Outcomes of Prophylactic Enoxaparin Against Venous Thromboembolism in Hospitalized Children

We conducted a retrospective observational study of children who were admitted to a 300-bed, freestanding children’s hospital in the western United States from January 2016 to December 2018 and received prophylactic enoxaparin against HA-VTE. At the commencement of the study period, the hospital initiated a pharmacist-driven protocol for prophylactic enoxaparin. This study was approved by the university’s institutional review board (#00104983).

The goal of the protocol was to standardize the dosing and monitoring of hospitalized children receiving prophylactic enoxaparin against HA-VTE via a pharmacist collaborative practice agreement (Tables 1 and 2). The pharmacist ordered laboratories, including hemoglobin, platelet count, activated partial thromboplastin time, international normalized ratio, and serum creatinine, before starting enoxaparin. The initial dose was 0.75 mg/kg per dose for infants aged <12 months and 0.5 mg/kg per dose for older children aged ≥12 months, with a maximum dose of 30 mg and rounded to the nearest mg or manufactured syringe size. Enoxaparin was administered subcutaneously every 12 hours. During the second half of the study period, the protocol was revised to require monitoring of LMWH anti-Xa level, which was not previously routinely done. Additionally, the dose was adjusted to a goal range of 0.2 to 0.5 IU/mL. Anti-Xa level was measured 4 to 6 hours after the second dose when enoxaparin was initiated and after dose adjustments.

The protocol did not provide indications or contraindications for prophylactic enoxaparin. Scoring systems or alerts in the electronic medical records were also not used to identify children at risk for HA-VTE. Children were started on the protocol at the provider’s discretion. As such, certain populations, such as children with congenital heart disease or adolescents, were more commonly started on the protocol on the basis of their perceived risk of HA-VTE.

Hospitalized children aged <18 years who received prophylactic enoxaparin were eligible for this study. Patient encounters were identified through electronic medical records by searching for the terms “enoxaparin” and “prophylactic enoxaparin.” Individual patient encounters were manually reviewed to confirm that the child received prophylactic enoxaparin against HA-VTE. We excluded patient encounters wherein children received enoxaparin every 24 hours. For children with multiple admissions, only the first admission was analyzed.

Data were manually abstracted from the electronic medical records and entered into a deidentified database. Patient information included demographic and historical data, interventions the child received, laboratories before initial dose of enoxaparin, and clinical outcomes. Data relevant to enoxaparin included the initial dose, timing of initial dose relative to admission to the hospital, dose immediately before achievement of goal anti-Xa range, number of anti-Xa levels drawn until achievement of goal range, and anti-Xa levels.

We defined obesity as BMI >95% percentile for age and sex.^10,11  Renal insufficiency was defined as estimated glomerular filtration rate <60 mL/min per 1.73 m² based on the Schwartz formula.^12,13  CVC included untunneled catheter inserted in the internal jugular, femoral or subclavian vein, umbilical venous catheter, tunneled or implanted catheter, and upper and lower extremity peripherally inserted central catheter.

The primary biochemical outcome was achievement of the goal anti-Xa range of 0.2 to 0.5 IU/mL. The primary clinical outcome was the development of HA-VTE while on prophylactic enoxaparin. We defined HA-VTE as incident DVT in an extremity or PE because these manifestations of VTE are the primary targets of pharmacologic prophylaxis. HA-VTE was detected clinically by symptomatology, including swelling or firmness of an extremity, inability to draw from or flush a CVC, chest pain, or shortness of breath with or without hypoxia. HA-VTE was then confirmed radiologically using ultrasound, computed tomography with or without angiography, echocardiogram, or MRI. The secondary clinical outcome was clinically relevant bleeding as defined by the International Society on Thrombosis and Haemostasis.¹⁴  Clinically relevant bleeding included bleeds associated with a hemoglobin drop ≥2 g/dL in 24 hours, fatal bleeds, bleeds that required medical or surgical intervention for hemostasis, or bleeds that were retroperitoneal, pulmonary, or in the central nervous system.

