Mortality prediction models in the general trauma population: A systematic review

A Prospective Multicenter Comparison of Trauma and Injury Severity Score, American Society of Anesthesiologists Physical Status, and National Surgical Quality Improvement Program Calculator's Ability to Predict Operative Trauma Outcomes

BACKGROUND

Trauma outcome prediction models have traditionally relied upon patient injury and physiologic data (eg, Trauma and Injury Severity Score [TRISS]) without accounting for comorbidities. We sought to prospectively evaluate the role of the American Society of Anesthesiologists physical status (ASA-PS) score and the National Surgical Quality Improvement Program Surgical Risk-Calculator (NSQIP-SRC), which are measurements of comorbidities, in the prediction of trauma outcomes, hypothesizing that they will improve the predictive ability for mortality, hospital length of stay (LOS), and complications compared to TRISS alone in trauma patients undergoing surgery within 24 hours.

METHODS

A prospective, observational multicenter study (9/2018-2/2020) of trauma patients ≥18 years undergoing operation within 24 hours of admission was performed. Multiple logistic regression was used to create models predicting mortality utilizing the variables within TRISS, ASA-PS, and NSQIP-SRC, respectively. Linear regression was used to create models predicting LOS and negative binomial regression to create models predicting complications.

RESULTS

From 4 level I trauma centers, 1213 patients were included. The Brier Score for each model predicting mortality was found to improve accuracy in the following order: 0.0370 for ASA-PS, 0.0355 for NSQIP-SRC, 0.0301 for TRISS, 0.0291 for TRISS+ASA-PS, and 0.0234 for TRISS+NSQIP-SRC. However, when comparing TRISS alone to TRISS+ASA-PS ( P = .082) and TRISS+NSQIP-SRC ( P = .394), there was no significant improvement in mortality prediction. NSQIP-SRC more accurately predicted both LOS and complications compared to TRISS and ASA-PS.

CONCLUSIONS

TRISS predicts mortality better than ASA-PS and NSQIP-SRC in trauma patients undergoing surgery within 24 hours. The TRISS mortality predictive ability is not improved when combined with ASA-PS or NSQIP-SRC. However, NSQIP-SRC was the most accurate predictor of LOS and complications.

Utilizing ultra-early continuous physiologic data to develop automated measures of clinical severity in a traumatic brain injury population

Determination of prognosis in the triage process after traumatic brain injury (TBI) is difficult to achieve. Current severity measures like the Trauma and injury severity score (TRISS) and revised trauma score (RTS) rely on additional information from the Glasgow Coma Scale (GCS) and the Injury Severity Score (ISS) which may be inaccurate or delayed, limiting their usefulness in the rapid triage setting. We hypothesized that machine learning based estimations of GCS and ISS obtained through modeling of continuous vital sign features could be used to rapidly derive an automated RTS and TRISS. We derived variables from electrocardiograms (ECG), photoplethysmography (PPG), and blood pressure using continuous data obtained in the first 15 min of admission to build machine learning models of GCS and ISS (ML-GCS and ML-ISS). We compared the TRISS and RTS using ML-ISS and ML-GCS and its value using the actual ISS and GCS in predicting in-hospital mortality. Models were tested in TBI with systemic injury (head abbreviated injury scale (AIS) ≥ 1), and isolated TBI (head AIS ≥ 1 and other AIS ≤ 1). The area under the receiver operating characteristic curve (AUROC) was used to evaluate model performance. A total of 21,077 cases (2009-2015) were in the training set. 6057 cases from 2016 to 2017 were used for testing, with 472 (7.8%) severe TBI (GCS 3-8), 223 (3.7%) moderate TBI (GCS 9-12), and 5913 (88.5%) mild TBI (GCS 13-15). In the TBI with systemic injury group, ML-TRISS had similar AUROC (0.963) to TRISS (0.965) in predicting mortality. ML-RTS had AUROC (0.823) and RTS had AUROC 0.928. In the isolated TBI group, ML-TRISS had AUROC 0.977, and TRISS had AUROC 0.983. ML-RTS had AUROC 0.790 and RTS had AUROC 0.957. Estimation of ISS and GCS from machine learning based modeling of vital sign features can be utilized to provide accurate assessments of the RTS and TRISS in a population of TBI patients. Automation of these scores could be utilized to enhance triage and resource allocation during the ultra-early phase of resuscitation.

Investigation of statistical methods used in prognostic prediction models for obstetric care: A 10 year-span cross-sectional study

INTRODUCTION

Obstetric care is a highly active area in the development and application of prognostic prediction models. The development and validation of these models often require the utilization of advanced statistical techniques. However, failure to adhere to rigorous methodological standards could greatly undermine the reliability and trustworthiness of the resultant models. Consequently, the aim of our study was to examine the current statistical practices employed in obstetric care and offer recommendations to enhance the utilization of statistical methods in the development of prognostic prediction models.

MATERIAL AND METHODS

We conducted a cross-sectional survey using a sample of studies developing or validating prognostic prediction models for obstetric care published in a 10-year span (2011-2020). A structured questionnaire was developed to investigate the statistical issues in five domains, including model derivation (predictor selection and algorithm development), model validation (internal and external), model performance, model presentation, and risk threshold setting. On the ground of survey results and existing guidelines, a list of recommendations for statistical methods in prognostic models was developed.

RESULTS

A total of 112 eligible studies were included, with 107 reporting model development and five exclusively reporting external validation. During model development, 58.9% of the studies did not include any form of validation. Of these, 46.4% used stepwise regression in a crude manner for predictor selection, while two-thirds made decisions on retaining or dropping candidate predictors solely based on p-values. Additionally, 26.2% transformed continuous predictors into categorical variables, and 80.4% did not consider nonlinear relationships between predictors and outcomes. Surprisingly, 94.4% of the studies did not examine the correlation between predictors. Moreover, 47.1% of the studies did not compare population characteristics between the development and external validation datasets, and only one-fifth evaluated both discrimination and calibration. Furthermore, 53.6% of the studies did not clearly present the model, and less than half established a risk threshold to define risk categories. In light of these findings, 10 recommendations were formulated to promote the appropriate use of statistical methods.

CONCLUSIONS

The use of statistical methods is not yet optimal. Ten recommendations were offered to assist the statistical methods of prognostic prediction models in obstetric care.

Barriers and facilitators associated with the upscaling of the Transmural Trauma Care Model: a qualitative study

BACKGROUND

To assess the barriers and facilitators associated with upscaling the Transmural Trauma Care Model (TTCM), a multidisciplinary and patient-centred transmural rehabilitation care model.

METHODS

Semi-structured interviews were conducted with eight trauma surgeons, eight hospital-based physiotherapists, eight trauma patients, and eight primary care physiotherapists who were part of a trauma rehabilitation network. Audio recordings of the interviews were made and transcribed verbatim. Data were analysed using a framework method based on the "constellation approach". Identified barriers and facilitators were grouped into categories related to structure, culture, and practice.

RESULTS

Various barriers and facilitators to upscaling were identified. Under structure, barriers and facilitators belonged to one of five themes: "financial structure", "communication structure", "physical structures and resources", "rules and regulations", and "organisation of the network". Under culture, the five themes were "commitment", "job satisfaction", "acting as a team", "quality and efficiency of care", and "patients' experience". Under practice, the two themes were "practical issues at the outpatient clinic" and "knowledge gained".

CONCLUSION

The success of upscaling the TTCM differed across hospitals and settings. The most important prerequisites for successfully upscaling the TTCM were adequate financial support and presence of "key actors" within an organisation who felt a sense of urgency for change and/or expected the intervention to increase their job satisfaction.

TRIAL REGISTRATION

NL8163 The Netherlands National Trial Register, date of registration 16-11-2019.

Assessing optimal methods for transferring machine learning models to low-volume and imbalanced clinical datasets: experiences from predicting outcomes of Danish trauma patients

Introduction

Accurately predicting patient outcomes is crucial for improving healthcare delivery, but large-scale risk prediction models are often developed and tested on specific datasets where clinical parameters and outcomes may not fully reflect local clinical settings. Where this is the case, whether to opt for de-novo training of prediction models on local datasets, direct porting of externally trained models, or a transfer learning approach is not well studied, and constitutes the focus of this study. Using the clinical challenge of predicting mortality and hospital length of stay on a Danish trauma dataset, we hypothesized that a transfer learning approach of models trained on large external datasets would provide optimal prediction results compared to de-novo training on sparse but local datasets or directly porting externally trained models.

Methods

Using an external dataset of trauma patients from the US Trauma Quality Improvement Program (TQIP) and a local dataset aggregated from the Danish Trauma Database (DTD) enriched with Electronic Health Record data, we tested a range of model-level approaches focused on predicting trauma mortality and hospital length of stay on DTD data. Modeling approaches included de-novo training of models on DTD data, direct porting of models trained on TQIP data to the DTD, and a transfer learning approach by training a model on TQIP data with subsequent transfer and retraining on DTD data. Furthermore, data-level approaches, including mixed dataset training and methods countering imbalanced outcomes (e.g., low mortality rates), were also tested.

Results

Using a neural network trained on a mixed dataset consisting of a subset of TQIP and DTD, with class weighting and transfer learning (retraining on DTD), we achieved excellent results in predicting mortality, with a ROC-AUC of 0.988 and an F2-score of 0.866. The best-performing models for predicting long-term hospitalization were trained only on local data, achieving an ROC-AUC of 0.890 and an F1-score of 0.897, although only marginally better than alternative approaches.

Conclusion

Our results suggest that when assessing the optimal modeling approach, it is important to have domain knowledge of how incidence rates and workflows compare between hospital systems and datasets where models are trained. Including data from other health-care systems is particularly beneficial when outcomes are suffering from class imbalance and low incidence. Scenarios where outcomes are not directly comparable are best addressed through either de-novo local training or a transfer learning approach.

In-hospital complications after trauma due to road traffic accidents

PURPOSE

The study aimed to verify the association between in-hospital complications and characterization and clinical variables including hospital care and trauma severity.

METHODS

This analysis with the prospective cohort data was conducted at a municipal hospital in São Paulo, Brazil, and included participants aged 14 years or older, with traumatic injuries from traffic accidents. Data was collected from January 2015 to July 2016 and included demographics variables, type of traumatic event, clinical parameters, length of stay in the Emergency department and in the Intensive Care Unit, length of hospital stay, survival probability, trauma severity and mortality.

