utilise the more comprehensive PROGRESS‐Plus framework to guide a deeper consideration of equity issues in their SWATs [126]. Where collection of these data is not possible or permitted (for example, ethnicity data in some jurisdictions), this should be clearly reported. Future research must prioritise evaluating recruitment strategies in the Global South and low‐ and middle‐income countries (LMICs). Addressing this stark geographic imbalance is essential to determine the transferability of these findings across diverse healthcare systems, cultural, and socio‐economic contexts, ensuring the recruitment evidence base is globally representative and applicable. Researchers should aim to understand how contextual factors – such as healthcare infrastructure, digital access, literacy levels, and cultural norms – affect the feasibility and effectiveness of recruitment strategies. Funding bodies and trial networks should actively support and incentivise recruitment methodology research in under‐represented settings to reduce the current Global North bias and to support greater involvement in trials of currently under‐served groups. Collaborative, international SWATs embedded within multinational trials should be encouraged, as they would enable comparative analyses of recruitment strategies across diverse settings and enhance the external validity of findings. Methodological evolution and impact This update introduced several changes to enhance rigour, directly impacting on our findings. Focusing exclusively on randomised trials ensured our meta‐analyses reflected high‐certainty evidence, but this refined scope also highlighted that many previously identified strategies lack the data required to reach robust conclusions. Similarly, including retention, cost‐effectiveness, and EDI provided deeper insights into the long‐term sustainability of recruitment strategies. However, this mostly revealed a significant reporting deficit for these outcomes. Ultimately, these shifts replaced over‐optimistic assessments with a more cautious stance, highlighting a thin evidence base and an urgent need for standardised reporting of retention, costs, and EDI data. Limitations of the evidence included in the review While over half of all included studies were new to this review update, we still have surprisingly little robust evidence supporting strategies that actually improve recruitment to trials. The overall picture of effective strategies to improve trial recruitment remains largely unchanged from the 2018 review [11], and is similar to that reported in 2010 [9, 127]. In other words, almost two decades of research into the effect of strategies to improve trial recruitment has not substantially reduced our uncertainty with regard to which strategies make recruitment more likely. Five recruitment strategies with high‐certainty evidence – three of which do not improve recruitment and two of which present substantial challenges – is thin gruel after 20 years of recruitment research. We should all be concerned. The chief reasons behind this state of affairs include methodology researchers' inclination to evaluate new strategies rather than replicate evaluations of existing strategies. This means the evidence base is heavily fragmented; many strategies, including several with moderate‐certainty evidence, rely on single studies. Consequently, quantitative estimates such as risk differences (RDs) must be interpreted with caution, as they may reflect specific trial contexts rather than a universal effect. This reliance on single studies introduces significant uncertainty regarding the transferability of findings across different clinical areas, trial contexts, and demographics. A strategy’s apparent effectiveness is not guaranteed in new settings, which underscores the urgent need to move beyond single evaluations in single contexts toward co‐ordinated replication to achieve the stability and precision required for high‐certainty evidence. Although we used random‐effects models to provide conservative effect estimates, substantial heterogeneity remains a challenge. We planned subgroup analyses including study setting, disease area, and trial design. However, these were not feasible for most comparisons due to the scarcity of trials per strategy. One exception was monetary incentives, where data allowed us to investigate allocation concealment. In this instance, the effect remained consistent (RD 4%, 95% CI 2% to 7%) across studies with adequate concealment. For most other strategies, the lack of granular data is a limitation, as effectiveness likely depends on the host trial context. Consequently, our pooled estimates represent average effects that may vary in practice. This inability to fully account for heterogeneity reinforces the need for SWAT replications to identify the conditions under which these strategies succeed. Poor reporting also leads to uncertain risk of bias assessments, and we recommend that researchers start to use the reporting standard for SWATs [128]. We strongly urge the trial methodology research community to strengthen the evidence base by prioritising replication of existing strategies over the development and testing of new ones. We urgently need less recruitment strategy innovation and more replication to reach conclusive answers. Future research should replicate promising strategies with moderate‐certainty evidence to strengthen the evidence base, including: SMS messages mentioning trial place scarcity or including participant quotes, telephone reminders, enclosing questionnaires with invitations, multimedia information via digital links or QR codes, and remote site initiation visits. While currently supported by low‐certainty evidence, monetary incentives are also a priority because they are frequently used by trial teams and may improve recruitment. The literature search was conducted in February 2023, and some relevant studies may have been published since then. However, recruitment research continues to be characterised by isolated