RISK FACTOR INFORMATION FOUND IN MEDICAL RECORDS OF LUNG AND PROSTATE-CANCER CASES, NEW-MEXICO-TUMOR-REGISTRY (UNITED-STATES)

The conduct of public health surveillance and epidemiologic investigat ions would be enhanced by a cost-effective method for the collection o f population-based cancer risk-factor data. We assessed the feasibilit y of ascertaining such data from hospital medical records during routi ne abstraction of information from a cancer registry. We examined the medical records of a representative sample of prostate and lung cancer cases diagnosed during 1992 and 1993 from the New Mexico (United Stat es) Tumor Registry (NMTR) database. Registry personnel abstracted anci llary information from the medical records of 575 prostate and 273 lun g cancer cases. Of medical records from 848 cases, 90 percent document ed a history of tobacco use, 77 percent contained any family medical h istory, and 48 percent documented occupational information sufficient to allow coding into standard occupational groups. Availability of occ upation and industry varied by patient ethnicity, age at diagnosis, ma rital status, and type of hospital, indicating the potential for bias in studies using occupational information from medical records. Our fi ndings suggest that risk factor information abstracted from medical re cords may provide valuable information for public health surveillance, but is generally too incomplete to serve as a single source of exposu re information for etiologic studies. Family histories in medical reco rds may be sufficient for population-based ascertainment of affected r elative pairs for genetic epidemiology studies for some cancer types.