Cancer Survival of Newly Diagnosed Cases, Quebec, 1992
The main purpose of this study is to describe the quality of the data used to calculate survival in Quebec. The discussion will focus on the quality of incidence and mortality data, the kind of file linkage needed to determine a person’s vital status, and the influence that this data quality has on survival results. This study is also part of a Canada-wide initiative to quantify cancer survival. Statistics Canada and the Canadian Cancer Survival Analysis Group have produced a Canadian Cancer Survival Protocol. To facilitate comparison, the protocol used in this survival study is largely based on the Canadian protocol.
The survival indicator is based primarily on the vital status of the cases reported. Vital status is not reported systematically in the Tumours File. It must be determined by matching the Tumours File against the Deaths File. Unfortunately, the Deaths File is not sufficient. A supplementary linkage with the administrative file of RAMQ subscribers shows that about 4 % of the new cases being studied died in the five years following diagnosis without being recorded in the Deaths File and that half of these were reported only in the FIPA, while the other half were reported in both the FIPA and the QTF. If we limited ourselves only to deaths drawn from the Deaths File, these unrecorded cases would be considered still alive, and survival would be overestimated as a result.
We have also noted that the linkage criteria are very important factors in determining the vital status of reported cases, since the HIN is not always recorded in the Deaths File. In general, Quebec's age-standardized five-year relative survival rates for the major cancer sites are comparable to those in other provinces even though survival duration in Quebec is possibly underestimated as a result of two significant problems. First, the number of new cancer cases is underestimated in the Tumours File because only hospitalized cases and cases treated in day surgery are recorded. Cases diagnosed and treated entirely on an out-patient basis are missing or reported at a more advanced stage of the disease when hospitalization is required. Second, the date of diagnosis in the Tumours File is the date of discharge following the first hospitalization for cancer. These two problems complicate the comparison of survival with other parts of Canada.
The underreporting problem and the survival duration problem could be largely remedied if the Tumours File managers had access to the reports of the laboratories that analyze fluid and tissue samples and confirm pathologies and diagnoses. This solution would not only capture all tumour cases but would also provide accurate data on the diagnosis and actual date of diagnosis.
Also, awareness of the magnitude of underreporting and the characteristics of cases unreported in the QTF as well as of their probability of survival would help to correct and improve the underestimation of survival in Quebec. Our results lead us to believe that hospitalization depends on the morbidity and gravity of the cancer, on duration of survival and on the age of the cancer patient. Therefore, cases not hospitalized and consequently not reported to the QTF are probably those of relatively young people who present a type of cancer with a relatively long survival duration.