The Addition of the Geriatric Nutritional Risk Index to the Prognostic Scoring Systems Did Not Improve Mortality Prediction in Trauma Patients in the Intensive Care Unit

Background

Malnutrition is prevalent among critically ill patients and has been associated with a poor prognosis. This study sought to determine whether the addition of a nutritional indicator to the various variables of prognostic scoring models can improve the prediction of mortality among trauma patients in the intensive care unit (ICU).

Methods

This study's cohort included 1,126 trauma patients hospitalized in the ICU between January 1, 2018, and December 31, 2021. Two nutritional indicators, the prognostic nutrition index (PNI), a calculation based on the serum albumin concentration and peripheral blood lymphocyte count, and the geriatric nutritional risk index (GNRI), a calculation based on the serum albumin concentration and the ratio of current body weight to ideal body weight, were examined for their association with the mortality outcome. The significant nutritional indicator was served as an additional variable in prognostic scoring models of the Trauma and Injury Severity Score (TRISS), the Acute Physiology and Chronic Health Evaluation (APACHE II), and the mortality prediction models (MPM II) at admission, 24, 48, and 72 h in the mortality outcome prediction. The predictive performance was determined by the area under the receiver operating characteristic curve.

Results

Multivariate logistic regression revealed that GNRI (OR, 0.97; 95% CI, 0.96-0.99; p=0.007), but not PNI (OR, 0.99; 95% CI, 0.97-1.02; p=0.518), was independent risk factor for mortality. However, none of these predictive scoring models showed a significant improvement in prediction when the GNRI variable is incorporated.

Conclusions

The addition of GNRI as a variable to the prognostic scoring models did not significantly enhance the performance of the predictors.

Recognizing Preventable Death: Is There a Role of Survival Prediction Algorithms?

Monitoring the quality of trauma care is important but particularly challenging. Preventable death assessment aims to identify those cases where the patient's death would have not occurred if the patient had been treated differently. Determination of preventable death in trauma care is often based on calculated probability of survival, commonly by using the Trauma and Injury Severity Score (TRISS). TRISS is not suited for identifying all cases with opportunities for improvement. Combined with other methods such as morbidity and mortality conferences, however, it might be a valid approach if a complete review of all trauma deaths is not feasible at an institution.

The Reverse Shock Index Multiplied by Glasgow Coma Scale Score (rSIG) and Prediction of Mortality Outcome in Adult Trauma Patients: A Cross-Sectional Analysis Based on Registered Trauma Data

The reverse shock index (rSI) multiplied by Glasgow Coma Scale (GCS) score (rSIG), calculated by multiplying the GCS score with systolic blood pressure (SBP)/hear rate (HR), was proposed to be a reliable triage tool for identifying risk of in-hospital mortality in trauma patients. This study was designed to externally validate the accuracy of the rSIG in the prediction of mortality in our cohort of trauma patients, in comparison with those that were predicted by the Revised Trauma Score (RTS), shock index (SI), and Trauma and Injury Severity Score (TRISS). Adult trauma patients aged ≥20 years who were admitted to the hospital from 1 January 2009 to 31 December 2017, were included in this study. The rSIG, RTS, and SI were calculated according to the initial vital signs and GCS scores of patients upon arrival at the emergency department (ED). The end-point of primary outcome is in-hospital mortality. Discriminative power of each score to predict mortality was measured using area under the curve (AUC) by plotting the receiver operating characteristic (ROC) curve for 18,750 adult trauma patients, comprising 2438 patients with isolated head injury (only head Abbreviated Injury Scale (AIS) ≥ 2) and 16,312 without head injury (head AIS ≤ 1). The predictive accuracy of rSIG was significantly lower than that of RTS in all trauma patients (AUC 0.83 vs. AUC 0.85, p = 0.02) and in patients with isolated head injury (AUC 0.82 vs. AUC 0.85, p = 0.02). For patients without head injury, no difference was observed in the predictive accuracy between rSIG and RTS (AUC 0.83 vs. AUC 0.83, p = 0.97). Based on the cutoff value of 14.0, the rSIG can predict the probability of dying in trauma patients without head injury with a sensitivity of 61.5% and specificity of 94.5%. The predictive accuracy of both rSIG and RTS is significantly poorer than that of TRISS, in all trauma patients (AUC 0.93) or in patients with (AUC 0.89) and without head injury (AUC 0.92). In addition, SI had the significantly worse predictive accuracy than all of the other three models in all trauma patients (AUC 0.57), and the patients with (AUC 0.53) or without (AUC 0.63) head injury. This study revealed that rSIG had a significantly higher predictive accuracy of mortality than SI in all of the studied population but a lower predictive accuracy of mortality than RTS in all adult trauma patients and in adult patients with isolated head injury. In addition, in the adult patients without head injury, rSIG had a similar performance as RTS to the predictive risk of mortality of the patients.

