Understanding and intervening on preventable factors that provoke venous thromboembolism (VTE) in pediatrics is a leading initiative for children’s hospitals (1). As the second leading cause of hospital-acquired morbidity (preventable harm) for children in the U.S., VTE significantly increases hospitalization costs. Using the nationwide inpatient sample, one recent analysis estimated increased mean hospital costs of $27,686 and mean length of stay (LOS) extension of 8.1 days in children with hospital-acquired VTE (HA-VTE) compared to controls (2).

The foundation of our current knowledge of pediatric VTE has emerged primarily through registries, administrative databases, and retrospective cohort studies (3–8). Differences in the patient population included and analysis make comparison of these studies challenging. The relative contributions of genetic, anatomic, and acquired risk factors in pediatrics have been less studied than in adults (9, 10). Sequelae of VTE, namely post-thrombotic syndrome (PTS) and recurrence risk, have not been fully investigated in children, but efforts to standardize outcome measures are underway (11–13). This pediatric VTE compendium contains mini-reviews on VTE in the setting of cancer, congenital heart disease, inflammatory states, central venous catheters (CVCs), trauma, and thrombophilia. This mini-review will focus on epidemiology of pediatric VTE as well as patient and acquired risks. For additional discussion of pediatric VTE in a specific disease or clinical state, see the accompanying mini-reviews.

The estimated incidence of pediatric VTE in developed countries ranges from 0.07 to 0.49 per 10,000 children (3, 14). VTE rates are notably higher in hospitalized children, 4.9–21.9 per 10,000 hospital admissions (3, 14, 15). A bimodal distribution is evident. The most prominent peak is in early infancy accounting for up to 20% of pediatric VTE. A second peak occurs during adolescence with about 50% of VTE events occurring in children 11–18 years old. Reported incidence rates vary due to differences in study design, cohort inclusion criteria (e.g., all events versus symptomatic events, whether neonates are included, and/or if age-specific sub-analyses are performed), whether the source of information is a database based on billing codes or whether radiological confirmation was required for inclusion. For example, VTE incidence rates in the Netherlands decreased from 0.14 per 10,000 children to 0.05 per 10,000 children when neonates and non-extremity VTE events were excluded (8).

While the annual burden of thrombosis is greater in adults (5.6–16 per 10,000 adults per year) (9, 16), the rate of increase in observed incidence in children discharged from tertiary care hospitals is notable. Cohort studies targeting HA-VTE demonstrate a steep rise in incidence, increasing from 0.3 to 28.8 cases per 10,000 admissions (1992–2005) in one study and from 34 to 58 cases per 10,000 admissions (2001–2007) in another (17, 18). In contrast to the majority of studies focused on hospital-based populations, longitudinal data from a population-based cohort study in Québec, QC, Canada found a stable VTE incidence of 0.29 events per 10,000 person-years in children less than 18 years old over the 11-year study period ending in 2004 (5). Notably, this study excluded patients less than 1 year of age, and rates were calculated based on person-years, as opposed to VTE rate per hospital admissions.

The majority of VTE diagnosed in children arise proximal to hospitalization and are considered provoked. At least one risk factor is identified in the majority of patients (19–21). This stands in contrast to adults where 30–50% of VTE events are idiopathic or spontaneous. CVC-associated VTE predominates in pediatrics. In the absence of a CVC, however, location frequency of VTE varies by age group. Renal vein VTE comprises a significant proportion of events in neonates but is exceedingly rare in older children where lower extremity VTE is more likely. Although rare, non-extremity VTE in children may arise in portal, splenic, mesenteric, pulmonary vessels, or cerebral sinuses (22). No racial or ethnic variations have been described, and data on gender differences are conflicting.

Pediatric subpopulations have higher VTE incidence including children with critical illness, neoplasm, renal disease (nephrotic syndrome), congenital heart disease, inflammatory bowel disease, or obesity and neonates. Children admitted to the intensive care unit (ICU) have a 2% higher risk of VTE if they had a short-term CVC and LOS greater than 7 days (23). Occurrence of VTE in neonates has been estimated as high as 24 per 10,000 neonatal intensive care admissions (20). There are presently no standard thromboprophylaxis guidelines to mitigate increased VTE risk.

Many patient attributes, medical diagnoses, and elements of hospitalized care have been shown or suggested to confer HA-VTE risk. Risk factor data are predominately derived from single-institution, retrospective cohort, or case–control studies (10).

Several risk-assessment models have been published for pediatric HA-VTE (Table 1) (23, 36, 49–52). Branchford et al. showed independent risk with mechanical ventilation, systemic infection, and hospital stay ≥5 days, and that these three factors co-occurring yielded a posttest probability of 3.1% for HA-VTE (36). Sharathkumar et al. derived six independent risk factors and assigned points from beta coefficients in the logistic regression model: immobilization (3), LOS ≥ 7 days (2), OCPs (2), CVC (1), bacteremia (1), and direct ICU admission (0.5). A cumulative score of ≥3 yielded a positive predictive value of 2.45% for HA-VTE at a prevalence of 0.71% (27).

