Air ambulance delivery and administration of 4-factor prothrombin complex concentrate is feasible and decreases time to anticoagulation reversal

Abstract

Objectives

To evaluate the feasibility, safety, and preliminary efficacy of four-factor prothrombin complex concentrate (4-factor PCC) administration by an air ambulance service prior to or during transfer of patients with warfarin-associated major hemorrhage to a tertiary care center for definitive management (interventional arm) compared to patients receiving 4-factor PCC following transfer by air ambulance or ground without 4-factor PCC treatment (conventional arm).

Methods

Retrospective chart review of patients presenting to a large academic medical center. All patients presenting to the emergency department (ED) treated with 4-factor PCC from April 1^st 2014 through June 30^th 2016 were identified For this study, only transfer patients with an INR >1.5 actively treated with warfarin were included. The primary outcome was the proportion of patients with an INR ≤1.5 upon tertiary care hospital arrival, and the secondary efficacy outcome was difference in time to achievement of INR ≤1.5. Additional safety and efficacy objectives included difference in thromboembolic complications, length of stay, ICU length of stay and in-patient mortality between groups.

Results

Of the 72 included patients, a higher proportion of patients in the interventional group had an INR ≤1.5 on ED arrival (proportion difference 0.82, 95% CI 0.64 to 0.92; p < 0.0001) and significantly reduced time to observed INR ≤1.5 (181 vs 541 minutes; p = 0.001). No differences were observed in thromboembolic complications or patient-centered outcomes with the exception of mortality, which was significantly higher in patients in the interventional group. This group was also observed to have lower Glasgow Coma Scale and higher intubation rates prior to transfer and treatment.

Conclusions

Dispatch of an air ambulance carrying 4-factor PCC with administration prior to transfer is feasible and leads to more rapid improvement in INR among patients with warfarin-associated major hemorrhage.

Introduction

Four-factor prothrombin complex concentrate (4-factor PCC) is a non-activated agent that contains vitamin K-dependent coagulation factors II, VII, IX, and X and antithrombotics C and S. Four-factor PCC has been FDA approved for patients requiring the rapid reversal of anticoagulation and has been shown to be as clinically effective, or even superior, in its reversal of Vitamin K Antagonists (VKA) within 30 minutes of administration compared to fresh frozen plasma (FFP).

Timely administration of treatments in the emergent reversal of anticoagulation therapy is a priority that practitioners continue to try to optimize. As a trend towards the provision of time-sensitive critical interventions in the pre-hospital setting continues to emerge, the impact of this expedited care has been postulated to extend to patients who suffer a hemorrhagic stroke or a major bleed.(1–5)

Like other rural states, many smaller critical access hospitals face significant challenges and limitations in the level of emergency care they can provide. FFP is unavailable at many critical access hospitals. Even at larger hospital with access to FFP, the time to thaw may exceed the time to arrange transport out of the facility. Meanwhile, 4-factor PCC and other reversal agents are frequently not stocked in the pharmacy of these smaller institutions. These challenges prompted our institution to develop a protocol with stakeholders from the department of emergency medicine, the tertiary care facility – owned helicopter transport service, the major emergency referral service for the state, and the state poison control center. The entities involved in this protocol allowed for a comprehensive statewide pre-hospital screening of patients in need of an emergent reversal agent due to a significant bleed.

The aim of this study was to describe the feasibility, efficacy, and safety of pre-hospital administration of 4-factor PCC to resource-limited hospitals compared to a contemporaneous cohort of patients who received treatment following conventional (either air or ground) transport without 4-factor PCC administration.

Methods

The study was conducted and is reported following the Strengthening The Reporting of OBservational studies in Epidemiology (STROBE) guidelines.(6)

Study Design

This study is a retrospective cohort study of two identified groups of patients requiring 4-factor PCC for the reversal of warfarin-induced anticoagulation. The study was approved by the local Institutional Review Board (IRB), which waived patient consent due to the retrospective chart review nature of this study.

Study Setting and Population

Eligible patients presented to a large academic medical center and the primary referral center for a predominantly rural state. The study period included patients presenting from April 1^st 2014 through June 30^th 2016. Patients were eligible for inclusion if they were at least 18 years of age, had an INR > 1.5, received 4-factor PCC, and met criteria for a significant bleed (defined as known ongoing blood loss, known ICH, known intra-abdominal, retroperitoneal, or pelvic blood loss, prolonged altered mental status prior to CT, and presumptive evidence of ongoing blood loss based on persistent hypotension or tachycardia resistant to fluid). For the purposes of this study, patients were excluded if they were not being actively treated with warfarin (due to ambiguity and heterogeneity for the definition of reversal of anticoagulation for the direct oral anticoagulants), if a pre- or post-PCC INR was not documented, or if the patient received 4-factor PCC after being admitted to the hospital. Patients presenting directly to the tertiary care facility were excluded, as the primary goal of this analysis was to determine the effect of air ambulance dispatch with 4-factor PCC on reversal of anticoagulation compared to a similar control group.

