Optimizing the Dosing of Heparin for Therapeutic Anticoagulation in Neurocritical Care Patients at High Risk of Bleeding: Report on a Quality Improvement Initiative

Abstract

Background: Heparin infusions are used to treat and prevent thromboembolic complications in neurocritical care, but optimal dosing in patients with acute intracranial pathology or recent neurosurgery is uncertain, due to elevated risk of hemorrhage. Many institutions customize heparin nomograms for such patients but fail to methodically evaluate their effectiveness. Context: Neurocritical care unit in an academic medical center in the United States. Problem: Several incidents of heparin infusions failing to reach their partial thromboplastin time (PTT) goal within 24 hours of initiation occurred. This created a concern that existing heparin dosing protocol should be adjusted to attain goal PTT more rapidly to better treat thrombotic events. Objective: To reduce time to therapeutic effectiveness of weight-based heparin in neurocritical care patients at high risk of bleeding. Study Design: Quality improvement initiative, comparing data from a retrospective chart review (historical comparison cohort) and a prospective observational quality improvement initiative (QI cohort). Patients: Adult patients with acute intracranial pathology and acute indications for therapeutic anticoagulation but considered at high risk of intracranial hemorrhage. Interventions: Increase heparin dosing nomogram from 12 units/kg/h (historical cohort) to 18 units/kg/h (quality improvement cohort), without an initial bolus in either. Measurements: Primary endpoint was time to therapeutic activated partial thromboplastin time (aPTT) in hours, assessed with a Kaplan-Meier curve. Any known bleeding or thrombotic complications were recorded. Results: Time to reach therapeutic target aPTT was shorter in quality improvement cohort than in historical cohort (see Figure 1 in full text for details). Bleeding complications occurred in 3 of 21 patients in each cohort. Conclusions: Quality improvement initiatives such as this can make documented improvements in health care provided to neurocritical care patients.

Introduction

Unfractionated heparin infusion is used to treat and prevent thromboembolic complications in the acute setting for conditions such as prosthetic heart valve, deep venous thrombosis, atrial fibrillation, and pulmonary embolism. Heparin infusion is typically targeted to activated partial thromboplastin time (aPTT) with use of boluses to quickly get the patient into the therapeutic range. There is some evidence suggesting that earlier attainment of therapeutic targets helps in reducing mortality and morbidity from acute coronary syndrome and pulmonary thromboembolism,^1,2 but there is also some evidence refuting an association.^3,4 Weight-based nomograms have superior efficacy^5,6 and effectiveness^7,8 for achieving therapeutic levels of anticoagulation compared with intuitive physician dosing^6,8 and standard heparin dosing nomograms.^5,7

Patients with acute cerebrovascular injury (such as acute ischemic stroke, intracranial hemorrhage, or intracranial tumors) and patients who undergo neurosurgery may need therapeutic anticoagulation for similar indications, but they are at high risk of bleeding complications. The relative safety of heparin in these patients remains uncertain, leading to uncertainty about the optimal timing and dosage. One small case series and literature review suggest that excess heparinization of stroke patients often quickly leads to intracerebral hemorrhage, ⁹ with the apparent implication that dosing should be cautious, reduced, or deferred. On the contrary, a large multicenter study of stroke patients with atrial fibrillation has shown no effect of heparin dosage (high vs low vs none) on the 6-month mortality. ¹⁰ If patients with acute cerebrovascular injury or recent neurosurgery have emergent indications for therapeutic anticoagulation with risk: benefit ratios favoring heparin administration, they are usually treated with conservative nomograms without boluses that target a lower therapeutic range to 1.5 to 2x baseline aPTT or an aPTT of 50 to 70 seconds instead of the recommended 70 to 90 seconds, using a lower weight-based dosing rate to minimize the risk of hemorrhage. Despite their widespread use, there is limited evidence of the efficacy or safety of using conservative no-bolus heparin protocols with lower aPTT targets, and to our awareness, there is no guidance on the optimal dosing protocol. Nonetheless, a previous study has shown that the heparin dosage used must be customized at each hospital based on past patient data at that specific hospital, ⁸ and a Guidance Statement recommends “internal audits to determine the dose requirement to produce therapeutic anticoagulation.” ¹¹

