The primary aim of this study is to assess the effectiveness of an innovative and replicable model for disseminating practice guidelines emanating from NIH Consensus Development Conferences on the Treatment of Early Breast Cancer. We will use a randomized controlled design in a voluntary consortium of hospitals (N = 40) in Minnesota. The intervention will consist of (a) technical assistance to locally recruited oncology opinion leaders who will provide presentations and informal peer consultations, (b) dissemination of attractively illustrated and concise practice guidelines, and (c) group performance feedback by local opinion leaders to physicians. The design and content of this intervention will be informed by an initial data collection phase conducted to document the extent to which practice does not correspond with guidelines and to identify factors associated with non-compliance. During this initial phase we will: (1) obtain baseline data on the extent to which current practice corresponds to treatments identified as effective in the 1990 guidelines for care of breast cancer patients as well as to treatments recognized in earlier (1979, 1980, 1985) guidelines; (2) determine the degree to which a) special medical circumstances, b) patient preferences, c) lack of physician knowledge, and d) physician disagreement account for noncompliance with these guidelines; and (3) identify patient, provider, and hospital characteristics associated with care that does not comply with guidelines. Using the information gained from the first phase of the study we will formulate the content of our intervention and training materials which will be used by opinion leaders in their presentations to and discussions with providers. Following the implementation of our intervention we will assess whether the proportion of patients who receive care that corresponds with the guidelines is greater at intervention hospitals than at non-intervention hospitals, controlling for baseline performance, and determine the cost of the dissemination program.
