Analyses of prospective observational cohort studies

Database 1 (London) was generated from the Activation of Coagulation & Inflammation in Trauma (ACIT) study, a prospective observational cohort study conducted at the Royal London Hospital, London, UK. Patient samples used in this study were collected between January 2008 and September 2013. The ACIT study was reviewed and approved by the UK Regional Ethics Committee [15]. All adult trauma patients (>15 years) who met the local criteria for full trauma team activation were eligible for enrollment and recruited into the study when research personnel were present (08:00-00:00 daily). Exclusion criteria were: arrival in the emergency department 2 h after injury; the administration of 2000 mL of intravenous fluid before emergency department arrival; transfer from another hospital; and burns covering 5% of the total body surface area. Patients were retrospectively excluded if they declined to give consent to use research samples collected, were found to be taking anticoagulant medications, had moderate or severe liver disease or a known bleeding diathesis.

Database 2 (Innsbruck) was generated from the Diagnosis and Treatment of Trauma-Induced Coagulopathy (DIA-TRE-TIC) study, a single center prospective cohort study conducted at Innsbruck Medical University Hospital between July 2005 and July 2008. Adult poly-trauma patients, who were admitted to the Level I Trauma Center at Innsbruck Medical University Hospital were eligible for enrollment and recruitment into the study. Starting in January 2006, patients with isolated traumatic brain injury were enrolled as well. Severe poly-trauma was defined as an Injury Severity Score (ISS) of ≥15 resulting from injury of at least two body regions. Isolated head injury was defined as a Glasgow Coma Score of ≤14 after blunt head trauma in patients with an Abbreviated Injury Score (AIS) of <3 in any other body region. Exclusion criteria were: patients <18 years, penetrating injuries, admittance to the study hospital later than 12 h after trauma, pre-existing coagulopathy, burn injury, malignant disease, avalanche victims, or exhibition of non-head single trauma. The study protocol was approved by the Ethics Committee of Innsbruck Medical University. The need for written informed consent was waived because study-related blood sampling was judged a minimal-risk intervention and all patients were treated according to routine institutional treatment guidelines [16].

Data have been collected by each trauma center and locally available methods have been used for sample analyses. Prothrombin concentration for database 1 was measured in the hospitals central lab with a prothrombin clot assay and a pooled normal plasma standard calibrated to an international standard for prothrombin. In both hospitals ROTEM analyses were performed locally to the emergency department, whilst PT was analysed in the hospitals central lab. In order to evaluate the predictive ability of admission coagulation biomarker data under real conditions no attempt was made to monitor variation or standardize sample analyses between hospitals.

The main outcomes considered were prothrombin concentration (IU/dL) measured at admission, total amount of PRBC (units) delivered during the first 24 h, and the proportion survival (%) at 24 h after admission. Three binary variants of these outcomes were also defined: low prothrombin (indicator of concentration <60 IU/dL at admission), massive transfusion (indicator of >10 PRBC units) and mortality at 24 h (indicator of death within the first 24 h). Biomarkers included in analyses were fibrinogen concentration, EXTEM CT, EXTEM MCF and PT, all measured at admission.

The relationship between prothrombin concentration at admission with survival at 24 h and, the amount PRBC delivered during the first 24 h after admission, as well as with fibrinogen concentration, EXTEM CT, EXTEM MCF and PT, was investigated graphically for database 1. Furthermore, Spearman’s correlation coefficient was used to quantify the relationships between prothrombin concentration and fibrinogen concentration, EXTEM CT, EXTEM MCF and PT values. Prothrombin and fibrinogen concentration, EXTEM CT, EXTEM MCF and PT are all admission values.

For each of the three binary outcomes the predictive ability of EXTEM CT, of EXTEM MCF and of PT was assessed using receiver operating characteristic (ROC) analyses. Comparison between EXTEM CT, EXTEM MCF and PT was quantified by the area under the ROC (AUC). DeLong’s test [17] was used to assess statistical difference between the AUCs. ROC analyses were performed using R package pROC [18]. Graphs and analyses were performed using R version 3.1.0 and above (R Foundation for Statistical Computing, Vienna, Austria). The significance level used was 0.05.