Anticoagulation in Hospitalized Patients With Renal Insufficiency^*

Materials and Methods

The investigational review board of our institution approved the study protocol. A list of all patients who had ever received therapeutic treatment doses of enoxaparin or UFH from June 1, 2000, through June 30, 2001, at our hospital was obtained from a computerized database of dispensed medications that is maintained by our pharmacy unit. The glomerular filtration rates (GFRs) of all the patients were estimated using the Modified Diet in Renal Disease Study Group equation.⁸ Computerized laboratory data for each patient from the period of anticoagulation therapy was used in the GFR estimation. The full medical records of all patients with an estimated GFR of < 60 mL/min were obtained, and selected data were abstracted using a uniform data acquisition sheet. The abstracted data included patient demographics, weight, date of hospital admission, indications for anticoagulation therapy, anticoagulant type/dose, duration of anticoagulation therapy, platelet count, coadministration of medications that might affect bleeding tendency, and disposition from the hospital (ie, discharged from the hospital alive or dead). Data regarding bleeding complications and any blood or plasma transfusions administered during the days of therapy with either UFH or enoxaparin were collected. Patients who received only prophylactic doses of either enoxaparin or UFH were excluded. UFH was administered by continuous IV infusion after an initial bolus. Dosages and titration rates were based on a weight-based normogram. The activated partial thromboplastin time (APTT) was monitored for all patients before and after 4 h of initiation of UFH therapy. Adjustments in the titration rates and the frequency of further APTT determinations were guided by a standardized protocol (available on request). The dose of enoxaparin was 1 mg/kg body weight administered subcutaneously twice a day. Doses were not adjusted based on the degree of renal insufficiency. Also, enoxaparin levels using an anti-factor Xa-based assay were not monitored.

Data Analysis

For purposes of analyses, all patients were categorized into groups with mild renal insufficiency (estimated GFR, 41 to 60 mL/min), moderate renal insufficiency (estimated GFR, 21 to 40 mL/min), or severe renal insufficiency (estimated GFR, ≤ 20 mL/min). Bleeding complications were classified as being major or minor according to previously described criteria.^{9 10} Major bleeding was defined as bleeding resulting in a hemoglobin drop of ≥ 3 g/dL, the requirement of two or more units of packed RBCs given within 48 hours or intraocular, retroperitoneal, or intracranial hemorrhage. All other bleeding episodes, including overt GI bleeding without a decrease in hemoglobin of ≥ 3 g/dL, were classified as minor bleeds. Each bleeding complication was assigned to the type of anticoagulation therapy that the patient was receiving at the time of the bleeding episode. The duration of anticoagulation therapy was calculated in days. Bleeding incidence rate for each type of anticoagulation therapy was calculated as new episodes of bleeding per 1,000 person-days of anticoagulant therapy. For patients receiving anticoagulation therapy with both agents, the duration of therapy with each agent was individually computed. Also, bleeding complications were assigned to the specific anticoagulant that the patient’s condition was being managed with at the time of the bleeding event.

Incidence density ratios (IDRs) and risk ratios (RRs) were computed where appropriate. Comparisons were made across categories of renal insufficiency, between bleeders and nonbleeders, and between decedents and those patients discharged from the hospital alive. Predictors of bleeding complications were ascertained using stratified analysis, Cox proportional hazards analysis, and logistic regression analysis. A p value of < 0.05 was used to signify statistical significance.

To adjust for potential selection bias in the nonrandomized choice of UFH or enoxaparin by individual physicians for their patients, propensity score analyses were performed. In general, the propensity score is the chance of receiving one treatment compared to another with given observed prognostic variables.¹¹ The propensity scores were estimated using logistic regression analyses with assignment to enoxaparin as the outcome of interest, and the potential indicators for anticoagulation, gender, estimated GFR, comorbid conditions, and concurrent medications as the predictor variables. Patients were categorized into empiric quartiles on the basis of their individual propensity scores. The Cox proportional hazards regression and logistic regression analyses (for determinants of major bleeding) were performed using the propensity score groups as covariates in the regression model or by using the propensity scores as a stratifying variable.¹²

