Longitudinal analysis of the relationship between changes in smoking and changes in drinking in a community sample: the Winnipeg Health and Drinking Survey.

Smoking cigarettes and drinking alcohol are positively correlated in cross-sectional studies of the general population. However, it is unclear whether changes in quantity of drinking over time are related to changes in amount of smoking over time. This investigation examined, with structural equation modeling, the relationship of changes in drinking to changes in smoking over 2 years among 344 adults who reported cigarette smoking and alcohol use at baseline in 1989-1990 or at follow-up in 1991-1992 or both. Surprisingly, no significant relationships were found between changes in smoking and changes in drinking. This lack of effect suggests that changes in the quantity or intensity of drinking and of smoking are not related in any important way in nonclinical populations.