Since the 2000s, many prognostic scores were developed to predict traumatic haemorrhage. Most of these studies were retrospectives based on registers. Due to missing data on death due to bleeding, these studies chose to predict the massive transfusion risk as a surrogate of haemorrhagic death. These scores include clinical parameters (vital signs), laboratory values (Haemoglobin, lactate, Base excess) and/or imaging (CT or ultrasound) values. The scores showing best performance are the Trauma Associated Severe Haemorrhage (TASH) score, developed and validated on the German register (DGU-Register) and the ABC score developed and validated in the United States of America. However, the majority of these scores cannot be applied at the trauma scene due to the unavailability of laboratory and imaging values. Therefore, their clinical utility remains unclear. To overcome the need for diagnostic tests, authors have developed and recently validated a clinical prognostic score in identifying trauma patients with, or at risk of, significant haemorrhage based on predicted probabilities of death due to bleeding: BATT score. This score was developed from an international cohort using data from 271 Trauma Centres in 41 countries on 5 continents and uses first clinical parameters at initial assessment. The BATT score predicts death due to bleeding and has been validated on a large population in England and Wales. It could also predict massive transfusion, as a surrogate of haemorrhagic death, earlier at the trauma scene. Its feasibility and external validation would make its clinical utility superior to other scores while identifying a greater number of patients requiring early management. Our study is an external validation of pre-existing prognostic scores of traumatic haemorrhages (TASH , ABC and BATT score) at different times of care (Scene of Injury, admission at the trauma room) in order to assess their overall performance, discrimination and calibration in the prediction of massive transfusion, and haemorrhagic death. The objective of the study is to assess a comparison of score performances (Overall performance, discrimination and calibration). Due to the study population (STR), which is partly integrated into the German DGU-Register, the investigators expect good transportability of the TASH score to the Swiss Trauma Registry in terms of overall performance, discrimination and calibration. The ABC score should show lowers results in terms of discrimination due to its validation on small cohorts exclusively in North America. The new BATT score predicting death due to bleeding has been validated on a large English cohort of more than 100,000 patients. It identifies all patients with haemorrhage and not only patients who have received a massive transfusion subject to survival bias. In this context, the BATT score provides good discrimination with only simple physiological variables available at the trauma scene. In case of its external validation on the STR as part of our study, its feasibility would make its clinical utility superior to other pre-existing scores, while identifying a greater number of patients requiring early management. Its application would activate a massive transfusion plan directly at the trauma scene and save precious time.
1 Study Method 1.1 Study Design The MTP STUDY is a retrospective, observational, non-interventional study based on a multicentric, anonymised Register. The Study is conducted by the emergency department of the University Hospital (CHUV) in Lausanne, Switzerland. The study design is observational, and no intervention is applied as part of the study protocol. 1.2 Sample Size and Power consideration As his study is retrospective, the sample size is fixed. The number of participants will depend on the STR Database. The investigators anticipate using the data of 10'000 participants included in the Swiss Trauma Registry from 1st January 2015 to 31st December 2019. A post-hoc power calculation will be performed. 1.3 Timing of Final analysis This statistical analysis plan is added before receiving the dataset and before any analyses have been conducted. After receiving the dataset, the investigators will check for data consistency. Once the database has been checked, statistical analysis will be performed (October 2020) 1.4 Baseline patient characteristics 1.4.1 Collected baseline patient characteristics The observational study is designed to record a set of demographical data, clinical examination in prehospital or in-hospital setting, in-hospital biochemical values and imaging variables for each included patient. The investigators will extract the data from the Swiss trauma registry (STR). The Investigators plan to extract the biological parameters from the register at the trauma room stage. The clinical examinations will be the first measures at the pre and in-hospital stage. 1.4.1 Descriptive summary of baseline patient characteristics The investigators will list general patient characteristics in a baseline characteristics table. Data will be presented as mean with standard deviation (SD) when normally distributed or as median with interquartile range in case of skewed data. Dichotomous and categorical data will be presented in proportions. 