Quantifying the absolute number of cancer deaths that would be avoided if cancers were diagnosed prior to progressing to distant metastasis, New South Wales, Australia 1985‐2014

Our study measures the impact of diagnosing cancers early before they metastasise on reducing the burden of cancer death. A cohort of 716 501 people aged 15 to 89 years diagnosed with a solid cancer in New South Wales, Australia, during 1985 to 2014 were followed-up to December 2015. Crude probabilities of cancer death by stage at diagnosis were calculated for all solid cancers combined and five individual cancers using flexible parametric relative survival models. These probabilities were used to estimate the number of avoided cancer deaths within 10 years of diagnosis in three 10-year diagnostic periods if all cases with known distant stage were instead diagnosed at an earlier stage. Cancers are known to be diagnosed at distant stage composed ~16% of all solid cancers diagnosed during 2005 to 2014. Assuming all these cases were instead diagnosed at regional stage, an annual average of 2064 cancer deaths would have been potentially avoided within 10 years of diagnosis. This equated to ~21% of modelled observed deaths. Alternatively, if half of all known distant cases diagnosed during 2005 to 2014 were diagnosed as regional and half as localised, the average number of deaths avoided per year would increase to 2677 (~28%). Estimates varied by diagnostic period, sex and cancer type, reflecting both the different stage distributions for the cancer types, and the respective survival differences between cancer stages. While prevention is the most effective pillar of cancer control, these findings quantify the potential benefits of diagnosing all cancer types when they are less advanced to reduce the burden of cancer mortality.