Overall, the criteria for variable inclusion in the various regression models were based on biological plausibility, current clinical practice, and evidence of association in the univariate analysis. The backward elimination approach was used in arriving at the final main-effects models. Effect modification and confounding were assessed using standard methods. Summary measures of goodness of fit were evaluated.¹³

Results

A total of 620 distinct hospitalizations fulfilled our inclusion criteria. Of these, 331 patients (53.4%) received anticoagulation therapy with UFH, 250 patients (40.3%) with enoxaparin, and 39 (6.3%) with both UFH and enoxaparin (not simultaneously) during the same hospitalization. The patients were predominantly white (81.6%) and elderly (83.7% were > 65 years of age) [Table 1 ]. The median age was 80 years. Half of the cohort (50.2%) had mild renal insufficiency, 34% had moderate renal insufficiency, and 15.8% had severe renal insufficiency. Of those with severe renal insufficiency, 38 were receiving hemodialysis and 3 were receiving peritoneal dialysis. The median estimated GFR was 40.1 mL/min for the entire cohort (interquartile range, 26.4 to 50.9 mL/min).

When stratified by the type of anticoagulant used, the patients did not differ by gender, race/ethnicity, age, or severity of renal insufficiency (Table 1) . Overall, patients receiving both UFH and enoxaparin had longer durations of anticoagulation therapy (53.8% of all patients received therapy for > 7 days; 85% of patients receiving UFH only received therapy for > 7 days; 9.6% of patients receiving enoxaparin only received therapy for > 7 days). Overall, there was a total of 2,537 person-days of anticoagulation therapy (enoxaparin therapy, 1,206 person-days; UFH therapy, 1,331 person-days).

The major indications for anticoagulation therapy were acute coronary syndrome (331 patients; 53.4%), atrial fibrillation (200 patients; 32.3%), deep vein thrombosis with or without pulmonary embolism (113 patients; 18.2%), ischemic stroke (22 patients; 3.5%), and others, including cardiomyopathies with low ejection fractions (70 patients; 11.3%). One hundred ten patients (17.7%) had more than one indication for anticoagulation therapy. Patients with acute coronary syndrome were more likely to receive enoxaparin as opposed to UFH (adjusted odds ratio, 1.6; 95% confidence interval [CI], 1.1 to 2.3). In contrast, patients with stroke were less likely to receive enoxaparin (adjusted odds ratio, 0.2; 95% CI, 0.1 to 0.9). Enoxaparin use did not differ from the other indications for anticoagulation.

In addition to the use of UFH or enoxaparin, 484 patients (78.1%) received other antithrombotic agents. Three hundred thirty-five patients (54.0%) were receiving aspirin, 151 patients (24.4%) were receiving warfarin, 104 patients (16.8%) were receiving clopidrogel, and 51 patients (8.2%) were receiving other antiplatelet medications. One hundred twenty-one patients (19.5%) were given two or more antithrombotic agents. Patients receiving enoxaparin were less likely to be given clopidrogel or agents inhibiting platelet glycoprotein IIB/IIIA function.

There was a total of 149 bleeding complications among 125 (of the 620) hospital admissions. Sixty of these (40.2%) were major bleeding and 89 (59.7%) were minor bleeding episodes. The GI tract was the most common site of major bleeding (47%). Table 2 and Figure 1 show the incidence and IDRs of major and minor bleeding across categories of renal insufficiency, respectively. Overall, the frequency of bleeding increased with worsening renal insufficiency, irrespective of the agent used. Among patients with mild or moderate renal insufficiency, the observed incidence rates of major bleeding appeared to be greater for patients receiving anticoagulation therapy with UFH compared with those receiving therapy with enoxaparin. This trend was reversed in patients with severe renal insufficiency. However, none of these differences achieved statistical significance (summary IDR, 0.7; 95% CI, 0.4 to 1.2; mild renal insufficiency IDR, 0.7; 95% CI, 0.3 to 2.0; moderate renal insufficiency IDR, 0.5; 95% CI, 0.3 to 1.1; severe renal insufficiency IDR, 1.2; 95% CI, 0.4 to 3.3). Among the subgroup of patients with severe renal insufficiency, there was a significantly higher rate of minor bleeding in those treated with enoxaparin vs those treated with UFH (IDR, 2.5; 95% CI, 1.01 to 6.36). Similar results were obtained when patients who received both (sequential) enoxaparin and UFH were excluded from the analysis (summary IDR, 0.8; 95% CI, 0.4 to 1.3).

