Abstract
BACKGROUND: With improvements in cancer treatment and early detection, the number of people living with cancer is increasing. OBJECTIVE: This study aimed to investigate the association between combined prediagnostic lifestyle factors, assessed by a Healthy Lifestyle Index (HLI) score, and mortality among women diagnosed with breast cancer (BC), colorectal cancer (CRC) and lung cancer (LC). DESIGN: Prospective cohort. SETTING: Women residing in Norway, general population. PARTICIPANTS: Our analysis included 5032, 2468 and 1594 women from the Norwegian Women and Cancer study diagnosed with BC, CRC and LC, respectively, who responded to a questionnaire between 1996 and 2014. EXPOSURE MEASURES: HLI score measured prior to cancer diagnosis. The HLI combines physical activity level, body mass index, smoking behaviour, alcohol consumption and dietary habits. OUTCOME MEASURES: We estimated HRs and 95% CIs using Cox proportional hazard models for all-cause and site-specific cancer mortality. RESULTS: After median follow-up times of 9.8, 7.1 and 5.9 years for BC, CRC and LC cases, respectively, there were 912, 902 and 1094 all-cause deaths; and 509 BC deaths, 679 CRC deaths and 961 LC deaths. For women diagnosed with BC, a 1-point HLI score increment was associated with a 6% lower all-cause mortality rate (HR: 0.94, 95% CI: 0.92 to 0.97). The data were compatible with no association for the estimated 3% lower BC mortality rate (HR: 0.97, 95% CI: 0.94 to 1.00) among women diagnosed with BC, and for the estimated 3% lower all-cause (HR: 0.97, 95% CI: 0.95 to 1.00) and 2% lower CRC mortality rates (HR: 0.98, 95% CI: 0.95 to 1.01) among women diagnosed with CRC. For women diagnosed with LC, prediagnostic HLI score was not associated with all-cause (HR: 1.00, 95% CI: 0.98 to 1.02) or LC mortality rates (HR: 1.00, 95% CI: 0.98 to 1.03). CONCLUSIONS: We observed that a higher HLI score measured before cancer diagnosis was associated with lower all-cause and, possibly, lower BC mortality among Norwegian women diagnosed with BC; and a possible lower all-cause and CRC mortality among those diagnosed with CRC. Smoking was likely responsible for the observed associations.