Role of the User in a Combined “Man and Machine” Effort to Improve the Quality of Data in a Large Cancer Registry

Studies using the newly created Ontario Cancer Registry (OCR) identified a number of duplicate records. These could be traced to incorrect spelling of surnames which did not permit incoming records to be correctly compared during the routine computerized record linkage process. The method of elimination of the majority of these duplicate records and the results will be presented. Test files were established. Files were sorted initially to facilitate manual identification of duplicates. Patterns used manually to link duplicate records were adopted into the computerized record linkage system. This modified computer linkage was then applied to the master incidence file with the elimination of more than 2.5% duplicate records. The proportion of duplicate records varied by sex and year, being greatest for the more recent years. A similar approach could be developed by other registries where computer facilities are available.