Are recent attempts to quit smoking associated with reduced drinking in England? A cross-sectional population survey

This study found that smokers who reported attempting to stop within the last week had lower levels of alcohol consumption, especially bingeing, and were less likely to be classified as having higher risk alcohol consumption (AUDIT-C ?5) compared with those who did not report an attempt to quit smoking in the last week. Among those with higher risk alcohol consumption, smokers who reported attempting to stop smoking within the last week compared with those who reported no attempt were more likely to report also currently trying to restrict their alcohol consumption.

This study adds to the literature on the close relationship between smoking and alcohol consumption [5]. One component of the relationship is that alcohol consumption is associated with lapse and relapse to smoking [13–17], which has resulted in smokers being widely advised to restrict their consumption during quit attempts [13, 18, 19]. In the current study, the association between a recent attempt to quit smoking and reduced alcohol consumption indicates that smokers in England may be following this best-practice advice to restrict their alcohol consumption during a smoking cessation attempt. It is not possible with cross-sectional epidemiological data to rule out reverse causation i.e., the possibility that the association between quitting and consumption may actually be driven by people with lower alcohol consumption being more likely to attempt to quit smoking. If this were the explanation, the association would remain important because it would suggest the need for smokers with higher alcohol consumption to be targeted for further encouragement to attempt to quit smoking. However, another finding in this study indicates that the association is unlikely to be driven exclusively by lighter drinkers being more likely to attempt quit smoking: among smokers who were also heavier drinkers, those who had made an attempt to quit smoking within the last week compared with those who had not were also more likely to report a current attempt to restrict their alcohol consumption. The present study cannot determine whether attempts to quit smoking tend to precede attempts to restrict alcohol consumption, or vice versa.

These findings have possible implications for policy evaluation and development: there appears to be a need for greater attention to possible crossover effects when evaluating the cost-effectiveness of alcohol and tobacco interventions and more reason for a coordinated strategy on alcohol and tobacco control. Policy on brief intervention by health professionals is one example. Brief intervention for smoking and alcohol are both effective and cost-effective interventions [30–34]. Delivery of smoking brief intervention is much more common in England than is alcohol [35], and there is a need to increase the rate of screening and brief intervention on alcohol [36, 37]. The current study suggests that smokers may be more likely to reduce their alcohol consumption when attempting to stop smoking than when they are not. While these findings cannot speak to the effectiveness of brief interventions, they do suggest that a smoking brief intervention may be a good opportunity to intervene also on alcohol: smokers may be likely to plan to reduce their alcohol consumption regardless and may therefore be particularly receptive to intervention on alcohol. However, this is an empirical question for which there is currently sparse experimental evidence [38–40], and until such evidence is forthcoming, other strategies to increase alcohol brief intervention may warrant greater resource [41–45]. In the meantime, the current findings could be simply disseminated to health professionals to reassure them that many smokers may be planning to cut down on their drinking anyway and their intervention on alcohol may be therefore unlikely to compromise the patient relationship: the GP-patient relationship is a regularly cited barrier to a greater rate of brief intervention, albeit one of several including inadequate training, and lack of time or financial incentives [46–52].

There are three important limitations of this study. The limitation on interpretation of cross-sectional design in relation to direction of causation has been discussed. A second limitation is that as an observational study it is possible that unmeasured confounding could have influenced the results. For example, it is possible that the diagnosis of a health problem led to attempts to cut down on both drinking and smoking (i.e., cross-behavioural sick-quitter effects). Our adjustment for a self-reported disability may not have sufficiently accounted for this possibility. Another limitation is that the study relied on self-reported data with the risk of socially desirable responding. However, in population surveys the social pressure and related misreporting of smoking is low and it is generally considered acceptable to rely on self-reported data [53], while we used an abbreviated version of a high quality tool that has been widely validated to assess alcohol use disorder (AUDIT-C; [27, 54]). The full version of the AUDIT questionnaire is more widely validated but includes questions across a longer reference period that would have rendered the scale less sensitive to recent changes, while AUDIT-C has good validity, excellent reliability and responsiveness to change [54].