RESULTS

Of the 327 patients, 25.1% had in-hospital complications and their occurrence was statistically associated with higher mean age, run-overs and higher trauma severity. The length of stay in the emergency room, hospital stay, ICU stay, percentage of deaths, and hospital readmission were higher in patients with complications. The number of complications was correlated with trauma severity, ICU stay, and mortality.

CONCLUSION

Complications were associated with older age, run-overs, greater trauma severity, length of stay and readmission after hospital discharge.

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.

Hospital mortality prediction in traumatic injuries patients: comparing different SMOTE-based machine learning algorithms

BACKGROUND

Trauma is one of the most critical public health issues worldwide, leading to death and disability and influencing all age groups. Therefore, there is great interest in models for predicting mortality in trauma patients admitted to the ICU. The main objective of the present study is to develop and evaluate SMOTE-based machine-learning tools for predicting hospital mortality in trauma patients with imbalanced data.

METHODS

This retrospective cohort study was conducted on 126 trauma patients admitted to an intensive care unit at Besat hospital in Hamadan Province, western Iran, from March 2020 to March 2021. Data were extracted from the medical information records of patients. According to the imbalanced property of the data, SMOTE techniques, namely SMOTE, Borderline-SMOTE1, Borderline-SMOTE2, SMOTE-NC, and SVM-SMOTE, were used for primary preprocessing. Then, the Decision Tree (DT), Random Forest (RF), Naive Bayes (NB), Artificial Neural Network (ANN), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost) methods were used to predict patients' hospital mortality with traumatic injuries. The performance of the methods used was evaluated by sensitivity, specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), accuracy, Area Under the Curve (AUC), Geometric Mean (G-means), F1 score, and P-value of McNemar's test.

RESULTS

Of the 126 patients admitted to an ICU, 117 (92.9%) survived and 9 (7.1%) died. The mean follow-up time from the date of trauma to the date of outcome was 3.98 ± 4.65 days. The performance of ML algorithms is not good with imbalanced data, whereas the performance of SMOTE-based ML algorithms is significantly improved. The mean area under the ROC curve (AUC) of all SMOTE-based models was more than 91%. F1-score and G-means before balancing the dataset were below 70% for all ML models except ANN. In contrast, F1-score and G-means for the balanced datasets reached more than 90% for all SMOTE-based models. Among all SMOTE-based ML methods, RF and ANN based on SMOTE and XGBoost based on SMOTE-NC achieved the highest value for all evaluation criteria.

CONCLUSIONS

This study has shown that SMOTE-based ML algorithms better predict outcomes in traumatic injuries than ML algorithms. They have the potential to assist ICU physicians in making clinical decisions.

Intercontinental validation of a clinical prediction model for predicting 90-day and 2-year mortality in an Israeli cohort of 2033 patients with a femoral neck fracture aged 65 or above

PURPOSE

Mortality prediction in elderly femoral neck fracture patients is valuable in treatment decision-making. A previously developed and internally validated clinical prediction model shows promise in identifying patients at risk of 90-day and 2-year mortality. Validation in an independent cohort is required to assess the generalizability; especially in geographically distinct regions. Therefore we questioned, is the SORG Orthopaedic Research Group (SORG) femoral neck fracture mortality algorithm externally valid in an Israeli cohort to predict 90-day and 2-year mortality?

METHODS

We previously developed a prediction model in 2022 for estimating the risk of mortality in femoral neck fracture patients using a multicenter institutional cohort of 2,478 patients from the USA. The model included the following input variables that are available on clinical admission: age, male gender, creatinine level, absolute neutrophil, hemoglobin level, international normalized ratio (INR), congestive heart failure (CHF), displaced fracture, hemiplegia, chronic obstructive pulmonary disease (COPD), history of cerebrovascular accident (CVA) and beta-blocker use. To assess the generalizability, we used an intercontinental institutional cohort from the Sheba Medical Center in Israel (level I trauma center), queried between June 2008 and February 2022. Generalizability of the model was assessed using discrimination, calibration, Brier score, and decision curve analysis.

RESULTS

The validation cohort included 2,033 patients, aged 65 years or above, that underwent femoral neck fracture surgery. Most patients were female 64.8% (n = 1317), the median age was 81 years (interquartile range = 75-86), and 80.4% (n = 1635) patients sustained a displaced fracture (Garden III/IV). The 90-day mortality was 9.4% (n = 190) and 2-year mortality was 30.0% (n = 610). Despite numerous baseline differences, the model performed acceptably to the validation cohort on discrimination (c-statistic 0.67 for 90-day, 0.67 for 2-year), calibration, Brier score, and decision curve analysis.

CONCLUSIONS

The previously developed SORG femoral neck fracture mortality algorithm demonstrated good performance in an independent intercontinental population. Current iteration should not be relied on for patient care, though suggesting potential utility in assessing patients at low risk for 90-day or 2-year mortality. Further studies should evaluate this tool in a prospective setting and evaluate its feasibility and efficacy in clinical practice. The algorithm can be freely accessed: https://sorg-apps.shinyapps.io/hipfracturemortality/ .

LEVEL OF EVIDENCE

Level III, Prognostic study.

Protocolized approach saves the limb in peripheral arterial injury: A decade experience

PURPOSE

Outcomes of peripheral arterial injury (PAI) depend on various factors, such as warm ischemia time and concomitant injuries. Suboptimal prehospital care may lead to delayed presentation, and a lack of dedicated trauma system may lead to poorer outcome. Also, there are few reports of these outcomes. The aim of this study was to review our experience of PAI management for more than a decade, and identify the predictors of limb loss in these patients.

METHODS

This is a retrospective analysis of prospectively maintained database of trauma admissions at a level I trauma center from January 2008 to December 2019. Patients with acute upper limb arterial injuries or lower limb arterial injuries at or above the level of popliteal artery were included. Association of limb loss with ischemia time, mechanism of injury and concomitant injuries was studied using multiple logistic regressions. Statistical analysis was performed using STATA version 15.0 (Stata Corp LLC, Texas).

RESULTS

Out of 716 patients with PAI, the majority (92%) were young males. Blunt trauma was the most common mechanism of injury. Median ischemia time was 4 h (interquartile range 2-7 h). Brachial artery (28%) was the most common injured vessel followed by popliteal artery (18%) and femoral artery (17%). Limb salvage rate was 78%. Out of them, 158 (22%) patients needed amputation, and 53 (7%) had undergone primary amputation. The majority (86%) of patients who required primary or secondary amputations had blunt trauma. On multivariate analysis, blunt trauma, ischemia time more than 6 h and concomitant venous, skeletal, and soft tissue injuries were associated with higher odds of amputation.

CONCLUSION

Over all limb salvage rates was 78% in our series. Blunt mechanism of injury and associated skeletal and soft tissue injury, ischemia time more than 6 h portend a poor prognosis. Injury prevention, robust prehospital care, and rapid referral to specialized trauma center are few efficient measures, which can decrease the morbidity associated with vascular injury.

PREHOSPITAL SHOCK INDEX MULTIPLIED BY AVPU SCALE AS A PREDICTOR OF CLINICAL OUTCOMES IN TRAUMATIC INJURY

ABSTRACT

Objectives: Many prehospital trauma triage scores have been proposed, but none has emerged as a criterion standard. Therefore, a rapid and accurate tool is necessary for field triage. The shock index (SI) multiplied by the AVPU (Alert, responds to Voice, responds to Pain, Unresponsive) score (SIAVPU) reflected the hemodynamic and neurological conditions through a combination of the SI and AVPU. This study aimed to investigate the prediction performance of SI multiplied by the AVPU and to compare the prediction performance of other prehospital trauma triage scores in a population with traumatic injury. Patients and Methods: This study included 6,156 patients with trauma injury from the Taipei Tzu Chi trauma database. We investigated the accuracy of four scoring systems in predicting mortality, intensive care unit (ICU) admission, and prolonged hospital stay (defined as a duration of hospitalization >14 days). In the subgroup analysis, we also analyzed the effects of age, injury mechanism and severity, underlying diseases, and traumatic brain injury. Results: The predictive accuracy of SIAVPU for mortality, ICU admission, and prolonged hospital stay was significantly higher than that of SI, modified SI, and SI multiplied by age in the traumatic injury population, with an area under the receiver operating characteristic curve of 0.738 for mortality, 0.641 for ICU admission, and 0.606 for prolonged hospital stay. In the subgroup analysis, the prediction accuracy of mortality, ICU admission, and prolonged hospital stay of SIAVPU was also better in patients with younger age, older age, major trauma (Injury Severity Score ≥16), motor vehicle collisions, fall injury, healthy, cardiovascular disease, mixed traumatic brain injury, and isolated traumatic brain injury. The best cutoff levels of SIAVPU score to predict mortality, ICU admission, and total length of stay ≥14 days in trauma injury patients were 0.90, 0.82, and 0.80, with accuracies of 88.56%, 79.84%, and 78.62%, respectively. Conclusions: In conclusion, SIAVPU is a rapid and accurate field triage score with better prediction accuracy for mortality, ICU admission, and prolonged hospital stay than SI, modified SI, and SI multiplied by age in patients with trauma. Patients with SIAVPU ≥0.9 should be considered for the highest-level trauma center available within the geographic constraints of regional trauma systems.

Shock index, modified shock index, age shock index score, and reverse shock index multiplied by Glasgow Coma Scale predicting clinical outcomes in traumatic brain injury: Evidence from a 10-year analysis in a single center

OBJECTIVES

Early identification of traumatic brain injury (TBI) patients at a high risk of mortality is very important. This study aimed to compare the predictive accuracy of four scoring systems in TBI, including shock index (SI), modified shock index (MSI), age-adjusted shock index (ASI), and reverse shock index multiplied by the Glasgow Coma Scale (rSIG).

PATIENTS AND METHODS

This is a retrospective analysis of a registry from the Taipei Tzu Chi trauma database. Totally, 1,791 patients with TBI were included. We investigated the accuracy of four major shock indices for TBI mortality. In the subgroup analysis, we also analyzed the effects of age, injury mechanism, underlying diseases, TBI severity, and injury severity.

