Future Burden of Cancer Attributable to Current Modifiable Behaviours: A Pooled Study of Seven Australian Cohorts

Abstract

Background: The Population Attributable Fraction (PAF) quantifies the fraction of cancer cases attributable to specific exposures. PAF estimates for the future burden of cancer preventable through modifications to current exposure distributions are lacking. Previous PAF studies have also not compared population subgroup differences. Aim: To apply a novel PAF method and i) assess the future burden of cancer in Australia preventable through modifications to current behaviors, and ii) compare the distribution of the preventable cancer burden between population subgroups. Methods: We harmonized and pooled data from seven Australian cohort studies (N=367058) and linked them to national registries to identify cancers and deaths. We estimated the strength of the associations between behaviors and cancer incidence and death using a proportional hazards model, adjusting for age, sex, study and other risk factors. Exposure prevalence was estimated from contemporary national health surveys. We then combined these estimates to calculate PAFs and their 95% confidence intervals for both individual and joint behavior modifications using a novel method accounting for competing risk of death and risk factor interdependence. We also compared PAFs between population subgroups by calculating the 95% confidence interval of the difference in PAF estimates. Results: During the first 10 years of follow-up, there were 22078 deaths and 27483 incident cancers, including 2025 lung, 3471 colorectal, 640 premenopausal and 2632 postmenopausal breast cancers. The leading preventable cause for lung cancer is current smoking (PAF = 53.7%), for colorectal and postmenopausal breast cancer body fatness or BMI ≥ 25 kg/m2 (PAF = 11.1% and 10.9% respectively), and for premenopausal breast cancer regular alcohol intake (PAF = 12.3%). Three in five lung cancers, but only one in five colorectal and breast cancers, are jointly attributable to potentially modifiable exposures, which also included physical inactivity and inadequate fruit intake for lung, excessive alcohol intake and current smoking for colorectal, regular alcohol intake and current menopausal hormone therapy for 1 year or more for postmenopausal breast and current oral contraceptive use for 5 years or more for premenopausal breast cancer. The cancer burden attributable to modifiable factors is markedly higher in certain population subgroups, including men (lung, colorectal), people with risk factor clustering (lung, colorectal, breast), and individuals with low educational attainment (lung, breast). Conclusion: We provided up-to-date estimates of the future Australian cancer burden attributable to modifiable risk factors, and identified population subgroups that experience the highest preventable burden. Application of the novel PAF method can inform timely public health action to improve health and health equity, by identifying those with the most to gain from programs that support behavior change and early detection.