In developed countries, up to 2% of the general population receives chronic oral anticoagulation with vitamin K antagonists (VKA) 1 . The rate of severe bleeding complications is above 3.8% per person-year and the associated mortality rate ranges between 10 to 60%, especially in case of intracranial hemorrhage (ICH), in the first week following the onset of the hemorrhage 1 2 3 4 . In case of severe hemorrhage, the two main predictors of poor prognosis are over anticoagulation, that is, an international normalized ratio (INR) above the therapeutic range, and late VKA reversal (over eight hours after presentation) 5 6 .

For several decades the evidence supporting appropriate clinical use of VKA antidotes has remained scarce 7 . In a recent randomized clinical trial, Sarode et al. demonstrated the superiority of prothrombin complex concentrate (PCC) versus fresh frozen plasma (FFP) to normalize coagulation within one hour (62.2 versus 9.6%; P <0.05) 8 . The role of vitamin K in association with PCC, in maintaining the normalized coagulation over six hours has been emphasized 9 . The importance of rapid hemorrhage control has led international guidelines to recommend the use of PCC rather than FFP 10 11 12 . For complete VKA reversal, all guidelines recommend an infusion of PCC (at least 20 IU/kg factor IX equivalent) related to admission INR value in combination with at least 5 mg of vitamin K to rapidly achieve a post-reversal INR ≤1.5 and maintain a normal coagulation profile over six hours. So as to save time, French guidelines suggest, on the basis of a previous study, the possibility of administering a probabilistic single-regimen dose of 25 IU/kg of PCC and 10 mg of vitamin K as soon as a severe hemorrhage is diagnosed 9 10 . In all situations, post-reversal INR must be measured 30 minutes after the infusion to evaluate the efficacy of the treatment and make any necessary adjustments thereafter, and measured six hours later to control the efficacy of vitamin K 10 11 12 . French guidelines were published in 2008 and have become a standard of care for the management of these patients in France 10 . Treatment time frames were not specified in the international guidelines and the effect of early reversal on mortality has not been studied. At a time when new oral anticoagulant agents without specific antidotes are emerging as promising alternatives to VKAs, it appears important to better define the prognosis benefit of anticoagulant reversal in the management of patients on oral anticoagulants with severe hemorrhage 13 .

The aim of this observational study was to evaluate the impact of guideline-concordant administration of PCC and vitamin K (GC-PCC-K) on early mortality in VKA-treated patients with severe hemorrhage. We tested the hypothesis that GCA-PCC-K is associated with improved early outcome.

The study flow chart is shown in Figure 1. A total of 833 patients were included in the study (detailed information on patient recruitment in the participating centers is provided in the Acknowledgment section). Eleven patients were excluded because they did not meet the criteria for severe hemorrhage. At Day 7, mortality was recorded for all patients included. Therefore, the analyzed population comprised 822 patients, including 262 (32%) patients with ICH. The main characteristics of the whole cohort and patients with ICH are shown in Table 1. A total of 110 patients (13%) died within seven days, including 86 ICH patients, representing 10% of the overall population and 33% of the ICH subgroup. Deaths in the ICH subgroup accounted for 78% of deaths in the overall population (Table 1).

Data are mean ± SD, median [25^th to 75^th percentiles], or number (percentage). P-values refer to comparison between alive and dead patients at Day 7.

INR, international normalized ratio; SAP, systolic arterial blood pressure; GCS, Glasgow coma scale; VKA, vitamin K antagonist.

*The sum is higher than 100% because two indications can be established for the same patient.

**Patients with no antiplatelet treatment or no history of severe hemorrhage are deducted from the presented results.

***“Other” types include hemopericardium (N = 1), hemothorax (N = 15), vascular bound (N = 4), emergency procedures (N = 126) and shock or red blood cells transfusion for epistaxis (N = 55), hematuria (N = 34) or wound of the scalp (N = 4).

VKA reversal treatment comprised PCC in 509 patients (62%) but only 379 (46%) received the recommended doses (≥20 IU/kg). Only 27 patients (3%), including 2 in the ICH subgroup, received FFP. Vitamin K was administered in 583 patients (71%) and the recommended dose (≥5 mg) was used in 531 patients (65%). The combination of PCC (≥20 IU/kg) and vitamin K (≥5 mg), regardless of administration time, was given to 336 patients (41%). Guideline-concordant administration of PCC was uncertain in 15 patients (total dose of PCC without the patient’s body weight), and two experts agreed on 13 cases (80% agreement with a κ of 0.60). Finally, treatment was considered concordant with guidelines in 313 patients (38%).