RESULTS

The predictive accuracy of rSIG was significantly higher than those of SI, MSI, and ASI in all the patients [area under the receiver operating characteristic curve (AUROC), 0.710 vs. 0.495 vs. 0.527 vs. 0.598], especially in the moderate/severe TBI (AUROC, 0.625 vs. 0.450 vs. 0.476 vs. 0.529) and isolated head injury populations (AUROC 0.689 vs. 0.472 vs. 0.504 vs. 0.587). In the subgroup analysis, the prediction accuracy of mortality of rSIG was better in TBI with major trauma [Injury Severity Score (ISS) ≥ 16], motor vehicle collisions, fall injury, and healthy and cardiovascular disease population. rSIG also had a better prediction effect, as compared to SI, MSI, and ASI, both in the non-geriatric (age < 65 years) and geriatric (age ≥ 65 years).

CONCLUSION

rSIG had a better prediction accuracy for mortality in the overall TBI population than SI, MSI, and ASI. Although rSIG have better accuracy than other indices (ROC values indicate poor to moderate accuracy), the further clinical studies are necessary to validate our results.

A Prognostic Model for In-Hospital Mortality in Critically Ill Patients with Pneumonia

Purpose

To determine the utility of a novel serum biomarker for the outcome prediction of critically ill patients with pneumonia.

Patients and Methods

A retrospective analysis of critically ill patients was performed at an emergency department. The expression and prediction value of parameters were assessed. Binary logistic regression analysis was utilized to determine the indicators associated with in-hospital mortality of pneumonia patients. The Last Absolute Shrinkage and Selection Operator was used to further determine the independent predictors, which were validated by multiple logistic regression. The receiver operator characteristic curve was performed to assess their prediction values. A prognostic nomogram model was finally established for the outcome prediction for critically ill patients with pneumonia.

Results

Retinol-binding protein (RBP) was significantly reduced in non-survived and pneumonia patients. CURB-65 score, levels of RBP, and blood urea nitrogen (BUN) were associated with in-hospital mortality of critically ill patients with pneumonia. Their combination was determined to be an ideal prognostic predictor (area under the curve of 0.762) and further developed into a nomogram prediction model (c-index 0.764).

Conclusion

RBP is a novel in-hospital mortality predictor, which well supplements the CURB-65 score for critical pneumonia patients.

Clustering of trauma patients based on longitudinal data and the application of machine learning to predict recovery

Predicting recovery after trauma is important to provide patients a perspective on their estimated future health, to engage in shared decision making and target interventions to relevant patient groups. In the present study, several unsupervised techniques are employed to cluster patients based on longitudinal recovery profiles. Subsequently, these data-driven clusters were assessed on clinical validity by experts and used as targets in supervised machine learning models. We present a formalised analysis of the obtained clusters that incorporates evaluation of (i) statistical and machine learning metrics, (ii) clusters clinical validity with descriptive statistics and medical expertise. Clusters quality assessment revealed that clusters obtained through a Bayesian method (High Dimensional Supervised Classification and Clustering) and a Deep Gaussian Mixture model, in combination with oversampling and a Random Forest for supervised learning of the cluster assignments provided among the most clinically sensible partitioning of patients. Other methods that obtained higher classification accuracy suffered from cluster solutions with large majority classes or clinically less sensible classes. Models that used just physical or a mix of physical and psychological outcomes proved to be among the most sensible, suggesting that clustering on psychological outcomes alone yields recovery profiles that do not conform to known risk factors.

Inferring Tissue-Specific, TLR4-Dependent Type 17 Immune Interactions in Experimental Trauma/Hemorrhagic Shock and Resuscitation Using Computational Modeling

Trauma/hemorrhagic shock followed by resuscitation (T/HS-R) results in multi-system inflammation and organ dysfunction, in part driven by binding of damage-associated molecular pattern molecules to Toll-like Receptor 4 (TLR4). We carried out experimental T/HS-R (pseudo-fracture plus 2 h of shock followed by 0-22 h of resuscitation) in C57BL/6 (wild type [WT]) and TLR4-null (TLR4^-/-) mice, and then defined the dynamics of 20 protein-level inflammatory mediators in the heart, gut, lung, liver, spleen, kidney, and systemic circulation. Cross-correlation and Principal Component Analysis (PCA) on data from the 7 tissues sampled suggested that TLR4^-/- samples express multiple inflammatory mediators in a small subset of tissue compartments as compared to the WT samples, in which many inflammatory mediators were localized non-specifically to nearly all compartments. We and others have previously defined a central role for type 17 immune cells in human trauma. Accordingly, correlations between IL-17A and GM-CSF (indicative of pathogenic Th17 cells); between IL-17A and IL-10 (indicative of non-pathogenic Th17 cells); and IL-17A and TNF (indicative of memory/effector T cells) were assessed across all tissues studied. In both WT and TLR4^-/- mice, positive correlations were observed between IL-17A and GM-CSF, IL-10, and TNF in the kidney and gut. In contrast, the variable and dynamic presence of both pathogenic and non-pathogenic Th17 cells was inferred in the systemic circulation of TLR4^-/- mice over time, suggesting a role for TLR4 in efflux of these cells into peripheral tissues. Hypergraph analysis - used to define dynamic, cross compartment networks - in concert with PCA-suggested that IL-17A was present persistently in all tissues at all sampled time points except for its absence in the plasma at 0.5h in the WT group, supporting the hypothesis that T/HS-R induces efflux of Th17 cells from the circulation and into specific tissues. These analyses suggest a complex, context-specific role for TLR4 and type 17 immunity following T/HS-R.

Dynamic Bayesian network for predicting physiological changes, organ dysfunctions and mortality risk in critical trauma patients

BACKGROUND

Critical trauma patients are particularly prone to increased mortality risk; hence, an accurate prediction of their conditions enables early identification of patients' mortality status. Thus, we aimed to develop and validate a real-time prediction model for physiological changes, organ dysfunctions and mortality risk in critical trauma patients.

METHODS

We used Dynamic Bayesian Networks (DBNs) to model complicated relationships of physiological variables across time slices, accessing data of trauma patients from the Medical Information Mart for Intensive Care database (MIMIC-III) (n = 2915) and validated with patients' data from ICU admissions at the Changhai Hospital (ICU-CH) (n = 1909). The DBN model's evaluation included the predictive ability of physiological changes, organ dysfunctions and mortality risk.

RESULTS

Our DBN model included two static variables (age and sex) and 18 dynamic physiological variables. The differences in ratios between the real values and the 24- and 48-h predicted values of most physiological variables were within 5% in the two datasets. The accuracy of our DBN model for predicting renal, hepatic, cardiovascular and hematologic dysfunctions was more than 0.8.The calculated area under the curve (AUC) from receiver operating characteristic curves and 95% confidence interval for predicting the 24- and 48-h mortality risk were 0.977 (0.967-0.988) and 0.958 (0.945-0.971) in the MIMIC-III and 0.967 (0.947-0.987) and 0.946 (0.925-0.967) in ICU-CH.

CONCLUSIONS

A DBN is a promising method for predicting medical temporal data such as trauma patients' mortality risk, demonstrated by high AUC scores and validation by a real-life ICU scenario; thus, our DBN prediction model can be used as a real-time tool to predict physiological changes, organ dysfunctions and mortality risk during ICU admissions.

Validation of the Trauma and Injury Severity Score for Prediction of Mortality in a Greek Trauma Population

BACKGROUND

Although the Trauma and Injury Severity Score (TRISS) has been extensively used for mortality risk adjustment in trauma, its applicability in contemporary trauma populations is increasingly questioned.

OBJECTIVE

The study aimed to evaluate the predictive performance of the TRISS in its original and revised version and compare these with a recalibrated version, including current data from a Greek trauma population.

METHODS

This is a retrospective cohort study of admitted trauma patients conducted in two tertiary Greek hospitals from January 2016 to December 2018. The model algorithm was calculated based on the Major Trauma Outcome Study coefficients (TRISSMTOS), the National Trauma Data Bank coefficients (TRISSNTDB), and reweighted coefficients of logistic regression obtained from a Greek trauma dataset (TRISSGrTD). The primary endpoint was inhospital mortality. Models' prediction was performed using discrimination and calibration statistics.

RESULTS

A total of 8,988 trauma patients were included, of whom 854 died (9.5%). The TRISSMTOS displayed excellent discrimination with an area under the curve (AUC) of 0.912 (95% CI 0.902-0.923) and comparable with TRISSNTDB (AUC = 0.908, 95% CI 0.897-0.919, p = .1195). Calibration of both models was poor (Hosmer-Lemeshow test p < .001), tending to underestimate the probability of mortality across almost all risk groups. The TRISSGrTD resulted in statistically significant improvement in discrimination (AUC = 0.927, 95% CI 0.918-0.936, p < .0001) and acceptable calibration (Hosmer-Lemeshow test p = .113).

CONCLUSION

In this cohort of Greek trauma patients, the performance of the original TRISS was suboptimal, and there was no evidence that it has benefited from its latest revision. By contrast, a strong case exists for supporting a locally recalibrated version to render the TRISS applicable for mortality prediction and performance benchmarking.

External validation of International Classification of Injury Severity Score to predict mortality in a Greek adult trauma population

INTRODUCTION

The International Classification of diseases- based Injury Severity Score (ICISS) obtained by empirically derived diagnosis-specific survival probabilities (DSPs) is the best-known risk-adjustment measure to predict mortality. Recently, a new set of pooled DSPs has been proposed by the International Collaborative Effort on Injury Statistics but it remains to be externally validated in other cohorts. The aim of this study was to externally validate the ICISS using international DSPs and compare its prognostic performance with local DSPs derived from Greek adult trauma population.

MATERIALS AND METHODS

This retrospective single-center cohort study enrolled adult trauma patients (≥ 16 years) hospitalized between January 2015 and December 2019 and temporally divided into derivation (n = 21,614) and validation cohorts (n = 14,889). Two different ICISS values were calculated for each patient using two different sets of DSPs: international (ICISSint) and local (ICISSgr). The primary outcome was in-hospital mortality. Models' prediction was performed using discrimination and calibration statistics.