Table 3 shows selected characteristics of patients with and without major bleeding. Worsening renal insufficiency and female gender emerged as the independent predictors of major bleeding complications in the multivariate analyses. Patients with serum creatinine concentrations of ≥ 1.5 mg/dL had an absolute increase of 13.4 episodes of major bleeding per 1,000 person-days of anticoagulation therapy (adjusted hazards ratio, 1.3; 95% CI, 1.0 to 1.6; adjusted RR, 1.8; 95% CI, 1.1 to 2.9). Similarly, female patients had an absolute increase of 7.6 episodes of major bleeding per 1,000 person-days of anticoagulation therapy compared to men (adjusted hazards ratio, 1.3; 95% CI, 1.1 to 1.5; p = 0.009; adjusted RR, 1.8; 95% CI, 1.1 to 3.0). Similar hazard ratios were observed after adjusting for the propensity scores.

Patients receiving anticoagulation therapy for > 3 days experienced a 180% excess risk of major bleeding compared to patients receiving anticoagulation therapy for between 1 and 3 days (adjusted RR, 3.0; 95% CI, 1.8 to 5.0). However, expressed per number of days of anticoagulation therapy, the incidence rate was slightly lower in the group receiving prolonged anticoagulation therapy (ie, 22.0 vs 28.3 per 1,000 person-days; IDR, 0.8; 95% CI, 0.5 to 1.3). The duration of anticoagulation therapy was of exceptional significance for the occurrence of bleeding in the subgroup of patients treated with enoxaparin. Whereas only 3.1% of patients receiving anticoagulation therapy between 1 and 3 days had a major bleed, 15.5% of those treated for > 3 days had bleeding (RR, 5.0; 95% CI, 1.9 to 12.9). For those treated with heparin, 48.6% of major bleeding and 65.3% of minor bleeding occurred in the first 3 days of anticoagulation therapy.

Twenty-four percent of heparin-related major bleeding were temporally related to an APTT > 85 s. There was no association between the specific indication for anticoagulation therapy and the occurrence of a major or minor bleed. Similarly, the type or number of additional antithrombotic agents that a patient received was not associated with bleeding occurrence.

There were 68 deaths (11.0%) in the whole cohort. There was a strong association between the degree of renal insufficiency and in-hospital, all-cause mortality (p < 0.0001). Also, there was a strong association between the occurrence of a major bleeding complication and all-cause mortality, even after excluding deaths that were probably attributable to bleeding (p < 0.0001). There was no association between gender, race/ethnicity, or age and the all-cause mortality.

Discussion

Contrary to our expectations, there was no statistically significant difference in the incidence of major bleeding in patients “therapeutically anticoagulated” with twice-daily dosing of enoxaparin compared with dosing with UFH across all levels of renal insufficiency. Among patients with severe renal insufficiency, there was a 150% excess incidence of minor bleeding in the enoxaparin group. No similar excess was seen in the patients with mild or moderate renal insufficiency. Also, the data demonstrated an increase in the risk of bleeding with increasing severity of renal insufficiency, irrespective of the agent used.

UFH is rapidly metabolized by a saturable, zero-order mechanism, mainly by the reticuloendothelial system. This is followed by a slower first-order renal clearance.^{14 15} Less than 10% is excreted in urine unchanged.¹⁶ The mean half-life is dependent on the administered dose and is unchanged with abnormal renal function.¹⁷ Enoxaparin, on the other hand, is excreted mainly by the kidneys. In experimental animals, the biological half-life of enoxaparin is increased with renal failure.^{18 19 20} The effect of renal function on the half-life of enoxaparin in humans appears less certain.^{21 22 23 24} A single-dose pharmacokinetic study demonstrated a twofold prolongation in heparin half-life based on the results of a anti-factor Xa assay in patients with severe renal insufficiency compared to normal control subjects.²¹ All other pharmacokinetic parameters were similar in both groups of patients. These led the authors to conclude that “end stage renal disease has little effect on the pharmacokinetics of enoxaparin, and dosing adjustments are unnecessary.” However, in another study²² in which daily prophylactic doses of enoxaparin were given for 4 days, the elimination half-life increased with the degree of renal impairment and was more evident after repeated dosing. The enoxaparin clearance on day 4 was 39% lower in patients with severe renal impairment than in healthy volunteers. A decrease in the number of patients with mild or moderate renal insufficiency was not statistically significant.