Two separate risk-assessment models, one for critically ill children and another for non-critically ill children, were created from a single institution during the same time period by retrospective, case–control study designs (23, 49). In non-critically ill children, scores of 8, 7, and ≤6 correlated to risk of HA-VTE of 12.5, 1.1, and 0.1%, respectively (49). In critically ill children, scores of 15, 7–14, and ≤6 correlated to risk of HA-VTE of 8.8, 1.3, and 0.03%, respectively (23).

Kerlin et al. derived an equation that uses 0 or 1 for absence or presence of certain factors (53): VET probability={e^[1.086 (male)+0.595 (asymmetric extremity) +0.643 (CVC)+0.549 (active cancer)−1.11 (alternative diagnosis)−2.03]}/{1+e^[1.086 (male)+0.595 (asymmetric extremity)+0.643 (CVC)+0.549 (active cancer)−1.11 (alternative diagnosis)−2.03]}

Reiter et al. created an ICU-specific risk-assessment model (excluded cardiac ICU) for “pre-hospital” and “in-hospital” thromboses but included both venous and arterial events (52). They derived 12 risk factors from existing literature and weighted them equally with 1 point. They found for every 1 point increase in total score, the risk of a symptomatic thrombosis increased by 1.57-fold (95% confidence interval 0.132–5.49) to 2.12-fold (95% confidence interval 0.175–18.34) for “pre-hospital” and “in-hospital” thrombi, respectively (p < 0.05) (52).

The model by Prentiss describes levels of risk, Risk Score 1–3, but neither details which risk factors comprise the scoring system nor their individual point values (51). Additional models have been derived for children with acute lymphoblastic leukemia (54), pediatric oncology patients with a CVC-related VTE (55), and pediatric trauma (56).

To date, a multi-institutional risk-assessment model has been lacking in the literature, but current research is underway to address this gap (57). The Children’s Hospital-Acquired Thrombosis (CHAT) Registry is a multi-institution registry with three planned phases of research. The first phase is a retrospectively derived risk-assessment model based on logistic regression. There are seven institutions contributing cases and controls (1:2), and as of January 1, 2017, there are 647 unique subjects with VTE in the registry (enrollment ongoing). Once the risk-assessment model is developed, it will be prospectively validated it in a separate multi-institutional study, i.e., phase 2. If validated, a randomized controlled trial investigating the utility of the CHAT risk-assessment model will be designed and conducted as the third phase.

Acute VTE sequelae depend on location and severity. Pulmonary thromboses may lead to pulmonary hypertension or cardiovascular instability. Thrombosis of the superior vena cava (SVC) or thoracic vessels may result in SVC syndrome. Extremity DVTs induce pain and swelling. VTE in patients with renal disease has been associated with 2-fold increase in hospital admissions and 10-fold increase in LOS (53). Neonates have increased risk of chronic renal insufficiency following renal vein thrombosis.

Risk of VTE recurrence has been variably reported but is estimated around 10% in children and 3% in neonates (9, 18). Comorbidities such as cardiovascular disease, malignancy, and neurovascular disorders have been associated with increased likelihood of recurrence. Patients with identified thrombophilia variants, e.g., prothrombin gene mutation and antithrombin deficiency, may also be at increased risk of recurrent VTE. PTS, chronic venous insufficiency associated extremity pain, edema, and stasis dermatitis, appear less frequent and severe in children (12.4%) compared to adults (30%); however, this likely represents underreporting as outcomes studies are limited and most have relatively short durations of follow-up (58). PTS occurs most commonly in the lower extremities. Recently, the Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis recommended use of a pediatric modification of the Villalta score for adult PTS to standardize reporting in children (11).

Venous thromboembolism in children is associated with increased mortality. VTE mortality rates of 11.4 per 1,000 child-years have been reported, greater than age-specific mortality estimated at 6.4 per 1,000 child-years in a population-based study conducted in Canada (5). This finding is consistent with data from several registries that report all-cause mortality in those with VTE of 9–17%. In one study, over 20% of deaths were noted to occur within 30 days of VTE diagnosis (8). The highest mortality rate is in the youngest patients. Fortunately, mortality rates directly attributed to VTE are low, 1.5–2.2%. Risk of fatality associated with pulmonary embolism is less in pediatrics compared to adults (59).

Pediatric VTE is increasing in incidence. Key contrasting points between adults and children are that pediatric VTE is more likely to be diagnosed in a hospital and have recognizable antecedent provocation. As incidence increases, associated morbidities, mortality, and health-care costs do likewise. Clearer understanding of pediatric-specific risk factors and validated risk-assessment models are needed to reduce preventable harm and investigate the efficacy of targeted prophylaxis interventions. A multitude of risk factors have been identified for pediatric VTE, but many need further elucidation. Similarly, risk-assessment models to date provide an initial approach but ultimately lack prospective validation and correlation to outcomes. The multi-institution CHAT registry is poised to overcome these limitations and provide details regarding the magnitude of risk attributable to patient, disease, and interventional factors. Initial results are anticipated in 2017. If successful, the CHAT Registry may serve as the foundation of a risk-stratified, randomized controlled trial to evaluate thromboprophylaxis measures in pediatrics.

AM and SC cowrote the manuscript, and each revised and edited the entire manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.