1.5 Assumed confounding covariate The majority of the requested variables from the STR are inevitably correlated, as most relate to the haemodynamic status of the patient and the trauma severity. The values of the variables can be confounded by unmeasured factors, such as environmental, genetic or psychological influences. Therefore, the investigators provide an example of possible confounding variables: * Clinical examinations in prehospital settings (i.e. heart rate, systolic blood pressure, respiratory rate, Glasgow Coma scale) are assumed to be confounded by: \- Quality of the measurements, stress, pain and anxiety. These confounding covariates should be minor for the statistical analysis. * Clinical examinations at hospital admission (i.e. vitals signs) are assumed to be confounded by: \- Quality of the measurements, administration of inotropes and/or vasopressors during transport, administration of propofol (negative inotropic effect), induced comas and the need for mechanical ventilation. * Because some patients will not survive long enough to receive 10 red blood cell units, massive transfusion is subject to misclassification. To correct this misclassification, the investigators add in the massive transfusion definition the use of ≥ 3 RBC administered in the first hour (if the variable is available in the registry). * The trauma-induced coagulopathy as secondary outcome is assumed to be confounded by anticoagulant treatments. To counteract the confounding covariate, the investigators will include the fibrinogen \< 1.5 g/L in the definition of trauma-induced coagulopathy. The investigators will also define a subgroup of patients with anticoagulant treatment or not. The investigators acknowledge that there will be residual confounding in our dataset due to the presence of unmeasured confounding, some of which is listed above. However, the actual measured variables reflect daily practice and so are assumed to reflect similar confounding in daily assessments. 2\. Analysis 2.1. Analysis methods 2.1.1. Efficacy analyses of primary outcome First, the investigators will assess the accuracy (overall performance), discrimination and calibration of ABC, TASH and BATT score for the prediction of massive transfusion in trauma patients at the trauma scene and at the hospital admission. 2.1.2. Accuracy The accuracy will be assessed using the Brier score. Where Y is the observed outcome and p is the prediction of the model The Brier score depends on the prevalence of the outcome, the investigators will also calculate the scaled Brier score to account for the baseline risk of Massive transfusion. The scaled Brier score ranges from 0% to 100% and indicates the degree of error in prediction. A scaled Brier score of 0% shows perfect accuracy. 2.1.3. Discrimination Discrimination is the ability of the score to correctly identify patients with the outcome. The investigators will estimate the sensitivity, specificity, positive and negative likelihood ratio for the defined threshold of each score (ABC, TASH, BATT). The likelihood ratio is the likelihood of a positive score in a patient with the outcome compared to the likelihood of a positive score in a patient without the outcome. The positive likelihood ratio is the ratio of sensitivity to 1-specificity. The negative likelihood ratio is the ratio of 1-sensitivity to specificity. A positive likelihood ratio of 10 or above will result in a large increase in the probability of the outcome. A negative likelihood ratio of 0.1 or less will result in a large decrease in the probability of the outcome. The investigators will plot the Receiving Operating Characteristic (ROC) curve which is the sensitivity (true positives) on 1-specificity (false positives) for each defined threshold of each score. An ideal score will reach the upper left corner (all true positive with no false positive). The investigators will estimate the area under the ROC curve (AUROC) that corresponds to the concordance statistic (C- Statistic) for binary outcome. A C-statistic of 1.0 shows perfect discrimination ability. 2.1.4. Calibration Calibration is the agreement between observed and predicted outcomes. The investigators will mostly estimate calibration as the difference between the mean predicted and observed probabilities and the ratio of the predicted and observed number of events (P/O). The investigators will plot the observed and predicted probabilities of massive transfusion by decile of the score and with local regression based on LOESS algorithm. The investigators will estimate the calibration intercept and slope of the calibration plot as a measure of spread between predicted and observed outcome. Ideally, the intercept would be zero indicating that the predictions are neither systematically too low or too high and the slope would be 1. Unfortunately, the investigators cannot estimate the calibration of the BATT score, because of its different outcome (death due to bleeding and not the massive transfusion as the TASH and ABC score). For the BATT, calibration will be assessed with the outcome of death due to bleeding or early death. 2.1.1.1 Efficacy analyses of secondary outcome The investigators will perform the same analysis for secondary outcomes as the primary outcome. 