A systematic review finds methodological improvements necessary for prognostic models in determining traumatic brain injury outcomes

Influence of Sociodemographic, Premorbid, and Injury-Related Factors on Post-Traumatic Stress, Anxiety, and Depression after Traumatic Brain Injury

Psychopathological symptoms are common sequelae after traumatic brain injury (TBI), leading to increased personal and societal burden. Previous studies on factors influencing Post-traumatic Stress Disorder (PTSD), Generalized Anxiety Disorder (GAD), and Major Depressive Disorder (MDD) after TBI have produced inconclusive results, partly due to methodological limitations. The current study investigated the influence of commonly proposed factors on the clinical impairment, occurrence, frequency, and intensity of symptoms of PTSD, GAD, and MDD after TBI. The study sample comprised 2069 individuals (65% males). Associations between psychopathological outcomes and sociodemographic, premorbid, and injury-related factors were analyzed using logistic regression, standard, and zero-inflated negative binomial models. Overall, individuals experienced moderate levels of PTSD, GAD, and MDD. Outcomes correlated with early psychiatric assessments across domains. The clinical impairment, occurrence, frequency, and intensity of all outcomes were associated with the educational level, premorbid psychiatric history, injury cause, and functional recovery. Distinct associations were found for injury severity, LOC, and clinical care pathways with PTSD; age and LOC:sex with GAD; and living situation with MDD, respectively. The use of suitable statistical models supported the identification of factors associated with the multifactorial etiology of psychopathology after TBI. Future research may apply these models to reduce personal and societal burden.

Glasgow coma scale pupil score (GCS-P) and the hospital mortality in severe traumatic brain injury: analysis of 1,066 Brazilian patients

BACKGROUND

 Pupil reactivity and the Glasgow Coma Scale (GCS) score are the most clinically relevant information to predict the survival of traumatic brain injury (TBI) patients.

OBJECTIVE

 We evaluated the accuracy of the GCS-Pupil score (GCS-P) as a prognostic index to predict hospital mortality in Brazilian patients with severe TBI and compare it with a model combining GCS and pupil response with additional clinical and radiological prognostic factors.

METHODS

 Data from 1,066 patients with severe TBI from 5 prospective studies were analyzed. We determined the association between hospital mortality and the combination of GCS, pupil reactivity, age, glucose levels, cranial computed tomography (CT), or the GCS-P score by multivariate binary logistic regression.

RESULTS

 Eighty-five percent (n = 908) of patients were men. The mean age was 35 years old, and the overall hospital mortality was 32.8%. The area under the receiver operating characteristic curve (AUROC) was 0.73 (0.70-0.77) for the model using the GCS-P score and 0.80 (0.77-0.83) for the model including clinical and radiological variables. The GCS-P score showed similar accuracy in predicting the mortality reported for the patients with severe TBI derived from the International Mission for Prognosis and Clinical Trials in TBI (IMPACT) and the Corticosteroid Randomization After Significant Head Injury (CRASH) studies.

CONCLUSION

 Our results support the external validation of the GCS-P to predict hospital mortality following a severe TBI. The predictive value of the GCS-P for long-term mortality, functional, and neuropsychiatric outcomes in Brazilian patients with mild, moderate, and severe TBI deserves further investigation.

Mechanical power of ventilation is associated with mortality in neurocritical patients: a cohort study

