Maintaining abstinence from smoking after a period of enforced abstinence - systematic review, meta-analysis and analysis of behaviour change techniques with a focus on mental health.

ABSTRACT

BACKGROUND: Smoking prevalence is doubled among people with mental health problems and reaches 80% in inpatient, substance misuse and prison settings, widening inequalities in morbidity and mortality. As more institutions become smoke-free but most smokers relapse immediately post-discharge, we aimed to review interventions to maintain abstinence post-discharge. METHODS: MEDLINE, EMBASE, PsycINFO, CINAHL and Web of Science were searched from inception to May 2016 and randomised controlled trials (RCTs) and cohort studies conducted with adult smokers in prison, inpatient mental health or substance use treatment included. Risk of bias (study quality) was rated using the Effective Public Health Practice Project Tool. Behaviour change techniques (BCTs) were coded from published papers and manuals using a published taxonomy. Mantel-Haenszel random effects meta-analyses of RCTs used biochemically verified point-prevalence smoking abstinence at (a) longest and (b) 6-month follow-up. RESULTS: Five RCTs (n = 416 intervention, n = 415 control) and five cohort studies (n = 471) included. Regarding study quality, four RCTs were rated strong, one moderate; one cohort study was rated strong, one moderate and three weak. Most common BCTs were pharmacotherapy (n = 8 nicotine replacement therapy, n = 1 clonidine), problem solving, social support, and elicitation of pros and cons (each n = 6); papers reported fewer techniques than manuals. Meta-analyses found effects in favour of intervention [(a) risk ratio (RR) = 2.06, 95% confidence interval (CI) 1.30-3.27; (b) RR = 1.86, 95% CI 1.04-3.31]. CONCLUSION: Medication and/or behavioural support can help maintain smoking abstinence beyond discharge from smoke-free institutions with high mental health comorbidity. However, the small evidence base tested few different interventions and reporting of behavioural interventions is often imprecise.