Admission INRs are shown in Figure 2. Post-reversal INR values were available for 417 patients (51%); however, they were obtained within one hour in only 61 (7%) patients and within six hours in 225 (27%) patients. Considering all post-reversal INR that we had collected, the proportion of patients in which coagulation was normalized (post-reversal INR ≤1.5) was significantly greater with GC-PCC-K. Nobody received a complementary PCC dose in case of post-reversal INR >1.5.

In multivariate analysis, we observed that early mortality was significantly associated with no GC-PCC-K in the whole cohort (OR = 2.15 (1.20-3.88); P = 0.011) and ICH patients (OR = 3.23 (1.53-6.79); P = 0.002) (Table 2). Using bootstrapping, the OR of no guideline-concordant treatment was 2.44 (1.09-5.74) in the whole cohort leading to an optimism of 0.24.

CI, confidence interval; ICH, intracranial hemorrhage; GCS, Glasgow coma scale; NS, not statistically significant; OR, odds ratio; SAP, systolic arterial blood pressure; trt: treatment.

*“Other” types include hemopericardium (N = 1), hemothorax (N = 15), vascular bound (N = 4), emergency procedures (N = 126) and shock or red blood cell transfusion for epistaxis (N = 55), hematuria (N = 34) or a wound of the scalp (N = 4).

Within the first week after admission, 10 thromboembolic events with five associated deaths were recorded: pulmonary embolism (N = 5), cerebral or coronary ischemic events (N = 3), venous (N = 1) and peripheral artery thrombosis (N = 1). Hemorrhagic recurrence was observed in 83 patients (11%) with four associated deaths, including 14 patients with two deaths in the ICH subgroup.

This is the first study showing that GC-PCC-K (that is, administration of recommended doses of PCC and vitamin K) within eight hours after patient admission for severe hemorrhage is associated with a decrease in seven-day mortality. We found a two-fold decrease in mortality (OR = 2.2 (1.2-3.9); P = 0.011) when guideline-concordant treatment was performed. Mortality among ICH patients has been reported to be approximately 20 to 30% in non-anticoagulated patients and 40 to 65% in VKA-treated patients 5 17 . In this study, GC-PCC-K was associated with a three-fold reduction of mortality in the ICH subgroup (OR = 3.2 (1.5 to 6.8); P = 0.002), suggesting that recommended doses of PCC and vitamin K administered in the first eight hours, can decrease mortality to levels close to those observed in non-anticoagulated ICH patients. The very low mortality rate (4%) observed in patients with other types of severe hemorrhage does not allow us to draw conclusions with regard to these patients due to lack of statistical power. Hypotension was associated with increased mortality in the whole cohort, but this association was not observed in ICH patients. GCS was also strongly associated with early mortality. As a result of population ageing, the prevalence of VKA-related hemorrhagic complications (ICH in particular) increases worldwide 1 5 . The median age of 80 years in our study is a clear reminder of this fact. However, we did not observe a statistically significant association between age and mortality.

Guideline-concordant treatment doses for normalization of coagulation and a fixed time frame within which reversal is considered useful are probably the two conditions required to improve prognosis. To achieve a post reversion INR ≤1.5, most pharmaceutical companies recommend a PCC dose related to admission INR value. However, INR is not a linear representation of the coagulation and a dose-response curve has not been demonstrated in the literature 18 19 . Moreover, in the absence of bedside INR monitoring, waiting for an admission INR value from the laboratory in bleeding patients may result in increased blood losses and coagulation factor consumption. Based on a previous small cohort study of VKA-treated patients with ICH suggesting that a single-regimen dose of PCC (administered as an intravenous bolus) is effective in achieving a post-reversal INR ≤1.5, French guidelines recommend, in order to save time when admission INR is not available, the administration of a single-regimen dose of 25 IU/kg factor IX equivalent PCC (that is, 1 mL/kg with available four-factor PCCs) in combination with 10 mg of vitamin K 9 10 . Use of the recommended single-regimen dose has not yet become routine practice, as shown by the fact that only 205 patients (25%) received a dose of 25 IU/kg in our study. Guidelines also strongly recommend measuring the INR after reversal to control the degree of reversal and further normalize the coagulation 30 minutes after the first administration if the PCC dose was insufficient 10 . Unfortunately, post-reversal INR was obtained within less than one hour in only 7% of our patients, underscoring the need for improved guideline adherence. In our study, bedside INR monitoring was never used. Adoption of this technique might help increase the rate of post-reversal INR measurement within one hour after treatment administration.

Time between hemorrhage onset and VKA reversal is a key variable to be considered when evaluating treatment efficacy. In patients with ICH, hematoma volume is crucial for the prognosis and is known to increase during the first day following admission 20 . Not surprisingly, it has been shown that late reversal, over eight hours after admission, does not modify the outcome 5 . Based on a previous study, we chose a time frame of eight hours to define timely treatment, which was considered realistic taking into consideration the time needed to confirm the diagnosis of hemorrhage, which is not always easy, for example, in patients with deep-muscle hematoma, gastrointestinal hemorrhage or ICH.

