Evaluation of Enoxaparin for Inpatient Venous Thromboembolism Prophylaxis in End-Stage Renal Disease Patients on Hemodialysis

Abstract

Background: Enoxaparin is not recommended for venous thromboembolism (VTE) prophylaxis in the end-stage renal disease (ESRD) on hemodialysis (HD) population due to concerns for drug accumulation and increased bleeding risk. Due to the paucity of literature with clinical outcomes to support this theoretical safety concern, the purpose of this study was to compare the risks of bleeding of enoxaparin and unfractionated heparin (UFH) in hospitalized, HD-dependent patients. Methods: This retrospective cohort study examined ESRD on HD patients who received either subcutaneous enoxaparin or UFH for VTE prophylaxis and were admitted for at least 48 hours. The primary outcome was major bleeding or clinically relevant non-major bleeding (CRNMB) as guided by definitions from the International Society of Thrombosis and Haemostasis. Results: A total of 322 enoxaparin and 10 UFH patients were analyzed. All enoxaparin patients were dosed 30 mg subcutaneous daily. Twenty-two (6.8%) enoxaparin and zero UFH patients experienced major or CRNMB (P = .498). Three enoxaparin patients suffered fatal hemorrhages. Multiple logistic regression demonstrated thrombocytopenia was associated with bleeding (odds ratio 4.23, P = .004). Conclusion: The difference in major or CRNMB rates between both anticoagulants was not statistically significant. However, the 6.8% bleed rate is concerning for inpatient enoxaparin usage, and caution should be applied when considering this drug for VTE prophylaxis in the ESRD on HD population.

Introduction

Enoxaparin is a low-molecular-weight heparin (LMWH) commonly used for venous thromboembolism (VTE) prophylaxis. Although end-stage renal disease (ESRD) on hemodialysis (HD) is not officially listed as a contraindication, enoxaparin usage for systemic anticoagulation is not recommended in these patients due to concerns for drug accumulation, increased bleeding risk, and lack of extensive studies. ¹ Enoxaparin is not removed from plasma during HD. ² Unfractionated heparin (UFH) is the drug of choice for VTE prophylaxis in hospitalized HD patients due to widespread clinical usage and independence from renal excretion. ³

There is some literature to support enoxaparin use in ESRD on HD. In a retrospective cohort study (n = 225), Green et al concluded enoxaparin may be as safe and effective as UFH for VTE prophylaxis in medically ill patients receiving HD. ⁴ An observational study (n = 7721) by Chan et al found serious bleeding risk to be similar between subcutaneous (SQ) enoxaparin and UFH when given to chronic dialysis patient for thromboprophylaxis. ⁵ Although these observational studies had similar conclusions, there were limitations.^4,5 Chan et al analyzed non-hospitalized patients from outpatient dialysis centers, and the exact indications for thromboprophylaxis were not listed for the entire study population. ⁵ Moreover, Green et al excluded those who resided in the intensive care unit (ICU) or underwent surgery. ⁴ This can underestimate bleeding complications, especially if they required escalated levels of care due to anticoagulant-associated hemorrhage.

In spite of safety issues with enoxaparin in severe renal dysfunction, there are possible advantages. For VTE prophylaxis, patients may prefer once daily enoxaparin dosing as compared to 2 to 3 times daily UFH dosing. Enoxaparin accumulation may lead to decreased need for extracorporeal circulation (ECC) anticoagulation in HD. Also, the incidences of heparin-induced thrombocytopenia (HIT) and hyperkalemia with LMWH are lower than that of UFH.^6,7 Nevertheless, these listed benefits are vastly overshadowed by theoretical risk of major, possibly fatal, bleeding in ESRD on HD. At the study institution, many hospitalists, surgeons, and nephrologists have years of experience and comfort utilizing enoxaparin at prophylactic and therapeutic doses in the ESRD population as there is lack of robust literature to oppose its use. Due to the paucity of published literature regarding enoxaparin for VTE prophylaxis in hospitalized ESRD on HD patients, the purpose of this study was to compare the risks of bleeding of enoxaparin and UFH in the general inpatient, HD-dependent population (including ICU and surgical admissions).

Methods

Study Design

This was a single-center retrospective cohort study at a 716-bed academic medical center. Ethical approval to conduct this research was obtained from the Broward Health Institutional Review Board (ie, IRB# 2019-096). Informed consent was not obtained from participants because Institutional Review Board review found this research to meet pre-specified criteria, such as minimal risk, and in accordance with 45 CFR 46.116 (d) waived the requirement for documentation of consent.

