Objectives Recruitment in randomised controlled trials (RCTs) is a complex process and includes several stages including the assessment of potential participants for eligibility and successively converting the number of eligible people into consenting, randomised participants. In behavioural RCTs of smoking cessation interventions, poor recruitment is common. In our pool of trials with data available eligibility rates were 61% (95% confidence interval 54% to 66%) and recruitment rates were 69% (95% confidence interval 64% to 73%). However, evidence on which factors predicts poor recruitment in behavioural RCTs of smoking cessation interventions is limited. Individual studies have suggested that participant characteristics (e.g. older age), recruitment strategies (e.g. via internet) and intervention characteristics (e.g. face-to-face) may influence the recruitment process. Therefore, our aim was to identify predictors of eligibility and recruitment rates in smoking cessation RCTs by performing meta-regression analyses on the studies within the IC-SMOKE systematic review project.
Method The IC-SMOKE project is a systematic reviews project of 173 behavioural RCTs of smoking cessation interventions; the protocol is published on PROSPERO (CRD42019121453). Relevant studies were identified in the Cochrane Tobacco Addiction Group Specialized Register. We identified RCTs testing behavioural interventions (with or without pharmacological support) directed at adult smokers (> 18years old) and assessing smoking cessation with a biochemically verified outcome at a minimum of 6 months follow-up between 1996 and 2018. We identified 20 potential predictors based on previous evidence of their predictive value in the published literature and consultation with experts in the field. Predictors were grouped into participant characteristics, recruitment strategy and intervention characteristics and were pre-specified with the meta-regression analyses (https://osf.io/5er49/). Univariate meta-regression analyses were used to investigate the association of predictors on logit-transformed eligibility rates ([#randomised + #declined]/#assessed for eligibility) and recruitment rates (#randomised/#eligible) as the outcomes.
Results 137 and 118 out of 173 studies had complete data and were included in the analyses for eligibility and recruitment rates, respectively. We found no evidence for the recruitment strategy (classified as targeted vs. non-targeted and direct vs. indirect or a combination of both) to predict eligibility rates. Studies including self-help interventions for smoking cessation were associated with 89% lower odds of recruiting eligible participants than person-delivered interventions for smoking cessation (OR 0.11, 95% confidence interval 0.02 to 0.54, p-value 0.007). Additionally, studies recruiting participants with an ongoing mental condition (OR 2.48, 95% confidence interval 0.93 to 6.61, p-value 0.070) and older age (OR 1.04, 95% confidence interval 0.99 to 1.09, p-value 0.094) were associated with higher, but not statistically significant, odds. Participants characteristics such as having a low socioeconomic status or being motivated to quit smoking and the recruitment strategy were not associated with recruitment rates.
Conclusions Our findings suggest that behavioural RCTs of smoking cessation that test interventions without person-to-person contact have more difficulties converting the number of eligible people into consenting, randomised participants. No other participant, recruitment strategy or intervention characteristics were found to be associated with recruitment rates. Due to considerable heterogeneity in trial recruitment and eligibility rates, this study provides little guidance other than intervention mode of delivery, to enhance these rates. Furthermore, we noted that 32% and 21% of eligible RCTs did not report enough data to allow the investigation of recruitment and eligibility rates, respectively. We encourage researchers to follow available guidelines on reporting RCTs (e.g. CONSORT-SPI) to improve reporting of recruitment data, such as the number of people approached or assessed for eligibility, so that their data can contribute to evidence on optimising RCT participation.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.