Long term smoking and quitting among people with severe mental illness: 3-year follow-up of the SCIMITAR+ Trial

Simon Gilbody,The SCIMITAR collaborative
DOI: https://doi.org/10.1101/2024.06.03.24308386
2024-06-04
Abstract:Background People with severe mental illnesses (SMIs) are three times more likely to smoke than the wider population, contributing to widening health inequalities. Here we report the long term [3 year] outcomes for the SCIMITAR+ trial ( ), which compared usual care to a bespoke smoking cessation package. Methods We recruited 526 heavy smokers with bipolar illness or schizophrenia who were randomly allocated to a bespoke smoking cessation intervention (n=265) or to usual care (n=261) between October 2015 and December 2016. We measured long-term quitting by carbon monoxide-verified smoking status at 3 years post randomisation, and a range of secondary outcomes. Results 261 of the original 526 were followed up. 209 participants (80.1% of followed-up; 39.7% of randomised) could be defined as a smoker or non-smoker for the 36-month primary analysis. Forty-three participants (16.5% of 261) were determined to be a non-smoker via self-reporting and CO confirmation; 21 in the intervention arm (16.3% of 129; 19.6% of those who provided both self-reported and CO measure) and 22 in the control arm (16.7% of 132; 21.6% of those who provided both self-reported and CO measure), adjusted OR of 0.89 (95% CI 0.45, 1.77, p = 0.74). For secondary outcomes, there was no sustained between group difference in reduction in nicotine dependence (measured using the Fagerstrom Test for Nicotine Dependence) or motivation to quit. Some short-term improvements in physical health (measured by the Short Form 12) were present at 36 months. Conclusions A bespoke intervention represents the model for care for mental health services in the UK, but long term quit rates cannot be assumed. Sustained attention to smoking and relapse is likely to be needed to ensure that short term gains are maintained. The certainty of these results is tempered by loss to follow up and low statistical power.
What problem does this paper attempt to address?