Introduction
Venous thromboembolism (VTE) and bleeding are significant complications for hospitalized patients. While prophylactic anticoagulation is effective in preventing VTE, it carries an inherent risk of bleeding. To assist clinicians in this delicate balancing act, Risk Assessment Models (RAMs)—such as the Padua, Caprini, and HAS-BLED scores—have been developed to standardize risk stratification. Despite endorsement by clinical guidelines, the adoption of these tools in inpatient settings remains inconsistently documented. This study aims to identify clinician characteristics associated with RAM utilization to better understand barriers to their use.
Methodology
This study is a secondary analysis of a national, multidisciplinary electronic survey conducted between August 2021 and August 2022. The survey collected data from a convenience sample of 612 healthcare providers, including attending physicians, residents, pharmacists, and advanced practice providers. Researchers utilized log binomial regression to calculate adjusted risk ratios (aRR), allowing for the identification of professional and demographic factors independently associated with self-reported RAM use.
Results
The study population was primarily composed of physicians (70.6%) and pharmacists (20.6%), with a majority practicing in academic hospital settings.
The most significant finding was the widespread underutilization of RAMs. Over half of the respondents (54.2%) reported using no RAM for either VTE or bleeding risk assessment. Only a small minority (10.4%) utilized RAMs for both types of risk.
Several clinician characteristics were significantly associated with RAM usage:
- Medical Specialty: Internal medicine clinicians were significantly more likely to use VTE RAMs than their counterparts in hematology/oncology (aRR 0.43; 95% CI 0.30–0.62).
- Research Interest: Clinicians with a self-reported research interest in thrombosis or hemostasis were 61% more likely to utilize a VTE RAM (aRR 1.61; 95% CI 1.24–2.09).
- Practice Setting: Practitioners in community hospitals were more likely to use bleeding risk RAMs than those in academic centers (aRR 1.47).
Interestingly, factors such as the clinician's age, specific clinical role (e.g., attending vs. resident), and years of clinical experience did not show statistically significant associations with RAM adoption.
Discussion
The uniformly low adoption of RAMs across most clinician demographics suggests a systemic gap in clinical practice rather than an issue isolated to specific groups. The higher usage among internal medicine specialists may be due to the broad nature of their patient population, where standardized tools provide a necessary framework for decision-making. In contrast, hematology/oncology patients often have specific risk profiles (e.g., active malignancy) that may lead clinicians to rely on specialty-specific protocols rather than general RAMs.
The increased use of bleeding risk RAMs in community hospitals suggests that site-specific factors, such as standardized admission order sets, may play a crucial role in promoting the use of these tools. However, the overall lack of adoption highlights the need for more integrated clinical decision support systems.
Conclusion and Future Directions
The findings underscore a significant opportunity to improve patient safety through increased RAM adoption. Because underutilization is widespread, broad interventions are necessary. Future strategies should focus on integrating automated RAMs into electronic health records, enhancing clinician education, and leveraging artificial intelligence to provide real-time clinical decision support. Additionally, further research is needed to validate bleeding RAMs specifically for the inpatient population to increase clinician confidence in these tools.
Summary updated