Objective: To describe the hospital activities of radiology groups in the United States, including the characteristics of hospitals at which groups practice, workload and staffing at these hospitals, and groups' professional and business arrangements at these hospitals.
Materials and methods: The American College of Radiology surveyed a stratified random sample of radiology groups, defining a radiology group as any practice with two or more radiologists or radiation oncologists, including private groups, academic or multispecialty groups, and staffs of government facilities. All groups were directed to complete a detailed, two-page hospital information sheet (HIS) for each hospital at which they practiced. Sixty-nine percent of groups responded, and we obtained HISs for 86% of their hospitals. These responses were weighted to make our results representative of all U.S. hospitals at which radiology groups practice.
Results: Radiology groups reported performing on average 47,800 diagnostic procedures annually at hospitals. The average number of diagnostic procedures (inpatient plus outpatient) performed was 0.8 per inpatient day. This number varied greatly from hospital to hospital. A full-time equivalent (FTE) diagnostic radiologist at a hospital performed on average 11,400 procedures annually. Again, variation was large; the 25th percentile was 7000 procedures annually per FTE, and the 75th percentile was 14,100 procedures. Radiation oncology services were mainly provided at larger hospitals (more than 300 beds). On average, the annual number of new oncology patients at hospitals offering radiation oncology services was 380, and these patients received an average of 8435 fractions. Groups accepted nonreferred patients for mammography at 42% of hospitals at which they provided diagnostic services, and they provided 24-hr coverage for diagnostic radiology at 80% of the hospitals at which they provided diagnostic services.
Conclusions: Radiologists can use the findings from this survey to evaluate their hospital practice. The workload data show considerable variation, and averages should not be taken as standards.