This study aimed to determine the predictive relevance of mechanical power in the clinical outcomes (such as ICU mortality, hospital mortality, 90-day mortality, length of ICU stay, and number of ventilator-free days at day 28) of neurocritical patients. This is a retrospective cohort analysis of an open-access clinical database known as MIMIC-III. The study included patients who had sustained an acute brain injury and required invasive ventilation for at least 24 h. Demographic parameters, disease severity scores (Glasgow coma scale), comorbidities, vital signs, laboratory parameters and ventilator parameters were collected within the first 24 h of ICU admission. The main outcome was the relationship between MP and ICU mortality. A total of 529 patients were selected for the study. The critical value of MP was 12.16 J/min, with the area under the curve (AUC) of the MP was 0.678 (95% CI 0.637-0.718), and compared to the GCS scores, the MP performed significantly better in discrimination (DeLong's test: p < 0.001). Among these patients elevated MP was associated to higher ICU mortality (OR 1.11; 95% CI 1.06-1.17; p < 0.001), enhanced the risk of hospital mortality, prolonged ICU stay, and decreased the number of ventilator-free days. In the subgroup analysis, high MP was associated with ICU mortality regardless of ARDS (OR 1.01, 95% CI 1.00-1.02, p = 0.009; OR 1.01, 95% CI 1.00-1.02, p = 0.018, respectively) or obesity (OR 1.01, 95% CI 1.00-1.02, p = 0.012; OR 1.01, 95% CI 1.01-1.02, p < 0.001, respectively). In neurocritical care patients undergoing invasive ventilation, elevated MP is linked to higher ICU mortality and a variety of other clinical outcomes.

Predicting mortality in moderate-severe TBI patients without early withdrawal of life-sustaining treatments including ICU complications: The MYSTIC-score

PURPOSE

To develop and internally validate the MortalitY in Moderate-Severe TBI plus ICU Complications (MYSTIC)-Score to predict in-hospital mortality of msTBI patients without early (<24 h) withdrawal-of-life-sustaining treatments.

METHODS

We analyzed data from a Neuro-Trauma Intensive Care Unit prospectively collected between 11/2009-5/2019. Consecutive adult msTBI patients were included if Glasgow Coma Scale≤12, and neither died nor had withdrawal-of-life-sustaining treatments within 24 h of admission (n = 485). Using univariate and multivariable logistic regression in a random-split cohort approach (2/3 derivation;1/3 validation), we identified independent predictors of in-hospital mortality while adjusting for validated predictors of mortality (IMPACT-variables). We constructed the MYSTIC-Score and examined discrimination and calibration.

RESULTS

The MYSTIC-Score included the ICU complications brain edema, herniation, systemic inflammatory response syndrome, sepsis, acute kidney injury, cardiac arrest, and urinary tract infection. In the derivation cohort(n = 324), discrimination and calibration were excellent (area-under-the-receiver-operating-curve [AUC-ROC] = 0.95;Hosmer-Lemeshow p-value = 0.09, with p > 0.05 indicating good calibration). Internal validation revealed an AUC-ROC = 0.93 and Hosmer-Lemeshow-p-value = 0.76 (n = 161).

CONCLUSIONS

Certain ICU complications are independent predictors of in-hospital mortality and strengthen outcome prediction in msTBI when combined with validated admission predictors of mortality. However, external validation is needed to determine robustness and practical applicability of our model given the high potential for residual confounders.

External validation of the Glasgow coma scale-pupils in low- to middle-income country patients with traumatic brain injury: Could “motor score-pupil” have higher prognostic value?

Background:

The objective of this study is to validate the admission Glasgow coma scale (GCS) associated with pupil response (GCS-P) to predict traumatic brain injury (TBI) patient’s outcomes in a low- to middle-income country and to compare its performance with that of a simplified model combining the better motor response of the GCS and the pupilar response (MS-P).

Methods:

This is a prospective cohort of patients with TBI in a tertiary trauma reference center in Brazil. Predictive values of the GCS, GCS-P, and MS-P were evaluated and compared for 14 day and in-hospital mortality outcomes and length of hospital stay (LHS).

Results:

The study enrolled 447 patients. MS-P demonstrated better discriminative ability than GCS to predict mortality (AUC 0.736 × 0.658; P < 0.001) and higher AUC than GCS-P (0.736 × 0.704, respectively; P = 0.073). For hospital mortality, MS-P demonstrated better discrimination than GCS (AUC, 0.750 × 0.682; P < 0.001) and higher AUC than GCS-P (0.750 × 0.714; P = 0.027). Both scores were good predictors of LHS (r2 = 0.084 [GCS-P] × 0.079 [GCS] × 0.072 [MS-P]).

