The completeness and accuracy of health authority and cancer registry records according to a study of ovarian neoplasms.

The completeness and accuracy of Hospital Activity Analysis (HAA) and Regional Cancer Registry (RCR) records were investigated in a series of 868 histologically reviewed cases in which primary ovarian neoplasms had been diagnosed according to one or more of seven data sources including HAA and RCR. All the women concerned were residents of Manchester and Salford who had presented in 1979-83 aged 15 years or more. The histological review confirmed the diagnosis of ovarian neoplasia in 829 of these women and excluded it in 39. Among the 829 confirmed cases, 333 were malignant or of borderline malignancy and therefore eligible for registration with the RCR, and 496 were benign. Only 611 (74%) of the 829 cases were listed as ovarian neoplasms or cysts in HAA records of hospital admissions from the study area during the study period, and the HAA diagnosis was incorrect in 40% of the 611 listed cases. Among the 333 borderline or malignant cases, only 241 (72%) appeared among the RCR's registrations of ovarian neoplasms for the study period and area. The RCR record of histological diagnosis was inaccurate in over 20% of these 241 registered cases, although most of the inaccuracies did not affect whether the neoplasm was classified as borderline or malignant. Five per cent of the cases listed as ovarian neoplasms in the HAA file and 15% of those listed as registrable ovarian neoplasms by the RCR should not have been so listed. The findings highlight the limitations of these routine health information systems, both as sources of cases for research and as National Health Service management tools.