The impact of excluding or including Death Certificate Initiated (DCI) cases on estimated cancer survival: A simulation study

Population-based cancer registries strive to cover all cases diagnosed within the population, but some will always be missed and no register is 100 % complete. Many use death certificates identify additional not captured through other routine sources, hopefully add a large proportion of cases. Cases notified this route, who would have been without certificate information, are referred as Death Certificate Initiated (DCI) Inclusion DCI in increases completeness important for estimating incidence. However, inclusion generally lead biased estimates survival, same often also true if excluding Missed probably random sample cases, rather with poor prognosis. Further, poorer prognosis than general, since they died mentioned on certificates. We performed simulation study estimate impact including or survival estimates, under different scenarios. demonstrated that underestimates survival. The exclusion gives unbiased while overestimated these In our most extreme scenarios, 25 initially missed, usual practice underestimated 5-year by at 3 percentage points.