Comparative performance of khorana, padua, and improve-VTE risk assessment models for venous thromboembolism risk stratification in hospitalized oncology patients

Abstract Background: Venous thromboembolism (VTE) occurs in up to 8% of patients with cancer and is associated with increased morbidity and mortality (Van Es et al., 2020). While the Khorana risk assessment model (RAM) has demonstrated efficacy in VTE risk stratification to identify patients with cancer who may benefit from primary VTE prophylaxis, its validity is limited to the ambulatory setting (Overvad et al., 2022). The primary objective of this study is to compare the performance of the Khorana, Padua, and IMPROVE-VTE RAMs for determining VTE risk in hospitalized medical oncology patients. Methods: This was a single-center, retrospective analysis on the performance of the Khorana, Padua, and IMPROVE-VTE RAMs in patients ≥ 18 years of age with a cancer diagnosis admitted to a medical oncology service or intensive care unit at The Johns Hopkins Hospital in Baltimore, MD. Patients prescribed anticoagulation prior to admission were excluded. Our primary outcomes were the sensitivity and specificity of each scoring tool using the validated cutoffs. The validated cutoffs were defined as ≥ 2 for IMPROVE-VTE, ≥ 4 for Padua, and ≥ 2 for Khorana. Results: This study included 788 patients with a median age of 64 years. The most common primary cancer types included pancreatic (9.9%), sarcoma (9.5%), lung (8.5%), colorectal (8.5%), breast (7.9%), head and neck (7.8%), and CNS lymphoma (7.3%). Prophylaxis was prescribed upon admission for 75.4% of patients. A total of 42 VTE events occurred (5.3%); including pulmonary embolism (45.3%) and lower extremity deep vein thrombosis [DVT] (35.7%). All 42 patients that experienced a VTE event were classified as high risk for VTE by the IMPROVE-VTE RAM. The Padua RAM and the Khorana RAM appropriately identified 78.5% and 33.3% of patients who experienced a VTE event as high risk for the development of a VTE, respectively. Finally, the IMPROVE-VTE, Padua, and Khorana RAMs provided 100%, 78.0%, and 33.3% sensitivity and 0%, 26.8%, and 60.9% specificity, respectively. Conclusions: Overall, the Khorana RAM demonstrated poor sensitivity and would not be beneficial for use in the inpatient medical oncology setting. The Padua RAM provided the best sensitivity but still leaves a chance of one in five patients being misidentified as low risk for VTE. Finally, the IMPROVE-VTE RAM was unable to distinguish patients who are at low risk for the development of VTE. This model is consistent with the NCCN and ASCO guidelines that recommend universal prophylaxis for hospitalized patients with cancer. However, this approach may expose low-risk patients to the bleeding risk associated with pharmacologic VTE prophylaxis. Overall, these results suggest the need for a better RAM in hospitalized patients with cancer to appropriately distinguish between high versus low VTE risk.