evaluations rather than systematic replication. In the absence of major ongoing replication studies, it is unlikely that individual studies published since 2023 would materially change the overall conclusions of this review. SWAT initiatives such as MRC START [122, 129], TRECA [130], PROMETHEUS [125], and Implement SWATs have played, or are playing, a central role in the co‐ordinated evaluation of recruitment strategies across multiple trials. Notably, two recruitment strategies rated as having high‐certainty evidence and supported by the largest number of studies – multimedia information provided via link or QR code (Analysis 46.1; Table) and user‐tested participant information leaflets (Analysis 24.1; Table) – were developed through MRC START and TRECA, respectively. PROMETHEUS and Implement SWATs are expected to contribute additional studies in future updates of this review. Reducing uncertainty around recruitment strategies in a timely manner requires sustained focus and co‐ordination across diverse trial contexts. All recruitment strategies have a cost, and generally authors do not tell us what this is. We strongly encourage researchers to report cost data to support formal cost‐effectiveness analyses (see guidance from Gkekas and colleagues [131]), equity, diversity, and inclusion data to inform applicability, as well as retention data to assess potential impacts on longer‐term trial engagement. Collectively, these elements are essential for enabling researchers to make evidence‐based decisions about which recruitment strategies are most appropriate for their contexts. Potential biases in the review process Despite conducting a comprehensive search and screening nearly 21,000 records in this update, we identified three relevant studies outside the formal search process. Two were published after we completed the search, so our search strategy missed only one eligible study. We welcome feedback on any studies we may have overlooked, and on newly published research, to help ensure future updates remain complete and up to date. Authors' conclusions Implications for systematic reviews and evaluations of healthcare Researchers should prioritise well‐designed evaluations of strategies to improve recruitment in randomised trials. We strongly urge the methodology research community to strengthen the evidence base by prioritising replication of existing strategies over the development and testing of new ones. What is needed urgently now is less innovation and more replication. Conclusions about cost‐effectiveness, participant retention, as well as equity, diversity, and inclusion (EDI) were severely limited by sparse, low‐quality data. Most strategies lacked the reporting necessary to assess their cost‐effectiveness or impact on retention. Similarly, the evidence for recruiting under‐served groups is currently too weak to support clear recommendations. To avoid overinterpretation and build a robust evidence base, future Studies Within A Trial (SWATs) must move beyond simple recruitment rates to consistently report cost, retention, and participant characteristics data that support EDI judgements. Implications for methodological research The implications for methodological research are clear. Replication, replication, replication. We need many more replications of SWATs and less innovation. After 20 years of research, only five of 65 comparisons are supported by GRADE high‐certainty evidence. The research community should therefore focus on testing the strategies already identified in this review to strengthen the evidence base, rather than developing new strategies. There are resources available to support trial teams to undertake replications of SWATs, which we outline below. The research community should focus on testing priority recruitment strategies. To accelerate the growth of the evidence base, the research community should replicate strategies identified as priorities for evaluation. The UK's National Institute for Health and Care Research (NIHR)‐funded Implement SWATs project, working together with the SWAT Network, has produced a list of 11 priority recruitment and retention SWATs for trial teams to undertake, which can be found here: https://www.trialforge.org/2024/02/a-list-of-11-priority-recruitment-and-retention-swats/[132]. These SWATs were prioritised based on the frequency with which recruitment and retention strategies were used (assessed using a technique developed by Way and colleagues [133]), evidence from Cochrane systematic reviews [11, 18], and findings from two James Lind Alliance priority‐setting exercises on recruiting and retaining trial participants [134, 135]. Evaluations of these SWATs would be highly valuable. There are promising strategies that need further replications. Future research should prioritise replicating promising strategies with moderate GRADE certainty that probably increase recruitment, such as SMS messages mentioning trial place scarcity or including participant quotes, telephone reminders versus SMS reminders, and enclosing questionnaires with invitations. Multimedia information via digital links or QR codes and remote site initiation visits also warrant testing as potentially cost‐effective methods. Although evidence is currently of low certainty, monetary incentives are a key priority because they are widely used and may improve recruitment. These strategies are the focus of the NIHR‐funded Implement SWATs project, which is currently undertaking coordinated replication SWATs to build a more robust evidence base. The research community should make use of available template SWAT protocols and resources. To support trial teams to undertake replications of SWATs, template protocols, ethics guidance, and other resources for testing priority SWATs have been developed as part of the PRESS project, which are available here: https://www.trialforge.org/trial-diversity/press/. The PRESS project developed protocols and resources to assist and encourage trial teams to evaluate recruitment and retention strategies