National and hospital-based initiatives are being undertaken to reduce the frequency of venous thromboembolism (VTE) in children (1). The initiatives target critically ill adolescents due to their high risk of VTE, and similarities with adults in their coagulation system and risk factors for VTE (1–6). However, in contrast to adults in whom pharmacologic prophylaxis has been shown to reduce the risk of VTE by half without increasing the risk of clinically relevant bleeding (CRB), the efficacy of prophylaxis is unproven in children (178). Furthermore, the risk of CRB in critically ill adolescents is unclear. In uncontrolled single-center studies, the risk of CRB among mostly critically ill adolescents who received prophylaxis was 2–5% (910). Thus, the net clinical benefit of prophylaxis, that is, the trade-off between the absolute risk reduction in VTE and absolute risk increase in CRB, is uncertain in critically ill adolescents (11). This uncertainty, together with the heterogeneity in the risks of VTE and bleeding and lack of definitive pediatric guidelines, results in the highly variable practice of prophylaxis in critically ill adolescents (12).
In a single-center nested case-control study of risk factors for VTE, we reported that the risk of CRB in a cohort of children admitted to the PICU was 9.1% (13). CRB was associated with a five-fold longer stay in the PICU, but not with pharmacologic prophylaxis. Although this study provided seminal information regarding the epidemiology of CRB in critically ill children, the study was weighted toward children unlikely to receive pharmacologic prophylaxis against VTE, such as those who are young or at low risk of VTE. In preparation for a randomized clinical trial (RCT) of pharmacologic prophylaxis against VTE in critically ill adolescents, we conducted the THrombosis Epidemiology in Ill Adolescents (THEIA) Study in which adolescents admitted to the PICU who were at high risk of VTE were prospectively followed with serial ultrasounds for the development of deep venous thrombosis (DVT) in the lower extremities (2). In this post hoc analysis of data from the THEIA Study, we characterized the epidemiology of CRB in a cohort of critically ill adolescents who were at high risk of VTE.
MATERIALS AND METHODS
THrombosis Epidemiology in Ill Adolescents Study
The THrombosis Epidemiology in Ill Adolescents (THEIA) Study was previously reported in detail (2). In brief, it was a multicenter, prospective, cohort study of adolescents receiving cardiopulmonary support. The study was conducted in the PICUs of Yale-New Haven Children’s Hospital, Maria Fareri Children’s Hospital, Joseph M. Sanzari Children’s Hospital at Hackensack University Medical Center, Children’s Hospital at Montefiore, University of Rochester Golisano Children’s Hospital, and Hasbro Children’s Hospital from September 2014 to June 2017. None of the ICUs had a formal policy or protocol regarding the use of pharmacologic prophylaxis against VTE. Institutional review boards at each hospital approved the study.
Adolescents 13 to 17 years old who were admitted to the ICU were eligible if they were anticipated to receive cardiopulmonary support, defined as vasopressor or respiratory support, for at least 48 hours. Vasopressor included greater than or equal to 5 μg/kg/min of dopamine or any dose of dobutamine, phenylephrine, epinephrine, norepinephrine, or milrinone (14). Respiratory support included mechanical ventilation or noninvasive ventilation, which included high flow nasal cannula with flow of at least 10 L/min. Adolescents on pharmacologic prophylaxis against VTE were included because of its unclear efficacy in children. Those with any of the following were excluded: 1) radiologically confirmed VTE at admission, 2) receiving or anticipated to receive therapeutic anticoagulation, 3) the family or clinical team had decided not to provide full medical support, or 4) previously enrolled in the study (12).
Parental permission and assent, as appropriate, were obtained. Enrolled adolescents were monitored daily for any bleeding event until discharge from the ICU or a maximum of 28 days after enrollment while admitted to the ICU. For this report, we only considered the first occurrence of CRB. A member of the research team queried the clinical care team and reviewed the medical records daily for bleeding events.
Demographic data and outcomes were collected from the medical records. Adolescents admitted after trauma or surgery were categorized as surgical admissions, whereas the rest were categorized as medical admissions. Selected medications received, interventions performed, and laboratory tests obtained at any time while the adolescent was on study that may be associated with bleeding were recorded. These included unfractionated heparin, enoxaparin, warfarin, aspirin, nonsteroidal anti-inflammatory drugs, vasopressors, mechanical ventilation, noninvasive ventilation, blood transfusion, platelet count, prothrombin time, activated partial thromboplastin time, and creatinine. Predicted risk of mortality was calculated based on the Pediatric Index of Mortality (PIM) 2 (15). Prophylaxis against VTE included unfractionated heparin, except at doses to maintain patency of a vascular catheter, enoxaparin, and warfarin. Cryoprecipitate was included in plasma transfusion. For laboratory tests, which were not done daily, the last value was carried forward.