Emergency referral services

All transfer requests to the tertiary care facility are communicated through an emergency referral service. Using a standardized protocol, all patients are screened to determine if there is presence of a significant bleed (defined below) and thus the need for rapid reversal of anticoagulation. If a patient was identified as having a significant bleed via this screening process, communication specialists would ask if the patient was taking any of the following medications: warfarin, rivaroxaban, apixaban, dabigatran, enoxaparin, fondaparinux, aspirin, clopidogrel, prasugrel, cilostazol, dipyridamole, or ticagrelor, and request an INR and complete blood count from the sending provider.

Air ambulance services

Our state has sixteen air medical bases. The majority of those bases are private bases (10) and the remainder are hospital based services. Only our institutions’ helicopters were involved in this protocol (4 bases), though the other helicopter services transported some of the patients in the control group but did not administer 4-factor PCC prior to transport. Our helicopters provide services statewide and the protocol had no limitations or restrictions anywhere in the state. The average flight time for our service is forty five minutes

Description of the groups

Two groups of patients were defined. The interventional group was comprised of patients transferred via the tertiary care hospital’s air ambulance service. Upon arrival to the transferring facility, the air ambulance personnel measured a INR using a point-of-care device (Coaguchek XS Plus; Roche Diagnostics, Mannheim, Germany) if not already performed, and 4-factor PCC was initiated either at the transferring facility or in the air ambulance after a verbal order from the tertiary hospital’s Emergency Department attending. The conventional group (controls) was comprised of a retrospectively identified contemporaneous cohort of all patients transferred via ground or other air-ambulance (not tertiary facility-operated) services and administered 4-factor PCC upon arrival to the tertiary care hospital. Patients in this group were identified via the tertiary hospital’s Data Enterprise Warehouse, which provides de-identified data sets, limited data sets, and data sets containing protected health information (PHI) with proper approval and authorizations. All charts were manually reviewed to ensure no misallocation of treatment group. The conventional group was primarily transported by ground ambulance (32/38, 84%), though in certain cases were transported via another air ambulance service not associated with the tertiary care facility (6/38, 16%) and without the capability of pre-hospital 4-factor PCC administration.

Study Protocol and Measurements

Patients’ electronic medical records were accessed by a single abstracter (CV) to collect patient demographics and the following data collection points: source of bleed, reason for anticoagulation, 4-factor PCC dose, use of vitamin K, use and volume of blood products (defined as packed red blood cells, fresh frozen plasma, platelets, or cryoprecipitate), use of endotracheal intubation, presence of altered mental status, Glasgow Coma Scale (GCS) score, length of ICU and hospital stay, and mortality. INR values were recorded at the following predefined time points: pre-transfer INR (defined as the first INR available either by air ambulance point of care testing or outside hospital records prior to administration of 4-factor PCC), post-transfer INR (defined as the first INR measured at the tertiary care center), and first corrected INR (defined at the first value ≤1.5). The date and time of each of these measurements were recorded, and the time to INR normalization was defined as the difference between the times of first corrected and pre-transfer INR. If additional INR values were available prior to the first corrected INR, these values and time points were also recorded. To ensure validity of the abstraction process and minimize bias, a random sample of 10% of charts identified by using a computer-generated random number selection were abstracted by a second reviewer (ST or KC) for the key variables of initial outside hospital, first post-transfer, and first normal INR. In the case of any disagreements, a third reviewer (MAP) evaluated the record and the authors came to a consensus. This information was then de-identified prior to analysis.

Outcomes

The primary outcome of the study was the difference in proportion of patients with an INR ≤ 1.5 on arrival to the tertiary care ED, while the secondary efficacy outcome was time to INR ≤ 1.5. Additional safety and efficacy outcomes included volume of blood products administered, occurrence of thromboembolic events, length of hospital stay, and in-hospital mortality.

Data Analysis

Continuous data were described medians (interquartile range) as the data were not normally distributed. Categorical variables were compared between groups using chi-square testing, while continuous data were compared using Wilcoxon rank sum test. Interobserver agreement of chart abstraction data was determined using a kappa statistic. A time to event curve was created to determine the difference in time to achieving the primary outcome and differences between groups were determined using log-rank test. Data were analyzed using commercially available statistical software (STATA 10.0, College Station, TX). All statistical tests were two sided with p<0.05 being considered significant.

