Purpose: The aim of our project was to compare two different VTE risk stratification tools, Padua and IMPROVE, for their accuracy of predicting in-hospital VTE events to determine which would be most appropriate to incorporate in our electronic health record EHR.
Description: A multidisciplinary team, including physicians, nurses, quality specialists, data analysts, and information technologists, was assembled to develop and verify an EHR embedded, automated VTE risk score. Padua was initially selected and built into the EHR, however, on validation, multiple data points of the tool were difficult to consistently confirm, such as mobility and recent trauma or surgery. To improve the use of the selected tool prior to hospital-wide implementation, the test characteristics of the IMPROVE score were subsequently evaluated using a retrospective review of information available in the EHR. A summary of the risk factors included in each score are listed in Table 1. The study population included patients admitted to general medicine units at an urban tertiary referral center and suburban community hospital between January 2014 and August 2016. Padua and IMPROVE scores were calculated for each patient and in-hospital VTE events were identified based on an internally validated diagnosis algorithm. Of 51,417 hospitalizations reviewed, 60.8% of the patients were categorized as low risk for VTE using the Padua score compared to 72.3% for IMPROVE. There was a total of 667 VTE events (1.3%). For both risk tools, the rate of in-hospital VTE events for the low risk population was 0.3%. The specificity for IMPROVE was significantly higher than the Padua (72.9% vs. 61.4%, p<0.001). Additional test characteristics are displayed in Table 2.
Conclusions: The utility of automated inpatient VTE risk stratification tools is dependent on the reliability and consistency of data within the EHR. Within our health system, we found the IMPROVE score to have a higher specificity and included more reliable data points compared to Padua.