The primary outcome measure was the development of a CRB. For each bleeding event, the severity was assessed based on the definitions of the International Society on Thrombosis and Haemostasis (ISTH) for pediatric RCTs of anticoagulation (16). ISTH categorized any overt or macroscopic evidence of bleeding as major, clinically relevant nonmajor or minor based on the severity of physiologic derangement, intervention needed to restore hemostasis, and site of the bleed. For purposes of the THEIA Study, a CRB (i.e., fatal, hemoglobin decreased ≥ 2 g/dL in 24 hr, required medical or surgical intervention for hemostasis, or in the retroperitoneum, pulmonary, intracranial, or CNS) included both major and clinically relevant nonmajor bleeds (1316). Any bleeding event that did not fulfill these criteria was considered minor bleed. Given that blood-tinged secretions from the endotracheal or tracheostomy tube were mostly due to trauma from suctioning, we considered them minor bleeds despite the definition from ISTH (13).
The frequency of CRB was expressed as proportion of adolescents and as rate per 100 patient-days. Cox regression model was used to identify exposures that could be associated with the probability rate of CRB. Patient characteristics at admission, including the quadratic form of age to test the nonlinear effect of age, were examined individually but were adjusted for the ICU in which the adolescent was admitted, that is, ICU of admission, by accounting for residual autocorrelation (17). Adjusting for ICU of admission controlled for correlations in clinical care within ICUs that may have affected the risk of CRB. Unadjusted associations that were significant at a two-sided level of significance of 0.10 were included in the final model. The proportional hazards assumption was tested on the basis of Schoenfeld residuals (18). Interactions between the significant baseline patient characteristics to test the dependence of the effect of a characteristic on another on time to first CRB were assessed at a two-sided level of significance of 0.10.
Time-varying factors and covariates, which included the medications, interventions, and laboratory tests, were assessed in separate models adjusted for the significant baseline characteristics and ICU of admission. Based on pharmacokinetics, adolescents were considered at risk of CRB within 7 days after the last dose of aspirin, 5 days for warfarin, 2 days for enoxaparin, and 1 day for unfractionated heparin and vasopressors (13). The time at risk of CRB after surgery or trauma and after a minor bleed was set at 7 days. For other interventions, adolescents were considered at risk of CRB only on the day that they received the intervention. Unadjusted time-varying associations that were significant at a two-sided level of significance of 0.10 were included in the final model. Analyses were performed on the log-transformed values of the laboratory tests. The median duration of stay in the ICU and the hospital, adjusted for age, PIM 2 score, and ICU of admission were compared between adolescents with and without CRB using quantile regression.
Data were expressed as median (interquartile range [IQR]), count (percentage), and hazard ratio (HR, 95% CI). All analyses were performed in Stata 14 (StataCorp, College Station, TX). A two-sided p value of less than 0.05 was considered statistically significant, unless otherwise specified.
A total of 88 adolescents who were enrolled in the THEIA Study contributed 710 patient-days to the present report. The median age and PIM 2 score were 15.2 years (IQR, 14.3–16.4 yr) and 0.035 (IQR, 0.011–0.077), respectively, with sex distributed equally (Table 1). There were 26 (29.6%) surgical admissions. At any time during the study period, 35 (39.8%), 47 (53.4%), and 63 (71.6%) received vasopressor, noninvasive ventilation, and mechanical ventilation, respectively. In addition, 11 adolescents (12.5%) received prophylaxis against VTE with enoxaparin (n = 5) or subcutaneous unfractionated heparin (n = 1). A total of nine adolescents (10.2%) developed DVT in the lower extremities, all of which occurred in those who received mechanical ventilation. Of these, six were in the site of a femoral central venous catheter. The median duration of stay in the ICU and hospital were 9 days (IQR, 5–20 d) and 13 days (IQR, 8–137 d), respectively.