RESULTS

ICISSint displayed good discrimination in derivation (AUC = 0.836 CI 95% 0.819-0.852) and validation cohort (AUC = 0.817 CI 95% 0.797-0.836). Calibration using visual analysis showed accurate prediction at patients with low mortality risk, especially below 30%. ICISSgr yielded better discrimination (AUC = 0.834 CI 95% 0.814-0.854 vs 0.817 CI 95% 0.797-0.836, p ˂ .05) and marginally improved overall accuracy (Brier score = 0.0216 vs 0.0223) compared with the ICISSint in the validation cohort. Incorporation of age and sex in both models enhanced further their performance as reflected by superior discrimination (p ˂ .05) and closer calibration curve to the identity line in the validation cohort.

CONCLUSION

This study supports the use of international DSPs for the ICISS to predict mortality in contemporary trauma patients and provides evidence regarding the potential benefit of applying local DSPs. Further research is warranted to confirm our findings and recommend the widespread use of ICISS as a valid measure that is easily obtained from administrative data based on ICD-10 codes.

Electronic health record machine learning model predicts trauma inpatient mortality in real time: A validation study

INTRODUCTION

Patient outcome prediction models are underused in clinical practice because of lack of integration with real-time patient data. The electronic health record (EHR) has the ability to use machine learning (ML) to develop predictive models. While an EHR ML model has been developed to predict clinical deterioration, it has yet to be validated for use in trauma. We hypothesized that the Epic Deterioration Index (EDI) would predict mortality and unplanned intensive care unit (ICU) admission in trauma patients.

METHODS

A retrospective analysis of a trauma registry was used to identify patients admitted to a level 1 trauma center for >24 hours from October 2019 to July 2020. We evaluated the performance of the EDI, which is constructed from 125 objective patient measures within the EHR, in predicting mortality and unplanned ICU admissions. We performed a 5 to 1 match on age because it is a major component of EDI, then examined the area under the receiver operating characteristic curve (AUROC), and benchmarked it against Injury Severity Score (ISS) and new injury severity score (NISS).

RESULTS

The study cohort consisted of 1,325 patients admitted with a mean age of 52.5 years and 91% following blunt injury. The in-hospital mortality rate was 2%, and unplanned ICU admission rate was 2.6%. In predicting mortality, the maximum EDI within 24 hours of admission had an AUROC of 0.98 compared with 0.89 of ISS and 0.91 of NISS. For unplanned ICU admission, the EDI slope within 24 hours of ICU admission had a modest performance with an AUROC of 0.66.

CONCLUSION

Epic Deterioration Index appears to perform strongly in predicting in-patient mortality similarly to ISS and NISS. In addition, it can be used to predict unplanned ICU admissions. This study helps validate the use of this real-time EHR ML-based tool, suggesting that EDI should be incorporated into the daily care of trauma patients.

LEVEL OF EVIDENCE

Prognostic, level III.

Development of a new score for early mortality prediction in trauma ICU patients: RETRASCORE

Background

Severity scores are commonly used for outcome adjustment and benchmarking of trauma care provided. No specific models performed only with critically ill patients are available. Our objective was to develop a new score for early mortality prediction in trauma ICU patients.

Methods

This is a retrospective study using the Spanish Trauma ICU registry (RETRAUCI) 2015–2019. Patients were divided and analysed into the derivation (2015–2017) and validation sets (2018–2019). We used as candidate variables to be associated with mortality those available in RETRAUCI that could be collected in the first 24 h after ICU admission. Using logistic regression methodology, a simple score (RETRASCORE) was created with points assigned to each selected variable. The performance of the model was carried out according to global measures, discrimination and calibration.

Results

The analysis included 9465 patients: derivation set 5976 and validation set 3489. Thirty-day mortality was 12.2%. The predicted probability of 30-day mortality was determined by the following equation: 1/(1 + exp (− y)), where y = 0.598 (Age 50–65) + 1.239 (Age 66–75) + 2.198 (Age > 75) + 0.349 (PRECOAG) + 0.336 (Pre-hospital intubation) + 0.662 (High-risk mechanism) + 0.950 (unilateral mydriasis) + 3.217 (bilateral mydriasis) + 0.841 (Glasgow ≤ 8) + 0.495 (MAIS-Head) − 0.271 (MAIS-Thorax) + 1.148 (Haemodynamic failure) + 0.708 (Respiratory failure) + 0.567 (Coagulopathy) + 0.580 (Mechanical ventilation) + 0.452 (Massive haemorrhage) − 5.432. The AUROC was 0.913 (0.903–0.923) in the derivation set and 0.929 (0.918–0.940) in the validation set.

Conclusions

The newly developed RETRASCORE is an early, easy-to-calculate and specific score to predict in-hospital mortality in trauma ICU patients. Although it has achieved adequate internal validation, it must be externally validated.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-021-03845-6.

Comparison of Machine Learning Techniques for Mortality Prediction in a Prospective Cohort of Older Adults

As global demographics change, ageing is a global phenomenon which is increasingly of interest in our modern and rapidly changing society. Thus, the application of proper prognostic indices in clinical decisions regarding mortality prediction has assumed a significant importance for personalized risk management (i.e., identifying patients who are at high or low risk of death) and to help ensure effective healthcare services to patients. Consequently, prognostic modelling expressed as all-cause mortality prediction is an important step for effective patient management. Machine learning has the potential to transform prognostic modelling. In this paper, results on the development of machine learning models for all-cause mortality prediction in a cohort of healthy older adults are reported. The models are based on features covering anthropometric variables, physical and lab examinations, questionnaires, and lifestyles, as well as wearable data collected in free-living settings, obtained for the "Healthy Ageing Initiative" study conducted on 2291 recruited participants. Several machine learning techniques including feature engineering, feature selection, data augmentation and resampling were investigated for this purpose. A detailed empirical comparison of the impact of the different techniques is presented and discussed. The achieved performances were also compared with a standard epidemiological model. This investigation showed that, for the dataset under consideration, the best results were achieved with Random UnderSampling in conjunction with Random Forest (either with or without probability calibration). However, while including probability calibration slightly reduced the average performance, it increased the model robustness, as indicated by the lower 95% confidence intervals. The analysis showed that machine learning models could provide comparable results to standard epidemiological models while being completely data-driven and disease-agnostic, thus demonstrating the opportunity for building machine learning models on health records data for research and clinical practice. However, further testing is required to significantly improve the model performance and its robustness.

Injured patients who would benefit from expedited major trauma centre care: a consensus-based definition for the United Kingdom

INTRODUCTION

Despite the importance of treating the 'right patient in the right place at the right time', there is no gold standard for defining which patients should receive expedited major trauma centre (MTC) care. This study aimed to define a reference standard applicable to the United Kingdom (UK) National Health Service major trauma networks.

METHODS

A one-day facilitated roundtable expert consensus meeting was conducted at the University of Sheffield, UK, in September 2019. An expert panel of 17 clinicians was purposively sampled, representing all specialities relevant to major trauma management. A consultation process was subsequently held using focus groups with Public and Patient Involvement (PPI) representatives to review and confirm the proposed reference standard.

RESULTS

Four reference standard domains were identified, comprising: need for critical interventions; presence of significant individual anatomical injuries; burden of multiple minor injuries; and important patient attributes. Specific criteria were defined for each domain. PPI consultation confirmed all aspects of the reference standard. A coding algorithm to allow operationalisation in Trauma Audit and Research Network data was also formulated, allowing classification of any case submitted to their database for future research.

CONCLUSIONS

This reference standard defines which patients would benefit from expedited MTC care. It could be used as the target for future pre-hospital injury triage tools, for setting best practice tariffs for trauma care reimbursement and to evaluate trauma network performance. Future research is recommended to compare patient characteristics, management and outcomes of the proposed definition with previously established reference standards.

Preventable death in trauma: A systematic review on definition and classification

PURPOSE

Trauma-related preventable death (TRPD) has been used to assess the management and quality of trauma care worldwide. However, due to differences in terminology and application, the definition of TRPD lacks validity. The aim of this systematic review is to present an overview of current literature and establish a designated definition of TRPD to improve the assessment of quality of trauma care.

METHODS

A search was conducted in PubMed, Embase, the Cochrane Library and the Web of Science Core Collection. Including studies regarding TRPD, published between January 1, 1990, and April 6, 2021. Studies were assessed on the use of a definition of TRPD, injury severity scoring tool and panel review.

RESULTS

In total, 3,614 articles were identified, 68 were selected for analysis. The definition of TRPD was divided in four categories: I. Clinical definition based on panel review or expert opinion (TRPD, trauma-related potentially preventable death, trauma-related non-preventable death), II. An algorithm (injury severity score (ISS), trauma and injury severity score (TRISS), probability of survival (Ps)), III. Clinical definition completed with an algorithm, IV. Other. Almost 85% of the articles used a clinical definition in some extend; solely clinical up to an additional algorithm. A total of 27 studies used injury severity scoring tools of which the ISS and TRISS were the most frequently reported algorithms. Over 77% of the panels included trauma surgeons, 90% included other specialist; 61% emergency medicine physicians, 46% forensic pathologists and 43% nurses.

CONCLUSION

The definition of TRPD is not unambiguous in literature and should be based on a clinical definition completed with a trauma prediction algorithm such as the TRISS. TRPD panels should include a trauma surgeon, anesthesiologist, emergency physician, neurologist, and forensic pathologist.