Patient data were extracted from January 2017 to August 2019. In late 2016, the anticoagulation monitoring protocol of the study institution was updated, requiring chart review of all patients on anticoagulation (eg, enoxaparin at any dose) by clinical pharmacists. Patients were included into the study if they were at least 18 years of age with ESRD on HD, admitted inpatient for at least 48 hours, and received at least 1 SQ dose of either enoxaparin 30 mg or UFH 5000 units during admission. They were excluded if they were not ESRD, previously on therapeutic anticoagulation before initiating prophylactic doses of enoxaparin or UFH, pregnant, prisoners, and/or received both enoxaparin and UFH for VTE prophylaxis.

Data were de-identified and collected via manual review of the electronic health record. Collected baseline characteristics included age, weight, sex, race, hospital length of stay, history of bleeding or VTE (eg, deep vein thrombosis [DVT] or pulmonary embolism [PE]), concomitant acid-suppressive therapy (eg, proton pump inhibitor, histamine H2-receptor antagonist), UFH bolus before HD, thrombocytopenia during admission, recent surgery or procedure, and concomitant antiplatelet therapy (eg, aspirin, clopidogrel, prasugrel, ticagrelor). Platelet count of less than 150 000/μl was considered thrombocytopenia. Recent surgery or procedure was defined as surgery or procedure during admission or within the previous 30 days.

Outcomes

The primary outcome was major bleeding or clinically relevant non-major bleeding (CRNMB) as guided by definitions from the International Society of Thrombosis and Haemostasis.^8,9 For non-surgical patients, major bleeding was considered symptomatic bleeding: causing a hemoglobin drop of at least 2 g/dl, leading to transfusion of at least 2 units of whole blood or red cells, occurring in a critical area or organ, and/or leading to death. ⁸ For surgical patients, in addition to fatal bleeds and those in critical areas or organs, major bleeding included extrasurgical site bleeding that met the aforementioned hemoglobin or transfusion threshold, surgical site bleeding that required a second intervention, and unexpected surgical site bleeding that met the prior hemoglobin or transfusion threshold, was prolonged and/or caused hemodynamic instability. ⁹ For all, CRNMB was considered to be any sign or symptom of hemorrhage that does not fit major bleeding criteria but it must require medical intervention by a healthcare professional, lead to increased level of care, and/or prompt face-to-face evaluation. ⁸

Every confirmed or suspected bleeding event was further assessed by both authors for contributing factors: related surgery/procedure and underlying coagulopathy. Adapted from a prior study, underlying coagulopathy was defined as prothrombin time greater than 20 seconds, international normalized ratio greater than 1.5, and/or platelets less than 100 000/μl. ¹⁰ Partial thromboplastin time was excluded from this definition due to possible influence by heparins. Secondary outcomes included major bleeding and CRNMB as separate outcomes as well as new-onset thromboembolic events (eg, DVT, PE, arterial thromboembolism, ischemic stroke, transient ischemic attack, myocardial infarction).

Statistical Analysis

Baseline characteristics were evaluated using Mann-Whitney U, chi-square, and Fisher’s exact tests, as appropriate, and data normality was assessed by the D’Agostino-Pearson test. The primary and secondary outcomes were analyzed using Fisher’s exact test. Multiple logistic regression was implemented to determine if confounders influenced the primary bleed outcome. A post-hoc sensitivity analysis was performed for the primary outcome. With an α-level of .05 for statistical significance, MedCalc statistical software was used to analyze the data (Ostend, Belgium).

Results

As displayed in Figure 1, 940 patients were screened. After evaluation using aforementioned criteria, 332 patients (10 UFH, 322 enoxaparin) were analyzed. The most common exclusion reason was non-ESRD (eg, acute kidney injury requiring HD).

Figure 1.

Screening flowchart for hospitalized dialysis patients.

Note. ESRD = end-stage renal disease; UFH = unfractionated heparin.

