Health behaviors of cancer survivors: data from an Australian population-based survey

Objective With increases in cancer survival, promotion of healthy lifestyle behaviors among survivors is receiving considerable attention. This study compared health behaviors among a large sample of Australian adult cancer survivors with an age- and sex-matched cohort of people with no cancer history. Methods Using the Australian National Health Survey, 968 cancer survivors were identified, and randomly matched by age and sex to 5,808 respondents without a history of cancer. Six health behaviors were compared (smoking, physical activity, servings of vegetables, servings of fruit, alcohol use, skin checks), along with overweight and obesity, using polytomous logistic regression analyses controlling for selected chronic conditions. Models were applied across both groups and by tumor site. Results Compared to the non-cancer comparison group, cancer survivors were significantly more likely to be current (OR = 1.35) smokers, particularly those under 40 years (OR = 1.69), and more likely to have regular skin checks (OR = 1.76). Although not significant, there was consistent evidence that cancer survivors were slightly more likely to be overweight or obese (p = 0.065) and have higher levels of alcohol consumption (p = 0.088). There was no evidence of differences between survivors and controls for levels of physical inactivity, vegetable consumption or fruit consumption. Women with a history of gynecological cancers were much more likely to be current smokers (OR = 2.37), while other differences by sex and cancer site were consistent with overall patterns. Cancer survivors were also significantly more likely to report having a range of co-morbid chronic medical conditions. Conclusion Given their increased risk of second cancers and co-morbid chronic conditions, the lack of difference in behavioral risk factors confirms the need for a focus on improving the health behaviors of cancer survivors. Collaborative chronic disease management models may be particularly appropriate in this regard.