Background There is a common perception that smoking generally helps people to manage stress, and may be a form of 'self-medication' in people with mental health conditions. However, there are biologically plausible reasons why smoking may worsen mental health through neuroadaptations arising from chronic smoking, leading to frequent nicotine withdrawal symptoms (e.g. anxiety, depression, irritability), in which case smoking cessation may help to improve rather than worsen mental health. Objectives To examine the association between tobacco smoking cessation and change in mental health. Search methods We searched the Cochrane Tobacco Addiction Group's Specialised Register, Cochrane Central Register of Controlled Trials, MEDLINE, Embase, PsycINFO, and the trial registries clinicaltrials.gov and the International Clinical Trials Registry Platform, from 14 April 2012 to 07 January 2020. These were updated searches of a previously-conducted non-Cochrane review where searches were conducted from database inception to 13 April 2012. Selection criteria We included controlled before-aJer studies, including randomised controlled trials (RCTs) analysed by smoking status at follow-up, and longitudinal cohort studies. In order to be eligible for inclusion studies had to recruit adults who smoked tobacco, and assess whether they quit or continued smoking during the study. They also had to measure a mental health outcome at baseline and at least six weeks later. Data collection and analysis We followed standard Cochrane methods for screening and data extraction. Our primary outcomes were change in depression symptoms, anxiety symptoms or mixed anxiety and depression symptoms between baseline and follow-up. Secondary outcomes included change in symptoms of stress, psychological quality of life, positive aFect, and social impact or social quality of life, as well as new incidence of depression, anxiety, or mixed anxiety and depression disorders. We assessed the risk of bias for the primary outcomes using a modified ROBINS-I tool. For change in mental health outcomes, we calculated the pooled standardised mean diFerence (SMD) and 95% confidence interval (95% CI) for the diFerence in change in mental health from baseline to follow-up between those who had quit smoking and those who had continued to smoke. For the incidence of psychological disorders, we calculated odds ratios (ORs) and 95% CIs. For all meta-analyses we used a generic inverse variance random-eFects model and quantified statistical heterogeneity using I-2. We conducted subgroup analyses to investigate any diFerences in associations between sub-populations, i.e. unselected people with mental illness, people with physical chronic diseases. We assessed the certainty of evidence for our primary outcomes (depression, anxiety, and mixed depression and anxiety) and our secondary social impact outcome using the eight GRADE considerations relevant to non-randomised studies (risk of bias, inconsistency, imprecision, indirectness, publication bias, magnitude of the eFect, the influence of all plausible residual confounding, the presence of a dose-response gradient). Main results We included 102 studies representing over 169,500 participants. Sixty-two of these were identified in the updated search for this review and 40 were included in the original version of the review. Sixty-three studies provided data on change in mental health, 10 were included in meta-analyses of incidence of mental health disorders, and 31 were synthesised narratively. For all primary outcomes, smoking cessation was associated with an improvement in mental health symptoms compared with continuing to smoke: anxiety symptoms (SMD -0.28, 95% CI -0.43 to -0.13; 15 studies, 3141 participants; I-2 = 69%; low-certainty evidence); depression symptoms: (SMD -0.30, 95% CI -0.39 to -0.21; 34 studies, 7156 participants; I-2 = 69%' very low-certainty evidence); mixed anxiety and depression symptoms (SMD -0.31, 95% CI -0.40 to -0.22; 8 studies, 2829 participants; I-2 = 0%; moderate certainty evidence). These findings were robust to preplanned sensitivity analyses, and subgroup analysis generally did not produce evidence of diFerences in the eFect size among subpopulations or based on methodological characteristics. All studies were deemed to be at serious risk of bias due to possible time-varying confounding, and three studies measuring depression symptoms were judged to be at critical risk of bias overall. There was also some evidence of funnel plot asymmetry. For these reasons, we rated our certainty in the estimates for anxiety as low, for depression as very low, and for mixed anxiety and depression as moderate. For the secondary outcomes, smoking cessation was associated with an improvement in symptoms of stress (SMD -0.19, 95% CI -0.34 to -0.04; 4 studies, 1792 participants; I2 = 50%), positive aFect (SMD 0.22, 95% CI 0.11 to 0.33; 13 studies, 4880 participants; I2 = 75%), and psychological quality of life (SMD 0.11, 95% CI 0.06 to 0.16; 19 studies, 18,034 participants; I2 = 42%). There was also evidence that smoking cessation was not associated with a reduction in social quality of life, with the confidence interval incorporating the possibility of a small improvement (SMD 0.03, 95% CI 0.00 to 0.06; 9 studies, 14,673 participants; I2 = 0%). The incidence of new mixed anxiety and depression was lower in people who stopped smoking compared with those who continued (OR 0.76, 95% CI 0.66 to 0.86; 3 studies, 8685 participants; I2 = 57%), as was the incidence of anxiety disorder (OR 0.61, 95% CI 0.34 to 1.12; 2 studies, 2293 participants; I2 = 46%). We deemed it inappropriate to present a pooled estimate for the incidence of new cases of clinical depression, as there was high statistical heterogeneity (I2 = 87%). Authors' conclusions Taken together, these data provide evidence that mental health does not worsen as a result of quitting smoking, and very low- to moderatecertainty evidence that smoking cessation is associated with small to moderate improvements in mental health. These improvements are seen in both unselected samples and in subpopulations, including people diagnosed with mental health conditions. Additional studies that use more advanced methods to overcome time-varying confounding would strengthen the evidence in this area.