Sample size

As there were no previous studies available for comparison, and the primary goal of this investigation was to describe the feasibility and preliminary efficacy of a prehospital 4-factor PCC reversal protocol, an a priori determination of sample size was not performed. In a post-hoc power analysis, given our observed proportion difference of 0.82, sample size of 72 and sampling ratio of 1.12, and an alpha of 0.05, the power to detect the observed difference was >0.99.

Results

During the study period, 186 patients were identified as having received 4-factor PCC, with 114 excluded for the reasons demonstrated on the study flow diagram, leaving a total of 72 patients for inclusion in this analysis (Figure 1). Of these, 34 patients were in the intervention group and received pre-transfer 4-factor PCC, while 38 were in the conventional group and received treatment with 4-factor PCC following transport.

Figure 1.

Study flow diagram

The patient demographics and clinical characteristics for the intervention and conventional groups are available in Table 1. Patients arriving by conventional transport traveled a shorter distance than those in the intervention group (62 vs 92 miles; p=0.002). The interventional group received vitamin K less frequently but received slightly higher dosing of 4-factor PCC per kilogram, though neither of these results met statistical significance.

Table 1.

Patient demographics, clinical characteristics, and treatments comparing the interventional and conventional groups

Parameter	Interventional (n=34)	Conventional (n=38)	p-value
Age, years (IQR)	65 (58, 77)	72 (62, 82)	0.18
Gender, males (%)	19 (56)	15 (39)	0.16
Weight, kg (IQR)	82 (75, 110)	83 (64, 96)	0.27
Distance, miles (IQR)	95 (77, 117)	65 (6, 95)	<0.01
Reason for anticoagulation, n (%)*
Atrial fibrillation	14 (41)	19 (50)	0.45
Valve replacement	1 (3)	2 (5)	0.62
Venous thromboembolism	12 (35)	14 (37)	0.89
Other	8 (24)	7 (18)	0.59
Type of Bleed+, n (%)
Intracranial	19 (56)	23 (61)	0.59
Gastrointestinal	11 (32)	12 (32)	0.94
Trauma	5 (15)	7 (18)	0.67
Other	4 (12)	3 (7)	0.58
Anticoagulation use
Warfarin alone	22 (65)	22 (58)	0.55
Warfarin + aspirin	9 (26)	13 (34)	0.48
Warfarin + aspirin + clopidigrel	3 (9)	2 (5)	0.55
Warfarin + aspirin + LMWH	0 (0)	1 (3)	0.34
Pre-transfer INR	3.7 (2.6, 7.0)	3.6 (2.8, 6.0)	0.66
Initial Glasgow Coma Score (IQR)	14 (10, 15)	15 (14, 15)	0.03
Intubated prior to transfer, n (%)	11 (32)	3 (8)	<0.01

INR: International normalized ratio; IQR: Interquartile range; IU: International Units; kg: kilogram; LMWH: Low molecular weight heparin

^*Some patients had multiple indications for anticoagulant use, so percentages sum to >100%

⁺Some patients had both intracranial and multi-system trauma, so percentages sum to >100%

In the random sample of charts with a second abstractor for key outcome data, there were no disagreements between abstractors (κ = 1.0). The baseline and first corrected INR did not differ significantly between the two groups [3.7 (IQR 2.6, 7.0) versus 3.6 (IQR 2.8, 6.0), p = 0.66; and 1.1 (IQR 1.0, 1.3) versus 1.2 (IQR 1.0, 1.3), p = 0.56, respectively]. In regards to our primary outcome, we observed a significant difference in INR values at the post-transfer time point [1.2 (IQR 1.0, 1.4) versus 3.2 (IQR 2.1, 4.4), p <0.001], with a significantly higher proportion of patients in the intervention group meeting the target INR (82% vs 0%, proportion difference 82%, 95% CI 64% to 92%; p < 0.0001).

As this observation could have potentially been due to more rapid measurement of INR in patients undergoing pre-transfer anticoagulation reversal, we assessed the relationship between timing of the post-transfer INR and initial request for transfer, shown in Figure 2.

Figure 2.

Scatterplot demonstrating the first INR measured post-transfer in the conventional (black diamonds) and interventional (grey circles) groups. Few of the conventional group reached their goal INR at first recheck, demonstrating the faster time to observed INR normalization is not a byproduct of prolonged time of transport or delayed measurement.

As expected, 4-factor PCC was administered earlier in the intervention group (159 versus 366 minutes; p=0.0001). Patients in the interventional groups were observed to have a significantly faster time to anticoagulant reversal (181 versus 541 minutes, p = 0.001; Figure 3.