Baseline characteristics (Table 1) were similar between groups. Approximately half of both groups were female, and the cohort majority was black. For thrombocytopenia, 138 enoxaparin (42.9%) and 4 UFH (40%) patients had median platelet nadir of 105 000 and 66 000/μl, respectively. Four thrombocytopenic patients (3 enoxaparin, 1 UFH) were tested for HIT, and none had a positive functional assay. Of 322 enoxaparin patients, 158 (49.1%) were on concomitant antiplatelet therapy compared to 5 for UFH (50%). There were 158 enoxaparin and 3 UFH patients (49.1 vs 30%, P = .193) who underwent recent surgery or procedure. Recent surgeries and procedures encompassed interventions of varying bleed risk. These included HD access placement, cardiac catheterization, orthopedic surgery, soft tissue debridement, limb amputation, coronary artery bypass graft surgery, and different modes of thrombectomy. The enoxaparin group received 30 mg SQ daily, and UFH dosing was 5000 units SQ every 8 and 12 hours for 6 and 4 patients, respectively.

Table 1.

Baseline Characteristics.

	Enoxaparin (n = 322)	UFH (n = 10)	P-value
Age (y), median (IQR)	63 (56-69)	61.5 (52-71)	.918
Weight (kg), median (IQR)	75 (65.5-90.9)	82.6 (73.5-90.9)	.378
Female, n (%)	168 (52.2)	5 (50)	.892
Race, n (%)
Black	250 (77.6)	10 (100)	.582
White	56 (17.4)	0 (0)
Hispanic	14 (4.3)	0 (0)
Asian	1 (0.3)	0 (0)
Other	1 (0.3)	0 (0)
Length of stay (d), median (IQR)	7.8 (4.5-14.9)	8.4 (4.2-11.4)	.286
History of major bleeding, n (%)	9 (2.8)	1 (10)	.967
History of CRNMB, n (%)	9 (2.8)	0 (0)	.756
History of VTE, n (%)	17 (5.3)	1 (10)	.903
History of gastrointestinal bleeding, n (%)	9 (2.8)	1 (10)	.967
Concomitant PPI or H2RA, n (%)	220 (68.3)	8 (80)	.874
UFH bolus before HD, n (%)	105 (32.6)	2 (20)	.322
Thrombocytopenia during admission, n (%)	138 (42.9)	4 (40)	.562
Concomitant antiplatelet, n (%)	158 (49.1)	5 (50)	.953
Recent surgery or procedure, n (%)	158 (49.1)	3 (30)	.193

Note. UFH = unfractionated heparin; CRNMB = clinically relevant non-major bleeding; VTE = venous thromboembolism; PPI = proton pump inhibitor; H2RA = histamine H2-receptor antagonist; HD = hemodialysis.

The primary outcome of major or CRNMB occurred more frequently in the enoxaparin group (6.8%, 22/322) as compared to the UFH group (0%, 0/10) but it did not reach statistical significance (P = .498, Table 2). Eight of 22 bleeding events (36.4%) were suspected or confirmed gastrointestinal hemorrhages. Furthermore, 3 of 22 bleeding events (13.6%) were fatal. After orthopedic surgery, a thrombocytopenic patient developed a thigh hematoma while on enoxaparin and continued to bleed after anticoagulation cessation. Ultimately, he required over 40 units of blood products and expired. After improving from septic shock and being extubated, the second patient presented with altered mental status the day after enoxaparin was discontinued, and imaging revealed a large spontaneous intracranial hemorrhage with herniation despite no lab evidence of coagulopathy. The third one experienced a substantial hemoglobin decrease, new-onset thrombocytopenia, and coffee ground output from the nasogastric tube while on enoxaparin. She subsequently expired secondary to septic shock (ie, bacteremia) and possibly some aspect of hemorrhagic shock.

Table 2.

Primary and Secondary Outcomes.

	Enoxaparin (n = 322)	UFH (n = 10)	P-value
Major bleeding or CRNMB, n (%)	22 (6.8)	0 (0)	.498
Major bleeding alone, n (%)	14 (4.3)	0 (0)	.646
New onset thromboembolic event, n (%)
Overall	8 (2.5)	1 (10)	.973
DVT alone	5 (1.6)	0 (0)
PE alone	1 (0.3)	1 (10)
DVT and PE	1 (0.3)	0 (0)
Arterial thrombosis	1 (0.3)	0 (0)

Note. UFH = unfractionated heparin; CRNMB = clinically relevant non-major bleeding; DVT = deep vein thrombosis; PE = pulmonary embolism.