Association of Platelets and White Blood Cells Subtypes with Trauma Patients' Mortality Outcome in the Intensive Care Unit

Background: White blood cell (WBC) subtypes have been suggested to reflect patients’ immune-inflammatory status. Furthermore, the derived ratio of platelets and WBC subtypes, including monocyte-to-lymphocyte ratio (MLR), neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR), is proposed to be associated with patient outcome. Therefore, this study aimed to identify the association of platelets and white blood cells subtypes with the mortality outcome of trauma patients in the intensive care unit (ICU). Method: The medical information from 2854 adult trauma patients admitted to the ICU between 1 January 2009 and 31 December 2019 were retrospectively retrieved from the Trauma Registry System and classified into two groups: the survivors group (n = 2524) and the death group (n = 330). The levels of monocytes, neutrophils, lymphocytes, platelets, and blood-drawn laboratory data detected upon patient arrival to the emergency room and the derived MLR, NLR, and PLR were calculated. Multivariate logistic regression analysis was used to determine the independent effects of univariate predictive variables on mortality occurrence. Result: The results revealed the patients who died had significantly lower platelet counts (175,842 ± 61,713 vs. 206,890 ± 69,006/μL, p < 0.001) but higher levels of lymphocytes (2458 ± 1940 vs. 1971 ± 1453/μL, p < 0.001) than the surviving patients. However, monocyte and neutrophil levels were not significantly different between the death and survivor groups. Moreover, dead patients had a significantly lower PLR than survivors (124.3 ± 110.3 vs. 150.6 ± 106.5, p < 0.001). However, there was no significant difference in MLR or NLR between the dead patients and the survivors. Multivariate logistic regression revealed that male gender, old age, pre-existing hypertension, coronary artery disease and end-stage renal disease, lower Glasgow Coma Scale (GCS), higher Injury Severity Score (ISS), higher level of lymphocytes and lower level of red blood cells and platelets, longer activated partial thromboplastin time (aPTT), and lower level of PLR were independent risk factors associated with higher odds of trauma patient mortality outcome in the ICU. Conclusion: This study revealed that a higher lymphocyte count, lower platelet count, and a lower PLR were associated with higher risk of death in ICU trauma patients.

The advanced machine learner XGBoost did not reduce prehospital trauma mistriage compared with logistic regression: a simulation study

Background

Accurate prehospital trauma triage is crucial for identifying critically injured patients and determining the level of care. In the prehospital setting, time and data are often scarce, limiting the complexity of triage models. The aim of this study was to assess whether, compared with logistic regression, the advanced machine learner XGBoost (eXtreme Gradient Boosting) is associated with reduced prehospital trauma mistriage.

Methods

We conducted a simulation study based on data from the US National Trauma Data Bank (NTDB) and the Swedish Trauma Registry (SweTrau). We used categorized systolic blood pressure, respiratory rate, Glasgow Coma Scale and age as our predictors. The outcome was the difference in under- and overtriage rates between the models for different training dataset sizes.

Results

We used data from 813,567 patients in the NTDB and 30,577 patients in SweTrau. In SweTrau, the smallest training set of 10 events per free parameter was sufficient for model development. XGBoost achieved undertriage rates in the range of 0.314–0.324 with corresponding overtriage rates of 0.319–0.322. Logistic regression achieved undertriage rates ranging from 0.312 to 0.321 with associated overtriage rates ranging from 0.321 to 0.323. In NTDB, XGBoost required the largest training set size of 1000 events per free parameter to achieve robust results, whereas logistic regression achieved stable performance from a training set size of 25 events per free parameter. For the training set size of 1000 events per free parameter, XGBoost obtained an undertriage rate of 0.406 with an overtriage of 0.463. For logistic regression, the corresponding undertriage was 0.395 with an overtriage of 0.468.

Conclusion

The under- and overtriage rates associated with the advanced machine learner XGBoost were similar to the rates associated with logistic regression regardless of sample size, but XGBoost required larger training sets to obtain robust results. We do not recommend using XGBoost over logistic regression in this context when predictors are few and categorical.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12911-021-01558-y.

Development and validation of a scoring system for mortality prediction and application of standardized W statistics to assess the performance of emergency departments

Background

In-hospital mortality and short-term mortality are indicators that are commonly used to evaluate the outcome of emergency department (ED) treatment. Although several scoring systems and machine learning-based approaches have been suggested to grade the severity of the condition of ED patients, methods for comparing severity-adjusted mortality in general ED patients between different systems have yet to be developed. The aim of the present study was to develop a scoring system to predict mortality in ED patients using data collected at the initial evaluation and to validate the usefulness of the scoring system for comparing severity-adjusted mortality between institutions with different severity distributions.

Methods

The study was based on the registry of the National Emergency Department Information System, which is maintained by the National Emergency Medical Center of the Republic of Korea. Data from 2016 were used to construct the prediction model, and data from 2017 were used for validation. Logistic regression was used to build the mortality prediction model. Receiver operating characteristic curves were used to evaluate the performance of the prediction model. We calculated the standardized W statistic and its 95% confidence intervals using the newly developed mortality prediction model.

Results

The area under the receiver operating characteristic curve of the developed scoring system for the prediction of mortality was 0.883 (95% confidence interval [CI]: 0.882–0.884). The Ws score calculated from the 2016 dataset was 0.000 (95% CI: − 0.021 – 0.021). The Ws score calculated from the 2017 dataset was 0.049 (95% CI: 0.030–0.069).

Conclusions

The scoring system developed in the present study utilizing the parameters gathered in initial ED evaluations has acceptable performance for the prediction of in-hospital mortality. Standardized W statistics based on this scoring system can be used to compare the performance of an ED with the reference data or with the performance of other institutions.

Prehospital Variables Alone Can Predict Mortality After Blunt Trauma: A Novel Scoring Tool

BACKGROUND

We sought to develop a novel Prehospital Injury Mortality Score (PIMS) to predict blunt trauma mortality using only prehospital variables.

STUDY DESIGN

The 2017 Trauma Quality Improvement Program database was queried and divided into two equal sized sets at random (derivation and validation sets). Multiple logistic regression models were created to determine the risk of mortality using age, sex, mechanism, and trauma activation criterion. The PIMS was derived using the weighted average of each independent predictor. The discriminative power of the scoring tool was assessed by calculating the area under the receiver operating characteristics (AUROC) curve. The PIMS ability to predict mortality was then assessed by using the validation cohort. The score was compared to the Revised Trauma Score (RTS) using the AUROC curve, including a subgroup of patients with normal vital signs.

RESULTS

The derivation and validation groups each consisted of 163 694 patients. Seven independent predictors of mortality were identified, and the PIMS was derived with scores ranging from 0 to 20. The mortality rate increased from 1.4% to 43.9% and then 100% at scores of 1, 10, and 19, respectively. The model had very good discrimination with an AUROC of .79 in both the derivation and validation groups. When compared to the RTS, the AUROC were similar (.79 vs. .78). On subgroup analysis of patients with normal prehospital vital signs, the PIMS was superior to the RTS (.73 vs. .56).

CONCLUSION

The PIMS is a novel scoring tool to predict mortality in blunt trauma patients using prehospital variables. It had improved discriminatory power in blunt trauma patients with normal vital signs compared to the RTS.

RISC II is superior to TRISS in predicting 30-day mortality in blunt major trauma patients in Hong Kong

PURPOSE

Hong Kong (HK) trauma registries have been using the Trauma and Injury Severity Score (TRISS) for audit and benchmarking since their introduction in 2000. We compare the mortality prediction model using TRISS and Revised Injury Severity Classification, version II (RISC II) for trauma centre patients in HK.

METHODS

This was a retrospective cohort study with all five trauma centres in HK. Adult trauma patients with Injury Severity Score (ISS) > 15 suffering from blunt injuries from January 2013 to December 2015 were included. TRISS models using the US and local coefficients were compared with the RISC II model. The primary outcome was 30-day mortality and the area under the receiver operating characteristic curve (AUC) for tested models.

RESULTS

1840 patients were included, of whom 1236/1840 (67%) were male. Median age was 59 years and median ISS was 25. Low falls were the most common mechanism of injury. The 30-day mortality was 23%. RISC II yielded a superior AUC of 0.896, compared with the TRISS models (MTOS: 0.848; PATOS: 0.839; HK: 0.858). Prespecified subgroup analyses showed that all the models performed worse for age ≥ 70, ASA ≥ III, and low falls. RISC II had a higher AUC compared with the TRISS models in all subgroups, although not statistically significant.

CONCLUSION

RISC II was superior to TRISS in predicting the 30-day mortality for Hong Kong adult blunt major trauma patients. RISC II may be useful when performing future audit or benchmarking exercises for trauma in Hong Kong.

A Comparative Study Between Two Combat Injury Severity Scores

INTRODUCTION

In recent years, specific trauma scoring systems have been developed for military casualties. The objective of this study was to examine the discrepancies in severity scores of combat casualties between the Abbreviated Injury Scale 2005-Military (mAIS) and the Military Combat Injury Scale (MCIS) and a review of the current literature on the application of trauma scoring systems in the military setting.

METHODS

A cross-sectional, descriptive, and retrospective study was conducted between May 1, 2005, and December 31, 2014. The study population consisted of all combat casualties attended in the Spanish Role 2 deployed in Herat (Afghanistan). We used the New Injury Severity Score (NISS) as reference score. Severity of each injury was calculated according to mAIS and MCIS, respectively. The severity of each casualty was calculated according to the NISS based on the mAIS (Military New Injury Severity Score-mNISS) and MCIS (Military Combat Injury Scale-New Injury Severity Score-MCIS-NISS). Casualty severity were grouped by severity levels (mild-scores: 1-8, moderate-scores: 9-15, severe-scores: 16-24, and critical-scores: 25-75).

RESULTS

Nine hundred and eleven casualties were analyzed. Most were male (96.37%) with a median age of 27 years. Afghan patients comprised 71.13%. Air medevac was the main casualty transportation method (80.13). Explosion (64.76%) and gunshot wound (34.68%) mechanisms predominated. Overall mortality was 3.51%. Median mNISS and MCIS-NISS were similar in nonsurvivors (36 [IQR, 25-49] vs. [IQR, 25-48], respectively) but different in survivors, 9 (IQR, 4-17) vs. 5 (IQR, 2-13), respectively (P < .0001). The mNISS and MCIS-NISS were discordant in 34.35% (n = 313). Among cases with discordant severity scores, the median difference between mNISS and MCIS-NISS was 9 (IQR, 4-16); range, 1 to 57.

CONCLUSION

Our study findings suggest that discrepancies in injury severity levels may be observed in one in three of the casualties when using mNISS and MCIS-NISS.

Effectiveness and cost-effectiveness of the Transmural Trauma Care Model investigated in a multicenter trial with a controlled before-and-after design: A study protocol

Objective

The rehabilitation of trauma patients in primary care is challenging, and there are no guidelines for optimal treatment. Also, the organization of care is not well‐structured. The Transmural Trauma Care Model (TTCM) has been developed in the Netherlands, aiming to improve patient outcomes by optimizing the organization and the quality of the rehabilitation process in primary care. A recent feasibility study showed that implementation of the TTCM at a Dutch Level 1 trauma center was feasible, patient outcomes were improved, and costs were reduced. This study aims to assess the effectiveness and cost‐effectiveness of the TTCM compared to the usual care in a multicenter trial.