Figure 3.

Time to event curves demonstrating the time to achieving Warfarin reversal (≤1.5) comparing the intervention versus conventional groups demonstrating significantly faster time to goal achievement in the pre-transfer administration group (p = 0.001). While the interventional group reaches goal more quickly, this difference is eliminated within 12–24 hours. Several failures to achieve the primary outcome (y-axis) are attributable to early mortality.

Summaries of all of the studied clinical outcomes between the groups are shown in Table 3. Overall, thromboembolic complications were rare, with only 2 events noted, both in the conventional group. We observed no differences in the need for blood products or length of stay. We unexpectedly observed a higher mortality rate in the interventional group compared to the conventional transport group (p =0.049). The interventional group was also observed to have a lower GCS score and higher rate of endotracheal intubation prior to transport.

Table 3.

Differences in outcomes between study groups.

Result	Interventional n=34	Conventional n=38	p-value
Timing of INR reversal
INR <1.5 on tertiary care facility arrival, n (%)	28 (82)	0 (0)	<0.01
Time to first INR <1.5, minutes (IQR)	181 (152, 254)	541 (341, 737)	<0.01
Time from referral facility INR to 4-factor PCC administration, minutes (IQR)	115 (79, 189)	322 (210, 436)	<0.01
Secondary outcomes
Length of hospital stay, days (IQR)	5 (3, 8)	5 (4, 9)	0.62
ICU length of stay, days (IQR)	2 (0, 3)	1.5 (1, 3)	0.62
Administration of blood products, n (%)	10 (29)	7 (18)	0.27
Thromboembolic complication, n (%)	0 (0)	2 (5)	0.18
In-hospital mortality, n (%)	7 (21)	2 (5)	0.05

ICU: Intensive care unit; IQR: Interquartile range; PCC: Prothrombin complex concentrate;

Discussion

This study demonstrates the feasibility of a protocol for the delivery and administration of 4-factor PCC in smaller, outlying rural facilities that lack the resources necessary for the rapid reversal of warfarin-induced anticoagulation in the setting of major hemorrhage. The primary finding of the study was a marked increase in the proportion of patients normalizing INR upon ED arrival and a faster time to normalization overall.

Given the retrospective nature of the study and non-standardized INR measurement times, it remained a possibility that shorter transport times led to the differences in outcomes between study groups. However, Figure 2 demonstrates the majority of the interventional patients normalized prior to arrival to the ED. On the other hand, very few patients in the conventional group reached the target INR upon arrival to the accepting facility. These data suggest routine care (including vitamin K and in some cases FFP) was insufficient to reverse warfarin-induced anticoagulation in the study time frame and are not a likely explanation for our study findings. Similarly, treatment with vitamin K and FFP did not differ significantly between groups, and is not a likely explanation for these differences. Dosing also differed slightly between groups. The reasons for this relate to the fact that the intervention group was typically dosed based on the package labeling of a generic 500 units or 1000 units, not specifically factor IX units. In the conventional group, the pharmacy rounds the dose of factor IX to the nearest vial size. These differences in processes may account for the reported higher doses in the intervention group, though the study design limits this assessment.

In regards to safety, it is important to note that while we observed no difference in thromboembolic complications, our study was small and could have been affected by survivorship bias. Furthermore, all patients in this study received 4-factor PCC, and therefore we can only compare the relative incidence of complications when administered in different environments, and not the risk compared to treatment with FFP or other reversal agents alone. Importantly, we unexpectedly observed a higher mortality in the interventional group. On assessment of Table 1, patients in the interventional group were observed to have a significantly lower GCS score and higher rate of intubation prior to transfer, which may indicate a higher severity of illness in this cohort While this may be responsible for this unexpected finding, this hypothesis remains untested and deserves careful observation in any subsequent study of the topic.

In general, literature regarding rural treatment of critical illness and bleeding is quite limited. One prior study reported the feasibility of prothrombin complex concentrates using a national aeromedical service in rural areas, though only 3 patients received treatment in that report.(11) This study differed significantly from ours, as there was no centralized communications specialist, nor specific protocols designed to capture or treat these patients. Rather, their results reflect ad hoc provision of national guidelines for coumadin reversal instead of a systematic, centralized approach to the protocolized reversal of these patients. Our data extend these findings by providing additional safety and feasibility data in a significantly larger cohort of patients with a reasonable, though imperfect control group as a comparison.