A total of 47 adolescents (53.4%) developed a bleeding event of whom 26 had a CRB. Thus, the frequency of CRB among all enrolled adolescents was 29.5% (95% CI, 20.3–40.2%) or a rate of 3.7 events (95% CI, 2.5–5.4 events) per 100 patient-days. None of the events were fatal. Of the nine adolescents with DVT, five had CRB (55.6%) compared with 21 with CRB (26.6%) in those without DVT for a HR of CRB of 2.06 (95% CI, 1.08–3.94). The median number of days for the first CRB to occur was 1 day (IQR, 1–1 d). Excluding adolescents with CRB on the day of ICU admission, the frequency of CRB was 8.8% (95% CI, 3.3–18.2%) or a rate of 0.9 events (95% CI, 0.3–1.9 events) per 100 patient-days. Only three adolescents with CRB had a preceding minor bleed. The most common sites of bleeding in those with CRB were intracranial (n = 12, 46.2%), gastrointestinal (n = 7, 26.9%), surgical site (n = 5, 19.2%), and skin (n = 5, 19.2%) (Table 2).
Age, PIM 2 score, and surgical admission were associated with CRB (Table 3). The interactions among these characteristics were not statistically significant. Age was negatively associated with CRB with a HR of 0.61 for every 1-year increase in age (95% CI, 0.50–0.74; p < 0.001). In contrast, PIM 2 score and surgical admission were positively associated with CRB. For every 0.10 increase in PIM 2 score, the HR of CRB was 1.65 (95% CI, 1.32–2.08; p = 0.02), while the HR of CRB for a surgical admission was 4.68 (95% CI, 2.41–9.09; p = 0.008). The model did not violate the proportional hazards assumption with the global test and test for each covariate not statistically significant.
Among surgical admissions, 16 (61.5%; 95% CI, 40.6–79.8%) adolescents developed CRB for a rate of 13.1 events (95% CI, 8.0–21.4 events) per 100 patient-days. The median number of days for the first CRB to occur was 1 day (IQR, 1–1 d) (Fig. 1). Among medical admissions, 10 (16.1%; 95% CI, 8.0–27.7%) adolescents developed CRB for a rate of 1.7 events (95% CI, 0.9–3.2 events) per 100 patient-days. The median number of days for the first CRB to occur was 2 days (IQR, 1–6 d). The frequency of CRB after the first day among surgical admissions and after the second day among medical admissions was 9.1% and 8.8%, respectively.
Among the time-varying covariates, mechanical ventilation and plasma transfusion were statistically significant in separate models (Table 3). However, neither were statistically significant in the final model. The HR for mechanical ventilation was 5.32 (95% CI, 0.78–36.42) while that for plasma transfusion was 1.35 (95% CI, 0.95–1.91). The association between CRB and the other medications, interventions, or laboratory tests, including prophylaxis against VTE did not reach statistical significance. A total of four of 11 adolescents (36.4%) who received prophylaxis against VTE developed CRB compared with 22 adolescents (28.6%) who did not receive it (HR, 1.19; 95% CI, 0.93–1.52; p = 0.17). Of the nine adolescents with DVT, three received prophylaxis against VTE, of whom two developed CRB. The association of CRB with a preceding minor bleed also did not reach statistical significance (HR, 1.12; 95% CI, 0.97–1.29; p = 0.11).
Adolescents with CRB had prolonged duration of stay in the hospital but not in the ICU (Fig. 2). After adjusting for age, PIM 2 score and ICU of admission, the increase in the median duration of stay in the hospital with CRB was 11 days (95% CI, 5–17 d; p = 0.001) while that for the ICU was 5 days (95% CI, –1 to 10 d; p = 0.08).
In this post hoc analysis of the THEIA Study, we report that CRB is common in critically ill adolescents at high risk of VTE. The frequency of CRB is 29.5% (95% CI, 20.3–40.2%) or 3.7 events (95% CI, 2.5–5.4 events) per 100 patient-days. CRB is negatively associated with age, but positively associated with severity of illness and surgical admission. We did not find evidence that CRB is associated with medications, interventions, or laboratory tests, including mechanical ventilation and pharmacologic prophylaxis against VTE. Adolescents who develop CRB stay in the hospital longer than those without CRB. These findings suggest that the net clinical benefit of pharmacologic prophylaxis against VTE in critically ill adolescents should be carefully considered, particularly in the absence of RCTs demonstrating its efficacy and safety in this population.