Methods

A multicenter trial with a controlled before‐and‐after design will be performed at 10 hospitals in the Netherlands. First, participating hospitals will include 322 patients in the control group, receiving usual care as provided in these specific hospitals. Subsequently, the TTCM will be implemented in all participating hospitals, and hospitals will include an additional 322 patients in the intervention group. The TTCM consists of a multidisciplinary team at the outpatient clinic (trauma surgeon and hospital‐based physical therapist), an educated and trained network of primary care trauma physical therapists, and structural communication between them. Co‐primary outcomes will investigate generic and disease‐specific, health‐related quality of life. Secondary outcomes will include pain, patient satisfaction, perceived recovery, and patient‐reported physical functioning. For the economic evaluation, societal and healthcare costs will be measured. Measurements will take place at baseline and after 6 weeks, 3, 6, and 9 months. Analyses will be based on the intention‐to‐treat principle. Missing data will be handled using longitudinal data analyses in the effect analyses and by multivariate imputation in the economic evaluation.

Conclusion

This trial with a controlled before‐and‐after design will give insight into the effectiveness and cost‐effectiveness of the TTCM in a multicenter trial.

Computational Derivation of Core, Dynamic Human Blunt Trauma Inflammatory Endotypes

Systemic inflammation ensues following traumatic injury, driving immune dysregulation and multiple organ dysfunction (MOD). While a balanced immune/inflammatory response is ideal for promoting tissue regeneration, most trauma patients exhibit variable and either overly exuberant or overly damped responses that likely drive adverse clinical outcomes. We hypothesized that these inflammatory phenotypes occur in the context of severe injury, and therefore sought to define clinically distinct endotypes of trauma patients based on their systemic inflammatory responses. Using Patient-Specific Principal Component Analysis followed by unsupervised hierarchical clustering of circulating inflammatory mediators obtained in the first 24 h after injury, we segregated a cohort of 227 blunt trauma survivors into three core endotypes exhibiting significant differences in requirement for mechanical ventilation, duration of ventilation, and MOD over 7 days. Nine non-survivors co-segregated with survivors. Dynamic network inference, Fisher Score analysis, and correlations of IL-17A with GM-CSF, IL-10, and IL-22 in the three survivor sub-groups suggested a role for type 3 immunity, in part regulated by Th17 and γδ 17 cells, and related tissue-protective cytokines as a key feature of systemic inflammation following injury. These endotypes may represent archetypal adaptive, over-exuberant, and overly damped inflammatory responses.

Diagnostic accuracy, yield, and comparative effectiveness of whole-body computed tomography in blunt trauma: A protocol for systematic review and meta-analysis

OBJECTIVES

Controversies emerge over routine performances of whole-body computed tomography (WBCT) in patients with blunt polytrauma. The existing randomized and non-randomized evidence is inconclusive, and during observations of non-trauma, incidental findings, detected by WBCT, have left uncertainty regarding their consequences and optimal management. Additionally, previous meta-analyses have failed to address the limitations of primary studies and issues associated with incidental findings. Therefore, we planned a new systematic review to address these points.

METHODS

We will search the PubMed, EMBASE, and Cochrane Central databases from inception to December 31, 2020, with no language restriction and perform full-text evaluation of potentially relevant articles. We will include prospective and retrospective studies with a single-gate design that assessed diagnostic accuracy and/or yield of WBCT to detect traumatic injuries, and studies that assessed incidental findings detected by WBCT. Additionally, we will include randomized controlled trials and non-randomized comparative studies that assessed the effectiveness of WBCT against conventional care, including selective computed tomography (CT). Studies of patients of all ages with blunt traumatic injuries, assessed at an emergency department, will be included. Two reviewers will extract data and rate the study validity via standard quality assessment tools. The primary outcome of interest will be reduction in mortality. Our secondary outcomes will include diagnostic accuracy and yield, detection of incidental findings and clinical outcomes associated with these detections, and improvement in other non-mortality clinical outcomes. We will qualitatively assess study, patient, and intervention characteristics and clinical outcomes. If appropriate, we will perform random-effects model meta-analyses to obtain summary estimates. Finally, we will assess the certainty of evidence by the grading the quality of evidence and strength of recommendations.

ETHICS AND DISSEMINATION

Ethics approval is not applicable, as this is a secondary analysis of publicly available data. The review results will be submitted for publication in peer-reviewed journals.

PROSPERO REGISTRATION

CRD42020187852.

The Trajectory of Alterations in Immune-Cell Counts in Severe-Trauma Patients Is Related to the Later Occurrence of Sepsis and Mortality: Retrospective Study of 917 Cases

Background

Severe trauma is believed to disrupt the homeostasis of the immune system, and lead to dramatic changes in the circulating immune-cell count (ICC). The latter fluctuates widely over time. Knowledge about the relationship between these dramatic changes and dynamic fluctuations and the late prognosis of trauma patients is sparse. We investigated the relationship between the trajectory of alterations in the circulating ICC within 7 days in severe-trauma patients and subsequent sepsis and mortality.

Methods

A retrospective analysis of 917 patients with an Injury Severity Score ≥16 was undertaken. The absolute neutrophil, lymphocyte, and monocyte counts (ANC, ALC, and AMC, respectively) on days 1, 3, and 7 (D1, D3, and D7, respectively) after trauma, and whether sepsis or death occurred within 60 days, were recorded. As the disordered circulating ICCs fluctuated widely, their time-varying slopes (D3/D1 and D7/D3) were calculated. Patients were divided into “sepsis” and “non-sepsis” groups, as well as “alive” and “death” groups. Comparative studies were conducted between every two groups. Univariate and multivariate logistic regression analyses were used to identify variables related to the risk of sepsis and mortality. Receiver operating characteristic curves were plotted to assess the predictive value of various risk factors.

Results

More severe trauma caused more pronounced increases in the ANC and slower recovery of the ALC within 7 days. The ALC (D3), ANC (D7), ALC (D3/D1), and ANC (D7/D3) were independent risk factors for sepsis. The ALC (D3), ALC (D7), AMC (D7), and ALC (D3/D1) were independent risk factors for mortality. A combination of the ALC (D3) and ALC (D3/D1) exerted a good predictive value for sepsis and death.

Conclusions

The trajectory of alterations in the circulating ICC in the early stage after trauma is related to subsequent sepsis and mortality.

Geriatric Nutritional Risk Index as a Tool to Evaluate Impact of Malnutrition Risk on Mortality in Adult Patients with Polytrauma

BACKGROUND

Identification of malnutrition is especially important in severely injured patients, in whom hypermetabolism and protein catabolism following traumatic injury worsen their nutritional condition. The geriatric nutritional risk index (GNRI), based on serum albumin level and the current body weight/ideal body weight ratio, is useful for identifying patients with malnutrition in many clinical conditions. This study aimed to explore the association between admission GNRI and mortality outcomes of adult patients with polytrauma.

METHODS

From 1 January 2009 to 31 December 2019, a total of 348 adult patients with polytrauma, registered in the trauma database of a level I trauma center, were recognized and categorized into groups of death (n = 71) or survival (n = 277) and into four nutritional risk groups: a high-risk group (GNRI < 82, n = 87), a moderate-risk group (GNRI 82 to <92, n = 144), a low-risk group (GNRI 92-98, n = 59), and a no-risk group (GNRI > 98, n = 58). Univariate and multivariate logistic regression analyses were used to identify the independent risk factors for mortality. The mortality outcomes of patients at various nutritional risks were compared to those of patients in the no-risk group.

RESULTS

The comparison between the death group (n = 71) and the survival group (n = 277) revealed that there was no significant difference in gender predominance, age, pre-existing comorbidities, injury mechanism, systolic blood pressure, and respiratory rate upon arrival at the emergency room. A significantly lower GNRI and Glasgow Coma Scale score but higher injury severity score (ISS) was observed in the death group than in the survival group. Multivariate logistic regression analysis revealed that Glasgow Coma Scale (GCS), odds ratio (OR), 0.88; 95% confidence interval (CI), 0.83-0.95; p < 0.001), ISS (OR, 1.07; 95% CI, 1.04-1.11; p < 0.001), and GNRI (OR, 0.94; 95% CI, 0.91-0.97; p < 0.001) were significant independent risk factors for mortality in these patients. The mortality rates for the high-risk, moderate-risk, low-risk, and no-risk groups were 34.5%, 20.1%, 8.5%, and 12.1%, respectively. Unlike patients in the moderate-risk and low-risk groups, patients in the high-risk group had a significantly higher death rate than that of those in the no-risk group.

CONCLUSIONS

This study revealed that the GNRI may serve as a simple, promising screening tool to identify the high risk of malnutrition for mortality in adult patients with polytrauma.

A study protocol for a predictive algorithm to assess population-based premature mortality risk: Premature Mortality Population Risk Tool (PreMPoRT)

BACKGROUND

Premature mortality is an important population health indicator used to assess health system functioning and to identify areas in need of health system intervention. Predicting the future incidence of premature mortality in the population can facilitate initiatives that promote equitable health policies and effective delivery of public health services. This study protocol proposes the development and validation of the Premature Mortality Risk Prediction Tool (PreMPoRT) that will predict the incidence of premature mortality using large population-based community health surveys and multivariable modeling approaches.

METHODS

PreMPoRT will be developed and validated using various training, validation, and test data sets generated from the six cycles of the Canadian Community Health Survey (CCHS) linked to the Canadian Vital Statistics Database from 2000 to 2017. Population-level risk factor information on demographic characteristics, health behaviors, area level measures, and other health-related factors will be used to develop PreMPoRT and to predict the incidence of premature mortality, defined as death prior to age 75, over a 5-year period. Sex-specific Weibull accelerated failure time models will be developed using a Canadian provincial derivation cohort consisting of approximately 500,000 individuals, with approximately equal proportion of males and females, and about 12,000 events of premature mortality. External validation will be performed using separate linked files (CCHS cycles 2007-2008, 2009-2010, and 2011-2012) from the development cohort (CCHS cycles 2000-2001, 2003-2004, and 2005-2006) to check the robustness of the prediction model. Measures of overall predictive performance (e.g., Nagelkerke's R2), calibration (e.g., calibration plots), and discrimination (e.g., Harrell's concordance statistic) will be assessed, including calibration within defined subgroups of importance to knowledge users and policymakers.

