Background: Clinicians frequently use their experience to determine the pretest probability of pulmonary embolism (PE), although scoring systems are promoted as being more reliable. The Canadian Pulmonary Embolism Score (CPES) combines six objective questions and one subjective question. The CPES has been validated and appears to be useful for risk-stratifying patients. However, research suggests that subjective gestalt performs similarly to the CPES, and the influence of the subjective question on the predictive value of the CPES is not clear.
Objectives: To determine the test characteristics of the CPES, its subjective question, and the degree to which the predictive value of the CPES is influenced by its individual questions.
Methods: The authors performed a prospective observational study on a cohort of emergency department patients suspected of having PE. The authors compared patients' CPES results with the diagnosis of PE, calculated the test characteristics of the CPES, and determined the contribution of individual CPES questions to the score's overall predictive value.
Results: Of 607 patients, 61 (10%) had PE. Of low-risk patients (CPES < or =4), 5.54% (n = 449; 95% confidence interval [95% CI] = 3.64% to 8.11%) had PE. The sensitivity (59.0%; 95% CI = 47.4% to 69.8%) and the negative predictive value (94.4%; 95% CI = 92.8% to 95.9%) of the CPES were similar to the sensitivity (53.2%; 95% CI = 40.2% to 65.8%) and negative predictive value (93.5%; 95% CI = 90.7% to 95.5%) of the subjective question alone. In multivariable analysis, nearly all of the predictive value of the CPES was derived from the subjective question.
Conclusions: The predictive value of the CPES appears to be derived primarily from its subjective component.