Send letter to journal:
Re: Mental disorders and smoking: Correlation is not causality

A recent article in Tobacco Control 1 reported that 33% of cigarettes are consumed by smokers who had a current mental disorder. The title, abstract and discussion of that article stated that this 33% represented how much “mental disorders contribute to tobacco consumption in New Zealand.” This statement is misleading for at least two reasons. First, although 33% of smokers had a current mental disorder, 21% of nonsmokers also had a current mental disorder; thus, the actual excess that mental disorders could “contribute” is +12%, not 33%. Second, and more importantly, neither this study nor the prior literature has consistently shown that mental disorders per se cause the initiation of smoking, cause smokers to smoke more or interfere with cessation 2. For example, recent reviews have concluded that prior alcohol dependence and depression do not appear to impair smoking cessation 3,4. Since we cannot randomize smokers to mental illness, we must rely on associative data. Among the criteria for judging whether an association is causal 5, the plausibility, replicability, strength of the association, dose-responsivity of the association, and the consistency with other knowledge argue for causality. However, data on whether smoking remits if mental disorders remit, whether the association persists when all reasonable confounds are considered, evidence of specificity, and temporal relationship do not strongly argue causality. For example, a) most mental disorders temporally follow, not precede, smoking, b) those in remission from a psychiatric disorder have not been shown to stop smoking and c) smoking is associated with over 20 different mental disorders suggesting nonspecificity 2. Finally, even if the association was causal, it is unlikely that mental illness accounts for 100% of the reason these smokers smoke. If it accounted for only half, then the excess due to mental disorders would be only +6% (half of +12%). In summary, the “contribution” of psychiatric co-morbidity to the current prevalence of smoking is likely much less than the stated 33%.

References

1. Tobias M, Templeton R, Collings S. How much do mental disorders contribute to New Zealand's tobacco epidemic? Tobacco Control 2008;17:347- 350.

2. Hughes JR. Comorbidity and smoking. Nicotine and Tobacco Research 1999;1:S149-152.

3. Hughes JR, Kalman D. Do smokers with alcohol problems have more difficulty quitting? Drug Alcohol Depend 2005;82:91-102.

4. Hitsman B, Borrelli B, McChargue DE, Spring B, Niaura R. History of depression and smoking cessation outcome: A meta-analysis. J Consult Clin Psychol 2003;71:657-663.

5. Hill AB. The environment and disease: Association or causation? 295-300. 1965.