A Prediction Rule for Occurrence of Chronic Thromboembolic Disease After Acute Pulmonary Embolism

BACKGROUND: Occurrence of chronic thromboembolic disease (CTED) after 3 or 6 months of standard and effective anticoagulation is not uncommon in patients with acute pulmonary embolism (PE). To date, there has been no scoring model for the prediction of CTED occurrence. METHODS: A Prediction Rule for CTED (PRC) was established in the establishment cohort (n=1,124) and then validated in the validation cohort (n=211). Both original and simplified versions of the PRC score were provided by using different scoring and cut-offs. RESULTS: (2.186; 1.573-2.799; p=0.001), chronic immobility (2.135; 1.741-2.529; p=0.001), D-dimer >2,000 ng/mL (1.618; 1.274-1.962; p=0.005), PE with deep vein thrombosis (3.199; 2.356-4.042; p<0.001), previous venous thromboembolism (VTE) history (5.268; 3.472-7.064; p<0.001), thromboembolism besides VTE (4.954; 3.150-6.758; p<0.001), thrombophilia (3.438; 2.573-4.303; p<0.001), and unprovoked VTE (2.227; 1.471-2.983; p=0.001). In the establishment cohort, the sensitivity, specificity, Youden index (YI), and C-index were 85.5%, 79.7%, 0.652, and 0.821 (0.732-0.909) when using the original PRC score, whereas they were 87.9%, 74.6%, 0.625, and 0.807 (0.718-0.897) when using the simplified one, respectively (Kappa coefficient 0.819, p-value of McNemar's test 0.786). In the validation cohort, the sensitivity, specificity, YI, and C-index were 86.3%, 76.3%, 0.626, and 0.815 (0.707-0.923) when using the original PRC score, whereas they were 85.0%, 78.6%, 0.636, and 0.818 (0.725-0.911) when using the simplified one, respectively (Kappa coefficient 0.912, p-value of McNemar's test 0.937); both were better than that of the DASH score (72.5%, 69.5%, 0.420, and 0.621 [0.532-0.710]). CONCLUSIONS: A prediction score for CTED occurrence, termed PRC, predicted the likelihood of CTED occurrence after 3 or 6 months of standard anticoagulation in hospitalised patients with a diagnosis of acute PE.