Bayesian averaging over decision tree models: An application for estimating uncertainty in trauma severity scoring

INTRODUCTION

For making reliable decisions, practitioners need to estimate uncertainties that exist in data and decision models. In this paper we analyse uncertainties of predicting survival probability for patients in trauma care. The existing prediction methodology employs logistic regression modelling of Trauma and Injury Severity Score (TRISS), which is based on theoretical assumptions. These assumptions limit the capability of TRISS methodology to provide accurate and reliable predictions.

METHODS

We adopt the methodology of Bayesian model averaging and show how this methodology can be applied to decision trees in order to provide practitioners with new insights into the uncertainty. The proposed method has been validated on a large set of 447,176 cases registered in the US National Trauma Data Bank in terms of discrimination ability evaluated with receiver operating characteristic (ROC) and precision-recall (PRC) curves.

RESULTS

Areas under curves were improved for ROC from 0.951 to 0.956 (p = 3.89 × 10^-18) and for PRC from 0.564 to 0.605 (p = 3.89 × 10^-18). The new model has significantly better calibration in terms of the Hosmer-Lemeshow Hˆ statistic, showing an improvement from 223.14 (the standard method) to 11.59 (p = 2.31 × 10^-18).

CONCLUSION

The proposed Bayesian method is capable of improving the accuracy and reliability of survival prediction. The new method has been made available for evaluation purposes as a web application.

The modified rapid emergency medicine score: A novel trauma triage tool to predict in-hospital mortality

BACKGROUND

Trauma systems currently rely on imperfect and subjective tools to prioritize responses and resources, thus there is a critical need to develop a more accurate trauma severity score. Our objective was to modify the Rapid Emergency Medicine (REMS) Score for the trauma population and test its accuracy as a predictor of in-hospital mortality when compared to other currently used scores, including the Revised Trauma Score (RTS), the Injury Severity Score (ISS), the "Mechanism, Glasgow Coma Scale, Age and Arterial Pressure" (MGAP) score, and the Shock Index (SI) score.

METHODS

The two-part study design involved both a modification step and a validation step. The first step incorporated a retrospective analysis of a local trauma database (3680 patients) where three components of REMS were modified to more accurately represent the trauma population. Using clinical judgment and goodness-of-fit tests, systolic blood pressure was substituted for mean arterial pressure, the weighting of age was reduced, and the weighting of Glasgow Coma Scale was increased. The second part comprised validating the new modified REMS (mREMS) score retrospectively on a U.S. National Trauma Databank (NTDB) that included 429,711 patients admitted with trauma in 2012. The discriminate power of mREMS was compared to other trauma scores using the area under the receiver operating characteristic (AUC) curve.

RESULTS

Overall the mREMS score with an AUC of 0.967 (95% CI: 0.963-0.971) was demonstrated to be higher than RTS (AUC 0.959 [95% CI: 0.955-0.964]), ISS (AUC 0.780 [95% CI 0.770-0.791]), MGAP (AUC 0.964 [95% CI: 0.959-0.968]), and SI (AUC 0.670 [95% CI: 0.650-0.690]) in predicting in-hospital mortality on the NTDB.

