There was no evidence for a casual effect of smoking on suicidal ideation or suicide attempts, according to study findings published in The British Journal of Psychiatry.

The literature suggests a strong association between smoking and suicide-related behaviors, with a positive dose-response relationship after controlling for confounding factors such as psychiatric symptoms, familial risk, socioeconomic background, and alcohol consumption. However, the causality of the association between smoking and suicide-related behaviors is unknown.

In this study, researchers used Mendelian randomization to identify a potential causal relationship between genetic variants associated with smoking and suicide-related behaviors. Mendelian randomization reduces bias from confounding factors and reverse causation. The researchers performed observational analyses using data from the UK Biobank. Participants with available genetic data and of European ancestry (N=337,053) responded to questionnaires assessing smoking status, suicidal ideation, and suicide attempts.

Continue Reading

Smoking initiation was assessed as smoking at any point in participants’ lifetime. Among participants who had smoked at any point in life, smoking was further assessed by active status, cigarettes per day, and lifetime smoking score which was determined as a combination of smoking duration, cessation, and heaviness. Mendelian randomization used genome wide association study (GWAS) data, which included 2433 individuals with suicide attempts and data from the Sequencing Consortium of Alcohol and Nicotine use (GSCAN) on 378 SNPs associated with smoking initiation.

Related Articles

The observational analysis of UK Biobank participants who offered responses on suicidal ideation (n=109,649) showed that every smoking behavior increased the odds of suicidal attempts and ideation. Smoking initiation showed the strongest association with attempted suicide (odds ratio [OR], 2.07; 95% CI, 1.91-2.26, P <.001).

While the observational analysis showed an association between smoking and suicide-related behavior, the Mendelian randomization analyses showed that this association was not causal. Similar to the observational data, SNPs for smoking initiation increased the odds of attempted suicide (OR, 2.42; 95% CI, 1.53-3.83, P<.001). However, lifetime smoking was not associated with suicidal ideation (OR, 0.050, 95% CI, -0.027-0.127) or attempted suicide (OR, 0.053, 95% CI, -0.003-0.110).

The researchers next examined if the link between smoking initiation and attempted suicide could be explained by impulsivity as a common factor. A Mendelian randomization analysis between smoking initiation and risk-taking showed strong evidence of a bidirectional causal relationship. The researchers concluded that the smoking initiation SNPs were most likely overlapping with an impulsivity phenotype.

The study may be limited by its relatively small sample size and inclusion of only failed suicide attempts. Bias from reverse causation could be another limiting factor.

“Our results suggest that, despite observed associations . . . there was no clear evidence for a casual effect of smoking behavior,” the researchers concluded, “This supports recent evidence suggesting that suicidal ideation and suicide attempts are different processes.” They also noted that their findings emphasize the difficulty of separating behavioral and biological factors related to smoking.


Harrison R, Munafò MR, Davey Smith G, Wootton RE. Examining the effect of smoking on suicidal ideation and attempts: triangulation of epidemiological approaches [published online ahead of print, 2020 Apr 15]. Br J Psychiatry. 2020;1-7. doi:10.1192/bjp.2020.68.