For the 19 non-fatal bleeding events, 5 were related to surgery or biopsy and 5 occurred in those with underlying coagulopathy (see Appendix for detailed summary on all individuals with the primary outcome). Eight enoxaparin patients (2.5%) developed a new-onset thromboembolic event while 1 (10%) occurred in the UFH group (P = .973, Table 2).

Multiple logistic regression assessed for confounders that may influence bleeding as shown in Table 3. Unsurprisingly, thrombocytopenia and recent surgery/procedure were trending toward increased bleed risk (odds ratio 4.23 and 1.28, respectively). However only thrombocytopenia reached statistical significance (P = .004) which means thrombocytopenic patients had approximately 4 times higher odds of developing major or CRNMB versus those without thrombocytopenia.

Table 3.

Multiple Logistic Regression for Primary Outcome of Major Bleeding or CRNMB.

	Odds ratio	95% confidence interval	P-value
Age (y)	0.99	0.96-1.01	.309
Weight (kg)	0.98	0.96-1.00	.099
Length of stay (d)	1.00	0.98-1.03	.776
Concomitant antiplatelet	1.01	0.40-2.57	.975
Thrombocytopenia	4.23	1.57-11.44	.004
Recent surgery or procedure	1.28	0.51-3.22	.601

Note. CRNMB = clinically relevant non-major bleeding.

Post-hoc sensitivity analysis on the enoxaparin group showed: 8.2% (13/158) of those with recent surgery/procedure, 11.6% (16/138) of thrombocytopenic patients, and 12.2% (9/74) of those with both risk factors experienced the primary outcome. Of 100 patients who possessed neither of these risk factors (eg, no thrombocytopenia during admission or recent surgery/procedure), only 2 (2%) had major or CRNMB.

Discussion

The results did not indicate a statistically significant difference in bleeding or VTE occurrence between both anticoagulants in those with ESRD on HD. Despite the reduced dose of 30 mg SQ daily for creatinine clearance less than 30 ml/min, the 6.8% bleed rate is concerning; this indicates approximately 1 in every 15 enoxaparin patients in this HD-dependent cohort experienced clinically significant in-hospital bleeding. ¹ Notably, 3 (0.9%) suffered fatal hemorrhages, in which anticoagulation was likely a contributor.

Enoxaparin is infrequently used for therapeutic anticoagulation in HD patients as theoretical risks of accumulation and bleeding still exist. The published literature presents mixed results in this area. For VTE treatment, Klil-Drori et al concluded LMWH monotherapy was not associated with more bleeding as compared to vitamin K antagonist (eg, warfarin) monotherapy in ESRD on HD, but enoxaparin was only used by 19 (23.2%) of the 82 LWMH patients. ¹¹ A retrospective chart review of 164 hospitalized HD patients by Pon et al found the 30-day incidence of major bleeding to be similar for SQ enoxaparin and intravenous UFH (n = 82 for each group) when used for therapeutic anticoagulation. ¹² However, in dialysis patients who underwent percutaneous coronary intervention (PCI), Tsai et al found increased risks of in-hospital major bleeding and mortality (adjusted odds ratio 1.28 and 1.35, respectively) in those receiving enoxaparin versus those receiving antithrombotic therapy that was indicated in dialysis patients (eg, UFH, bivalirudin, abciximab). ¹³ For the subgroup that underwent PCI for acute coronary syndrome, enoxaparin was also associated with similar increases in in-hospital major bleeding and mortality risks when compared to UFH. The authors classified enoxaparin as contraindicated in dialysis patients, and these findings support the avoidance of this anticoagulant, especially at therapeutic doses, in dialysis patients undergoing PCI.

In addition, the risk of enoxaparin accumulation rises with repeated dosing, especially in severe renal dysfunction. Brumley et al conducted a retrospective study (n = 7) to determine if enoxaparin accumulation occurs during therapeutic anticoagulation (ie, 1 mg/kg actual body weight once daily dosing) in ESRD on HD. ¹⁴ Although the small study did not meet statistical significance, there was evidence supporting drug accumulation as greater anti-Xa levels, which correlate with enoxaparin’s anticoagulant activity, were obtained in patients with higher cumulative number of doses and those with higher total enoxaparin exposure prior to the initial level. At this time, it is unknown if significant drug accumulation occurs with lower enoxaparin dosing, such as 30 mg daily, in the ESRD on HD population as anti-Xa levels were not performed in this study’s 322 patients.