At our institution, our existing heparin protocol used for anticoagulation in neurocritical care patients at high risk of bleeding seemed suboptimal in achieving the therapeutic goals for which it was prescribed. Our perceived concerns for delayed treatment of known thrombotic events while on heparin infusion with subtherapeutic aPTT led us to investigate our heparin protocol for patients at high risk of bleeding. A safety event discussion at the Pharmacy and Therapeutics committee led to initiation and approval to do a pilot quality improvement initiative. Our goal was to assess and hopefully improve the effectiveness of our heparin nomogram in achieving conservative therapeutic aPTT targets in patients in our neurocritical care unit with acute indications for heparin infusions but also high risk of bleeding. We assumed that the high prevalence of obesity or a change in commercial heparin products led to less than therapeutic doses being delivered by our existing nomogram than previously observed. So, the newly proposed protocol had a higher dosing rate based on analyzed results of other nomograms. Current clinical practices continue to avoid boluses in heparin nomograms in neurocritical care patients and data regarding the safety of the boluses is conflicting, neither our old protocol nor our new protocol used a bolus.^{12 -16} This quality improvement pilot was conducted in our neurocritical care unit, under the close supervision of the clinical providers and pharmacy, with monthly review at the Neurocritical Care Quality Improvement meeting. This work was conducted for the sole purpose of quality improvement, and no aspects were outside of what would be considered routine clinical care.

Methods

Ethics

The quality improvement initiative was approved by our hospital’s Pharmacy and Therapeutics committee as an attempt to improve patient safety and therapeutic effectiveness. We obtained approval from our Institutional Review Board (IRB00034311, Optimizing Heparin Infusion in NeuroCritical Care Patients, 8/14/2015) to allow publication of results so others can learn from our initiative.

Study design

The work reported here is a single-center observational study of a pharmacy-supervised quality improvement pilot evaluating a cohort compared with a preceding historical comparison cohort. A retrospective chart review was performed to obtain the data for the patients treated with our original heparin protocol. Data were collected prospectively for the patients treated with our new QI heparin protocol.

Context

All data were collected in the neurocritical care unit of a tertiary-level academic medical center in the United States. The data for the new heparin dosing protocol were collected from February to July 2015 for the a priori approved period of 6 months of a pharmacy-supervised quality improvement project. The comparison data for the old protocol were extracted from the medical records of patients treated from May to October 2014 where same number of consecutive patients on heparin infusion were analyzed to match the number of patients (21) that received heparin infusions during the pilot period (February-July 2015).

Patients

These heparin protocols were used on patients with emergent indications for systemic anticoagulation but high risk of hemorrhage due to existing acute brain injury or a recent neurosurgical procedure. The decision to initiate a “high risk of bleeding” heparin protocol (targeting 50-70 aPTT) instead of the standard heparin protocol (targeting 70-90 aPTT) was at the treating physician’s discretion, based on an assessment of the patient’s risk for bleeding.

Intervention

For all patients reported here, our neurocritical care unit used a pharmacy-driven “high risk of bleeding” nomogram, with no initial bolus, and a target aPTT of 50 to 70. In the original nomogram, the heparin dose was 12 units/kg/h. In the new nomogram, the heparin dose was 18 units/kg/h. We used this dose because a coexisting protocol that targeted 70 to 90 aPTT without bolus was used infrequently. Preliminary investigation revealed that the new nomogram achieved 50 to 70 aPTT in most cases, so it would be the ideal dose for this target. There were no other changes made to the heparin protocol. The pharmacists managing the heparin infusion were aware of the increased baseline rate, but the treating team was unaware of the increased dose.

Outcomes and measurements

All aPTT values while on heparin were retrieved from the patient’s medical records. The primary endpoint was time to therapeutic aPTT in hours. The secondary endpoints were the percentage of time with aPTT ≤50, and percentage of time with aPTT ≥70. All bleeding episodes were captured and classified as major if overt bleeding resulted in fall in hemoglobin of 2 or more g/dL,^17,18 external consultation recommending cessation of heparin without re-initiation, required transfusion of blood based on institutional anemia cutoffs (transfuse for less than 7g/dL for critically ill, transfuse for less than 8 if acute coronary or cerebral ischemic), or any intracranial bleeding. Hemorrhagic conversion if present was classified by published criteria. We also gathered data on incidences of deep vein thrombosis (DVT) defined as new DVT diagnosed by ultrasound ordered as a part of standard of care, new embolic stroke diagnosed on incidental imaging or new clinical symptoms, pulmonary embolus diagnosed by a computed tomography chest angiogram ordered as a part of standard of care, or cerebral venous thrombosis.

Statistical analysis

Descriptive statistics were used to compare the patient characteristics. A Kaplan-Meier curve was used to compare the 2 cohorts on the time to therapeutic aPTT since initiation of infusion. A non-parametric χ² LogRank test was used to calculate the significance of the difference between these curves. A univariate linear regression analysis was performed to assess the influence of baseline PTT on the time to attain therapeutic aPTT, defined as aPTT values 50 to 70. The 2 cohorts were also compared on the percentage of time with a subtherapeutic aPTT, using a Mann-Whitney rank sum test, because these data did not have a normal distribution. The 2 cohorts were compared on the percentage of time with a supratherapeutic aPTT, using a t-test, because these data were normally distributed according to a Shapiro-Wilk test.

