—To examine whether specialty and system of care exert independent effects on resource utilization.
—Cross-sectional analysis of just over 20 000 patients (≥18 years of age) who visited providers' offices during 9-day periods in 1986. Patient- and physician-provided information was obtained by self-administered questionnaires.
—Offices of 349 physicians practicing family medicine, internal medicine, endocrinology, and cardiology within health maintenance organizations, large multispecialty groups, and solo practices or small single-specialty group practices in three major US cities.
—Indicators of the intensity of resource utilization were examined among four medical specialties (family practice, general internal medicine, cardiology, and endocrinology) and five systems of care (health maintenance organization, multispecialty group—fee-for-service, multispecialty group—prepaid; solo practice and single-specialty group—fee-for-service, and solo practice and single-specialty group—prepaid) before and after controlling for the mix of patients seen in these offices. The indicators of resource utilization were hospitalizations, annual office visits, prescription drugs, and common tests and procedures, with rates estimated on both a per-visit and per-year basis.
—Variation in patient mix was a major determinant of the large variations in resource use. However, increased utilization was also independently related to specialty (cardiology and endocrinology), fee-for-service payment plan, and solo and single-specialty group practice arrangements. After adjusting for patient mix, solo practice/single-specialty groups—fee-for-service had 41% more hospitalizations than health maintenance organizations. General internists had utilization rates somewhat greater than family physicians on some indicators.
—Although variations in patient mix should be a major determinant of variations in resource use, the independent effects of specialty training, payment system, and practice organization on utilization rates need further explication. The 2- and 4-year outcomes now being analyzed will provide information critical to interpretation of the variations reported herein.(JAMA. 1992;267:1624-1630)