Risk Factors Associated with Hospital-Acquired Venous Thromboembolism in the Pediatric Intensive Care Unit

Introduction: Venous thromboembolism (VTE) in Pediatric Intensive Care Unit (PICU) patients is associated with central venous catheter (CVC) use. However, risk factors for VTE development in PICU patients with CVCs are not well established. The impact of Hospital-Acquired VTE in the PICU on clinical outcomes needs to be studied in large multicenter databases to identify subjects that may benefit from screening and/or prophylaxis.

Method: With IRB approval, the Virtual Pediatric Systems, LLC database was interrogated for children < 18yo admitted between 01/2009-09/2014 who had PICU length of stay (LOS) <1 yr and a CVC present at some point during PICU care. The exact timing of VTE diagnosis was unavailable in the database, so VTE-PICU was defined as an "active" VTE that was not "present at admission". VTE-prior was defined as a VTE that was "resolved," "ongoing" or "present on admission." Variables extracted from the database included demographics, primary diagnosis category, and Pediatric Index of Mortality (PIM2) score. PICU LOS was divided into quintiles. Chi squared and Wilcoxon rank-sum were used to identify variables associated with outcomes, which were then included in multivariate models. Our primary outcome was diagnosis of VTE-PICU and our secondary outcome was PICU mortality. Children with VTE-prior were included in the mortality analyses, but not the VTE-PICU analyses. Data shown as median (IQR) and OR (95% CI).

Results: Among 143,524 subjects, the median age was 2.8 (0.47-10.31) years and 55% were male. Almost half (44%) of the subjects were post-operative. The median PIM2 score was -4.11. VTE-prior was observed in 2498 patients (1.78%) and VTE-PICU in 1741 (1.2%). The incidence of VTE-PICU were 852 (1.7%) in patients ≤ 1 year old, 560 (0.9%) in patients 1-12 years old, and 303 (1.1%) in patients ≥ 13 years old (p < 0.0001). In univariate analysis, variables associated with a diagnosis of VTE-PICU were post-operative state, four LOS quintiles (3-7, 7-14, and 14-21 and >21 days) and several primary diagnosis categories: cardiovascular, gastrointestinal, infectious, neurologic, oncologic, genetic, and orthopedic. Multivariate analysis showed increased risk of VTE with cardiovascular diagnosis, infectious disease diagnosis, and LOS > 3 d (Table 1). The odds increased with increasing LOS: 7 d < LOS ≤ 14 d (5.18 [4.27-6.29]), 14 d < LOS ≤ 21 d (7.96 [6.43-9.82]), and LOS > 21 d (20.73 [17.29-24.87]). Mortality rates were 7.1% (VTE-none), 7.2% (VTE-prior), and 10.1% (VTE-PICU) (p < 0.0001). In the multivariate model, VTE-PICU (1.25 [1.05-1.49]) and VTE-prior (1.18 [1.002-1.39]) were associated with death vs. VTE-none. PIM2 score, trauma, and several primary diagnosis categories were also independently associated with death (Table 2).

Conclusion: This large, multicenter database study identified several variables that are independently associated with diagnosis of VTE during PICU care of critically ill children with a CVC. Children with primary cardiovascular or infectious diseases, and those with PICU LOS >3 days may represent specific populations that may benefit from VTE screening and/or prophylaxis. Hospital-Acquired VTE in PICU was independently associated with death in our database. Additional analysis of this database, including adding specific diagnoses and secondary diagnoses, may further refine risk factors for Hospital-Acquired VTE among PICU patients with a CVC.