Cost-minimization analysis of two algorithms for diagnosing acute pulmonary embolism

Introduction: Pulmonary embolism is a common disorder that often requires extensive diagnostic testing. We hypothesized that an algorithmic approach to diagnosis of pulmonary embolism based upon clinical risk stratification and D-dimer testing would be less costly than a standard approach. Materials and methods: We constructed a decision tree based upon two published algorithms for diagnosing acute pulmonary embolism. Branch point probabilities were obtained from the best available published literature. Costs were based upon Medicare charges. From this we obtained a base-case analysis and conducted sensitivity analysis. Results: Our base-case analysis revealed that the cost-per-patient for diagnostic testing were US$216.52 for the algorithm based upon pre-test probability and D-dimer testing and US$538.62 for the standard algorithm. The cost difference per patient evaluated was US$322.10. One- and two-way sensitivity analyses did not reveal any instances in which the clinical risk algorithm was more costly than the standard algorithm. Two-way sensitivity analysis revealed several scenarios in which the standard algorithm would be less costly; however, the conditions required for these scenarios are rarely encountered in clinical practice. Conclusions: Costs of testing using an algorithm based on clinical pre-test probability and D-dimer testing are less than with a standard approach for evaluating suspected acute pulmonary embolism. This new algorithm has previously been shown to be safe and has the potential for large cost savings if widely applied.