[Comparison of two sources on mortality information in a retrospective cohort study on occupational cancer]

In the context of a historical cohort study among workers in the pulp and paper industry, we examined the agreement between the underlying cause of death obtained from municipal mortality files and from the mortality register of Catalonia (n = 50 pairs). The percentage of simple agreement (P0) for major causes of death was 76%. The agreement and the Kappa index was higher for external causes (P0 = 98%; K = 0.87) and for cancer (P0 = 90%; K = 0.79). The agreement was moderate for mortality from circulatory diseases (K = 0.57). Agreement between 3 digit ICD codes was 48% but was higher for deaths from cancer (P0 = 67%). These results indicate that information on major causes of death and on specific causes of cancer retrieved from municipal mortality files may be used for the follow-up of historical cohort studies.