DISCUSSION

Using routinely collected risk factor information, we anticipate that PreMPoRT will produce population-based estimates of premature mortality and will be used to inform population strategies for prevention.

Machine learning techniques for mortality prediction in critical traumatic patients: anatomic and physiologic variables from the RETRAUCI study

Background

Interest in models for calculating the risk of death in traumatic patients admitted to ICUs remains high. These models use variables derived from the deviation of physiological parameters and/or the severity of anatomical lesions with respect to the affected body areas. Our objective is to create different predictive models of the mortality of critically traumatic patients using machine learning techniques.

Methods

We used 9625 records from the RETRAUCI database (National Trauma Registry of 52 Spanish ICUs in the period of 2015–2019). Hospital mortality was 12.6%. Data on demographic variables, affected anatomical areas and physiological repercussions were used. The Weka Platform was used, along with a ten-fold cross-validation for the construction of nine supervised algorithms: logistic regression binary (LR), neural network (NN), sequential minimal optimization (SMO), classification rules (JRip), classification trees (CT), Bayesian networks (BN), adaptive boosting (ADABOOST), bootstrap aggregating (BAGGING) and random forest (RFOREST). The performance of the models was evaluated by accuracy, specificity, precision, recall, F-measure, and AUC.

Results

In all algorithms, the most important factors are those associated with traumatic brain injury (TBI) and organic failures. The LR finds thorax and limb injuries as independent protective factors of mortality. The CT generates 24 decision rules and uses those related to TBI as the first variables (range 2.0–81.6%). The JRip detects the eight rules with the highest risk of mortality (65.0–94.1%). The NN model uses a hidden layer of ten nodes, which requires 200 weights for its interpretation. The BN find the relationships between the different factors that identify different patient profiles. Models with the ensemble methodology (ADABOOST, BAGGING and RandomForest) do not have greater performance. All models obtain high values ​​in accuracy, specificity, and AUC, but obtain lower values ​​in recall. The greatest precision is achieved by the SMO model, and the BN obtains the best recall, F-measure, and AUC.

Conclusion

Machine learning techniques are useful for creating mortality classification models in critically traumatic patients. With clinical interpretation, the algorithms establish different patient profiles according to the relationship between the variables used, determine groups of patients with different evolutions, and alert clinicians to the presence of rules that indicate the greatest severity.

Performance of Prognostic Scoring Systems in Trauma Patients in the Intensive Care Unit of a Trauma Center

BACKGROUND

Prediction of mortality outcomes in trauma patients in the intensive care unit (ICU) is important for patient care and quality improvement. We aimed to measure the performance of 11 prognostic scoring systems for predicting mortality outcomes in trauma patients in the ICU.

METHODS

Prospectively registered data in the Trauma Registry System from 1 January 2016 to 31 December 2018 were used to extract scores from prognostic scoring systems for 1554 trauma patients in the ICU. The following systems were used: the Trauma and Injury Severity Score (TRISS); the Acute Physiology and Chronic Health Evaluation (APACHE II); the Simplified Acute Physiology Score (SAPS II); mortality prediction models (MPM II) at admission, 24, 48, and 72 h; the Multiple Organ Dysfunction Score (MODS); the Sequential Organ Failure Assessment (SOFA); the Logistic Organ Dysfunction Score (LODS); and the Three Days Recalibrated ICU Outcome Score (TRIOS). Predictive performance was determined according to the area under the receiver operator characteristic curve (AUC).

RESULTS

MPM II at 24 h had the highest AUC (0.9213), followed by MPM II at 48 h (AUC: 0.9105). MPM II at 24, 48, and 72 h (0.8956) had a significantly higher AUC than the TRISS (AUC: 0.8814), APACHE II (AUC: 0.8923), SAPS II (AUC: 0.9044), MPM II at admission (AUC: 0.9063), MODS (AUC: 0.8179), SOFA (AUC: 0.7073), LODS (AUC: 0.9013), and TRIOS (AUC: 0.8701). There was no significant difference in the predictive performance of MPM II at 24 and 48 h (p = 0.37) or at 72 h (p = 0.10).

CONCLUSIONS

We compared 11 prognostic scoring systems and demonstrated that MPM II at 24 h had the best predictive performance for 1554 trauma patients in the ICU.

Prognosis of the Trauma Patients According to the Trauma and Injury Severity Score (TRISS); A Diagnostic Accuracy Study

Objective

To investigate the prognosis and survival rates of a group of Iranian patients with traumatic injuries using the trauma and injury severity score (TRISS) model.

Methods

In this prospective cohort study, all the patients with multi-trauma referring to the Yasuj Shahid Beheshti hospital during 2018 were included. The patients' demographic information, trauma and history of previous illness were recorded. Vital symptoms including respiratory rate, heart rate, hypertension, pulse rate and Glasgow coma scale (GCS) score were assessed. The injury severity score (ISS) was calculated based on the type and location of the injuries and according to the abbreviated injury scale (AIS) classification. The survival probability of the patients was assessed according to the TRISS model.

Results

Overall, 252 trauma patients were evaluated out of whom, 195 (77.4%) were men and 57 (22.6%) women. If we consider the TRISS score probability above 0.5 as the chance of being alive, the mortality rate was 6.75%, that was lower than our series (7.1%). The ISS score and GCS had a positive significant relationship with other variables except respiratory rate, body temperature and hospitalization. Revised trauma score (RTS) was significantly associated with other variables including age, GCS, hemoglobin, systolic blood pressure and respiratory rate. TRISS had an area under curve (AUC) of 0.988 indicating a high prognostic accuracy.

Conclusion

The mortality rate was lower than that of being predicted by TRISS. This might be due to treatment effectiveness and care for traumatic patients leading to decreased mortality. TRISS had high prognostic accuracy in trauma patients. We also reported an association between hemoglobin and survival rate. Therefore, it seems that considering the laboratory parameters can be useful in patients with trauma.

The AdHOC (age, head injury, oxygenation, circulation) score: a simple assessment tool for early assessment of severely injured patients with major fractures

Purpose

We sought to develop a simple, effective and accurate assessment tool using well-known prognostic parameters to predict mortality and morbidity in severely injured patients with major fractures at the stage of the trauma bay.

Methods

European Data from the TraumaRegister DGU^® were queried for patients aged 16 or older and with an ISS of 9 and higher with major fractures. The development (2012–2015) and validation (2016) groups were separated. The four prognostic aspects Age, Head injury, Oxygenation and Circulation along with parameters were identified as having a relevant impact on the outcome of severely injured patients with major fractures. The performance of the score was analyzed with the area under the receiver operating characteristics curve and compared to other trauma scores.

Results

An increasing AdHOC (Age, Head injury, Oxygenation, Circulation) score value in the 17,827 included patients correlated with increasing mortality (0 points = 0.3%, 1 point = 5.3%, 2 points = 15.6%, 3 points = 42.5% and 4 points = 62.6%). With an AUROC of 0.858 for the development (n = 14,047) and 0.877 for the validation (n = 3780) group dataset, the score is superior in performance compared to the Injury Severity Score (0.806/0.815).

Conclusion

The AdHOC score appears to be easy and accessible in every emergency room without the requirement of special diagnostic tools or knowledge of the exact injury pattern and can be useful for the planning of further surgical treatment.

Assignment of pre-event ASA physical status classification by pre-hospital physicians: a prospective inter-rater reliability study

Background

Individualized treatment is a common principle in hospitals. Treatment decisions are made based on the patient’s condition, including comorbidities. This principle is equally relevant out-of-hospital. Furthermore, comorbidity is an important risk-adjustment factor when evaluating pre-hospital interventions and may aid therapeutic decisions and triage. The American Society of Anesthesiologists Physical Status (ASA-PS) classification system is included in templates for reporting data in physician-staffed pre-hospital emergency medical services (p-EMS) but whether an adequate full pre-event ASA-PS can be assessed by pre-hospital physicians remains unknown. We aimed to explore whether pre-hospital physicians can score an adequate pre-event ASA-PS with the information available on-scene.

Methods

The study was an inter-rater reliability study consisting of two steps. Pre-event ASA-PS scores made by pre- and in-hospital physicians were compared. Pre-hospital physicians did not have access to patient records and scores were based on information obtainable on-scene. In-hospital physicians used the complete patient record (Step 1). To assess inter-rater reliability between pre- and in-hospital physicians when given equal amounts of information, pre-hospital physicians also assigned pre-event ASA-PS for 20 of the included patients by using the complete patient records (Step 2). Inter-rater reliability was analyzed using quadratic weighted Cohen’s kappa (κ_w).

Results

For most scores (82%) inter-rater reliability between pre-and in-hospital physicians were moderate to substantial (κ_w 0,47-0,89). Inter-rater reliability was higher among the in-hospital physicians (κ_w 0,77 to 0.85). When all physicians had access to the same information, κ_w increased (κ_w 0,65 to 0,93).

Conclusions

Pre-hospital physicians can score an adequate pre-event ASA-PS on-scene for most patients. To further increase inter-rater reliability, we recommend access to the full patient journal on-scene. We recommend application of the full ASA-PS classification system for reporting of comorbidity in p-EMS.

Inclusion of lactate level measured upon emergency room arrival in trauma outcome prediction models improves mortality prediction: a retrospective, single-center study

Background

This study aimed to develop a model for predicting trauma outcomes by adding arterial lactate levels measured upon emergency room (ER) arrival to existing trauma injury severity scoring systems.

Methods

We examined blunt trauma cases that were admitted to our hospital during 2010– 2014. Eligibility criteria were cases with an Injury Severity Score of ≥9, complete Trauma and Injury Severity Score (TRISS) variable data, and lactate levels that were assessed upon ER arrival. Survivor and non-survivor groups were compared and lactate-based prediction models were generated using logistic regression. We compared the predictive performances of traditional prediction models (Revised Trauma Score [RTS] and TRISS) and lactate-based models using the area under the curve (AUC) of receiver operating characteristic curves.

Results

We included 829 patients, and the in-hospital mortality rate among these patients was 21.6%. The model that used lactate levels and age provided a significantly better AUC value than the RTS model. The model with lactate added to the TRISS variables provided the highest Youden J statistic, with 86.0% sensitivity and 70.8% specificity at a cutoff value of 0.15, as well as the highest predictive value, with a significantly higher AUC than the TRISS.