Reporting

This quality improvement initiative is reported according to the SQUIRE 2.0 guidelines.^19,20

Results

Study sample

There were 21 patients in each cohort with no missing data. All patients received aPTT levels when on heparin. The 2 cohorts were very similar demographically (Table 1), but they were very different in terms of the indications for anticoagulation (Table 2). The median baseline aPTT before initiating anticoagulation was somewhat lower in the patients under the original protocol (31.5 vs 42.2).

Table 1.

Demographics.

	Historical cohort (n = 21)	QI cohort (n = 21)
BMI (median kg/m2)	28.9	28.2
Weight (median kg)	82.4	73.5
Sex (n female)	14	15
Age (median years)	68	70

Table 2.

Indications for Heparin.

	Historical cohort (n = 21)	QI cohort (n = 21)
Deep venous thrombosis OR pulmonary embolus	4	7
Mechanical valve	7	0
Basilar artery thrombus	0	6
Atrial fibrillation	3	3
Continuous renal replacement therapy	4	0
Middle cerebral artery thrombus	0	2
Left ventricle thrombus	1	0
Common carotid thrombus	1	0
Left ventricular assist device	1	0
Atrial flutter	0	1
Flap preservation	0	1
Deep sinus venous thrombosis	0	1

Treatment

The median duration of heparin infusion was substantially longer in the historical cohort (161.5 vs 133.0 h). The median number of aPTTs drawn on each patient was also substantially higher in the historical cohort (15 vs 14).

Therapeutic effectiveness

Aside from one outlier, the patients in the QI cohort achieved therapeutic aPTT (median—15 vs 28 h) sooner than those in the historical cohort (Figure 1), yet this endpoint did not reach statistical significance (P = 0.07). The baseline PTT was a significant (P = 0.05) predictor of the time to reach the therapeutic aPTT range, with a regression equation: time-to-event = 35.7 – (0.286 * baseline PTT) hours, yet this model only explained a small amount of the outcome variance (adj. R² = 0.07). Given the median difference of baseline PTT of 10.7 mentioned above and the regression coefficient of −0.286, we estimate that the better baseline PTT in the QI cohort gave them a median 3-hour advantage to attain therapeutic aPTT. This would leave about 12 hours of median improvement in the time to therapeutic aPTT which could be due to the difference of dosing or other factors. The median percentage of aPTTs that were subtherapeutic was similar in the historical versus QI cohorts: 33% versus 30% (P = 0.8). The mean percentage of aPTTs that were supratherapeutic was similar in the historical versus QI cohorts: 19.9% versus 19.1% (P = 0.9).

Figure 1.

Kaplan-Meier time-to-event curve of the portion of patients reaching therapeutic anticoagulation. There was one patient in each cohort who never reached therapeutic anticoagulation. The endpoints of their available data are marked with an “x” on the respective curves.

Complications

Three patients in each cohort had an episode of bleeding (Table 3). There were no incidences of DVT propagation nor new stroke, pulmonary embolus, DVT, or cerebral venous thrombosis either in the historical or pilot cohort.

Table 3.

Bleeding Complications.

Historical cohort		QI cohort
Location	aPTT at time of event (s)	Location	aPTT at time of event (s)
Rectum	67	Rectum	64
Nasogastric tube	44	Tracheostomy site	54
Epistaxis	72	Tracheostomy site	127

Institutional decision

The Pharmacy and Therapeutics committee decided to not escalate the dose further; the new nomogram baseline dose was adopted for institutional use for high-risk patients in 2016 and replaced the existing nomogram. A repeat follow-up was performed in 12 months with similar results.

Due to administrative restructuring, the heparin protocol was changed to a nursing-driven protocol in 2018. After the initial washout for 2 years, we re-evaluated the results of the new nomogram in 2020 with no significant change. The nomogram was then implemented across the institute to be used in any patient with acute brain injury at risk of intracranial bleed as a part of a larger order set for heparin nomograms. The rate of heparin aPTT within the target range was also added as a reported quality measure in the critical care dashboard (Figure 2). Any out-of-range patients were evaluated for opportunities for improvement, with none found. A repeat evaluation necessitated by high use of heparin during COVID in neurocritical patients triggered a re-evaluation in April 2021 with manual review of the first 10 patients placed on heparin. One patient was excluded due to significant comorbidities affecting the interpretation of aPTT results. A sample of 9 patients with 120 aPTT measurements showed a mean time to therapeutic levels achieved in 14 hours, with 50% 60/120 measured aPTT therapeutic, 29% (37/128) subtherapeutic, and 23% (29/124) supratherapeutic, with no evidence of thrombosis or bleeding.

Figure 2.