We compared continuous variables between infants aged <1 year and older children aged ≥1 year using Student’s t test, discrete variables using Mann-Whitney U test, and categorical variables using χ² or Fisher exact tests, as appropriate. Similar tests were performed between children with and without anti-Xa level. Categorical variables were compared between children admitted to the hospital during the first and second halves of the study period using χ² or Fisher exact tests, as appropriate. We compared the child’s initial dose of enoxaparin and the adjusted dose that achieved the goal range to the recommended doses of 0.75 mg/kg per dose for infants and 0.50 mg/kg per dose for older children using 1 sample t test.

We used logistic regression to identify factors associated with HA-VTE. We evaluated biologically plausible risk factors for HA-VTE in children. We included early administration of prophylactic enoxaparin, defined as initial dose administered within 3 days of admission to the hospital.¹⁵  We used the Youden index to confirm that, for the current study, this was also the optimal threshold for HA-VTE.¹⁶  We evaluated renal insufficiency because enoxaparin is eliminated renally.^12,13  The proportion of children without anti-Xa level precluded the evaluation of its association with HA-VTE. Instead, we evaluated the association of presence of anti-Xa level and admission to the hospital during the second half of the study period with HA-VTE. We did not evaluate the association of the dose of enoxaparin with HA-VTE in the adjusted model because of the occurrence of Simpson paradox.¹⁷  Under this paradox, we obtained the biologically implausible positive association between risk of HA-VTE and dose of enoxaparin. Infants had higher risk of HA-VTE than older children and infants were administered higher doses of enoxaparin. Laboratory tests with missing values were also not evaluated. In the adjusted model, stepwise backward elimination was used with variables removed for P >.10. Given the differential efficacy of enoxaparin between infants and older children, infancy status was included in all models. We conducted similar analyses for clinically relevant bleed.

Data were presented as mean if normally distributed (SD), median if not normally distributed (interquartile range [IQR]), or count (percentage). Associations with HA-VTE and clinically relevant bleed were expressed as odds ratios (ORs) with 95% confidence intervals (CI). Statistical analyses were conducted using Stata 16.1 (StataCorp, College Station, TX). Hypothesis tests were conducted at a 2-sided α of 0.05.

During the study period, 228 patient encounters wherein prophylactic enoxaparin was administered were identified. Of these, 21 patient encounters with children receiving enoxaparin every 24 hours and 13 patient encounters for subsequent hospital admissions were excluded.

A total of 194 children were analyzed, of whom 109 (56.2%) and 85 (43.8%) were admitted to the hospital during the first and second halves of the study period, respectively. There were 13 (6.7%) infants and 181 (93.3%) older children in the cohort (Table 3 and Supplemental Fig 2). The proportions of infants during the first and second halves were not statistically different (8.3% vs 4.7%, P = .40). Mean ages were 0.1 (SD: 0.2) year for infants and 13.1 (SD: 3.9) years for older children (P <.001). Compared with older children, congenital heart disease, renal insufficiency, admission to the ICU, and CVC were proportionately higher in infants (Table 3). Conversely, admission for surgery, admission for trauma, and mechanical prophylaxis were proportionately lower in infants. Infants had higher hemoglobin.

Infants received a mean initial dose of enoxaparin of 0.80 (SD: 0.34) mg/kg per dose, which was not statistically different from the recommended dose of 0.75 mg/kg per dose (P = .62). Older children received a mean initial dose of 0.51 (SD: 0.15) mg/kg per dose, which was also not statistically different from the recommended dose of 0.50 mg/kg per dose (P = .31). Enoxaparin was administered early, within 3 days of admission to the hospital, in 3 (23.1%) infants and in 107 (59.1%) older children (P = .02).

LMWH anti-Xa level was measured in 116 (59.8%) children. The proportions of children on prophylactic enoxaparin with anti-Xa level was 41.3% during the first half and 83.5% during the second half of the study period (P <.001). Admission for trauma and mechanical thromboprophylaxis were proportionately higher, whereas cancer was proportionately lower, among children with than those without anti-Xa level. The proportion of children who achieved goal anti-Xa range was also higher during the second half of the study period (71.7% vs 30.3% during the first year; P <.001).

