Background: This investigation was undertaken to gain insight into the validity of the American Medical Association (AMA) Masterfile data.
Methods: Allopathic family physicians were chosen as the study population Omissions were picked up from by comparing the AMA list with the 1990 Ohio Academy of Family Physicians Foundation-Ohio Department of Health (OAFPF-ODH) census. Verification of the 1990 specialty and geographic location of allopathic family physicians not common to both files was achieved by sequentially (1) reviewing the AMA names against 1990 deletions from the 1985 OAFPF-ODH census, (2) contacting physicians directly by telephone, (3) verifying 1990 physician status with county medical personnel, and (4) mailing a brief questionnaire to each physician whose 1990 status remained unverified.
Results: The status of specialty and geographic location in 1990 was verified in 91 percent of names not common to both lists. Incorrect omissions (undercounts) and incorrect inclusions (overcounts) offset each other for both lists. Two groups of family physicians contribute to counting biases: family physicians who fulfill short-term goals by part-time practice in several locations, and family physicians who restrict their practice to a limited medical content area.
Conclusions: Because of nearly equal offsetting of overcounting (incorrect inclusions) and under counting (incorrect omissions), the 1990 Ohio family physician AMA Masterfile data is adequate for work-force projections and policy studies when the county data are aggregated at the state level. The overcounting and undercounting for smaller areas or categories must still be studied, however. Application of the AMA Masterfile data of other geographic areas requires a knowledge of the components of undercounts and overcounts of the population being studied.