Poor agreement of occupational data between a hospital-based cancer registry and interview

With occupation recognized as a risk factor for various cancers, collecting occupation and industry data by a number of vital registries, including cancer registries, has developed. Registries may be data sources for cancer etiology research and occupational disease surveillance, despite concerns that their data are fragmentary and may lack validity. To improve completeness and validity of occupational information in a hospital-based cancer registry, this study compared information obtained through abstracting medical records for the registry with information obtained through lung-cancer patient interviews. Employing the kappa statistic, agreement was generally poor, largely due to data missing in the medical record. Data quality of hospital-based cancer registries can be improved by employing trained cancer registrars to elicit occupational histories from patients.