Completeness of cancer registration in Denmark 1943-1966 and efficacy of record linkage procedures.

The completeness of registration in the national, population based cancer registry in Denmark was evaluated for the first 23 years (1943-1966) of operation. The registry was linked to a complementary data file on 5674 Danish invasive cervical cancer patients enrolled in an international, clinical follow-up study on the basis of identifying information including name, month and year of birth and date of cervical cancer diagnosis. The cancer register had a deficit of 2.2% (95% confidence interval, 1.8-2.6) cervical cancer patients; this figure is low compared to those of other cancer registers. Some 80% of the identified cases were retrieved by computerized matching alone, and an additional 15% were identified by combining computerized and visual matching procedures. Scrutiny of non-retrieved case records revealed that major errors in the identifiers of the cohort used for linkage were responsible for inability to identify an additional 1.7%. The presents study underlines the importance of meticulously recorded, high quality key identifiers in registers, and linkage of cohorts to establish the presence or absence of disease.