Dawkins 2020.

Study characteristics
Methods	Design: Prospective cohort 4‐center pragmatic cluster feasibility trial Recruitment: At homeless centres Setting: 4 homeless centres in the UK Study start date: 1 October 2018; Study end date: 31 March 2020
Participants	Total N: 80 N per arm: EC 48; UC 32 Inclusion criteria: adults (≥ 18 years) who smoke accessing homeless support services on a regular basis and also known to staff;  daily smokers; smoking status was also biochemically verified by exhaled CO breath. Exclusion criteria: non‐smokers, or using another smoking cessation aid; pregnancy, or unable to consent, e.g. currently intoxicated or unable to speak English; not well known to centre staff. Inclusion based on specific population characteristic: people accessing homeless centres 35% women; mean age 42.7; mean cpd 20; mean FTND: FTCD 5.51 Motivated to quit: “varied considerably; large majority expressed a desire to quit smoking in the near future” EC use at baseline: Not specified
Interventions	EC: Refillable Usual care: Written information on quitting smoking (adapted from NHS Choices); signposting to the local stop‐smoking service (SSS) by center staff Intervention: as usual care, plus refillable EC provided once with e‐liquid provided 1 x wk for 4 weeks, Aspire PockeX (tank style), choice of 3 flavors (fruit, menthol, tobacco) and 2 nicotine strengths (12 mg/mL or 18 mg/mL). Written info for EC use and support from center staff, who met once a week to provide e‐liquid and troubleshoot EC use
Outcomes	Weeks: 4, 12, 24; Clinic visits and self‐report Cessation: CO‐validated sustained at 24 weeks Adverse events and biomarkers: Self‐reported negative effects in EC arm only – each participant asked to rate on scale so cannot meta‐analyse; exhaled CO; unintended consequences Other outcomes measured: Qualitative process evaluation; costs; self‐reported positive and negative affects; recruitment rates; retention; EC/other tobacco/nicotine product use at study end; HRQoL; healthcare service utilization; other drug use/dependence; unintended consequences
Study funding	This study is funded by the National Institute for Health Research Public Health (project reference: 17/44/29)
Author declarations	SC, AF, JL, CB, AT, DR, IU, LB, SP have no competing interests. PH has received research grant from and provided consultancy to Pfizer. LD has provided consultancy for the pharmaceutical industry relating to the development of smoking cessation products
Notes	New for 2021 update. Authors provided information prior to peer review
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Intention was to randomize but were unable to due to practical constraints Quote: “Thus the actual allocation of centres to each arm was a pragmatic decision based on centre readiness and staff/researcher availability though we balance potential confounders and differences in environment by ensuring each cluster (EC and UC) contained one day centre and one residential unit.”
Allocation concealment (selection bias)	Unclear risk	Quote: “Participants joined after cluster randomisation… Allocation was concealed to participants until after the baseline assessment.” Comment: But unclear if allocation was concealed for those recruiting, and allocation would have been known to new participants
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not blinded and different levels of support between arms, so performance bias cannot be ruled out
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Cessation (primary outcome) biochemically‐validated
Incomplete outcome data (attrition bias) All outcomes	High risk	13/48 (27.1%) lost to follow‐up in the intervention arm and 20/32 (62.5%) lost to follow‐up in the control arm at 24 weeks
Selective reporting (reporting bias)	Low risk	All anticipated outcomes reported