The proportions of infants among those with (7.8%) and without (5.1%) anti-Xa levels were not statistically different (P = .57). Among the 9 infants with anti-Xa level, 6 (67%) achieved goal range compared with 88 (82.2%) of 107 older children with anti-Xa level (P = .37). However, after the initial enoxaparin dose, only 1 (11.1%) infant, but 62 (57.9%) older children, achieved goal range (P = .01; Fig 1). The median number of anti-Xa levels until goal range was achieved was 2 (IQR: 2–3) in infants and 1 (IQR: 1–2) in older children (P = .01; Fig 1). Immediately before achievement of goal range, the enoxaparin dose in infants was 1.04 (SD: 0.25) mg/kg per dose, which was statistically different from the recommended dose of 0.75 mg/kg per dose (P = .04). The comparable dose in older children was 0.50 (SD: 0.13) mg/kg per dose, which was not statistically different from the recommended dose of 0.50 mg/kg per dose (P = .84).

A total of 11 (5.7%) children developed HA-VTE while on prophylactic enoxaparin. There were 8 (7.3%) and 3 (3.5%) children who developed HA-VTE during the first and second halves of the study period, respectively (P = .26). HA-VTE developed in 2 (15.4%) infants and in 9 (5.0%) older children (P = .16). DVT developed in the upper extremities of 5 children and in the lower extremities of 6 children. Of these, 7 (63.6%) were associated with a CVC. PE also developed in 2 older children with DVT. HA-VTE developed after a median of 8 (IQR: 6–11) days after admission to the hospital. All children who developed HA-VTE on prophylactic enoxaparin were switched to therapeutic doses of enoxaparin.

Previous VTE (OR: 20.65; 95% CI: 3.31–128.89) was positively associated with HA-VTE, whereas early administration of prophylactic enoxaparin (OR: 0.16; 95% CI: 0.03–0.82) was negatively associated with it (Table 4). Among children with anti-Xa level, 2 (2.1%) of 94 children who achieved goal range developed HA-VTE compared with 3 (13.6%) of 22 children who did not achieve goal range (P = .046).

A total of 4 (2.1%) children developed clinically relevant bleed while on prophylactic enoxaparin. There were 2 cases during each half of the study period (P >.99). All bleeds occurred in older children. The sites of the bleeds were the gastrointestinal tract (n = 3) and intracranial (n = 1), with 1 child transfused with blood products because of the bleed. Among children with anti-Xa level, 3 (3.2%) of 94 children who achieved goal range developed a bleed compared with none of 22 children who did not achieve goal range (P >.99). Of those who bled in the gastrointestinal tract, 2 had an anti-Xa level within goal range, whereas 1 did not have any measured anti-Xa levels. The child with intracranial bleed had anti-Xa level within goal range. We did not identify any factors associated with clinically relevant bleeds (Table 5).

In this retrospective observational study, we report that, after the initial enoxaparin dose, only 11.1% of infants, but 57.9% of older children, achieved the LMWH anti-Xa goal range of 0.2 to 0.5 IU/mL. Median number of anti-Xa levels until goal range was 2 in infants and 1 in older children. The incidence of HA-VTE among hospitalized children who received prophylactic enoxaparin was 5.7%, whereas the incidence of clinically relevant bleeding was 2.1%. In the subset of children with anti-Xa level, those who achieved goal range had lower frequency of HA-VTE, but comparable frequency of bleeding, than those who did not achieve goal range. These findings provide the framework for future studies on the effectiveness of prophylactic enoxaparin in hospitalized children.

