Comparison of two prognostic models in trauma outcome

Validation of the China mortality prediction model in trauma based on the ICD-10-CM codes

The China mortality prediction model in trauma, based on the International Classification of Diseases, Tenth Revision, Clinical Modification lexicon (CMPMIT-ICD-10), is a novel model for predicting outcomes in patients who experienced trauma. This model has not yet been validated using data acquired from patients at other trauma centers in China. This retrospective study used data retrieved from the Peking University People's Hospital discharge database and included all patients admitted for trauma between 2012 and 2022 for model validation. Model performance was categorized into discrimination and calibration. In total, 23,299 patients were included in this study, with an overall mortality rate of 1.2%. CMPMIT-ICD-10 showed good discrimination and calibration, with an area under the curve of 0.84 (95% confidence interval: 0.82-0.87) and a Brier score of 0.02. The performance of the CMPMIT-ICD-10 during validation was satisfactory, and the application of the model will be scaled up in future studies.

A Retrospective Cohort Study: Predicting 90-Day Mortality for ICU Trauma Patients with a Machine Learning Algorithm Using XGBoost Using MIMIC-III Database

Objective

The aim of this study was to develop and validate a machine learning-based predictive model that predicts 90-day mortality in ICU trauma patients.

Methods

Data of patients with severe trauma were extracted from the Medical Information Mart for Intensive Care III (MIMIC-III) database. The performances of mortality prediction models generated using nine machine learning extreme gradient boosting (XGBoost), logistic regression, random forest, AdaBoost, multilayer perceptron (MLP) neural networks, support vector machine (SVM), light gradient boosting machine (GBM), k nearest neighbors (KNN) and gaussian naive bayes (GNB). The performance of the model was evaluated in terms of discrimination, calibration and clinical application.

Results

We found that the accuracy, sensitivity, specificity, PPV, NPV and F1 score of our proposed XGBoost model were 82.8%, 79.7%, 77.6%, 51.2%, 91.5% and 0.624, respectively. Among the nine models, the XGBoost model performed best. Compared with traditional logistic regression, the calibration curves of the XGBoost model and decision curve analysis (DCA) performed well.

Conclusion

Our study shows that the XGBoost model outperforms other machine learning models in predicting 90-day mortality in trauma patients. It can be used to assist clinicians in the early identification of mortality risk factors and early intervention to reduce mortality.

Adaptive Bayesian learning for making risk-aware decisions: A case of trauma survival prediction

Decision tree (DT) models provide a transparent approach to prediction of patient's outcomes within a probabilistic framework. Averaging over DT models under certain conditions can deliver reliable estimates of predictive posterior probability distributions, which is of critical importance in the case of predicting an individual patient's outcome. Reliable estimations of the distribution can be achieved within the Bayesian framework using Markov chain Monte Carlo (MCMC) and its Reversible Jump extension enabling DT models to grow to a reasonable size. Existing MCMC strategies however have limited ability to control DT structures and tend to sample overgrown DT models, making unreasonably small partitions, thus deteriorating the uncertainty calibration. This happens because the MCMC explores a DT model parameter space within a limited knowledge of the distribution of data partitions. We propose a new adaptive strategy which overcomes this limitation, and show that in the case of predicting trauma outcomes the number of data partitions can be significantly reduced, so that the unnecessary uncertainty of estimating the predictive posterior density is avoided. The proposed and existing strategies are compared in terms of entropy which, being calculated for predicted posterior distributions, represents the uncertainty in decisions. In this framework, the proposed method has outperformed the existing sampling strategies, so that the unnecessary uncertainty in decisions is efficiently avoided.

Early interventions for post-traumatic stress following musculoskeletal trauma: protocol for a systematic review and meta-analysis

INTRODUCTION

Post-traumatic stress symptoms (PTSS) can be triggered following exposure to a traumatic event, such as violence, disasters, serious accidents and injury. Little is known about which interventions provide the greatest benefit for PTSS. This systematic review aims to estimate the effects of early interventions on PTSS following musculoskeletal trauma.