Fear of thrombosis may be one of the reasons for not performing VKA reversal, particularly in patients (6%) with a normal initial INR (≤1.5) and a potentially high thrombotic risk if the single-regimen approach is considered. In severely bleeding patients with a high thromboembolic risk, total mortality (N = 110, 13%) was mainly related to hemorrhage, with only five deaths (0.6%) observed in patients with a thromboembolic event 21 . This provides a good example of the risk balance and reinforces the fact that the priority is to organize a rapid reversal as soon as the diagnosis of severe hemorrhage is confirmed.

Some limitations of our study deserve consideration. First, we studied mortality within seven days as it appears to be the period when the cause of death is a direct consequence of hemorrhage in all situations, even in ICH patients 22 . In a study of ICH-related mortality in warfarin users and non-users, Huhtakangas et al. observed a two-fold difference between death rates during the first week after stroke onset and no further increase after one week with parallel mortality curves 3 . We considered that many clinical complications or medical attitudes not related to VKA reversal could interfere with 30-day mortality in all groups. Second, we did not exclude patients with care limitation decisions, although such decisions are not rare in an aged population with severe neurological insults, such as in ICH, and may also be associated with “self-fulfilling prophecies” 23 . In an observational study conducted in emergency conditions it was not possible to clearly identify the timing of these decisions in relationship with VKA reversal and to precisely distinguish explicit and implicit decisions during this early phase.

It is interesting to consider that only 41% of patients received the recommended doses of PCC and vitamin K and only 38% of them did so within the first eight hours after the onset of bleeding. These results were similar to those of a previous study 16 . A large French study, which was performed in 33 hospitals during 2008 to 2011 showed the same difficulties in obtaining good guideline adherence for VKA reversal in patients with severe hemorrhage 24 . More important, in our study, only 27% of patients who had received PCC had an INR post-reversion during the six hours. It appears therefore necessary to promote guideline dissemination and implementation. It would also be beneficial to improve the current guidelines by including explicit time frames for treatment administration.

Guideline-recommended doses of PCC and vitamin K were associated with a reduced mortality in patients on VKA therapy presenting with severe hemorrhage, particularly among patients with ICH in whom we observed a three-fold decrease in mortality. VKA reversal should be initiated within eight hours after admission. Only 38% of our patients received appropriate treatment, indicating the need for improved guideline adherence.

● In patients on VKA therapy presenting with severe hemorrhage, international guidelines recommend, as soon as the diagnosis is confirmed, the administration of PCC (≥20 UI/kg) and vitamin K (≥5 mg) to normalize coagulation (post-reversal INR ≤1.5).

● A guideline-concordant administration dose of PCC and vitamin K administrated in the first eight hours was associated with a two-fold decrease in seven-day mortality overall and with a three-fold decrease in the ICH subgroup

● The guideline-concordant reversal was performed in 38% of the patients within eight hours after admission

● Whereas pre-reversal INR is not absolutely necessary, post-reversal INR is essential to evaluate treatment efficacy

● The post-reversal INR target must be performed systematically and immediately after PCC administration

CI: Confidence interval; EPAHK: Evaluation pronostique de l’antagonisation des hémorragies graves sous AVK; FFP: Fresh frozen plasma; GC-PCC-K: Guideline-concordant of administration PCC and Vitamin K; GCS: Glasgow coma scale score; HR: Hazard ratio; ICH: intracranial hemorrhage; INR: International normalized ratio; OR: Odds ratio; PCC: Prothombin concentrated complex; SAP: Systolic arterial blood pressure; VKA: Vitamin K antagonist.

KT has received consulting and speaker’s fees from Sanofi, Lilly and LFB. BR has received travel grants, consultancy fees, honoraria and study grants from Sanofi, GlaxoSmithKline, ThermoFisher, Octapharma, Sangart and LFB. BT report has received consulting fees from LFB. C.-MS has received grant monies (industry-related sources) from NovoNordisk, CSL Behring and LFB, and speaker’s fees from CSL Behring and LFB. EV has received travel grants, consultancy fees, honoraria and study grants from Pfizer, Eli Lilly, Sanofi, Abbott, Fresenius, Stallergene, Boehringer Ingelheim, Merk and LFB. BV has received grant monies (industry-related sources) from ThermoFisher and speaker’s fees from ThermoFisher, CSL Behring, Octapharma and LFB.

KT and BV contributed to the conception and design of the study, data collection and analysis, manuscript writing and final approval of the manuscript. BR contributed to the conception and design of the study, manuscript writing and final approval of the manuscript. BT, EV and CMS contributed to the conception and design of the study, revising the draft for important intellectual content, and final approval of the manuscript. All authors read and approved the final manuscript.