Conclusions

These findings indicate that lactate testing upon ER arrival may help supplement or replace traditional physiological parameters to predict mortality outcomes among Korean trauma patients. Adding lactate levels also appears to improve the predictive abilities of existing trauma outcome prediction models.

Prognostic factors for medical and productivity costs, and return to work after trauma

AIM

The aim of this study was to determine prognostic factors for medical and productivity costs, and return to work (RTW) during the first two years after trauma in a clinical trauma population.

METHODS

This prospective multicentre observational study followed all adult trauma patients (≥18 years) admitted to a hospital in Noord-Brabant, the Netherlands from August 2015 through November 2016. Health care consumption, productivity loss and return to work were measured in questionnaires at 1 week, 1, 3, 6, 12 and 24 months after injury. Data was linked with hospital registries. Prognostic factors for medical costs and productivity costs were analysed with log-linked gamma generalized linear models. Prognostic factors for RTW were assessed with Cox proportional hazards model. The predictive ability of the models was assessed with McFadden R2 (explained variance) and c-statistics (discrimination).

RESULTS

A total of 3785 trauma patients (39% of total study population) responded to at least one follow-up questionnaire. Mean medical costs per patient (€9,710) and mean productivity costs per patient (€9,000) varied widely. Prognostic factors for high medical costs were higher age, female gender, spine injury, lower extremity injury, severe head injury, high injury severity, comorbidities, and pre-injury health status. Productivity costs were highest in males, and in patients with spinal cord injury, high injury severity, longer length of stay at the hospital and patients admitted to the ICU. Prognostic factors for RTW were high educational level, male gender, low injury severity, shorter length of stay at the hospital and absence of comorbidity.

CONCLUSIONS

Productivity costs and RTW should be considered when assessing the economic impact of injury in addition to medical costs. Prognostic factors may assist in identifying high cost groups with potentially modifiable factors for targeted preventive interventions, hence reducing costs and increasing RTW rates.

Risk factors for death of trauma patients admitted to an Intensive Care Unit

OBJECTIVE

To analyze the risk factors for death of trauma patients admitted to the intensive care unit (ICU).

METHOD

Retrospective cohort study with data from medical records of adults hospitalized for trauma in a general intensive care unit. We included patients 18 years of age and older and admitted for injuries. The variables were grouped into levels in a hierarchical manner. The distal level included sociodemographic variables, hospitalization, cause of trauma and comorbidities; the intermediate, the characteristics of trauma and prehospital care; the proximal, the variables of prognostic indices, intensive admission, procedures and complications. Multiple logistic regression analysis was performed.

RESULTS

The risk factors associated with death at the distal level were age 60 years or older and comorbidities; at intermediate level, severity of trauma and proximal level, severe circulatory complications, vasoactive drug use, mechanical ventilation, renal dysfunction, failure to perform blood culture on admission and Acute Physiology and Chronic Health Evaluation II.

CONCLUSION

The identified factors are useful to compose a clinical profile and to plan intensive care to avoid complications and deaths of traumatized patients.

Reverse shock index multiplied by Glasgow Coma Scale (rSIG) predicts mortality in severe trauma patients with head injury

The reverse shock index (rSI), a ratio of systolic blood pressure (SBP) to heart rate (HR), is used to identify prognosis in trauma patients. Multiplying rSI by Glasgow Coma Scale (rSIG) can possibly predict better in-hospital mortality in patients with trauma. However, rSIG has never been used to evaluate the mortality risk in adult severe trauma patients (Injury Severity Score [ISS] ≥ 16) with head injury (head Abbreviated Injury Scale [AIS] ≥ 2) in the emergency department (ED). This retrospective case control study recruited adult severe trauma patients (ISS ≥ 16) with head injury (head AIS ≥ 2) who presented to the ED of two major trauma centers between January 01, 2014 and May 31, 2017. Demographic data, vital signs, ISS scores, injury mechanisms, laboratory data, managements, and outcomes were included for the analysis. Logistic regression and receiver operating characteristic analysis were used to evaluate the accuracy of rSIG score in predicting in-hospital mortality. In total, 438 patients (mean age: 56.48 years; 68.5% were males) were included in this study. In-hospital mortality occurred in 24.7% patients. The median (interquartile range) ISS score was 20 (17-26). Patients with rSIG ≤ 14 had seven-fold increased risks of mortality than those without rSIG ≤ 14 (odds ratio: 7.64; 95% confidence interval: 4.69-12.42). Hosmer-Lemeshow goodness-of-fit test and area under the curve values for rSIG score were 0.29 and 0.76, respectively. The sensitivity, specificity, positive predictive value, and negative predictive values of rSIG ≤ 14 were 0.71, 0.75, 0.49, and 0.89, respectively. The rSIG score is a prompt and simple tool to predict in-hospital mortality among adult severe trauma patients with head injury.

Machine Learning Models of Survival Prediction in Trauma Patients

BACKGROUND

We aimed to build a model using machine learning for the prediction of survival in trauma patients and compared these model predictions to those predicted by the most commonly used algorithm, the Trauma and Injury Severity Score (TRISS).

METHODS

Enrolled hospitalized trauma patients from 2009 to 2016 were divided into a training dataset (70% of the original data set) for generation of a plausible model under supervised classification, and a test dataset (30% of the original data set) to test the performance of the model. The training and test datasets comprised 13,208 (12,871 survival and 337 mortality) and 5603 (5473 survival and 130 mortality) patients, respectively. With the provision of additional information such as pre-existing comorbidity status or laboratory data, logistic regression (LR), support vector machine (SVM), and neural network (NN) (with the Stuttgart Neural Network Simulator (RSNNS)) were used to build models of survival prediction and compared to the predictive performance of TRISS. Predictive performance was evaluated by accuracy, sensitivity, and specificity, as well as by area under the curve (AUC) measures of receiver operating characteristic curves.

RESULTS

In the validation dataset, NN and the TRISS presented the highest score (82.0%) for balanced accuracy, followed by SVM (75.2%) and LR (71.8%) models. In the test dataset, NN had the highest balanced accuracy (75.1%), followed by the TRISS (70.2%), SVM (70.6%), and LR (68.9%) models. All four models (LR, SVM, NN, and TRISS) exhibited a high accuracy of more than 97.5% and a sensitivity of more than 98.6%. However, NN exhibited the highest specificity (51.5%), followed by the TRISS (41.5%), SVM (40.8%), and LR (38.5%) models.

CONCLUSIONS

These four models (LR, SVM, NN, and TRISS) exhibited a similar high accuracy and sensitivity in predicting the survival of the trauma patients. In the test dataset, the NN model had the highest balanced accuracy and predictive specificity.

Predicting health status in the first year after trauma

Background

Although mortality rates following major trauma are continuing to decline, a growing number of patients are experiencing long-term disability. The aim of this study was to identify factors associated with health status in the first year following trauma and develop prediction models based on a defined trauma population.

Methods

The Brabant Injury Outcome Surveillance (BIOS) study was a multicentre prospective observational cohort study. Adult patients with traumatic injury were included from August 2015 to November 2016 if admitted to one of the hospitals of the Noord-Brabant region in the Netherlands. Outcome measures were EuroQol Five Dimensions 5D-3L (EQ-5D™ utility and visual analogue scale (VAS)) and Health Utilities Index (HUI) 2 and 3 scores 1 week and 1, 3, 6 and 12 months after injury. Prediction models were developed using linear mixed models, with patient characteristics, preinjury health status, injury severity and frailty as possible predictors. Predictors that were significant (P &lt; 0·050) for one of the outcome measures were included in all models. Performance was assessed using explained variance (R2).

Results

In total, 4883 patients participated in the BIOS study (50·0 per cent of the total), of whom 3366 completed the preinjury questionnaires. Preinjury health status and frailty were the strongest predictors of health status during follow-up. Age, sex, educational level, severe head or face injury, severe torso injury, injury severity, Functional Capacity Index score, co-morbidity and duration of hospital stay were also relevant in the multivariable models predicting health status. R2 ranged from 35 per cent for EQ-VAS to 48 per cent for HUI 3.

Conclusion

The most important predictors of health status in the first year after trauma in this population appeared to be preinjury health status and frailty.

The development and internal validation of a model to predict functional recovery after trauma

Objective

To develop and internally validate the PROgnosis of functional recovery after Trauma (PRO-Trauma) prediction model.

Design

A prospective single-center longitudinal cohort study. Patients were assessed at 6 weeks and 12 months post-injury.

Methods

Patients that presented at the emergency department with an acute traumatic injury, were prompted for participation. Patients that completed the assessments at 6 weeks and 12 months post injury were included. Exclusion criteria: age < 18, age > 65, pathologic fractures, injuries that resulted in severe neurologic deficits. The predicted outcome, functional recovery, was defined as a Short Musculoskeletal Function Assessment (SMFA-NL) Problems with Daily Activities (PDA) subscale ≤ 12.2 points at 12 months post-injury (Dutch population norm). Predictors were: gender, age, living with partner, number of chronic health conditions, SMFA-NL PDA score 6 weeks post-injury, ICU admission, length of stay in hospital, injury severity score, occurrence of complications and treatment type. All predictors were obtained before 6 weeks post-injury. Missing data were multiply imputed. Predictor variables were selected using backward stepwise multivariable logistic regression. Hosmer-Lemeshow tests were used to evaluate calibration. Bootstrap resampling was used to internally validate the final model.

Results

A total of 246 patients were included, of which 104 (44%) showed functional recovery. The predictors in the final PRO-Trauma model were: living with partner, the number of chronic health conditions, SMFA-NL PDA subscale score at 6 weeks post-injury and length of stay in hospital. The apparent R² was 0.33 [0.33;0.34], the c-statistic was 0.79 [0.79;0.80]. Hosmer-Lemeshow test indicated good calibration (p = 0.92). Optimism-corrected R² was 0.28 [0.27;0.29] and the optimism-corrected Area Under the Curve was 0.77 [0.77;0.77].

Conclusion

The PRO-Trauma prediction model can be used to obtain valid predictions of attaining functional recovery after trauma at 12 months post-injury. The PRO-Trauma prediction model showed acceptable calibration and discrimination.