METHODS/ANALYSIS

Development of this review protocol was guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols checklist. This review will include randomised controlled trials and non-randomised controlled studies evaluating the effect of early (within 3 months of a traumatic event) non-pharmacological and non-surgical interventions on PTSS in adults (aged ≥18 years). MEDLINE, PsycINFO, Embase, CINAHL, Zetoc, PROSPERO, Web of Science, PubMed and Google Scholar, as well as key journals/grey literature, will be searched from inception to 31 July 2022. Only articles published in English will be considered. Two independent reviewers will search, screen studies, extract data and assess risk of bias using the Cochrane Risk of Bias tool V.2 (RoB 2) and the Risk Of Bias in Non-randomised Studies of Interventions (ROBINS-I), respectively. Mean difference or standardised mean difference (SMD) will be extracted with accompanying 95% CIs and p values where these are reported. Group effect size will be extracted and reported. Symptoms of PTSS will be ascertained using SMDs (continuous) and diagnosis of PTSS using risk ratio (dichotomous). If possible, study results will be pooled into a meta-analysis. A narrative synthesis of the results will be presented if heterogeneity is high. The overall quality of evidence and risk of bias will be assessed using the Grading of Recommendations Assessment, Development and Evaluation, RoB 2 and ROBINS-I guidelines, respectively.

ETHICS AND DISSEMINATION

Ethical approval is not required for this systematic review since data from published studies will be used. This review is expected to provide a better understanding of the effect of early intervention for PTSS following musculoskeletal trauma. Findings of this review will be disseminated in peer-reviewed publications and through national and international conferences.

PROSPERO REGISTRATION NUMBER

CRD42022333905.

Successful recovery following musculoskeletal trauma: protocol for a qualitative study of patients' and physiotherapists' perceptions

Background

Annually in the UK, 40,000–90,000 people are involved in a traumatic incident. Severity of injury and how well people recover from their injuries varies, with physiotherapy playing a key role in the rehabilitation process. Recovery is evaluated using multiple outcome measures for perceived levels of pain severity and quality of life. It is unclear however, what constitutes a successful recovery from injury throughout the course of recovery from the patient perspective, and whether this aligns with physiotherapists’ perspectives.

Methods

A qualitative study using two approaches: Interpretive Phenomenological Analysis (IPA) using semi-structured interviews and thematic analysis following the Kreuger framework for focus groups. A purposive sample of 20 patients who have experienced musculoskeletal trauma within the past 4 weeks and 12 physiotherapists who manage this patient population will be recruited from a single trauma centre in the UK. Semi-structured interviews with patients at 4 weeks, 6 and 12 months following injury, and 2 focus groups with physiotherapists will be undertaken at one time point. Views and perceptions on the definition of recovery and what constitutes a successful recovery will be explored using both methods, with a focus on the lived experience and patient journey following musculoskeletal trauma, and how this changes through the process of recovery. Data from both the semi-structured interviews and focus groups will be analysed separately and then integrated and synthesised into key themes ensuring similarities and differences are identified. Strategies to ensure trustworthiness e.g., reflexivity will be employed.

Discussion

Recovery following musculoskeletal trauma is complex and understanding of the concept of successful recovery and how this changes over time following an injury is largely unknown. It is imperative to understand the patient perspective and whether these perceptions align with current views of physiotherapists. A greater understanding of recovery following musculoskeletal trauma has potential to change clinical care, optimise patient centred care and improve efficiency and clinical decision making during rehabilitation. This in turn can contribute to improved clinical effectiveness, patient outcome and patient satisfaction with potential service and economic cost savings. This study has ethical approval (IRAS 287781/REC 20/PR/0712).

Supplementary Information

The online version contains supplementary material available at 10.1186/s12891-021-04035-9.

Machine learning combining CT findings and clinical parameters improves prediction of length of stay and ICU admission in torso trauma

OBJECTIVE

To develop machine learning (ML) models capable of predicting ICU admission and extended length of stay (LOS) after torso (chest, abdomen, or pelvis) trauma, by using clinical and/or imaging data.

MATERIALS AND METHODS

This was a retrospective study of 840 adult patients admitted to a level 1 trauma center after injury to the torso over the course of 1 year. Clinical parameters included age, sex, vital signs, clinical scores, and laboratory values. Imaging data consisted of any injury present on CT. The two outcomes of interest were ICU admission and extended LOS, defined as more than the median LOS in the dataset. We developed and tested artificial neural network (ANN) and support vector machine (SVM) models, and predictive performance was evaluated by area under the receiver operating characteristic (ROC) curve (AUC).