in their trials, especially those that have been prioritised for evaluation. We need better reporting of the cost‐effectiveness of recruitment strategies. We strongly encourage researchers to report cost data to support cost‐effectiveness analyses. Our findings highlight a significant gap in the evidence on the cost‐effectiveness of recruitment strategies. Of the 91 studies included in this review, we were only able to estimate cost‐effectiveness in nine comparisons. Health economic guidance for undertaking and reporting recruitment and retention SWATs has been developed as part of the PRESS project, and can be found here: https://www.trialforge.org/wp-content/uploads/2025/03/PRESS-Health-Economics-Guidance-for-randomised-recruitment-and-retention-SWATs-V1.0-24.03.2025.pdf. Evidence from the Global South and low‐ and middle‐income countries (LMICs) is needed. Current evidence is heavily skewed towards populations in high‐income, Global North countries (and the UK and USA in particular), with only one study including participants from middle‐income‐country settings. Future research must prioritise evaluating recruitment strategies in LMICs to ensure recruitment science is globally representative and applicable. Applicability and generalisability of findings. Current evidence is concentrated in healthcare settings, leaving strategy effectiveness in environments like social care or education uncertain. Transferability is also limited by the interplay between study context and participant characteristics. Recruitment effectiveness is not universal; it is influenced by a range of factors including disease severity, socioeconomic status, and perceived trial burden. Since most strategies have only been tested in specific clinical contexts, trial teams should exercise caution when applying these methods to different populations or settings. Better reporting of participant populations and contexts is needed. In many cases, limited or no demographic information (e.g. age, gender, ethnicity) is reported for SWAT participants, which restricts the ability to assess how applicable the findings are to specific population groups. Future recruitment SWATs should provide clearer information about who was included in the evaluation, and at a minimum, report the six mandatory PRO‐EDI characteristics, which are: age; sex; gender; race, ethnicity, and ancestry; socioeconomic status; and location (country/countries of data collection and site co‐ordination). More information is available at https://www.trialforge.org/trial-diversity/pro-edi/. We also need better reporting of the impact of recruitment strategies on participant retention. Retention outcomes were reported in only 13 comparisons. The evidence base is limited and highly uncertain, with most strategies supported by single studies and imprecise estimates. As it is entirely possible for recruitment interventions to also affect retention, future SWATs should report retention at the host trial’s primary outcome time point to align with core outcome standards and enhance comparability across studies. Register your recruitment SWAT. Researchers evaluating recruitment strategies should register their protocol on the SWAT Repository (go.qub.ac.uk/SWAT‐SWAR), in addition to registering the SWAT as part of the main trial on trial registries (such as ClinicalTrials.gov, EU‐CTR, and ISRCTN). Consider informing the Trial Forge SWATs Centre. Researchers should consider notifying the York Trial Forge SWATs Centre about their planned recruitment (and other trial process) evaluations to favour better co‐ordination and wider dissemination of evaluation efforts. Consider including a SWAT in all your trial funding applications. Researchers should aim to include SWAT evaluations – especially those identified as high‐priority – in all their trial funding applications. Guidance for researchers applying for funding to conduct high‐priority SWATs of recruitment and retention strategies has been developed as part of the PRESS project, and is available here: https://osf.io/w6zym/files/osfstorage. Funders should support this to avoid another decade with little progress regarding which interventions are effective in improving trial recruitment. Indeed, some funders are increasingly supporting SWATs (e.g. the NIHR and the Wellcome Trust in the UK; the Health Research Board Trial Methodology Network in Ireland), and we strongly encourage more funders to routinely support the embedding of SWATs in the trials they fund. Supporting Information Supplementary materials are available with the online version of this article: 10.1002/14651858.MR000013.pub7. Supplementary materials are published alongside the article and contain additional data and information that support or enhance the article. Supplementary materials may not be subject to the same editorial scrutiny as the content of the article and Cochrane has not copyedited, typeset or proofread these materials. The material in these sections has been supplied by the author(s) for publication under a Licence for Publication and the author(s) are solely responsible for the material. Cochrane accordingly gives no representations or warranties of any kind in relation to, and accepts no liability for any reliance on or use of, such material. Supplementary material 1 Search strategies Supplementary material 2 Characteristics of included studies Supplementary material 3 Characteristics of excluded studies Supplementary material 4 Analyses Supplementary material 5 Data package Supplementary material 6 Protocol Supplementary material 7 Reference list of reviews and other publications searched Supplementary material 8 Full list of strategies Supplementary material 9 Participant numbers per study Supplementary material 10 Data and analyses Supplementary material 11 Equity, diversity and inclusion characteristics of included participants (PRO‐EDI) Supplementary material 12 Cost‐effectiveness of recruitment strategies Supplementary material 13 Impact of recruitment strategies on retention rates