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 recommenda tions from a prospective, controlled interventional trial of a guideli ne to reduce hospital length of stay for patients admitted for chest p ain. Setting: A large community teaching hospital. Participants: Patie nts admitted with chest pain who were not discharged according to a pr actice guideline. Results: 79 (34%) of 230 patients with chest pain cl assified as being at low risk by concurrent or retrospective review we re not discharged by day 3 (the guideline recommendation). Of these 79 patients, 33 (42%) were misclassified at concurrent review (10 were f alsely 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 ca re 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-insensi tive 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 guid eline, physician refusal accounts for a small percentage (16%) of nonc ompliance. Implementation issues, health care system inefficiency, and severity of illness were the predominant reasons why physicians did n ot comply with guidelines. Our study further supports the principle th at clinical practice guidelines should complement rather than be a sub stitute for physician judgment.