The validity of the mortality to incidence ratio as a proxy for site-specific cancer survival

Background: The complement of the cancer mortality to incidence ratio [1-(M/I)] has been suggested as a valid proxy for 5-year relative survival. Whether this suggestion holds true for all types of cancer has not yet been adequately evaluated. Methods: We used publicly available databases of cancer incidence, cancer mortality and relative survival to correlate relative survival estimates and 1-(M/I) estimates from Denmark, Finland, Iceland, Norway, Sweden, the USA and the Netherlands. We visually examined for which tumour sites 5-year relative survival cannot simply be predicted by the 1-(M/I) and evaluated similarities between countries. Results: Country-specific linear regression analyses show that there is no systematic bias in predicting 5-year relative survival by 1-(M/I) in five countries. There is a small but significant systematic underestimation of survival from prognostically poor tumour sites in two countries. Furthermore, the 1-(M/I) overestimates survival from oral cavity and liver cancer with >10 in at least two of the seven countries. By contrast, the proxy underestimates survival from soft tissue, bone, breast, prostate and oesophageal cancer, multiple myeloma and leukaemia with >10 in at least two of the seven countries. Conclusion: The 1-(M/I) is a good approximation of the 5-year relative survival for most but not all tumour sites.