Applicability of decision analysis to everyday clinical practice

Abstract

To determine whether decision analysis is applicable to routine management of suspected pulmonary embolism in an emergency care setting. Controlled feasibility trial. Emergency center of a university hospital. Outpatients (n = 84) admitted with clinical and scintigraphic evidence of pulmonary embolism. Patients were treated either with the usual clinical work-up for pulmonary embolism (control group) or using a decision analysis model with three options: no action: angiography followed by treatment if positive; treatment without angiography. All six senior residents in the decision analysis group agreed to fully participate for the 16 months of the study. Summarizing the decision analysis model in a graph was critical to obtain acceptance from all the physicians. Decision analysis (n = 43) and control (n = 41) patients underwent similar numbers of angiographies. However, angiographies for patients who had intermediate clinical probabilities of pulmonary embolism, between 25 and 75%, were more frequent in the decision analysis group (9/13 = 69%) than in the control group (7/20 = 35%). Agreement between clinical probability and lung-scan result was stronger in the decision analysis group. Decision analysis was successfully used to manage all patients suspected of having pulmonary embolism admitted to an emergency center during the 16-month trial. There was no insuperable obstacle to acceptance of clinical decision analysis by the physicians. Decision analysis may have resulted in a better discrimination between low and intermediate clinical probabilities of pulmonary embolism.