Impact of a Risk Calculator Embedded in the Electronic Health Record vs. Standard Venous Thromboembolism Prophylaxis for Medical Patients: A Cluster Randomized Trial

From the 2021 HVPAA National Conference

Michael Rothberg (Cleveland Clinic), Aaron Hamilton, Bo Hu, Jacqueline Fox, Megan Sheehan, Toyomi Goto

Background

Venous thromboembolism (VTE) is a serious source of hospital morbidity and mortality. Chemoprophylaxis with heparin has been shown to reduce the occurrence of VTE, but it increases the risk of bleeding and heparin-induced thrombocytopenia and is uncomfortable to receive. It should therefore be reserved for patients at high risk of VTE.

Objective

To assess the impact of a VTE risk calculator embedded in an electronic health record (EHR) on prophylaxis prescribing and patient outcomes.

Methods

We embedded a locally validated risk calculator in our EHR. Ten hospitals were randomized to implement the calculator using a stepped-wedge design. Most calculator fields were autopopulated, but required physician confirmation. Physicians were encouraged to use the calculator but to prescribe prophylaxis according to their clinical judgment. We included all patients 18 years and older who were admitted to a medical service, including intensive care, between October 1, 2018 and March 31, 2019. We excluded patients not eligible to receive VTE prophylaxis because they were already receiving anticoagulation for another purpose, terminal patients receiving comfort care only, and surgical patients temporarily housed on a medical floor. Outcomes were determined through review of the EHR. Our primary outcome was proportion of patients receiving appropriate prophylaxis (high risk receiving + low risk not receiving/total patients). High risk patients were defined as those whose predicted VTE risk exceeded 0.75% at 14 days. Secondary outcomes included total proportion of patients receiving prophylaxis and rate of VTE among high risk patients at 14 days after admission. Outcomes were compared using mixed-effect models to account for the stepped-wedge design.

Results

The final sample included 26,547 patients (mean age 61, 73% white), including 11134 before and 15413 after implementation of the risk model. Prior to the intervention, the model identified 21% of patients as high risk for VTE, 2.8% of whom developed VTE, whereas physicians identified 75% of patients as high risk, 1.1% of whom developed VTE. During the intervention phase, physicians used the calculator for 13.8% of patients. Appropriate prophylaxis was prescribed to 37% of patients pre-intervention and to 44% of patients during the intervention (p<0.17). The total proportion of patients receiving prophylaxis declined from 71% pre-intervention to 62% during the intervention (p=0.03). The rate of VTE among high risk patients at 14 days was 3.3% in the pre-intervention phase and 2.4% in the intervention phase (p=0.70). Overall rates of VTE at 14 and 45 days also did not differ between the groups.

Conclusions

A locally validated risk calculator embedded in the EHR was used infrequently, but decreased overall use of prophylaxis without impacting the rate of VTE among high risk patients.

Clinical Implications

Clinicians tend to overestimate the risk of VTE in medical patients. A risk calculator embedded in the EHR can help them to reduce the unnecessary use of prophylaxis for low risk patients, thereby reducing cost, discomfort and complications.