The association between schizophrenia and tobacco smoking has been described in more than 1,000 articles, many with inadequate methodology. The studies on this association can focus on: (1) current smoking, ever smoking or smoking cessation; (2) non-psychiatric controls or controls with severe mental illness (e.g., bipolar disorder); and (3) higher smoking frequency or greater usage in smokers. The association with the most potential for genetic studies is that between ever daily smoking and schizophrenia; it may reflect a shared genetic vulnerability. To reduce the number of false-positive genes, we propose a three-stage approach derived from epidemiological knowledge. In the first stage, only genetic variations associated with ever daily smoking that are simultaneously significant within the non-psychiatric controls, the bipolar disorder controls and the schizophrenia cases will be selected. Only those genetic variations that are simultaneously significant in the three hypothesis tests will be tested in the second stage, where the prevalence of the genes must be significantly higher in schizophrenia than in bipolar disorder, and significantly higher in bipolar disorder than in controls. The genes simultaneously significant in the second stage will be included in a third stage where the gene variations must be significantly more frequent in schizophrenia patients who did not start smoking daily until their 20s (late start) versus those who had an early start. Any genetic approach to psychiatric disorders may fail if attention is not given to comorbidity and epidemiological studies that suggest which comorbidities are likely to be explained by genetics and which are not. Our approach, which examines the results of epidemiological studies on comorbidities and then looks for genes that simultaneously satisfy epidemiologically suggested sets of hypotheses, may also apply to the study of other major illnesses.

Document Type


Publication Date


Notes/Citation Information

Published in Human Genetics, v. 131, issue 6, p. 877-901.

The final publication is available at Springer via http://dx.doi.org/10.1007/s00439-011-1122-6.

Digital Object Identifier (DOI)