Conclusion:

The predictive value of the GCS, GCS-P, and MS-P scales was demonstrated, thus contributing to its external validation in low- to middle-income country.

Modeling Brain-Heart Crosstalk Information in Patients with Traumatic Brain Injury

BACKGROUND

Traumatic brain injury (TBI) is an extremely heterogeneous and complex pathology that requires the integration of different physiological measurements for the optimal understanding and clinical management of patients. Information derived from intracranial pressure (ICP) monitoring can be coupled with information obtained from heart rate (HR) monitoring to assess the interplay between brain and heart. The goal of our study is to investigate events of simultaneous increases in HR and ICP and their relationship with patient mortality..

METHODS

In our previous work, we introduced a novel measure of brain-heart interaction termed brain-heart crosstalks (ctnp), as well as two additional brain-heart crosstalks indicators [mutual information ([Formula: see text]) and average edge overlap (ωct)] obtained through a complex network modeling of the brain-heart system. These measures are based on identification of simultaneous increase of HR and ICP. In this article, we investigated the relationship of these novel indicators with respect to mortality in a multicenter TBI cohort, as part of the Collaborative European Neurotrauma Effectiveness Research in TBI high-resolution work package.

RESULTS

A total of 226 patients with TBI were included in this cohort. The data set included monitored parameters (ICP and HR), as well as laboratory, demographics, and clinical information. The number of detected brain-heart crosstalks varied (mean 58, standard deviation 57). The Kruskal-Wallis test comparing brain-heart crosstalks measures of survivors and nonsurvivors showed statistically significant differences between the two distributions (p values: 0.02 for [Formula: see text], 0.005 for ctnp and 0.006 for ωct). An inverse correlation was found, computed using the point biserial correlation technique, between the three new measures and mortality: - 0.13 for ctnp (p value 0.04), - 0.19 for ωct (p value 0.002969) and - 0.09 for [Formula: see text] (p value 0.1396). The measures were then introduced into the logistic regression framework, along with a set of input predictors made of clinical, demographic, computed tomography (CT), and lab variables. The prediction models were obtained by dividing the original cohort into four age groups (16-29, 30-49, 50-65, and 65-85 years of age) to properly treat with the age confounding factor. The best performing models were for age groups 16-29, 50-65, and 65-85, with the deviance of ratio explaining more than 80% in all the three cases. The presence of an inverse relationship between brain-heart crosstalks and mortality was also confirmed.

CONCLUSIONS

The presence of a negative relationship between mortality and brain-heart crosstalks indicators suggests that a healthy brain-cardiovascular interaction plays a role in TBI.

Does poor methodological quality of prediction modeling studies translate to poor model performance? An illustration in traumatic brain injury

BACKGROUND

Prediction modeling studies often have methodological limitations, which may compromise model performance in new patients and settings. We aimed to examine the relation between methodological quality of model development studies and their performance at external validation.

METHODS

We systematically searched for externally validated multivariable prediction models that predict functional outcome following moderate or severe traumatic brain injury. Risk of bias and applicability of development studies was assessed with the Prediction model Risk Of Bias Assessment Tool (PROBAST). Each model was rated for its presentation with sufficient detail to be used in practice. Model performance was described in terms of discrimination (AUC), and calibration. Delta AUC (dAUC) was calculated to quantify the percentage change in discrimination between development and validation for all models. Generalized estimation equations (GEE) were used to examine the relation between methodological quality and dAUC while controlling for clustering.

RESULTS

We included 54 publications, presenting ten development studies of 18 prediction models, and 52 external validation studies, including 245 unique validations. Two development studies (four models) were found to have low risk of bias (RoB). The other eight publications (14 models) showed high or unclear RoB. The median dAUC was positive in low RoB models (dAUC 8%, [IQR - 4% to 21%]) and negative in high RoB models (dAUC - 18%, [IQR - 43% to 2%]). The GEE showed a larger average negative change in discrimination for high RoB models (- 32% (95% CI: - 48 to - 15) and unclear RoB models (- 13% (95% CI: - 16 to - 10)) compared to that seen in low RoB models.

