Age-dependent diagnostic accuracy of clinical scoring systems and D-dimer levels in the diagnosis of pulmonary embolism with computed tomography pulmonary angiography (CTPA).

OBJECTIVE: The aim of this study was to compare the age-dependent diagnostic performance of clinical scores and D-dimer testing to identify patients with suspected pulmonary embolism (PE).

METHODS: Consecutive patients with suspected PE referred from the emergency department for computed tomography pulmonary angiography (CTPA) were retrospectively evaluated. Diagnostic scores (classic Wells score (WS), modified WS, simplified WS, revised Geneva score (GS), simplified GS, and YEARS score) were calculated from medical records. Results of D-dimer testing were retrieved from the laboratory database. CTPA was the diagnostic reference standard. Four age groups were analyzed (< 50, 50-64, 65-74, and ≥ 75 years). Statistical analysis used receiver operating characteristics as well as uni- and multivariate analyses with calculation of prediction models. The study was IRB approved.

RESULTS: One thousand consecutive patients were included. Areas under the curve (AUC) and accuracies were superior in patients < 50 years. For the classic WS, the AUC decreased by 11% with the optimal cutoff dropping 1.5 points in patients ≥ 75 years; for D-dimer levels, the optimal cutoff was 900 μg/L higher in both ≥ 65 years groups with a max. decrease of the AUC of 9%. In terms of accuracy, the YEARS score performed best across all groups. Classic WS and D-dimer level showed a significant interaction with patient age in prediction models.

CONCLUSION: D-dimer measurement and clinical scores perform best in patients < 50 years. The YEARS score performs best across all age groups and is therefore recommended.

KEY POINTS: • The probability of pulmonary embolism predicted by fibrin fibrinogen degradation products and clinical scores shows the highest accuracy in patients < 50 years. • The probability of pulmonary embolism predicted by the YEARS score shows the highest accuracy in each age group. • Classic Wells score and fibrin fibrinogen degradation products show a significant interaction with patient age in a logistic regression model.