Predicting difference in mean survival time from cause-specific hazard ratios for women diagnosed with breast cancer

Abstract

Background: Relative reduction in breast cancer mortality is the preferred outcome measure for evaluation of mammography screening. However, mean survival time has been advocated as a better and more intuitive outcome for risk communication. We have previously introduced a method to predict difference in mean survival time from empirical hazard ratios for all-cause mortality. In this article, we aim to investigate the association between hazard ratios for breast cancer mortality and the difference in mean survival time for women diagnosed with breast cancer. Methods: We retrieved data on all women diagnosed with first-time invasive breast cancer in Norway from 1960 through 2004. Women were followed until emigration or end of follow-up on 31 December 2015, whichever came first. Observed differences in mean survival times and hazard ratios for both breast cancer death and death from causes other than breast cancer were obtained for neighbouring time periods defined by women's age and year of diagnosis. Based on previously developed methods, we fitted a linear relationship between observed differences in mean survival and logarithmic hazard ratios. Results: A linear association was found between breast cancer-specific hazard ratios and difference in mean survival time for women diagnosed with breast cancer. This association was also estimated with adjustment for other causes of death than breast cancer. Conclusions: The change in mean survival time could be predicted from an estimated reduction in breast cancer mortality. This outcome measure can contribute to better and more understandable risk information about the effect of mammography screening programmes.