CONCLUSION

Lower methodological quality at model development associates with poorer model performance at external validation. Our findings emphasize the importance of adherence to methodological principles and reporting guidelines in prediction modeling studies.

Methodological conduct of prognostic prediction models developed using machine learning in oncology: a systematic review

BACKGROUND

Describe and evaluate the methodological conduct of prognostic prediction models developed using machine learning methods in oncology.

METHODS

We conducted a systematic review in MEDLINE and Embase between 01/01/2019 and 05/09/2019, for studies developing a prognostic prediction model using machine learning methods in oncology. We used the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) statement, Prediction model Risk Of Bias ASsessment Tool (PROBAST) and CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies (CHARMS) to assess the methodological conduct of included publications. Results were summarised by modelling type: regression-, non-regression-based and ensemble machine learning models.

RESULTS

Sixty-two publications met inclusion criteria developing 152 models across all publications. Forty-two models were regression-based, 71 were non-regression-based and 39 were ensemble models. A median of 647 individuals (IQR: 203 to 4059) and 195 events (IQR: 38 to 1269) were used for model development, and 553 individuals (IQR: 69 to 3069) and 50 events (IQR: 17.5 to 326.5) for model validation. A higher number of events per predictor was used for developing regression-based models (median: 8, IQR: 7.1 to 23.5), compared to alternative machine learning (median: 3.4, IQR: 1.1 to 19.1) and ensemble models (median: 1.7, IQR: 1.1 to 6). Sample size was rarely justified (n = 5/62; 8%). Some or all continuous predictors were categorised before modelling in 24 studies (39%). 46% (n = 24/62) of models reporting predictor selection before modelling used univariable analyses, and common method across all modelling types. Ten out of 24 models for time-to-event outcomes accounted for censoring (42%). A split sample approach was the most popular method for internal validation (n = 25/62, 40%). Calibration was reported in 11 studies. Less than half of models were reported or made available.

CONCLUSIONS

The methodological conduct of machine learning based clinical prediction models is poor. Guidance is urgently needed, with increased awareness and education of minimum prediction modelling standards. Particular focus is needed on sample size estimation, development and validation analysis methods, and ensuring the model is available for independent validation, to improve quality of machine learning based clinical prediction models.

Prognostic Research in Traumatic Brain Injury: Markers, Modeling, and Methodological Principles

Prognostic assessment in traumatic brain injury (TBI) is embedded deeply in clinical care. Considering the limitations of current prognostic indicators, there is increasing interest in understanding the role of new biomarkers, and in finding other prognostic indicators of long-term outcomes following TBI. New prognostic indicators may result in the development of more accurate prediction models that could be useful for both risk stratification and clinical decision making. We aimed to review methodological issues and provide tentative guidelines for prognostic research in TBI. Prognostic factor research focuses on the role of a specific patient or disease-related characteristic in relation to outcome. Typically, univariable relations of the prognostic factor are studied, followed by analyses adjusting for other variables related to the outcome. Following existing guidelines, we emphasize the importance of transparent reporting of patient and specimen characteristics, study design, clinical end-points, and statistical analysis. Prognostic model research considers combinations of predictors, with challenges for model specification, estimation, evaluation, validation, and presentation. We highlight modern approaches and opportunities related to missing values, exploration of non-linear effects, and assessing between-study heterogeneity. Prognostic research in TBI can be improved if key methodological principles are adhered to and when research is performed in collaboration among multiple centers to ensure generalizability.

Early diagnosis of mortality using admission CT perfusion in severe traumatic brain injury patients (ACT-TBI): protocol for a prospective cohort study

INTRODUCTION

Severe traumatic brain injury (TBI) is a catastrophic neurological condition with significant economic burden. Early in-hospital mortality (<48 hours) with severe TBI is estimated at 50%. Several clinical examinations exist to determine brain death; however, most are difficult to elicit in the acute setting in patients with severe TBI. Having a definitive assessment tool would help predict early in-hospital mortality in this population. CT perfusion (CTP) has shown promise diagnosing early in-hospital mortality in patients with severe TBI and other populations. The purpose of this study is to validate admission CTP features of brain death relative to the clinical examination outcome for characterizing early in-hospital mortality in patients with severe TBI.