CONCLUSION

In the trauma population, mREMS is an accurate predictor of in-hospital mortality, outperforming other used scores. Simple and objective, mREMS may hold value in the pre-hospital and emergency department setting in order to guide trauma team responses.

Validation of international trauma scoring systems in urban trauma centres in India

INTRODUCTION

In the Lower-Middle Income Country setting, we validate trauma severity scoring systems, namely Injury Severity Score (ISS), New Injury Severity Scale (NISS) score, the Kampala Trauma Score (KTS), Revised Trauma Score (RTS) score and the TRauma Injury Severity Score (TRISS) using Indian trauma patients.

PATIENTS AND METHODS

From 1 September 2013 to 28 February 2015, we conducted a prospective multi-centre observational cohort study of trauma patients in four Indian university hospitals, in three megacities, Kolkata, Mumbai and Delhi. All adult patients presenting to the casualty department with a history of injury and who were admitted to inpatient care were included. The primary outcome was in-hospital mortality within 30-days of admission. The sensitivity and specificity of each score to predict inpatient mortality within 30days was assessed by the areas under the receiver operating characteristic curve (AUC). Model fit for the performance of individual scoring systems was accomplished by using the Akaike Information criterion (AIC).

RESULTS

In a registry of 8791 adult trauma patients, we had a cohort of 7197 patients eligible for the study. 4091 (56.8%)patients had all five scores available and was the sample for a complete case analysis. Over a 30-day period, the scores (AUC) was TRISS (0.82), RTS (0.81), KTS (0.74), NISS (0.65) and ISS (0.62). RTS was the most parsimonious model with the lowest AIC score. Considering overall mortality, both physiologic scores (RTS, KTS) had better discrimination and goodness-of-fit than ISS or NISS. The ability of all Injury scores to predict early mortality (24h) was better than late mortality (30day).

CONCLUSION

On-admission physiological scores outperformed the more expensive anatomy-based ISS and NISS. The retrospective nature of ISS and TRISS score calculations and incomplete imaging in LMICs precludes its use in the casualty department of LMICs. They will remain useful for outcome comparison across trauma centres. Physiological scores like the RTS and KTS will be the practical score to use in casualty departments in the urban Indian setting, to predict early trauma mortality and improve triage.

The Applicability of Trauma and Injury Severity Score for a Blunt Trauma Population in Korea and a Proposal of New Models Using Score Predictors

PURPOSE

The purpose of this study was to verify the utility of existing Trauma and Injury Severity Score (TRISS) coefficients and to propose a new prediction model with a new set of TRISS coefficients or predictors.

MATERIALS AND METHODS

Of the blunt adult trauma patients who were admitted to our hospital in 2014, those eligible for Korea Trauma Data Bank entry were selected to collect the TRISS predictors. The study data were input into the TRISS formula to obtain "probability of survival" values, which were examined for consistency with actual patient survival status. For TRISS coefficients, Major Trauma Outcome Study-derived values revised in 1995 and National Trauma Data Bank-derived and National Sample Project-derived coefficients revised in 2009 were used. Additionally, using a logistic regression method, a new set of coefficients was derived from our medical center's database. Areas under the receiver operating characteristic (ROC) curve (AUC) for each prediction ability were obtained, and a pairwise comparison of ROC curves was performed.

RESULTS

In the statistical analysis, the AUCs (0.879-0.899) for predicting outcomes were lower than those of other countries. However, by adjusting the TRISS score using a continuous variable rather than a code for age, we were able to achieve higher AUCs [0.913 (95% confidence interval, 0.899 to 0.926)].

CONCLUSION

These results support further studies that will allow a more accurate prediction of prognosis for trauma patients. Furthermore, Korean TRISS coefficients or a new prediction model suited for Korea needs to be developed using a sufficiently sized sample.

