Update of the trauma risk adjustment model of the TraumaRegister DGU™: the Revised Injury Severity Classification, version II

Abstract

IntroductionThe TraumaRegister DGU™ (TR-DGU) has used the Revised Injury Severity Classification (RISC) score for outcome adjustment since 2003. In recent years, however, the observed mortality rate has fallen to about 2% below the prognosis, and it was felt that further prognostic factors, like pupil size and reaction, should be included as well. Finally, an increasing number of cases did not receive a RISC prognosis due to the missing values. Therefore, there was a need for an updated model for risk of death prediction in severely injured patients to be developed and validated using the most recent data.MethodsThe TR-DGU has been collecting data from severely injured patients since 1993. All injuries are coded according to the Abbreviated Injury Scale (AIS, version 2008). Severely injured patients from Europe (ISS ≥4) documented between 2010 and 2011 were selected for developing the new score (n = 30,866), and 21,918 patients from 2012 were used for validation. Age and injury codes were required, and transferred patients were excluded. Logistic regression analysis was applied with hospital mortality as the dependent variable. Results were evaluated in terms of discrimination (area under the receiver operating characteristic curve, AUC), precision (observed versus predicted mortality), and calibration (Hosmer-Lemeshow goodness-of-fit statistic).ResultsThe mean age of the development population was 47.3 years; 71.6% were males, and the average ISS was 19.3 points. Hospital mortality rate was 11.5% in this group. The new RISC II model consists of the following predictors: worst and second-worst injury (AIS severity level), head injury, age, sex, pupil reactivity and size, pre-injury health status, blood pressure, acidosis (base deficit), coagulation, haemoglobin, and cardiopulmonary resuscitation. Missing values are included as a separate category for every variable. In the development and the validation dataset, the new RISC II outperformed the original RISC score, for example AUC in the development dataset 0.953 versus 0.939.ConclusionsThe updated RISC II prognostic score has several advantages over the previous RISC model. Discrimination, precision and calibration are improved, and patients with partial missing values could now be included. Results were confirmed in a validation dataset.

Unconscious trauma patients: outcome differences between southern Finland and Germany—lesson learned from trauma-registry comparisons

“…This highlights methodological limitations of the SMR comparison between trauma registries. An updated version of the RISC appeared recently with better performance for outcome prediction in patients with head injuries [16]. Furthermore, the use of specifically developed prediction models (such as the IMPACT or the CRASH) for patients with TBI should be considered in future TBI benchmarking studies.…”

Unmatched Type O RhD+ Red Blood Cells in Multiple Injured Patients

“…Patient blood samples are taken before initiating the emergency transfusion, and regular blood group serology is performed on these samples. The decision to provide RhD+ (instead of RhD-) RBCs in the ER was initially based on the fact that 85% of all patients are expected to be RhD+; and 66% are expected to be male [4,5], indicating that the majority of multiple injured patients is at low risk for RhD immunization and/or long-term effects associated with such an immunization (e.g., pregnancy). In addition, based on historical data, it was calculated that less than 50% of ER patients require immediate transfusions and that the additional safety which comes with a RhD- strategy would result in approximately 50 unnecessary RhD- transfusions (to RhD+ recipients) per year.…”

A decision tree based approach with sampling techniques to predict the survival status of poly-trauma patients

“…This allows for a more precise description of the prognosis of trauma patients. However, several limitations of the RISC model have been identified, which have led the authors to develop a new updated version of the model known as RISC II 36 . However, doctors of the Hospital of Navarre conducted an study where they proved that the prediction capability of RISC II is less than their own method (introduced in the previous paragraph).…”