METHODS AND ANALYSIS

The Early Diagnosis of Mortality using Admission CT Perfusion in Severe Traumatic Brain Injury Patients study, is a prospective cohort study in patients with severe TBI funded by a grant from the Canadian Institute of Health Research. Adults aged 18 or older, with evidence of a severe TBI (Glasgow Coma Scale score ≤8 before initial resuscitation) and, on mechanical ventilation at the time of imaging are eligible. Patients will undergo CTP at the time of first imaging on their hospital admission. Admission CTP compares with the reference standard of an accepted bedside clinical assessment for brainstem function. Deferred consent will be used. The primary outcome is a binary outcome of mortality (dead) or survival (not dead) in the first 48 hours of admission. The planned sample size for achieving a sensitivity of 75% and a specificity of 95% with a CI of ±5% is 200 patients.

ETHICS AND DISSEMINATION

This study has been approved by the University of Manitoba Health Research Ethics Board. The findings from our study will be disseminated through peer-reviewed journals and presentations at local rounds, national and international conferences. The public will be informed through forums at the end of the study.

TRIAL REGISTRATION NUMBER

NCT04318665.

Improving patient selection for immuno-oncology phase 1 trials: External validation of six prognostic scores in a French Cancer Center

We compared the performance of six prognostic scores (Royal Marsden Hospital, MDACC: MD Anderson Clinical Center and MDACC + NLR: neutrophil-to-lymphocyte ratio, MD Anderson - immune checkpoint inhibitors (MDA-ICI), GRIm: Gustave Roussy Immune Score and LIPI: Lung Immune Prognostic Index) in predicting overall survival (OS) in phase I trial patients treated with immune checkpoint inhibitors (ICI). Medical records of patients with advanced solid tumors enrolled in ICI phase I trials between 2015 and 2018 at Institut Universitaire du Cancer de Toulouse-Oncopole were reviewed. The performance of prognostic scores on OS was compared using different criteria. A total of 259 patients were included. Median age was 63 years (range: 18-83). Main primary cancers were melanoma (19%), head and neck (16%), lung (13%) and bladder (10%). With a median follow-up of 15 months (95% confidence interval [CI] = [11.6;17.5]), median OS was 12.5 months (95% CI = [10.3;16.0]). All scores were associated with OS. The MDACC, LIPI and GRIm scores performed better than the others. Concordance of risk group assignment between the scoring systems was poor. According to our results, the MDACC, GRIm and LIPI scores better suited to ICI phase I settings. Adequate scoring would allow better patient selection in early ICI trials, especially during the critical period of dose escalation, and in proof-of-concept expansion cohorts.

Development of a Prognostic Model to Predict Mortality after Traumatic Brain Injury in Intensive Care Setting in a Developing Country

Objectives  We aimed to develop a prognostic model for the prediction of in-hospital mortality in patients with traumatic brain injury (TBI) admitted to the neurosurgery intensive care unit (ICU) of our institute.

Materials and Methods  The clinical and computed tomography scan data of consecutive patients admitted after a diagnosis TBI in ICU were reviewed. Construction of the model was done by using all the variables of Corticosteroid Randomization after Significant Head Injury and International Mission on Prognosis and Analysis of Clinical Trials in TBI models. The endpoint was in-hospital mortality.

Results  A total of 243 patients with TBI were admitted to ICU during the study period. The in-hospital mortality was 15.3%. On multivariate analysis, the Glasgow coma scale (GCS) at admission, hypoxia, hypotension, and obliteration of the third ventricle/basal cisterns were significantly associated with mortality. Patients with hypoxia had eight times, with hypotensions 22 times, and with obliteration of the third ventricle/basal cisterns three times more chance of death. The TBI score was developed as a sum of individual points assigned as follows: GCS score 3 to 4 (+2 points), 5 to 12 (+1), hypoxia (+1), hypotension (+1), and obliteration third ventricle/basal cistern (+1). The mortality was 0% for a score of “0” and 85% for a score of “4.”

