Can Electronic Alerts Improve VTE Prophylaxis Delivery?

The use of thromboprophylaxis among hospitalized patients at risk for a venous thromboembolism (VTE) – a major complication of hospitalization – remains inconsistent. Both overuse and underuse of VTE prophylaxis are common. In a report published in the Journal of Thrombosis and Haemostasis, David Spirk, MD, from the Institute of Pharmacology at the University of Bern in Switzerland, and co-authors explored whether inserting an electronic alert (e-alert) in the order entry system reminding physicians to calculate patients’ VTE risk could increase the rates of appropriate VTE prophylaxis.

Rates of appropriate prophylaxis in this open-label, randomized, controlled trial were higher than the researchers expected, but this was not attributable to the e-alert system. Many of the e-alerts were ignored by ordering physicians – hinting at the “alert fatigue” many physicians feel toward the overwhelming number of reminders and notifications associated with electronic medical record software.

In this study, a computer-based e-alert system automatically identified adult patients hospitalized at the University Hospital in Bern who were not receiving anticoagulation. Patients were randomized 1:1 to either the e-alert group (n=804; in which an alert and the Geneva Risk Score calculation tool were issued directly in the electronic patient chart) or the control group (n=789; in which no alert was issued). The Geneva Risk Score calculates a patient’s risk for VTE on a scale of one to 22 (22 representing highest risk for VTE), accounting for age, history of VTE, surgery, active malignancy, and other factors that might raise VTE risk.

Patients were hospitalized for a median of six days (range = 4-10 days), and patient characteristics and reasons for hospital admission (most commonly non-pulmonary infection and acute stroke) were similar between each group.

The e-alert system pop-up screen first asked the physician to complete the Geneva Risk Score. The system permitted physicians to postpone this action three times (an allowance to help increase compliance); once the physician agreed to calculate the Geneva Risk Score, the system generated the total.

If the risk score was ≥3 points, the e-alert system informed the physician that thromboprophylaxis was indicated, listing detailed recommendations for specific anticoagulants (enoxaparin or unfractionated heparin) or mechanical measures (compression stockings or intermittent pneumatic compression), depending on a patient’s bleeding risk. Thromboprophylaxis was not indicated for patients with a Geneva Risk Score <3 points.

Research nurses also reviewed patient records to gather data on baseline demographics, comorbid conditions, laboratory findings, individual items of the Geneva Risk Score, type and duration of thromboprophylaxis, and in-hospital clinical outcomes in the standardized electronic case report form.

According to the Geneva Risk Score calculated by study personnel, most patients were classified as having a high risk for VTE (52.2%), including 432 patients (53.7%) in the e-alert group and 400 patients (50.7%) in the control group. About one-quarter of patients had a high risk of bleeding (25.0% and 23.8%, respectively).

Two-thirds of patients in each study group received appropriate thromboprophylaxis: 536 patients (66.7%) in the alert group and 526 patients (66.7%) in the control group. Inappropriate thromboprophylaxis occurred in 16.8 percent of the overall study group, and was evenly split between overuse and underuse.

During the study period, rates of appropriate thromboprophylaxis remained stable in the overall group (65.4% in the first half vs. 67.9% in the second half; p=0.29) – suggesting that, while e-alerts did not improve appropriate thromboprophylaxis, they also did not interfere with the delivery of thromboprophylaxis. However, in the e-alert group, rates of appropriate thromboprophylaxis increased modestly from the first to the second half (63.1% vs. 70.4%; p=0.028).

Secondary endpoints (all-cause mortality, VTE, and bleeding complications) also did not differ between the e-alert and control groups, or between patients who did or did not receive appropriate thromboprophylaxis.

Compliance with the e-alert system was lower than the researchers anticipated: 446 patients (55.5%) either had no score calculation or had a calculated score result that was inconsistent with information from the research nurse–reviewed patient chart.

“Behavioral change is needed to motivate physicians to accurately calculate the risk score,” the authors concluded, noting that “simplification of the complex Geneva Risk Score, with its 19 items, may lead to improved physician compliance, and warrants further research.” An automatic risk-adapted alert system that issues alerts only for those patients with a high risk of VTE also may prevent the “alert fatigue” phenomenon, they added.

The study’s findings are limited by its single-center design and its reliance on in-hospital follow-up data, meaning that the results cannot be extrapolated to long-term effects of the e-alert system. Also, because the ordering physicians in the control group were aware of the ongoing study, they may have adjusted their rates of appropriate thromboprophylaxis, potentially obscuring any difference from the alert group.

The authors report financial relationships with Bayer Healthcare AG.


Spirk D, Stuck AK, Hager A, et al. Electronic alert system for improving appropriate thromboprophylaxis in hospitalized medical patients: a randomized controlled trial. J Thromb Haemo. 2017;15:2138-46.