The burden of cancer attributable to modifiable risk factors: the Australian cancer-PAF cohort consortium

Abstract

PurposeTo estimate the Australian cancer burden attributable to lifestyle-related risk factors and their combinations using a novel population attributable fraction (PAF) method that accounts for competing risk of death, risk factor interdependence and statistical uncertainty.Participants365 173 adults from seven Australian cohort studies. We linked pooled harmonised individual participant cohort data with population-based cancer and death registries to estimate exposure-cancer and exposure-death associations. Current Australian exposure prevalence was estimated from representative external sources. To illustrate the utility of the new PAF method, we calculated fractions of cancers causally related to body fatness or both tobacco and alcohol consumption avoidable in the next 10 years by risk factor modifications, comparing them with fractions produced by traditional PAF methods.Findings to dateOver 10 years of follow-up, we observed 27 483 incident cancers and 22 078 deaths. Of cancers related to body fatness (n=9258), 13% (95% CI 11% to 16%) could be avoided if those currently overweight or obese had body mass index of 18.5–24.9 kg/m2. Of cancers causally related to both tobacco and alcohol (n=4283), current or former smoking explains 13% (11% to 16%) and consuming more than two alcoholic drinks per day explains 6% (5% to 8%). The two factors combined explain 16% (13% to 19%): 26% (21% to 30%) in men and 8% (4% to 11%) in women. Corresponding estimates using the traditional PAF method were 20%, 31% and 10%. Our PAF estimates translate to 74 000 avoidable body fatness-related cancers and 40 000 avoidable tobacco- and alcohol-related cancers in Australia over the next 10 years (2017–2026). Traditional PAF methods not accounting for competing risk of death and interdependence of risk factors may overestimate PAFs and avoidable cancers.Future plansWe will rank the most important causal factors and their combinations for a spectrum of cancers and inform cancer control activities.

The burden of pancreatic cancer in Australia attributable to smoking

“…We analysed individual‐level data for the Australian cancer–PAF cohort consortium, a total pooled study population of 365 084 adults (aged 18 years or more) from seven Australian prospective cohort studies (Box 1). We obtained the most recent sex‐specific risk factor prevalence estimates from the Australian Bureau of Statistics 2014–15 National Health Survey (expanded confidentialised unit record file) and the Australian Institute of Health and Welfare (AIHW) 2013 National Drug Strategy Household Survey (Box 1).…”

“…We harmonised baseline data for these exposures across the cohort studies and health surveys (, table 1), both as continuous variables and classified according to Australian recommendations for healthy living . For smoking, we examined status (never, former, current); for former and current smokers we further evaluated the pancreatic cancer risk by time since quitting (in years) and smoking intensity (cigarettes per day).…”

The preventable burden of breast cancers for premenopausal and postmenopausal women in Australia: A pooled cohort study

“…Although the maximum follow‐up time in our study was 10 years, the median follow‐up time among postmenopausal women was only 5 years, which may have attenuated some of the associations. Furthermore, although the latest available prevalence estimates for both alcohol consumption and OC use date back to 1999–2000, the prevalence of alcohol consumption in Australian women (8% consumed >2 drinks/day in 2001 and 9% in 2014–2015) and the prevalence of hormonal contraception in other similar developed countries appears to have remained stable since then . Finally, the uncertainty in the exposure prevalence estimates should be incorporated in future PAF estimates, as per the uncertainty in the strength of the exposure‐cancer association estimates.…”

“…We used individual‐level data from the Australian cancer‐PAF cohort consortium, which comprises seven well‐established Australian prospective cohort studies, six of which included women: Melbourne Collaborative Cohort Study (MCCS), Blue Mountains Eye Study (BMES), Australian Longitudinal Study on Women's Health (ALSWH), Australian Diabetes, Obesity and Lifestyle Study (AusDiab), North West Adelaide Health Study (NWAHS) and the 45 and Up Study (45&Up). The final study sample comprised 207,505 women after excluding 1,959 who enrolled in more than one cohort, 1,820 who did not consent to record linkage and 7,031 with a history of breast cancer.…”

The future burden of kidney and bladder cancers preventable by behavior modification in Australia: A pooled cohort study

“…We harmonized the exposures across the cohort studies and external prevalence data sources (Supporting Information Table S1), both as continuous variables and classifying them in accordance with current evidence on dose–response relationships or current Australian recommendations for healthy living: not smoking, maintaining a healthy weight (BMI ≤25 kg/m 2 ), not drinking more than two standard alcoholic drinks per day and eating at least two serves of fruit and five serves of vegetables per day . However, as the probable protective effect of alcohol consumption on the risk of kidney cancer may be present at higher levels of consumption, we used a more detailed categorization (0, <1, 1–2, >2–3, >3 drinks/day).…”