There have been reports^{7 24} of excessive drug accumulation or bleeding complications in patients with renal insufficiency treated with the usual doses of enoxaparin. Bleeding complications were noted in two patients with very mild renal insufficiency (creatinine clearance rate, 60 to 70 mL/min) who received anticoagulation therapy with enoxaparin for several months.⁸ A retrospective review of the experience at one tertiary medical center revealed excess bleeding complications in patients with renal insufficiency compared with those with normal renal function.⁷ However, no comparative data were provided for patients receiving anticoagulation therapy with UFH. Finally, a recent review of the literature performed by Nagge and colleagues²⁵ concluded the following: “The use of a 30-mL/min (0.50-mL/s) cutoff (for use of low molecular weight heparins) is not justified, on the basis of currently available evidence.”

To our knowledge, this is the first study the purpose of which was to compare bleeding rates in patients with renal insufficiency who received anticoagulation therapy with therapeutic doses of enoxaparin or UFH. We had postulated excess bleeding incidence with enoxaparin compared to UFH on the basis of the known differences in renal clearance. The comparable rates of bleeding complications with either agent at the same levels of renal insufficiency suggest that factors other than drug clearance play a role in the bleeding complications. It is probable that the unfavorable clearance of enoxaparin in patients with renal insufficiency is offset by the decreased inhibition of platelet function and by less interference with platelet and vessel wall interaction compared to that of UFH.^{26 27} Renal failure by itself increases the risk of bleeding by impairing platelet adhesion and aggregation.^{28 29}

It is plausible that the lack of a statistically significant difference in the bleeding complications between the two groups is due, in part, to the relatively small sample sizes in each of the categories of renal insufficiency, and by a reversal of the trend in patients with severe renal insufficiency compared with those with mild or moderate renal insufficiency. Whereas there was an almost 30 to 50% reduction in the bleeding rates for the enoxaparin group in patients with mild or moderate renal insufficiency, there was a 20% excess in major bleeding complications for the enoxaparin group in patients with severe renal insufficiency.

The observation of increased bleeding rates in women receiving anticoagulation therapy with heparin is not new.^{30 31 32} This female preponderance has been proven to be independent of body mass and appears to be related to alterations in the pharmacokinetics of heparin in women.³⁰ Our study extends this observation to women receiving anticoagulation therapy with enoxaparin.

The study was limited by its retrospective cohort study design. Study subjects were not randomized to the type of anticoagulant used. Although modern statistical techniques, including propensity score analysis and multivariate modeling, were used to adjust for potential selection bias and confounding, residual (hidden) selection bias and confounding are still possible. The small numbers of intracranial bleeding and other site-specific major bleeding did not allow for a meaningful subgroup analysis using site-specific bleeding as the outcome of interest. Another important limitation of this study is that we did not ascertain the bleeding rates at variable doses of enoxaparin or with anti-factor Xa activity monitoring. Some have suggested²³ that such monitoring may prove useful in identifying the likelihood of bleeding during anticoagulation therapy with enoxaparin. However, future studies are required to examine this hypothesis.

These limitations notwithstanding, the study demonstrates that the use of both UFH and enoxaparin are associated with similar increases in major bleeding complications in patients with renal insufficiency, and that unmonitored, full-dose enoxaparin therapy (administered as twice-daily dosing) is no less safe than monitored, full-dose UFH therapy in patients with renal insufficiency. Both agents should be used with caution in patients with renal insufficiency. Prospective studies are needed to clarify the role of anti-factor Xa activity monitoring and/or enoxaparin dose adjustment in patients with severe renal insufficiency.