Conclusion  The outcome of patients treated in ICU was based on common admission variables. A simple clinical grading score allows risk stratification of patients with TBI admitted in ICU.

Development and internal validation of China mortality prediction model in trauma based on ICD-10-CM lexicon: CMPMIT-ICD10

Background:

Models to predict mortality in trauma play an important role in outcome prediction and severity adjustment, which informs trauma quality assessment and research. Hospitals in China typically use the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) to describe injury. However, there is no suitable prediction model for China. This study attempts to develop a new mortality prediction model based on the ICD-10-CM lexicon and a Chinese database.

Methods:

This retrospective study extracted the data of all trauma patients admitted to the Beijing Red Cross Emergency Center, from January 2012 to July 2018 (n = 40,205). We used relevant predictive variables to establish a prediction model following logistic regression analysis. The performance of the model was assessed based on discrimination and calibration. The bootstrapping method was used for internal validation and adjustment of model performance.

Results:

Sex, age, new region-severity codes, comorbidities, traumatic shock, and coma were finally included in the new model as key predictors of mortality. Among them, coma and traumatic shock had the highest scores in the model. The discrimination and calibration of this model were significant, and the internal validation performance was good. The values of the area under the curve and Brier score for the new model were 0.9640 and 0.0177, respectively; after adjustment of the bootstrapping method, they were 0.9630 and 0.0178, respectively.

Conclusions:

The new model (China Mortality Prediction Model in Trauma based on the ICD-10-CM lexicon) showed great discrimination and calibration, and performed well in internal validation; it should be further verified externally.

Predicting in-hospital mortality after traumatic brain injury: External validation of CRASH-basic and IMPACT-core in the national trauma data bank

BACKGROUND

Traumatic brain injury (TBI) prognostic prediction models offer value to individualized treatment planning, systematic outcome assessments and clinical research design but require continuous external validation to ensure generalizability to different settings. The Corticosteroid Randomization After Significant Head Injury (CRASH) and International Mission on Prognosis and Analysis on Clinical Trials in TBI (IMPACT) models are widely available but lack robust assessments of performance in a current national sample of patients. The purpose of this study is to assess the performance of the CRASH-Basic and IMPACT-Core models in predicting in-hospital mortality using a nationwide retrospective cohort from the National Trauma Data Bank (NTDB).

METHODS

The 2016 NTDB was used to analyze an adult cohort with moderate-severe TBI (Glasgow Coma Scale [GCS] ≤ 12, head Abbreviated Injury Scale of 2-6). Observed in-hospital mortality or discharge to hospice was compared to the CRASH-Basic and IMPACT-Core models' predicted probability of 14-day or 6-month mortality, respectively. Performance measures included discrimination (area under the receiver operating characteristic curve [AUC]) and calibration (calibration plots and Brier scores). Further sensitivity analysis included patients with GCS ≤ 14 and considered patients discharged to hospice to be alive at 14-days.

RESULTS

A total of 26,228 patients were included in this study. Both models demonstrated good ability in differentiating between patients who died and those who survived, with IMPACT demonstrating a marginally greater AUC (0.863; 95% CI: 0.858 - 0.867) than CRASH (0.858; 0.854 - 0.863); p < 0.001. On calibration, IMPACT overpredicted at lower scores and underpredicted at higher scores but had good calibration-in-the-large (indicating no systemic over/underprediction), while CRASH consistently underpredicted mortality. Brier scores were similar (0.152 for IMPACT, 0.162 for CRASH; p < 0.001). Both models showed slight improvement in performance when including patients with GCS ≤ 14.

CONCLUSION

Both CRASH-Basic and IMPACT-Core accurately predict in-hospital mortality following moderate-severe TBI, and IMPACT-Core performs well beyond its original GCS cut-off of 12, indicating potential utility for mild TBI (GCS 13-15). By demonstrating validity in the NTDB, these models appear generalizable to new data and offer value to current practice in diverse settings as well as to large-scale research design.

