Differences in the effectiveness of individual-level smoking cessation interventions by socioeconomic status.

Background: People from lower socioeconomic groups are more likely to smoke and less likely to succeed in achieving abstinence, making tobacco smoking a leading driver of health inequalities. Contextual factors affecting subpopulations may moderate the efficacy of individual-level smoking cessation interventions. It is not known whether any intervention performs differently across socioeconomically-diverse populations and contexts.

Objectives: To assess whether the effects of individual-level smoking cessation interventions on combustible tobacco cigarette use differ by socioeconomic groups, and their potential impact on health equalities.

Search Methods: We searched the Cochrane Database of Systematic Reviews from inception to 1 May 2023 for Cochrane reviews investigating individual-level smoking cessation interventions. We selected studies included in these reviews that met our criteria. We contacted study authors to identify further eligible studies.

Selection Criteria: We included parallel, cluster or factorial randomised controlled trials (RCTs) investigating any individual-level smoking cessation intervention which encouraged complete cessation of combustible tobacco cigarette use compared to no intervention, placebo, or another intervention in adults. Studies must have assessed or reported smoking quit rates, split by any measure of socioeconomic status (SES) at longest follow-up (≥ six months), and been published in 2000 or later.

Data Collection And Analysis: We followed standard Cochrane methods for screening, data extraction, and risk of bias assessment. We assessed the availability of smoking abstinence data by SES in lieu of selective reporting. The primary outcome was smoking cessation quit rates, split by lower and higher SES, at the longest follow-up (≥ six months). Where possible, we calculated ratios of odds ratios (ROR) with 95% confidence intervals (CIs) for each study, comparing lower to higher SES. We pooled RORs by intervention type in random-effects meta-analyses, using the generic inverse-variance method. We subgrouped by type of SES indicator and economic classification of the study country. We summarised all evidence in effect direction plots and categorised the intervention impact on health equality as: positive (evidence that the relative effect of the intervention on quit rates was greater in lower rather than higher SES groups), possibly positive, neutral, possibly neutral, possibly negative, negative, no reported statistically significant difference, or unclear. We evaluated certainty using GRADE.

Main Results: We included 77 studies (73 from high-income countries), representing 127,791 participants. We deemed 12 studies at low overall risk of bias, 13 at unclear risk, and the remaining 52 at high risk. Included studies investigated a range of pharmacological interventions, behavioural support, or combinations of these. Pharmacological interventions We found very low-certainty evidence for all the main pharmacological interventions compared to control. Evidence on cytisine (ROR 1.13, 95% CI 0.73 to 1.74; 1 study, 2472 participants) and nicotine electronic cigarettes (ROR 4.57, 95% CI 0.88 to 23.72; 1 study, 989 participants) compared to control indicated a greater relative effect of these interventions on quit rates in lower compared to higher SES groups, suggesting a possibly positive impact on health equality. CIs for both estimates included the possibility of no clinically important difference and of favouring higher SES groups. There was a lower relative effect of bupropion versus placebo on quit rates in lower compared to higher SES groups, indicating a possibly negative impact on health equality (ROR 0.05, 95% CI 0.00 to 1.00; from 1 of 2 studies, 354 participants; 1 study reported no difference); however, the CI included the possibility of no clinically important difference. We could not determine the intervention impact of combination or single-form nicotine replacement therapy on relative quit rates by SES. No studies on varenicline versus control were included. Behavioural interventions We found low-certainty evidence of lower quit rates in lower compared to higher SES groups for print-based self-help (ROR 0.85, 95% CI 0.52 to 1.38; 3 studies, 4440 participants) and text-messaging (ROR 0.76, 95% CI 0.47 to 1.23; from 3 of 4 studies, 5339 participants; 1 study reported no difference) versus control, indicating a possibly negative impact on health equality. CIs for both estimates included the possibility of no clinically important difference and of favouring lower SES groups. There was very low-certainty evidence of quit rates favouring higher SES groups for financial incentives compared to balanced intervention components. However, the CI included the possibility of no clinically important difference and of favouring lower SES groups (ROR 0.91, 95% CI 0.45 to 1.85; from 5 of 6 studies, 3018 participants; 1 study reported no difference). This indicates a possibly negative impact on health equality. There was very low-certainty evidence of no difference in quit rates by SES for face-to-face counselling compared to less intensive counselling, balanced components, or usual care. However, the CI included the possibility of favouring lower and higher SES groups (ROR 1.26, 95% CI 0.18 to 8.93; from 1 of 6 studies, 294 participants; 5 studies reported no difference), indicating a possibly neutral impact. We found very low-certainty evidence of a greater relative effect of telephone counselling (ROR 4.31, 95% CI 1.28 to 14.51; from 1 of 7 studies, 903 participants; 5 studies reported no difference, 1 unclear) and internet interventions (ROR 1.49, 95% CI 0.99 to 2.25; from 1 of 5 studies, 4613 participants; 4 studies reported no difference) versus control on quit rates in lower versus higher SES groups, suggesting a possibly positive impact on health equality. The CI for the internet intervention estimate included the possibility of no difference. Although the CI for the telephone counselling estimate only favoured lower SES groups, most studies narratively reported no clear evidence of interaction effects.

Authors' Conclusions: Currently, there is no clear evidence to support the use of differential individual-level smoking cessation interventions for people from lower or higher SES groups, or that any one intervention would have an effect on health inequalities. This conclusion may change as further data become available. Many studies did not report sufficient data to be included in a meta-analysis, despite having tested the association of interest. Further RCTs should collect, analyse, and report quit rates by measures of SES, to inform intervention development and ensure recommended interventions do not exacerbate but help reduce health inequalities caused by smoking.