RESULTS

The AUCs of SVM and ANN models to predict ICU admission were up to 0.87 ± 0.03 and 0.78 ± 0.02, respectively. The AUCs of SVM and ANN models to predict extended LOS were up to 0.80 ± 0.04 and 0.81 ± 0.05, respectively. Predictions based on imaging alone or imaging with clinical parameters were consistently more accurate than those based solely on clinical parameters.

CONCLUSIONS

The best performing models incorporated imaging findings and outperformed those with clinical findings alone. ML models have the potential to help predict outcomes in trauma by integrating clinical and imaging findings, although further research may be needed to optimize their performance.

KEY POINTS

• Artificial neural network and support vector machine-based models were used to predict the intensive care unit admission and extended length of stay after trauma to the torso. • Our input data consisted of clinical parameters and CT imaging findings derived from radiology reports, and we found that combining the two significantly enhanced the prediction of both outcomes with either model. • The highest accuracy (83%) and highest area under the receiver operating characteristic curve (0.87) were obtained for artificial neural networks and support vector machines, respectively, by combining clinical and imaging features in the prediction of intensive care unit admission.

Pull back the curtain: External data validation is an essential element of quality improvement benchmark reporting

BACKGROUND

Accurate and reliable data are pivotal to credible risk-adjusted modeling and hospital benchmarking. Evidence assessing the reliability and accuracy of data elements considered as variables in risk-adjustment modeling and measurement of outcomes is lacking. This deficiency holds the potential to compromise benchmarking integrity. We detail the findings of a longitudinal program to evaluate the impact of external data validation on data validity and reliability for variables utilized in benchmarking of trauma centers.

METHODS

A collaborative quality initiative-based study was conducted of 29 trauma centers from March 2010 through December 2018. Case selection criteria were applied to identify high-yield cases that were likely to challenge data abstractors. There were 127,238 total variables validated (i.e., reabstracted, compared, and reported to trauma centers). Study endpoints included data accuracy (agreement between registry data and contemporaneous documentation) and reliability (consistency of accuracy within and between hospitals). Data accuracy was assessed by mean error rate and type (under capture, inaccurate capture, or over capture). Cohen's kappa estimates were calculated to evaluate reliability.

RESULTS

There were 185,120 patients that met the collaborative inclusion criteria. There were 1,243 submissions reabstracted. The initial validation visit demonstrated the highest mean error rate at 6.2% ± 4.7%, and subsequent validation visits demonstrated a statistically significant decrease in error rate compared with the first visit (p < 0.05). The mean hospital error rate within the collaborative steadily improved over time (2010, 8.0%; 2018, 3.2%) compared with the first year (p < 0.05). Reliability of substantial or higher (kappa ≥0.61) was demonstrated in 90% of the 20 comorbid conditions considered in the benchmark risk-adjustment modeling, 39% of these variables exhibited a statistically significant (p < 0.05) interval decrease in error rate from the initial visit.

CONCLUSION

Implementation of an external data validation program is correlated with increased data accuracy and reliability. Improved data reliability both within and between trauma centers improved risk-adjustment model validity and quality improvement program feedback.

Cross-validation of two prognostic trauma scores in severely injured patients

Introduction

Trauma scoring systems are important tools for outcome prediction and severity adjustment that informs trauma quality assessment and research. Discrimination and precision of such systems is tested in validation studies. The German TraumaRegister DGU^® (TR-DGU) and the Trauma Audit and Research Network (TARN) from the UK agreed on a cross-validation study to validate their prediction scores (RISC II and PS14, respectively).

Methods

Severe trauma patients with an Injury Severity Score (ISS) ≥ 9 documented in 2015 and 2016 were selected in both registries (primary admissions only). The predictive scores from each registry were applied to the selected data sets. Observed and predicted mortality were compared to assess precision; area under the receiver operating characteristic curve was used for discrimination. Hosmer–Lemeshow statistic was calculated for calibration. A subgroup analysis including patients treated in intensive care unit (ICU) was also carried out.