CRB is an important concern among critically ill children because of the current focus to prevent VTE using pharmacologic prophylaxis in this population. However, until our single-center nested case-control study on CRB, there was paucity of information regarding its epidemiology in children admitted to the ICU (13). Children are less likely to receive prophylaxis if their risk of CRB is perceived to be high (1219). Yet, increased risk of CRB does not obviate the need for pharmacologic prophylaxis because these patients may also be at high risk of VTE, such as in critically ill adolescents who are admitted after surgery or trauma (20). Of note, the frequency of DVT in the THEIA Study is consistent with the frequency of femoral DVT in critically ill children with central venous catheters (2122). Given that CRB occurs early in the course of the admission of these patients, pharmacologic prophylaxis against VTE may simply need to be delayed until the risk of CRB has decreased. Our inability to find an association between pharmacologic prophylaxis against VTE and CRB is due to the study’s small sample size and should not be construed as proof of its safety.
In our prior single-center nested case-control study of CRB in critically ill children of all ages, we reported a frequency of CRB of 9.1% (95% CI, 6.5–12.4%), which was substantially lower than in our present report (13). A significant number of children with trauma or surgery were excluded in our prior publication because of presence of CRB at admission to the ICU. Although exclusion of these patients provided similar frequencies as in our prior publication, we did not exclude these patients in our present report because of the intent of the THEIA Study. As such, surgical admission is negatively associated with CRB in our prior publication but is positively associated with CRB in our present report. Similar to our prior publication, we found that CRB is associated with prolonged stay in the hospital.
The frequency of CRB in our present report is closer to the frequency of 20% that is reported among critically ill adults (23). Similar to adults, we also report that the gastrointestinal tract and surgical sites are common sites of CRB. Trauma is the most common cause of intracranial CRB in our study, which partly explains the preponderance of CRB in this site compared with adults. Severe illness and surgical admission are associated with CRB in critically ill adults. However, unlike in adults, we did not find evidence that prolonged partial thromboplastin time and depressed platelet count are associated with CRB. These findings suggest that, similar to VTE, the epidemiology of bleeding in critically ill adolescents may reflect that of adults, but with important differences, such that separate studies in adolescents need to be conducted.
We plan to conduct a RCT of pharmacologic prophylaxis against VTE in critically ill adolescents to inform its net clinical benefit. Based on our findings from the THEIA Study that DVT developed only in those who received mechanical ventilation, the RCT will enroll mechanically ventilated adolescents. To enhance the safety of this RCT, we will initially exclude critically ill adolescents who are admitted after surgery or trauma. In the absence of robust evidence of the net clinical benefit of pharmacologic prophylaxis against VTE in children who are at low risk of CRB, it will be difficult to justify enrolling adolescents at high risk of CRB despite the evidence and well-established practice in adults (24). We also found an association between predicted risk of mortality and CRB in our present report, which can potentially be used to stratify patients in the planned RCT. However, the magnitude of the association is small across the range of predicted risks of mortality that we observed in this patient population.
Certain limitations should be noted. A larger sample size would be needed to obtain more precise estimates of the frequency of CRB and to confirm or identify additional risk factors. Laboratory tests were not done daily. Therefore, we used the last available value for subsequent time intervals in the time-varying analyses. Regular measurement of these tests may identify the association between derangements in coagulation and CRB as was reported in critically ill adults (23). We did not collect information on other medications that may affect the risk of bleeding, such as antiplatelet drugs aside from aspirin, for example, clopidogrel, or direct-acting oral anticoagulants because they were rarely used in the participating ICUs during the study period. This information will be valuable when these medications are used more commonly in children. Last, we may have underestimated the risk of minor bleeds. Unlike CRBs, minor bleeds were less likely to be documented or remembered (13).
CRB is common in critically ill adolescents who are at high risk of VTE. Admission after trauma or surgery can be used to stratify the risk of CRB in this population. The frequency of CRB in this cohort of adolescents is comparable to that of critically ill adults in whom the safety of pharmacologic prophylaxis against VTE is well established. These notable findings deserve important consideration when pharmacologic prophylaxis against VTE is considered clinically for critically ill adolescents and in planning RCTs that will determine its net clinical benefit in critically ill adolescents.