The greatest strength of this study lies in the innovation and implementation of the protocol through a collaborative effort involving communications specialists, prehospital personnel, pharmacists, and poison control. When a referral facility called to request our facility-operated air transport, patients were screened for potential hemorrhage. If the patient was determined to have a significant bleed per protocol definitions, specialists asked if patient was taking anticoagulants and requested a PT/INR if not already obtained. Finally, specialist inquired about the availability of reversal agents or antidotes. If unavailable, the facility-operated air ambulance was dispatched with 4-factor PCC. Four-factor PCC was maintained on all facility-operated helicopters as well as being housed in an automated dispensing cabinet exclusive to the air ambulance service. Once the 4-factor PCC was administered to the patient, the team replenished their supply. This cabinet was maintained and stocked by the tertiary-care facility’s central pharmacy department. Rigorous training of the communication specialists and all air ambulance personnel was performed prior to and following implementation. In-services regarding proper mixture and administration of 4-factor PCC was performed frequently by the emergency department pharmacists, and pocket dosing cards was available on all aircrafts.

Limitations

There are several limitations of this study that deserve consideration. While we undertook several methods to ensure complete capture, it is possible patients may have been missed. Specifically, while all transfers meeting criteria for a significant bleed should have been screened as described by communications specialists, we did not record the number of transfers screened and excluded for various reasons, and this could have introduced bias into the study. Second, although patients were prospectively identified, the clinical data was collected retrospectively using chart abstraction techniques and typical concerns regarding retrospective study designs apply. Notably, 26 (14%) of cases needed to be excluded due to missing INR data, which could have affected our results. Furthermore, INR values were not checked at pre-specified time intervals. While Figure 2 suggests against this as a major source of bias, it remains a weakness of the study design that cannot be entirely eliminated. We submit that the face validity and magnitude of change observed in this study makes it unlikely these missing data would significantly affect the clinical relevance our findings. Third, we caution against over interpretation of secondary outcomes such as thrombotic events, transfusions, and length of stay due to potential survivorship bias. Fourth, It is also important to note that while alternative direct oral anticoagulants (DOACs) were included in our clinical protocol due to the potential that 4-factor PCC may help in their reversal(12), they were excluded from the present analysis due to difficulty in defining a uniform definition of anticoagulation reversal. Fifth, the conventional group serves as an imperfect comparison due to a mixture of both air and ground transport, differing transport times, and likely differing severities of illness. While we undertook methods to assess the effects of these differences on our results, these methods were limited and we present the data for the reader’s interpretation.

Critically, mortality was unexpectedly higher in the intervention group. While GCS score and intubation rates suggest this might be due to a higher severity of illness in the intervention group, the study was not designed to test this hypothesis and further study is warranted prior to widespread implementation of protocols such as this. It is conceivable that the increased mortality was due to unrecognized thromboembolic events in the intervention group, as we did not determine cause of death on chart review and patients did not undergo autopsies. However, no VTEs were diagnosed in the intervention group and all patients in both cohorts received treatment 4-factor PCC. While we believe it is unlikely that a several hour difference in administration of 4-factor PCC would be responsible for a significant increase in rates of fatal VTE, this was not formally evaluated in the study.

Our results may or may not be generalizable to other practice settings, as the time differences we report are highly likely to vary based upon the geography, use of air versus ground transport services, and critical access hospital capabilities in other practice settings. Although we observed a more rapid time to reversal of anticoagulation, we observed no differences in clinical outcomes. Our study was likely underpowered to detect such differences, and remains an area for future investigation as the more rapid reversal over the time frames we describe may or may not be clinically relevant. Finally, the costs of both 4-factor PCC and air ambulance transfer are significant, and the cost-effectiveness of this strategy remains an area for further investigation.

Conclusions

Dispatch of an air ambulance with 4-factor PCC and rural hospital pre-transfer administration is feasible and leads to a decreased time to INR reversal among patients with major hemorrhage in the setting of warfarin anticoagulation.

Table 2.

Concomitant treatments administered, by interventional versus conventional groups

Parameter	Interventional (n=34)	Conventional (n=38)	p-value
Fresh frozen plasma, n (%)	9 (26)	6 (16)	0.27
Units transfused, # (IQR)*	2 (1, 3)	2 (1.5, 2)	0.85
Vitamin K (any), n (%)	27 (79)	35 (92)	0.12
Oral	2 (6)	1 (3)	0.49
Intramuscular	6 (18)	7 (18)	0.93
Intravenous	13 (38)	13 (34)	0.61
Subcutaneous	4 (12)	12 (32)	0.04
4-factor PCC Dose, IU/kg	35 (25, 50)	25 (25, 35)	0.64

IQR: Interquartile range; IU: International Units; PCC: Prothrombin complex concentrate

^*Units transfused among patients given at least 1 unit