Utilization and Clinical Outcomes of an Electronic Best Practice Alert to Prevent Venous Thromboembolism in High-Risk Cancer Patients

Background

Cancer-associated venous thromboembolism (VTE) is associated with significant morbidity, increased mortality, and increased utilization of health care resources. The Khorana score (KS) is a validated tool used to identify ambulatory cancer patients at high risk of developing VTE. Several guidelines now recommend considering primary VTE prophylaxis with a low dose direct oral anticoagulant (DOAC) in patients with KS ≥2.

To help identify high-risk patients, an electronic best practice alert (BPA) was created for our ambulatory cancer center providers to identify cancer patients with a KS ≥2. The alert was integrated into the electronic health record of Cleveland Clinic’s ambulatory cancer center. We conducted a cohort study to identify utilization of BPA and patient outcomes.

Methods

The alert screened for cancer patients with an oncology plan applied within the past 30 days. Points were assigned for type of cancer, BMI, hemoglobin, platelet count, or white blood cell count as defined by the KS within 14 days of alert fire; patients with primary brain tumors, upper gastrointestinal or genitourinary cancers were not included in the BPA. Departmental education was provided prior to BPA launch on 3/2/2021. After BPA launch, we evaluated a consecutive cohort of patients seen at the Cleveland Clinic from 5/13/2021 to 2/7/2022. Variables and clinical outcomes identified included age, gender, site and stage of cancer, prior VTE, provider response to BPA, anticoagulation prescription, subsequent VTE, and bleeding outcomes. Bleeding outcomes were defined according to the International Society of Thrombosis and Hemostasis criteria. For this analysis, we excluded patients who were already on an anticoagulant at the time of BPA. Descriptive statistics and bi/multivariate inferential statistical tests including Chi-square test, Pearson correlation coefficient, logistic regression and multiple regression were planned as appropriate.

Results

The BPA was accurate in identifying cancer patients at high-risk of thrombosis with a KS ≥2. The study population was comprised 633 patients with a median age of 71 years and 308 patients (48.6%) were females. Lymphoma was the most frequent type of cancer (n=169, 26.7%) followed by lung cancer (n=115, 18.2%). Of 633 patients, 28 patients (4.4%) were prescribed prophylactic anticoagulant after the BPA fired. Of these, 18 (64.3%) were female and 19 (67.9%) were younger than 65 years old. All prescribed anticoagulants were direct oral anticoagulants including rivaroxaban (n=21, 75.0%) and apixaban (n=7, 25.0%). Most patients who were prescribed an anticoagulant were on cancer-directed treatment (n=22, 78.6%). Three patients (10.7%) had a prior history of VTE, and 9 patients (32.1%) had a port or other central line. About half (n=15, 53.6%) of patients had Stage IV disease. The breakdown of cancer types for those who started prophylactic anticoagulation is as follows: pancreatic cancer (n=13, 46.4%), lung cancer (n=8, 28.6%), lymphoma (n=3, 10.7%), and 1 each of colorectal, prostate, head and neck, and plasma cell. Despite receiving VTE prophylaxis, 1 patient had a subsequent episode of deep vein thrombosis. During the study period, of patients who were prescribed prophylactic anticoagulants after BPA fire, 16 (57.1%) were alive at a median follow up of 5 months. Of high-risk patients who were not prescribed prophylactic anticoagulation (n=605, 95.6%), 14 patients (2.3%) developed a VTE event. There were no major or clinically relevant non-major bleeding events in our entire cohort.

Discussion

This study demonstrates successful launch of a best practice alert to identify high risk cancer patients. However, there was a low uptake of BPA recommendation to start VTE prophylaxis from oncology providers compared to the previously published VTEPACC model from the University of Vermont Cancer Center which involves in-person consultation in addition to an electronic alert. There were no bleeding events in patients who started prophylactic anticoagulant, consistent with prior studies which showed good safety record. Efforts to mitigate BPA alert fatigue should be considered to encourage better BPA compliance and generate better outcome measures. Future studies should continue to explore options for implementation science including the need for further education or additional in-person interventions to prevent thrombosis in high-risk cancer patients.