E-poster Presentation 2014 World Cancer Congress

Population-level relevance of risk factors for cancer: pooled study of seven Australian cohorts (#798)

Maarit A Laaksonen 1 , Maria Arriaga 1 , Karen Canfell 1 , Robert MacInnis 2 , Emily Banks 3 , Graham Giles 2 , Paul Mitchell 4 , Robert Cumming 4 , Barbara-Ann Adelstein 1 , Julie Byles 5 , Dianna J Magliano 6 , Jonathan Shaw 6 , Janet Grant 7 , Anne Taylor 7 , Claire Vajdic 1
  1. University of New South Wales, Sydney, NSW, Australia
  2. Cancer Council Victoria, Melbourne, VIC, Australia
  3. Australian National University, Canberra, ACT, Australia
  4. University of Sydney, Sydney, NSW, Australia
  5. University of Newcastle, Newcastle, NSW, Australia
  6. Baker IDI Heart and Diabetes Institute, Melbourne, VIC, Australia
  7. University of Adelaide, Adelaide, SA, Australia

Background and Context: Avoidable cancer burden can be quantified using the Population Attributable Fraction (PAF) which estimates the fraction of cancer attributable to modifiable risk factors by combining estimates of the strength of the exposure-cancer association and the exposure prevalence in the population. PAF is best estimated from cohort studies. Competing risk of death must be accounted for if cancer and death share the same risk factors. Large-scale cohort studies or pooled cohort studies are needed for analysis of rarer cancers, risk factor combinations and specific sub-populations.

Aim: To evaluate and compare the cancer burden attributable to lifestyle-related risk factors and their combinations in the Australian population using the most accurate methodology.

Strategy/Tactics: Seven Australian cohort studies (N=369,727), with comprehensive data on exposure to lifestyle-related risk factors (e.g., smoking, alcohol consumption, overweight, diet, exercise) and up to 20 years follow-up, will be linked to the Australian Cancer Database and National Death Index, to collect cancer and death outcomes. Data will be modelled to estimatethe strength of the exposure-cancer association, in the presence of competing risk of death, using proportional hazards model with piecewise constant baseline hazard functions. The exposure prevalence estimates will be calculated from the Australian National Health Survey. Homogeneous cohort-specific PAF estimates will be pooled using inverse-variance weighting and the population-level relevance of the risk factors and their combinations for cancer in Australia, and in various population sub-groups (e.g., low SES), will be ranked.

Programme/Policy Process: We will provide the first estimates of the lifestyle-related avoidable cancer burden, including the most harmful risk factor combinations and the most vulnerable sub-groups, accounting for competing risk of death. 

Outcomes/What was learned:We expect to provide the most accurate information on preventable cancers and cancer prevention target groups in Australia and internationally, essential in prioritising cancer interventions and public health policies.