Does exclusion of cancers registered only from death-certificate information diminish socio-demographic disparities in recorded survival?

Background Death Certificate Only (DCO) cancer cases are commonly excluded from survival analyses due to unknown survival time. This study examines whether socio-demographic factors are associated with DCO diagnosis, and the potential effects of excluding DCO cases on socio-demographic cancer survival disparities in NSW, Australia. Methods NSW Cancer Registry data for cases diagnosed in 2000–2008 were used in this study. Logistic regression was used to estimate the odds of DCO registration by socio-demographic sub-group (socio-economic disadvantage, residential remoteness, country of birth, age at diagnosis). Cox proportional hazard regression was used to estimate the probability of death from cancer by socio-demographic subgroup when DCO cases were included and excluded from analyses. Results DCO cases consisted of 1.5% (n = 4336) of all cases (n = 299,651). DCO diagnosis was associated with living in socio-economically disadvantaged areas (most disadvantaged compared with least disadvantaged quintile: odds ratio OR 1.25, 95%CI 1.12–1.40), living in inner regional (OR 1.16, 95%CI 1.08–1.25) or remote areas (OR 1.48, 95%CI 1.01–2.19), having an unknown country of birth (OR 1.63, 95%CI 1.47–1.81) and older age. Including or excluding DCO cases had no significant impact on hazard ratios for cancer death by socio-economic disadvantage quintile or remoteness category, and only a minor impact on hazard ratios by age. Conclusion Socio-demographic factors were associated with DCO diagnosis in NSW. However, socio-demographic cancer survival disparities remained unchanged or varied only slightly irrespective of including/excluding DCO cases. Further research could examine the upper limits of DCO proportions that significantly alter estimated cancer survival differentials if DCOs are excluded.