(480) Automated Venous Thromboembolism (VTE) Prevalence Calculation in Children: A Comparative Analysis

Introduction: Currently there is no widely accepted method for automated calculation of pediatric hospital acquired (HA)-VTE. Prior studies have shown suboptimal detection using ICD codes only, and the current recommendation is to use at least two methods for detection (Solutions for Patient Safety [SPS]). The objective of this study was to compare the relative detection rate of four different automated methods.

Methods: This single-center retrospective study included all children admitted to the Children’s Hospital of Illinois at Peoria, IL between 2018 to 2023. ICD codes and key terms related to VTE, deep vein thrombosis and pulmonary embolism were used to create four automated methods to detect potential VTE events via review of the following: (1) ICD code [ICD] (2) radiology ultrasound/Doppler studies [USG] (3) hematology consults for VTE [Heme] and (4) receipt of enoxaparin and/or heparin at therapeutic doses [Meds]. Manual chart review was performed to confirm true HA-VTE events using SPS definitions. True HA-VTE rate was defined as positive VTE by manual review of the positive cases identified by any of the four methods. This was then used to calculate the accuracy of various automated methods.

Results: There were 35,445 admissions (37% ICU admissions). Of these patients, VTE was detected in 335 (0.94%) by ICD, 159 (0.45%) by Heme, 184 (0.52%) by Meds and 588 (1.7%) by USG. Overall, 804 patients were detected as a positive VTE by the composite of the four methods (VTE rate 2.3%). Manual review showed a false positive rate for each of the four methods of 62% [208/335], 60% [95/159], 56% [103/184] and 75% [442/588] (ICD, Heme, Meds and USG respectively), and a composite manually verified HA-VTE rate of 0.6% (197/35,445). Overall false positive rate for the composite of the four methods was 75% (608/804). Sensitivity for detection of HA VTE was highest for USG (74%), followed by ICD (64%), Heme (48%) and meds (32%), while specificity was highest for Heme (89%), followed by meds (80%), ICD (65%) and USG (27%).

Conclusions: The use of multiple diagnostic validation methods may increase the detection rate of HA-VTE but lead to high false positives. Amongst individual methods, compression limb US review was most sensitive (but least specific), while the review of hematology consults had highest specificity.