MEASURING AND IMPROVING PHYSICIAN COMPLIANCE WITH CLINICAL-PRACTICE GUIDELINES - A CONTROLLED INTERVENTIONAL TRIAL

Objective: To determine factors that may lead physicians not to comply with clinical practice guidelines. Design: Retrospective analysis of patients whose physicians were not compliant with discharge recommendations from a prospective, controlled interventional trial of a guideline to reduce hospital length of stay for patients admitted for chest pain. Setting: A large community teaching hospital. Participants: Patients admitted with chest pain who were not discharged according to a practice guideline. Results: 79 (34%) of 230 patients with chest pain classified as being at low risk by concurrent or retrospective review were not discharged by day 3 (the guideline recommendation). Of these 79 patients, 33 (42%) were misclassified at concurrent review (10 were falsely classified as being at high risk and 23 were falsely classified as being at low risk). Of 46 correctly classified patients, 11 (14%) were classified as having noncompliant physicians because of health care system inefficiencies. The status of 7 (9%) patients was changed to high risk between initial classification and potential discharge. For 15 patients (19%), no obvious reason for delayed discharge was found, but they had a higher severity of illness than did low-risk patients discharged according to the guideline as measured by mean time-insensitive predictive instrument scores (41.3% +/- [SD] 14.1% compared with 31.5% +/- 14.3%; P = 0.017). In 13 patients (16%), physicians refused to follow the guideline recommendations. Conclusions: In measuring and attempting to improve physician compliance with a length-of-stay guideline, physician refusal accounts for a small percentage (16%) of noncompliance. Implementation issues, health care system inefficiency, and severity of illness were the predominant reasons why physicians did not comply with guidelines. Our study further supports the principle that clinical practice guidelines should complement rather than be a substitute for physician judgment.