Article Text

Download PDFPDF
PP37 Use of mendelian randomisation to assess whether smoking is causally associated with alcohol consumption
  1. M Taylor1,2,3,
  2. M Hickman1,
  3. G Lewis4,
  4. M Munafò2,3
  1. 1School of Social and Community Medicine, University of Bristol, Bristol, UK
  2. 2MRC Integrative Epidemiology Unit (IEU), University of Bristol, Bristol, UK
  3. 3UK Centre for Tobacco and Alcohol Studies (UKCTAS), School of Experimental Psychology, University of Bristol, Bristol, UK
  4. 4Division of Psychiatry, University College London, London, UK


Background Many studies have shown that tobacco and alcohol use co-occur, with many reporting that tobacco use is a risk factor for later alcohol use. Determining the causality of this relationship from conventional epidemiological studies is difficult due to problems of confounding or reverse causation. Mendelian randomisation utilises the properties of genetic variants to assess causality in the absence of these problems.

Methods Using data from mothers in the Avon Longitudinal Study of Parents and Children (ALSPAC), we investigated observational associations between1 continued smoking (binary variable: yes/no) and alcohol consumption (binary variable: less than one drink per week/ one or more drinks per week) during pregnancy and2 cigarettes per day (continuous measure) in those who smoke and units of alcohol per week (continuous measure) before pregnancy. We also performed a Mendelian randomisation analysis using rs1051730 as the genetic instrument, as this variant has been shown to be associated with both heaviness of smoking in individuals who smoke and reduced ability to quit smoking during pregnancy.

Results Observationally, we found evidence of an association between continuing to smoke and alcohol consumption during pregnancy (N = 1,411; OR = 1.87, 95% CI 1.36 to 2.58; P < 0.001) and between cigarettes per day and units of alcohol consumed per week before pregnancy (N = 1,464; change in units per week for each additional cigarette per day smoked = 0.06, 95% CI 0.03 to 0.10; P < 0.001). In Mendelian randomisation analysis there was no evidence of an association between rs1051730 and any of the alcohol outcome measures: alcohol consumption during pregnancy (where rs1051730 is a proxy for continued smoking during pregnancy, N = 2,462; OR = 0.95, 95% CI 0.82 to 1.10; P = 0.482) and units of alcohol per week before pregnancy (where rs1051730 is a proxy for cigarettes per day in current smokers, N = 2,416; change in units per week for each extra copy of the risk allele = 0.06, 95% CI -0.24 to 0.35; P = 0.710).

Conclusion Observationally, there are strong links between both smoking behaviours and alcohol consumption. However Mendelian randomisation analysis suggests that the associations between smoking and alcohol (both here and in previously reported research) are not causal. This finding should be followed up in larger samples using the Mendelian randomisation approach.

  • Mendelian Randomisation
  • Tobacco
  • Alcohol

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.