2.2 Missing Data Due to the retrospective aspect of the study based on a multicentric registry, the investigators expect to have some missing data for some prehospital and in-hospital predictors. 2.2.1. Imputation method To estimate baseline risk for the full dataset, the investigators will replace missing predictors using multiple imputation by chained equations on sex, age, systolic blood pressure, respiratory rate, heart rate, Glasgow coma scale, Haemoglobin, base excess, type of injury (penetrating/blunt) Instable pelvis fracture and open/dislocated femur fracture with 20 imputed dataset. All analysis and results will be present in two subgroups: missing data imputed and missing data excluded. 2.2.2. Early deaths and early deaths with haemorrhage as a proxy for death due to bleeding Because the investigators don't know if the Swiss Trauma Registry record the cause of death, the investigators expect some missing data about death due to bleeding as a secondary outcome. In case of missing data on secondary outcomes, the investigators will use early deaths and early deaths with evidence of haemorrhage as a proxy for death due to bleeding. Specifically, the investigators will included deaths from all causes within 12 hours of injury (excluding massive destruction of skull or brain; asphyxia, drowning and hanging are already excluded from the STR) and deaths between 12 to 24 hours with evidence of bleeding (Activation of massive transfusion protocol or blood within 6 hours or an abbreviated injury scale (AIS) diagnosis associated with haemorrhage: Blood loss \>20%, Aorta \[OR\] Vena Cava \[OR\]carotid \[OR\]femoral \[OR\]Major arteries \[OR\]veins AND laceration, - Spleen \[OR\]liver \[OR\] Kidney \[OR\] Myocardium \[AND\] major laceration, major haemothorax, retroperitoneum haemorrhage). 2.3 Subgroup analyses If the sample size permits, the investigators will conduct subgroup analysis in different subpopulations for the primary and secondary outcomes. The investigators will create the following subgroups in our MTP study: * Subgroup 1: subdivide the population into two groups: with or without isolated severe traumatic brain Injury (TBI) with AIS HEAD ≥ 3 and AIS thorax/abdominal \< 3 and/or AIS lower extremity \< 4. * Subgroup 2: subdivide the population into two groups: with or without anticoagulation treatment before trauma. * Subgroup 3: subdivide the population into two groups: with or without trauma- induced coagulopathy. * Subgroup 4: subdivide the dataset in prehospital settings and in-hospital settings. * Subgroup 5: All analysis and results will be present into two subgroups: missing data imputed and missing data excluded with the complete case analysis. 2.4 Statistical Software All analyses will be performed using STATA software (version 16.0; Stata Corp, College Station, Texas, USA). 3 Ethical Approval As is mandatory in Swiss law (KVG), STR is authorised by the Human Research Act (HRA) as a quality registry for the Highly Specialised Medicine (HSM). Due to the retrospective aspect of our observational study based on an anonymized registry (identity, date of birth, trauma scene and hospital location unknown) and according to the swiss law by the HRA (Art. 2), the investigators don't need a protocol submission to an ethics committee. 4 Conclusion This Statistical analysis plan (SAP) presents the principles of analysis of the MTP study and discusses its major methodological and statistical concerns. The investigators hope that the results of the MTP study will be as transparent and robust as possible, so that the investigators minimized the risk of outcome reporting bias and data-driven results.
Study Type
OBSERVATIONAL
Enrollment
13,222
Our Study is an observational non-interventional study based on a multicentric anonymized registry. It's an external validation of pre-existing prognostic scores of traumatic haemorrhages at different times of care (Scene of Injury, admission at the trauma room) in order to assess their overall performance, discrimination and calibration in the prediction of massive transfusion and haemorrhagic death. The statistic intervention will be a comparison of score performances (Overall performance, discrimination and calibration).
Centre Hospitalier Universitaire Vaudois
Lausanne, Canton of Vaud, Switzerland
Presence of Massive transfusion
The primary outcome is the presence of massive transfusion (MT) in Swiss trauma patients, defined by a transfusion equal to or greater than 10 Red blood cell (RBC) in the first 24 hours or ≥ 3 RBC in the first hour if available in the Swiss Trauma Registry.
Time frame: 24 hours after trauma
Death due to bleeding
Secondary outcome is death due to bleeding if available in the registry (will be used for sensibility analysis)
Time frame: at 28 days after trauma
Early death (< 24 hours)
early death: Death in the first 24 hours (will be used for sensibility analysis)
Time frame: 24 hours after trauma
Coagulopathy at Hospital admission
Coagulopathy is defined by an INR \> 1.2 or TP \< 70% or Fibrinogen ≤ 1.5 g/L (will be used for sensibility analysis)
Time frame: within 2 hours of hospital admission
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.