• Journal Article

Comorbidity of depression with levels of smoking: An exploration of the shared familial risk hypothesis


Johnson, E., Rhee, S. H., Chase, G. A., & Breslau, N. (2004). Comorbidity of depression with levels of smoking: An exploration of the shared familial risk hypothesis. Nicotine and Tobacco Research, 6(6), 1029-1038. DOI: 10.1080/14622200412331324901


Comorbidity of depression and smoking is well recognized, but results from studies that have assessed alternative explanations have varied by the level of smoking and the study method. We examined all 13 etiology models of comorbidity described by Neale and Kendler (American Journal of Genetics, 57, 935–953, 1995) for depression and each of four levels of smoking to shed light on the role that differing definitions might have played in generating the conflicting findings. Data came from 979 young adults aged 26–35 years who participated in an epidemiological cohort study in southeastern Michigan. Respondent and family history data on parental smoking and depression were analyzed using the biometric modeling method for family data, which Rhee and colleagues (Journal of Child Psychology and Psychiatry and Allied Disciplines, 44, 612–636, 2003; Behavior Genetics, 34, 251–265, 2004) have shown to be valid more frequently than traditional prevalence analyses. Results of the biometric model fitting suggested that for ever smoking, the comorbidity with depression may be related to chance or a high liability threshold for smoking only. In contrast, a correlated liabilities model fit the data best for the comorbidity of depression with daily, heavy, and nicotine-dependent smoking. The familial correlations accounted for 73%–95% of the total variance shared between depression and these levels of smoking. These results differ from analyses of these data using a traditional prevalence approach, which found no evidence of shared familial liability. The conflicting findings of the studies that have examined the relationship between smoking and depression may be attributable to differences in definition of the disorders and the methods used to analyze them.