Square Pegs, Round Holes: Issues and considerations from running a cluster Randomised Control Trial in people who smoke and experience homelessness

Cluster randomised control trials (cRCT) are a type of RCT whereby instead of individuals being randomised to conditions, randomisation is performed at the setting known as ‘cluster’ level. cRCT’s are appropriate for studies when there is a risk of contamination and where individuals within a group may influence others behaviour (e.g., in health care, educational and community settings). cRCTs invite additional considerations, i) designing and planning the trial i.e., statistical issues, defining the clusters, randomisation, selection bias, and ii) when reporting trial outcomes i.e., reporting suitable statistical approaches to account for violations of the assumption of independence within clusters.
Following recommendations from our own PPI work, a cRCT design was adopted for our trial offering e-cigarettes or usual care to people accessing homeless services. It avoids disagreements between individuals within the same setting that otherwise might occur if randomisation occurred at an individual level, because e-cigarettes are offered which have a monetary value. This presentation will outline some of the key challenges of using a cRCT design for a smoking cessation study within the homeless sector. Examples include: 1) heterogeneity within and between services, and the implications for defining clusters and stratification; 2) Attempts were made to recruit participants via expressions of interest (EOI) before clusters were informed of their allocated arm, to reduce bias and minimise participants only signing up if the centre is known to be in the e-cigarette arm. Yet, issues around sporadic attendance and sometimes distrust and uncertainty of engaging with the unknown, EOI were not always possible and cross-checking EOIs against actual participation posed challenges.
There is a need for more trials to explore treatment approaches for people experiencing homelessness in order to address smoking-related health inequalities. Our experience highlights this is achievable. Applying a flexible and pragmatic approach to cRCT best practice guidelines is challenging but not insurmountable.