Mortality prediction using TRISS methodology in the Spanish ICU Trauma Registry (RETRAUCI)

OBJECTIVES

To validate Trauma and Injury Severity Score (TRISS) methodology as an auditing tool in the Spanish ICU Trauma Registry (RETRAUCI).

DESIGN

A prospective, multicenter registry evaluation was carried out.

SETTING

Thirteen Spanish Intensive Care Units (ICUs).

PATIENTS

Individuals with traumatic disease and available data admitted to the participating ICUs.

INTERVENTIONS

Predicted mortality using TRISS methodology was compared with that observed in the pilot phase of the RETRAUCI from November 2012 to January 2015. Discrimination was evaluated using receiver operating characteristic (ROC) curves and the corresponding areas under the curves (AUCs) (95% CI), with calibration using the Hosmer-Lemeshow (HL) goodness-of-fit test. A value of p<0.05 was considered significant.

MAIN VARIABLES OF INTEREST

Predicted and observed mortality.

RESULTS

A total of 1405 patients were analyzed. The observed mortality rate was 18% (253 patients), while the predicted mortality rate was 16.9%. The area under the ROC curve was 0.889 (95% CI: 0.867-0.911). Patients with blunt trauma (n=1305) had an area under the ROC curve of 0.887 (95% CI: 0.864-0.910), and those with penetrating trauma (n=100) presented an area under the curve of 0.919 (95% CI: 0.859-0.979). In the global sample, the HL test yielded a value of 25.38 (p=0.001): 27.35 (p<0.0001) in blunt trauma and 5.91 (p=0.658) in penetrating trauma. TRISS methodology underestimated mortality in patients with low predicted mortality and overestimated mortality in patients with high predicted mortality.

CONCLUSIONS

TRISS methodology in the evaluation of severe trauma in Spanish ICUs showed good discrimination, with inadequate calibration - particularly in blunt trauma.

Incorporating Recent Advances To Make the TRISS Approach Universally Available

BACKGROUND

The Trauma and Injury Severity Score (TRISS), used to garner predictions of survival from the Injury Severity Score (ISS), the Revised Trauma Score (RTS, for physiologic reserve), and age is difficult for many trauma facilities to compute because it requires 8 to 10 variables and ISS depends on the specialized Abbreviated Injury Scale (AIS) scale rather than the International Classification of Diseases scale (ICD-9). It has been shown that metrics describing a patient's worst injury (WORSTSRR) are a powerful predictor of survival (regardless of coding type, AIS versus ICD-9) and that the Glasgow Coma Scale (GCS) motor component contains the majority of the information found in the full GCS score. This study hypothesized that the TRISS approach could be made more predictive and efficient with fewer variables by incorporating these advances.

METHODS

A total of 310,958 patients with nonmissing TRISS variables were subset from the National Trauma Data Bank (NTDB). Logistic regression was used to model mortality as a function of anatomic, physiologic and age variables. A traditional TRISS model was computed (with NTDB-derived coefficients) that uses ISS, RTS, age index, and mechanism to predict survival. Four smaller three- or four-variable models employed the ICD-9 WORSTSRR, the GCS motor component, and age (both continuously and dichotomously). Two of the four models also use mechanism. These models were compared using the concordance index (c-index, a measure of model discrimination) and the pseudo-R statistic (estimates proportion of variance explained).

RESULTS

Each experimental model (two models with 3 variables and two models with 4 variables) have superior discrimination and explain more variance than the traditional TRISS model that employs 8-10 variables.

CONCLUSIONS

Recent advances in anatomic and physiologic scoring markedly simplify TRISS-type models at no cost to prediction. This approach uses routinely available data, requires up to seven fewer terms, and predicts at least as well as the original TRISS. These findings could increase the availability of accurate trauma scoring tools to smaller trauma facilities.