The Pulmonary Embolism Rule-out Criteria (PERC) identifies low-risk patients who are treated in the emergency department for suspected pulmonary embolism and for whom testing may be deferred. The purpose of this study is to develop a decision model to determine whether certain elements not included in the PERC methodology could better estimate the testing threshold for pulmonary embolism (ie, the pretest probability below which a patient should not be tested for pulmonary embolism). In addition, we determine which risks and benefits of pulmonary embolism evaluation and treatment have the greatest effect on the testing threshold.

We built decision models of low-risk patients with suspected pulmonary embolism, as determined by the PERC. We obtained model inputs from the literature or by using clinical judgment when data were unavailable. One-way sensitivity analysis derived the testing threshold, and 2-way sensitivity analysis was used to determine the main drivers of the testing threshold.

We found an average testing threshold of 1.4% across all age and sex cohorts. Two-way sensitivity analysis demonstrated that risk of major bleeding from anticoagulation, mortality from contrast-induced renal failure, risk of cancer from computed tomography scan, and mortality from both treated and untreated pulmonary embolism had the greatest effects on the testing threshold.

We found a testing threshold for the PERC similar to that calculated by the Pauker and Kassirer method, using somewhat different assumptions. The 5 major drivers for the testing threshold are variables for which there is a paucity of literature to assess accurately for low-risk patients.