Understanding Variability in Community Radiology Final rept. (Sept. 1, 2000-Aug. 31, 2005)

摘要

The purpose is to identify sources of variability associated with the interpretation of mammography in a population-based sample of radiologists and mammography facilities. Significant variability exists in the interpretation of mammograms. Despite strong evidence that mammography screening is an efficacious technology for reducing breast cancer mortality, it is unclear whether its full potential has been realized in the community. By better understanding sources of variability in mammography interpretation, we can identify areas for improvement. The ultimate goal is to enhance community mammography performance by reducing the recall rate and false positive interpretations while maintaining high levels of sensitivity and accuracy. Two surveys were developed and administered to community based radiologists and mammography facilities in three geographically distinct locations in the states of Washington, Colorado, and New Hampshire. The radiologist survey included questions on demographics, practice environment, and medical malpractice while the facility survey included questions on facility type, services offered, charges for services, information systems, and interpretation methods. The survey data was linked with performance data obtained from the Breast Cancer Surveillance Consortium (data at the individual mammogram level with patient characteristics, assessment, and cancer outcome). We gathered breast cancer outcome and interpretive data on approximately 500,000 mammograms from over 45 facilities and approximately 139 radiologists. The response rates for both of the surveys were very high with 45 of 53 (85%) facilities and 139 of 181 (77%) eligible radiologists completing the surveys. Important variation was noted at the patient, radiologist, and facility level. Using hierarchical modeling techniques, we explored the extent to which fiscal, legal, clinical, and personal characteristics of radiologists and facilities could be varied to lower the recall rates for community-based mammography while maintaining high levels of accuracy.