Introduction: Assessment of pretest probability should be the initial step in investigation of patients with suspected pulmonary embolism (PE). In teaching hospitals physicians in training are often the first physicians to evaluate patients.
Objective: To evaluate the accuracy of pretest probability assessment of PE by physicians in training using the Wells clinical model and to assess the safety of a diagnostic strategy including pretest probability assessment.
Patients and methods: 291 consecutive outpatients with clinical suspicion of PE were categorized as having a low, moderate or high pretest probability of PE by physicians in training who could take supervising physicians' advice when they deemed necessary. Then, patients were managed according to a sequential diagnostic algorithm including D-dimer testing, lung scan, leg compression ultrasonography and helical computed tomography. Patients in whom PE was deemed absent were followed up for 3 months.
Results: 34 patients (18%) had PE. Prevalence of PE in the low, moderate and high pretest probability groups categorized by physicians in training alone was 3% (95% confidence interval (CI): 1% to 9%), 31% (95% CI: 22% to 42%) and 100% (95% CI: 61% to 100%) respectively. One of the 152 untreated patients (0.7%, 95% CI: 0.1% to 3.6%) developed a thromboembolic event during the 3-month follow-up period.
Conclusion: Physicians in training can use the Wells clinical model to determine pretest probability of PE. A diagnostic strategy including the use of this model by physicians in training with access to supervising physicians' advice appears to be safe.