Reliability of Record Linkage in the Swedish Cancer-Environment Register

Abstract

The Swedish Cancer-Environment Register (CER) is intended to be used for studies on occupational cancer. CER was established by a computerized record linkage between the Swedish National Cancer Register (SCR) for 1961–1973 and the 1960 Population and Housing Census (PH 60) in order to obtain information on occupation, occupational status, economic activity (industry), place of work, domicile, etc. that might be useful for epidemiologic studies on cancer. Accurate inter-register linkage at individual level necessitates high quality of identification in both registers. The hit completeness in this respect was evaluated as 98.8 per cent, a figure regarded as acceptable for most studies based on the CER. Since a hit between SCR and PH 60 did not guarantee accuracy, the reliability of the record linkage was studied in random sample. With the aid of local and national population registries, 0.45 per cent (95% confidence interval 0.21-0.86%) false linkages were revealed in the sample. The CER was in 1982 supplemented with data notified to the SCR 1974–1979, and the hit completeness was 99.2 per cent for the total period. There are good grounds for assuming a lower rate of false linkages for the later period.