Guideline-concordant administration of prothrombin complex concentrate and vitamin K is associated with decreased mortality in patients with severe bleeding under vitamin K antagonist treatment (EPAHK study)

This prospective observational study was conducted from May 2009 to June 2010 in 44 emergency department hospitals in France. The protocol was approved by the Institutional Review Board (Comité d’évaluation de l’éthique des projets de recherche biomédicale, CEERB, Paris, France) on 5 June 2009. Electronic database permissions were obtained from the Comité consultatif sur le traitement de l’information en matière de recherche dans le domaine de la santé (CCTIRS) on 15 January 2009 and from the Commission nationale de l’informatique et des libertés (CNIL) on 3 April 2009. Because there was no randomization and only standard of care was performed, the Institutional Review Board waived the need for informed patient consent. Academic (N = 29) and general (N = 15) hospitals located in rural or urban areas were included. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for reporting observational studies were followed [14].

In each center consecutive VKA-treated patients, aged over 18 years with the exception of pregnant women, with a severe hemorrhage were included in the study. Hemorrhage was defined as severe if at least one of the following criteria was present: 1) externalized bleeding that cannot be stopped by applying conventional measures; 2) hemodynamic instability defined by systolic arterial blood pressure (SAP) <90 mmHg, or signs of shock; 3) need for an emergency procedure to stop bleeding; 4) need for red blood cell transfusion; 5) life-threatening bleeding or bleeding that compromises function, for example, intracranial or intraspinal hemorrhage, intraocular or retro-orbital bleeding, hemothorax, hemoretroperitoneum, hemopericardium, deep-muscle hematoma and/or neural compression syndrome, acute gastrointestinal bleeding and hemarthrosis [10].

Hemorrhages were classified into four categories: 1) gastrointestinal hemorrhages; 2) ICH; 3) deep-muscle and subperitoneal hematomas; and 4) “others”, including hemothorax, hemopericardium as well as hemorrhages for which anticoagulant reversal is not systematically required but may be urgently needed in case of shock or transfusion (for example, epistaxis or hematuria).

The data collected were patient characteristics, medical treatments, indication, duration and type of VKA treatment, history of thrombotic and bleeding events, Glasgow coma scale score (GCS), SAP, admission and post-reversal INR, and reversal treatment data (time from admission to reversal, PCC dose, vitamin K dose, surgical or radiological intervention, number of transfused red blood cell units, FFP units and platelet concentrates). GCS was divided into three strata: GCS ≤8 (severe), 8 < GCS ≤13 (moderate), and GCS >13 (mild). SAP was also divided into three strata: <90 mmHg (hypotension), 90 ≤ SAP ≤140 mmHg, and >140 mmHg (hypertension). Patients with an INR under 1.5 at admission were considered as having a normal coagulation profile. An INR between 1.5 and 4 was considered within the therapeutic range (including in patients with mechanical heart valves) and an INR >4 above the therapeutic range. When post-reversal INR was obtained, the value and the time from admission to the end of PCC administration were recorded. To provide biological evidence of coagulation normalization following reversal, the proportion of patients with a post-reversal INR ≤1.5 was determined. A four-factor PCC widely available in French hospitals was used as Kaskadil^© until September 2009 and Kanokad^© (Laboratoire Français du Biomédicament LFB, Les Ulis, France) or Octaplex^© (Octapharma, Swizerland). These products were a human plasma-derived four-factor PCC, including factors II, VII, IX and X, and have undergone detergent treatment and nanofiltration (Kanokad^© and Octaplex^©) for viral inactivation. These products also contain proteins C and S, two natural anticoagulant vitamin K dependent factors. For these two PCC, 1 ml corresponds to 25 IU of Factor IX. VKA reversal was considered guideline-concordant if a PCC dose ≥20 IU/kg and a vitamin K dose ≥5 mg were administered. Although no time frame was specified in the guidelines, we considered treatment administration within eight hours after hospital admission as a criterion for good practice, based on the median time from admission to treatment observed in a previous study [5, 15].

GC-PCC-K was considered only for levels 3 and 4. Administration of FFP was not considered as an appropriate treatment according to the guidelines [10‐12]. Patients were classified automatically according to this algorithm. If data were missing, the case was reviewed and classified by two experts designated by the scientific committee. In case of disagreement, a third expert was assigned to definitely categorize the patient. Agreement between experts was calculated (kappa score).

The primary outcome was seven-day mortality. Secondary outcome measures included thromboembolic events and hemorrhage recurrences within seven days.

Data are presented as mean ± standard deviation (SD), median (25^th to 75^th percentiles) or number (percentage).

The sample size was determined based on a preliminary study that provided a first estimate of mortality in the absence of appropriate reversal [16]. In the present study the sample size was calculated at N = 750 patients in order to allow a 80% power to detect (with a two-sided 5% α risk) a difference in mortality corresponding to 20% in the absence versus 12% in the presence of appropriate reversal, estimating that approximately 30% of the studied patients would receive appropriate reversal.

Factors associated with seven-day mortality were identified through univariate analysis (t-tests or χ² test). A multivariate analysis was then performed to identify independent factors in the overall study population and in the subgroup of patients with ICH. Variables included in the multivariate analysis were those significant (P <0.05) in univariate analysis and treatment appropriateness. Calibration of the multivariate model was assessed using the Hosmer-Lemeshow test and discrimination using c-statistics. To perform an internal validation, we used a bootstrapping procedure (N = 500 bootstrap samples) to evaluate optimism of the parameters of models and calculated the odds ratio (OR) associated with inappropriate treatment and its 95% confidence interval (CI). The difference between this OR and the observed OR in the cohort was defined as the optimism.

In view of the high mortality associated with ICH compared with other types of hemorrhages, we also prespecified a subgroup analysis of patients with ICH [5].

All P- values were two-tailed and statistical significance was set at P <0.05.