The anti-Xa goal range of 0.2 to 0.5 IU/mL is commonly used for children and accepted for adults.¹⁸  However, the optimal goal range for children remains unclear. The CRETE Trial suggests that, at least for older children, this goal range may be effective and safe.¹⁹  Infants in our study had lower proportion in goal range after the initial enoxaparin dose and needed more dose adjustments to achieve goal range than older children. Consistent with these, infants required a higher than the currently recommended dose of enoxaparin to achieve this goal range. Infants have larger volumes of distribution, lower levels of antithrombin activity, and increased clearance than older children, which may explain the difference in pharmacokinetics.^20,21  Delay in achieving goal range with the use of the currently recommended dose in infants may impact the effectiveness of prophylactic enoxaparin. We showed that early administration of prophylactic enoxaparin was associated with lower odds of HA-VTE (OR: 0.16; 95% CI: 0.03–0.82). This protective effect is compatible with the natural history of DVT in children. At least in critically ill children with CVC, most DVT occur within 4 days after insertion of a CVC.²²  Among hospitalized injured children, rates of VTE increased when prophylactic enoxaparin was started beyond 3 days after admission to the hospital.⁹ 

The incidence of HA-VTE of 5.7% that we report is higher than the typically reported incidence of <1% in hospitalized children.²³  Our pharmacy-driven protocol focused on the dosing management of prophylactic enoxaparin and not on its indications. Therefore, selection bias may have occurred with children perceived to be at high risk of HA-VTE being prescribed prophylactic enoxaparin.²⁴  Despite the potential bias, the higher proportions of children with anti-Xa level who achieved goal range and lower incidence of HA-VTE among this same subset of children with similar clinical demographics suggest that prophylactic enoxaparin may be effective. The incidence of bleeding in our study (2.1%) is comparable with the CRETE Trial (3.7%) and previously conducted meta-analyses.^25,26  Together with the lack of association between achievement of goal range and clinically relevant bleeding, our study also suggests the safety of prophylactic enoxaparin. However, in the absence of children who did not receive enoxaparin who would have served as controls, we cannot establish its effectiveness and safety.

Prospective controlled studies, preferably randomized controlled trials, are needed to definitively establish the effectiveness and safety of prophylactic enoxaparin. Our study provides critical elements needed for the design of these studies. Children will need to be stratified by age because of dosing requirement. Infants should receive a dose of at least 1.00 mg/kg per dose, whereas older children can use a dose of 0.50 mg/kg per dose. Intervention should be started as soon as possible, preferably within 3 days after admission to the hospital, and can aim for an anti-Xa goal range of 0.2 to 0.5 IU/mL. Consideration should be given to those with previous VTE. We showed that previous VTE was associated with higher odds of HA-VTE (OR: 20.65; 95% CI: 3.31–128.89). At the minimum, randomization should also be stratified by presence of previous VTE. Otherwise, alternative strategies, such as use of therapeutic enoxaparin, should be explored. Our study did not examine the use of therapeutic enoxaparin against HA-VTE. However, a recent observational study by Clark et al suggests that this strategy provided incremental reduction in CVC-associated DVT in hospitalized children with previous VTE without increasing the risk of bleeding.²⁷ 

Certain limitations should be considered. Only about half of the children had anti-Xa level. Thus, we were not able to evaluate its association with HA-VTE or bleeding. However, our analysis of the biochemical outcomes remains valid because children with and without anti-Xa level were generally comparable. We did not follow the children after discharge from the hospital. Although most HA-VTE in children are detected while hospitalized, some HA-VTE may have presented after discharge. Our sample size was relatively small, which emphasizes the need for adequately powered future studies. We may have missed important associations with HA-VTE, such as CVC and admission to the ICU. Our reported associations were also imprecise with wide CI. Lastly, there were significantly fewer infants in the cohort. Our findings may be more representative of older children.

Despite the development of HA-VTE in hospitalized children who received prophylactic enoxaparin, our findings suggest the effectiveness and safety of an anti-Xa level directed strategy of prophylactic enoxaparin. However, this strategy should be investigated in prospective controlled studies, preferably randomized controlled trials, before routine implementation in clinical practice. Our study provides critical elements needed for the design of these studies.