Results

From TR-DGU, 40,638 patients were included (mortality 11.7%). The RISC II predicted mortality was 11.2%, while PS14 predicted 16.9% mortality. From TARN, 64,622 patients were included (mortality 9.7%). PS14 predicted 10.6% mortality, while RISC II predicted 17.7%. Despite the identical cutoff of ISS ≥ 9, patient groups from both registries showed considerable difference in need for intensive care (88% versus 18%). Subgroup analysis of patients treated on ICU showed nearly identical values for observed and predicted mortality using RISC II.

Discussion

Each score performed well within its respective registry, but when applied to the other registry a decrease in performance was observed. Part of this loss of performance could be explained by different development data sets: the RISC II is mainly based on patients treated in an ICU, while the PS14 includes cases mainly cared for outside ICU with more moderate injury severity. This is according to the respective inclusion criteria of the two registries.

Conclusion

External validations of prediction models between registries are needed, but may show that prediction models are not fully transferable to other health-care settings.

A novel scoring system to predict the requirement for surgical intervention in victims of motor vehicle crashes: Development and validation using independent cohorts

Background

Given that there are still considerable number of facilities which lack surgical specialists round the clock across the world, the ability to estimate the requirement for emergency surgery in victims of motor vehicle crashes (MVCs) can ensure appropriate resource allocation. In this study, a surgical intervention in victims of MVC (SIM) score was developed and validated, using independent patient cohorts.

Methods

We retrospectively identified MVC victims in a nationwide trauma registry (2004–2016). Adults ≥ 15 years who presented with palpable pulse were included. Patients with missing data on the type/date of surgery were excluded. Patient were allocated to development or validation cohorts based on the date of injury. After missing values were imputed, predictors of the need for emergency thoracotomy and/or laparotomy were identified with multivariate logistic regression, and scores were then assigned using odds ratios. The SIM score was validated with area under the receiver operating characteristic curve (AUROC) and calibration plots of SIM score-derived probability and observed rates of emergency surgery.

Results

We assigned 13,328 and 12,348 patients to the development and validation cohorts, respectively. Age, motor vehicle collision and vital signs on hospital arrival were identified as independent predictors for emergency thoracotomy and/or laparotomy, and SIM score was developed as 0–9 scales. The score has a good discriminatory power (AUROC = 0.79; 95% confidence interval = 0.77–0.81), and both estimated and observed rates of emergency surgery increased stepwise from 1% at a score ≤ 1 to almost 40% at a score ≥ 8 with linear calibration plots.

Conclusions

The SIM score was developed and validated to accurately estimate the need for emergent thoracotomy and/or laparotomy in MVC victims.

Practice Variation in Vena Cava Filter Use Among Trauma Centers in the National Trauma Database

BACKGROUND

Agreement regarding indications for vena cava filter (VCF) utilization in trauma patients has been in flux since the filter's introduction. As VCF technology and practice guidelines have evolved, the use of VCF in trauma patients has changed. This study examines variation in VCF placement among trauma centers.

MATERIALS AND METHODS

A retrospective study was performed using data from the National Trauma Data Bank (2005-2014). Trauma centers were grouped according to whether they placed VCFs during the study period (VCF+/VCF-). A multivariable probit regression model was fit to predict the number of VCFs used among the VCF+ centers (the expected [E] number of VCF per center). The ratio of observed VCF placement (O) to expected VCFs (O:E) was computed and rank ordered to compare interfacility practice variation.

RESULTS

In total, 65,482 VCFs were placed by 448 centers. Twenty centers (4.3%) placed no VCFs. The greatest predictors of VCF placement were deep vein thrombosis, spinal cord paralysis, and major procedure. The strongest negative predictor of VCF placement was admission during the year 2014. Among the VCF+ centers, O:E varied by nearly 500%. One hundred fifty centers had an O:E greater than one. One hundred sixty-nine centers had an O:E less than one.

CONCLUSIONS

Substantial variation in practice is present in VCF placement. This variation cannot be explained only by the characteristics of the patients treated at these centers but could be also due to conflicting guidelines, changing evidence, decreasing reimbursement rates, or the culture of trauma centers.