1. Faustino EV, Raffini LJ. Prevention of hospital-acquired venous thromboembolism in children: A review of published guidelines. Front Pediatr 2017; 5:9
2. Faustino EV, Shabanova V, Pinto M, et al. Epidemiology of lower extremity deep venous thrombosis in critically ill adolescents. J Pediatr 2018; 201:176–183.e2
3. Biss T, Alikhan R, Payne J, et al. Venous thromboembolism occurring during adolescence. Arch Dis Child
4. Cook DJ, Crowther MA, Meade MO, et al.; VTE in the ICU Workshop Participants: Prevalence, incidence, and risk factors for venous thromboembolism in medical-surgical intensive care unit
patients. J Crit Care 2005; 20:309–313
5. Cushman M. Epidemiology and risk factors for venous thrombosis. Semin Hematol 2007; 44:62–69
6. Rosendaal FR. Thrombosis in the young: Epidemiology and risk factors. A focus on venous thrombosis. Thromb Haemost 1997; 78:1–6
7. Vidal E, Sharathkumar A, Glover J, et al. Central venous catheter-related thrombosis and thromboprophylaxis in children: A systematic review and meta-analysis. J Thromb Haemost 2014; 12:1096–1109
8. Alhazzani W, Lim W, Jaeschke RZ, et al. Heparin thromboprophylaxis in medical-surgical critically ill patients: A systematic review and meta-analysis of randomized trials. Crit Care Med 2013; 41:2088–2098
9. Stem J, Christensen A, Davis D, et al. Safety of prophylactic anticoagulation at a pediatric hospital. J Pediatr Hematol Oncol 2013; 35:e287–e291
10. Cavo M, Wang W, O’Brien SH. Use of low molecular weight heparin for thromboprophylaxis in a pediatric inpatient population: Reasons for use and incidence of bleeding complications. Thromb Res 2010; 125:370–372
11. Guyatt GH, Eikelboom JW, Gould MK, et al. Approach to outcome measurement in the prevention of thrombosis in surgical and medical patients: Antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 2012; 141(Suppl 2):e185S–e194S
12. Faustino EV, Hanson S, Spinella PC, et al.; PROphylaxis against ThRombosis prACTice (PROTRACT) Study Investigators of the PALISI BloodNet: A multinational study of thromboprophylaxis practice in critically ill children. Crit Care Med 2014; 42:1232–1240
13. White LJ, Fredericks R, Mannarino CN, et al. Epidemiology of bleeding in critically ill children. J Pediatr 2017; 184:114–119.e6
14. Goldstein B, Giroir B, Randolph A. International pediatric sepsis consensus conference: Definitions for sepsis and organ dysfunction in pediatrics
. Pediatr Crit Care Med 2005; 6:2–8
15. Slater A, Shann F, Pearson G; Paediatric Index of Mortality (PIM) Study Group: PIM2: A revised version of the Paediatric Index of Mortality. Intensive Care Med 2003; 29:278–285
16. Mitchell LG, Goldenberg NA, Male C, et al.; Perinatal and Paediatric Haemostasis Subcommittee of the SSC of the ISTH: Definition of clinical efficacy and safety outcomes for clinical trials in deep venous thrombosis and pulmonary embolism in children. J Thromb Haemost 2011; 9:1856–1858
17. Rogers WH. Regression standard errors in clustered samples. Stata Technical Bulletin 1993; 13:19–23
18. Schoenfeld D. Partial residuals for the proportional hazards regression model. Biometrika 1982; 69:239–241
19. Faustino EV, Patel S, Thiagarajan RR, et al. Survey of pharmacologic thromboprophylaxis in critically ill children. Crit Care Med 2011; 39:1773–1778
20. Landisch RM, Hanson SJ, Punzalan RC, et al. Efficacy of surveillance ultrasound for venous thromboembolism diagnosis in critically ill children after trauma
. J Pediatr Surg 2018; 53:2195–2201
21. Faustino EV, Spinella PC, Li S, et al. Incidence and acute complications of asymptomatic central venous catheter-related deep venous thrombosis in critically ill children. J Pediatr 2013; 162:387–391
22. Faustino EV, Li S, Silva CT, et al.; Northeast Pediatric Critical Care Research Consortium: Factor VIII may predict catheter-related thrombosis in critically ill children: A preliminary study. Pediatr Crit Care Med 2015; 16:497–504
23. Arnold DM, Donahoe L, Clarke FJ, et al. Bleeding during critical illness: A prospective cohort study using a new measurement tool. Clin Invest Med 2007; 30:E93–E102
24. Barrera LM, Perel P, Ker K, et al. Thromboprophylaxis for trauma
patients. Cochrane Database Syst Rev 2013; 3:CD008303