Critical care dashboard. Critical care dashboard showing the rate of heparin aPTT within the target range reported as a quality measure.

Discussion

Quality improvement initiatives can lead to demonstrable improvements in patient outcomes. ²¹ Here we have shown that increasing the infusion dosing rate (of our existing no-bolus heparin nomogram for high-risk patients) was associated with attaining the target aPTT more quickly (Figure 1), without additional bleeding complications (Table 3). The statistical significance of the difference between the 2 Kaplan-Meier curves was P = 0.07, which means there was a 7% probability that our results are due to random sampling chance and no such difference would be found in the larger population. We nonetheless believe that our results represent a real difference generalizable to the larger population, given that there is a strong mechanistic reason to believe that higher dosing leads to faster attainment of therapeutic levels. In any case, the benefit for our patient sample was not inferior to the other group in terms of complications, as seen in Figure 1.

The use of anticoagulation in neurocritical care patients at risk of hemorrhagic conversion in the presence of intracranial pathology may be necessary due to concomitant emergent indications for anticoagulation. Regrettably, there is a dearth of guidance in the literature about this. A Guidance Statement from 2016 acknowledged that “the optimal heparin therapeutic range is uncertain” and that “more research in defining and assessing the optimal dosage adjustment nomogram is needed.” ¹¹ These general statements on heparin treatment of venous thromboembolism are even more strongly applicable to the subgroup of patients who are deemed at high risk of bleeding. To our awareness, none of the neurocritical care guidelines make recommendations on the use of heparin in high-risk patients or the use of nomograms with no initial bolus and/or lower infusion dosing rates, and there are almost no clinical studies about this either. A recent retrospective study compared standard-intensity heparin (18 u/kg/h) in 219 patients to low-intensity heparin (12 u/kg/h) in 158 concurrent patients, treated mainly for venous thromboembolism, atrial fibrillation, or various other less frequent indications. ²² They found no statistically significant differences for thrombus, bleeding, or 3-month death rate, but their 95% CIs were wide, suggesting insufficient sample size and statistical power for these outcomes. Unlike in our study, most patients in that study received an initial bolus, and no subgroup results were presented for the patients who did not receive an initial bolus. We are not aware of any studies on heparin dosing nomograms without an initial bolus. Our study begins to address this knowledge gap and suggests that heparin dosing nomogram using 18 units/kg/h without an initial bolus is an effective way to reach the therapeutic levels of aPTT in shorter time compared with previous nomogram while still managing the risk of complications in patients at high risk of bleeding. Our follow-up evaluation reaffirmed time to target close to 14 hours.

This report has 2 important limitations. First, our study sample size was far too small to assess safety and complications. Although we observed an equal number of bleeding complications in both cohorts, we cannot statistically exclude the possibility that a higher dosing rate would have a higher bleeding complication rate in a population large enough to observe that difference. Similarly, this study does not show that the incidence of thrombosis (the original problem motivating the QI initiative) was reduced, as none were observed in either cohort. Instead, this study only shows that we reduced the time to reach a therapeutic level of aPTT, which is only a surrogate measure, not an actual patient-centered outcome. ²¹

Second, although our data show that the QI cohort achieved therapeutic aPTT faster (Figure 1) with seemingly comparable complication rates (Table 3), our study design does not permit the conclusion that this improvement was due entirely to the different dosages of heparin. Although the 2 cohorts were similar demographically (Table 1), they were not matched for clinical confounders such as body mass index, renal failure, or drug interactions that alter the monitoring of heparin. They also differed substantially on the indications for heparin (Table 2), and this might have contributed to the different outcomes. Most importantly, the historical cohort did have lower baseline aPTT, and using regression analysis we estimated that this accounted for about 3 of the 15 hours of median improvement in time-to-therapeutic-aPTT. Furthermore, the 2 cohorts were not treated concurrently, so there may have been other systemic changes, for example in the types of patients treated or the rest of the neurocritical care provided, that might have contributed to the improvement in outcomes. Last, despite improvement in time to therapeutic target to 14 to 15 hours which is hardly quick, we did not explore further interventions to shorten this time. Whether weight-based dosing needs to be replaced by alternate nomograms or adding initial boluses could have improved the results was not explored.

In conclusion, quality improvement initiatives can and often do lead to improvements in the health care received by patients. ²¹ Cerebrovascular centers and neurocritical care units using conservative heparin nomograms should periodically reassess the effectiveness of heparin protocols and time to therapeutic target.^8,11 Given that bolus doses should be avoided in patients with a high risk of bleeding, consciously optimizing the dosage rate of heparin should enable hospitals to achieve target aPTTs without an increase in complications. The safety and effectiveness of using conservative protocols in neurocritical care would be assessed best with a large multicenter patient registry, so everyone can learn from each other’s suboptimal versus best practices.