Tobacco smoking and all-cause mortality in a large Australian cohort study: findings from a mature epidemic with current low smoking prevalence

Journal Club Summary

18 March 2015

Facilitated by Research Fellow Amanda McCullough  



Smoking associated with increased risk of morbidity from conditions such as cancer, respiratory and cardiovascular disease. It is also associated with increased risk of premature death (1). Smoking is a potentially remediable factor whose influence on morbidity and mortality could be reduced through smoking cessation. The relationship between smoking and all-cause mortality is not known for the Australian population.

Paper presented

Banks et al (2) completed a prospective cohort study of Australians of 45 years or older recruited from New South Wales between 2006-2009.   The PICO principle was used to identify this paper’s research question:  

Participants: Australians >45 years old  

Intervention (Exposure): Smoking  

Comparison: Non-smoker or ever smoker  

Outcome: Risk of death  

Time: Mortality data were collected from recruitment until 30 June 2012.

The Cochrane ACROBAT-NRSI tool was used to evaluate risk of bias.(3)

1.   Was selection of exposed and non‐exposed cohorts drawn from the same population?

  • Graded as ‘Definitely yes (Low risk of bias)

2.    Can we be confident in the assessment of exposure?

  • Graded as ‘Definitely yes (Low risk of bias)’ for self-categorisation of smoker, never smoker or ex-smoker
  • Graded as ‘Probably yes’ for self-reported start and stop dates of smoking and for number of cigarettes smoked.

3.   Can we be confident that the outcome of interest was not present at start of study

  • Graded as ‘Definitely yes (Low risk of bias) for risk of death

4.   Did the study match exposed and unexposed for all variables that are associated with the outcome of interest or did the statistical analysis adjust for these prognostic variables?

  • Graded as ‘Definitely yes (Low risk of bias) as the authors had adjusted for prognostic variables in their analysis.

5.  Can we be confident in the assessment of the presence or absence of prognostic factors?

  • Graded as ‘Definitely yes (Low risk of bias) as prognostic factors were identified through a participant-completed survey along with location data.

6.   Can we be confident in the assessment of outcome?

  • Graded as ‘Probably yes.’ Citizens in Australia do not have a unique identification number linking their records, meaning the authors had to link mortality data probabilistically. The authors noted a very small risk (0.4%) of false-positive or false negative rates.

7.   Was the follow up of cohorts adequate?

  • Graded as ‘Definitely yes’ (Low risk of bias)

8.   Were co‐Interventions similar between groups?

  • Not applicable

Summary of results

Smokers were three times as likely to die as non-smokers in Australia (RR 2.96, 95%CI 2.69-3.25). Smokers have the same risk of death 10 years earlier than a non-smoker of 75 years old. From age 45, 45% of male smokers are estimated to die by 75 compared to 19% in male non-smokers, and in 33% and 12% of female smokers and non-smokers, respectively. In ex-smokers, mortality did not differ in those who quit smoking prior to 45 years of age compared to non-smokers.


This article offers an example of a high quality cohort study. Our discussion centred on grading the risk of bias of this study. We agreed that the risk of bias for this study was low; thus, the results can be considered valid. The findings also corroborate those reported in previous similar cohort studies.(4) This study provides an example of how to appropriately conduct a cohort study and we plan to use it as an example for teaching. The study findings could be used to inform the content of motivational messages to encourage smoking cessation. We noted that most of the participants started smoking in adolescence or early adulthood and we believed these age groups could be an ideal target for smoking cessation interventions. 


  1. US Department of Health and Human Services. The health consequences of smoking – 50 years of progress: a report of the surgeon general. Atlanta, GA; 2014.
  2. Banks et al. Tobacco smoking and all-cause mortality in a large Australian cohort study: findings from a mature epidemic with current low smoking prevalence. BMC Medicine 2015; 13: 38.
  3. ACROBAT-NRSI tool available at:
  4. Doll R et al. Mortality in relation to smoking: 50 years’ observations on male British doctors. Br Med J. 2004; 328: 1519–28.