Copyright by

Background: One of the central purposes of population-based cancer registries is 7 to identify patterns of cancer incidence and mortality that may be amenable to 8 intervention. To do so it is essential, at a minimum, to have information on tumor grade 9 and stage at diagnosis. Cases identified by a death certificate only, rather than through 10 medical records, do not provide this information thereby limiting the efficacy of the 11 registry. 12 Objective: To identify demographic differences between cancer patients whose 13 death certificates are the only source of information available to a cancer registry and 14 those for whom medical records are also accessible. 15 Sample: There were 330,961 records containing reports of tumors that were 16 diagnosed between January 1st 2007 and December 31st 2011 reported to the Ohio Cancer 17 Incidence Surveillance System and ~5 % were reported through a death certificate only. 18 Methods: We stratified records by cancer site and used logistic regression 19 models to identify demographic differences by death certificate only status. Bernoulli 20 models were used to identify spatial clusters of death certificate only cases compared to 21 medical record cases. 22 Results: Death certificate only cases were associated with older age at diagnosis 2