Interestingly, in contrast to Europe, Canada, and Australia, enoxaparin is not approved for ECC anticoagulation for HD in the United States (US).^1,15-17 A meta-analysis by Lazrak et al showed LMWH, which included enoxaparin, was as safe as UFH for ECC anticoagulation in chronic HD. ¹⁸ Outside the US, ECC enoxaparin dosing is 1 mg/kg or 0.5 to 0.75 mg/kg (in high bleed risk) into the dialysis circuit arterial line during HD initiation.^15-17 Theoretically, assuming no drug loss to the dialysis circuit, a 70-kg individual who receives either enoxaparin 30 mg SQ daily or 1 mg/kg into the dialysis circuit arterial line thrice weekly during HD sessions would have the same weekly cumulative dose of 210 mg in both scenarios. However, despite the same weekly cumulative dose, it would be inappropriate to infer daily enoxaparin dosing for VTE prophylaxis is as safe as thrice weekly enoxaparin dosing for ECC anticoagulation. Both scenarios would likely yield different thrombotic and hemorrhagic risks due to corresponding patient status and clinical setting (eg, acutely ill at the hospital vs not acutely ill at the outpatient dialysis center).

Regarding SQ UFH for VTE prophylaxis, the associated bleeding risk in the general inpatient HD-dependent population is not well-described in the literature. In a subgroup analysis of ESRD patients in the PROTECT trial, which compared dalteparin and UFH 5000 units SQ every 8 hours for VTE prophylaxis in the ICU, the VTE and major bleeding rates for UFH were 10% and 5%, respectively. ¹⁹ The major bleeding rate in ESRD patients with UFH within the PROTECT trial appears comparable to that of enoxaparin in this cohort (5% vs 4.3%), yet the PROTECT trial included ICU patients only, who are generally more prone to suffer major bleeding versus those on medical or surgical wards. Based on this reported major bleeding risk of 5% for ICU ESRD patients on SQ UFH for VTE prophylaxis, one may intuitively assume the corresponding bleeding risk of all hospitalized (ie, ICU and non-ICU) ESRD patients on this drug to be less than 5% but again there are no epidemiological or clinical studies to support this.

In particular, it is also unknown whether UFH or enoxaparin is more effective for preventing in-hospital VTE in HD-dependent patients. VTE risk scores and details on non-chemical forms of thromboprophylaxis, which influence VTE rate, were not collected in this study. For this indication, enoxaparin is the gold standard versus UFH in individuals with normal renal function, but the study was not primarily designed or powered to test for efficacy and the event rate was low. ²⁰

Likewise, there were other study limitations. Firstly, this was a retrospective chart review, so data were dependent on accuracy and validity of electronic health record documentation. Due to the study’s retrospective, non-randomized design, logistic regression was used to adjust for possible confounders. The bleed risk of each surgery or procedure was not formally assessed and input into statistical analysis, therefore this can serve as a significant confounder since these interventions certainly have variable bleed risks. Definitive conclusions and comparisons of bleeding risks between enoxaparin and UFH cannot be based on study findings because the UFH group was very small and the study was merely observational. Next, it is difficult to generalize the findings to everyone with ESRD as this was a single-center experience and did not include ESRD patients exclusively on peritoneal dialysis. Lastly, only in-hospital bleeding complications were recorded; therefore, ESRD patients who accumulated enoxaparin and bled after discharge may be omitted.

Conclusion

While the difference in bleeding rates between enoxaparin and UFH lacked statistical significance, it is concerning that approximately 1 in 15 enoxaparin patients (6.8%) experienced clinically significant hemorrhage in this cohort. In-hospital bleeding events can be significant sources of morbidity and mortality, resulting in increased length of stay and healthcare expenditure. ²¹ This appears to be the first study utilizing enoxaparin 30 mg SQ daily for VTE prophylaxis in a broader subset of hospitalized, HD-dependent patients, including those at highest risk: ICU and surgical admissions.^4,5 At this time, SQ UFH has the most clinical experience for the role of inpatient chemical thromboprophylaxis in ESRD and should be utilized first-line within this group. Serious caution should be applied when considering enoxaparin for inpatient VTE prophylaxis in ESRD on HD, particularly in those with considerable bleeding risk (eg, thrombocytopenia, certain procedures). Future prospective studies are needed to further investigate enoxaparin’s safety and efficacy for VTE prophylaxis within the ESRD population.