Design, methods, and reporting of impact studies of cardiovascular clinical prediction rules are suboptimal: a systematic review

OBJECTIVES

To evaluate design, methods, and reporting of impact studies of cardiovascular clinical prediction rules (CPRs).

STUDY DESIGN AND SETTING

We conducted a systematic review. Impact studies of cardiovascular CPRs were identified by forward citation and electronic database searches. We categorized the design of impact studies as appropriate for randomized and nonrandomized experiments, excluding uncontrolled before-after study. For impact studies with appropriate study design, we assessed the quality of methods and reporting. We compared the quality of methods and reporting between impact and matched control studies.

RESULTS

We found 110 impact studies of cardiovascular CPRs. Of these, 65 (59.1%) used inappropriate designs. Of 45 impact studies with appropriate design, 31 (68.9%) had substantial risk of bias. Mean number of reporting domains that impact studies with appropriate study design adhered to was 10.2 of 21 domains (95% confidence interval, 9.3 and 11.1). The quality of methods and reporting was not clearly different between impact and matched control studies.

CONCLUSION

We found most impact studies either used inappropriate study design, had substantial risk of bias, or poorly complied with reporting guidelines. This appears to be a common feature of complex interventions. Users of CPRs should critically evaluate evidence showing the effectiveness of CPRs.

Development and temporal external validation of a simple risk score tool for prediction of outcomes after severe head injury based on admission characteristics from level-1 trauma centre of India using retrospectively collected data

OBJECTIVE

To develop and validate a simple risk scores chart to estimate the probability of poor outcomes in patients with severe head injury (HI).

DESIGN

Retrospective.

SETTING

Level-1, government-funded trauma centre, India.

PARTICIPANTS

Patients with severe HI admitted to the neurosurgery intensive care unit during 19 May 2010-31 December 2011 (n=946) for the model development and further, data from same centre with same inclusion criteria from 1 January 2012 to 31 July 2012 (n=284) for the external validation of the model.

OUTCOMES

In-hospital mortality and unfavourable outcome at 6 months.

RESULTS

A total of 39.5% and 70.7% had in-hospital mortality and unfavourable outcome, respectively, in the development data set. The multivariable logistic regression analysis of routinely collected admission characteristics revealed that for in-hospital mortality, age (51-60, >60 years), motor score (1, 2, 4), pupillary reactivity (none), presence of hypotension, basal cistern effaced, traumatic subarachnoid haemorrhage/intraventricular haematoma and for unfavourable outcome, age (41-50, 51-60, >60 years), motor score (1-4), pupillary reactivity (none, one), unequal limb movement, presence of hypotension were the independent predictors as its 95% confidence interval (CI) of odds ratio (OR)_did not contain one. The discriminative ability (area under the receiver operating characteristic curve (95% CI)) of the score chart for in-hospital mortality and 6 months outcome was excellent in the development data set (0.890 (0.867 to 912) and 0.894 (0.869 to 0.918), respectively), internal validation data set using bootstrap resampling method (0.889 (0.867 to 909) and 0.893 (0.867 to 0.915), respectively) and external validation data set (0.871 (0.825 to 916) and 0.887 (0.842 to 0.932), respectively). Calibration showed good agreement between observed outcome rates and predicted risks in development and external validation data set (p>0.05).

CONCLUSION

For clinical decision making, we can use of these score charts in predicting outcomes in new patients with severe HI in India and similar settings.

Outcome Prediction after Moderate and Severe Traumatic Brain Injury: External Validation of Two Established Prognostic Models in 1742 European Patients

The International Mission on Prognosis and Analysis of Clinical Trials in Traumatic Brain Injury (IMPACT) and Corticoid Randomisation After Significant Head injury (CRASH) prognostic models predict functional outcome after moderate and severe traumatic brain injury (TBI). We aimed to assess their performance in a contemporary cohort of patients across Europe. The Collaborative European NeuroTrauma Effectiveness Research in Traumatic Brain Injury (CENTER-TBI) core study is a prospective, observational cohort study in patients presenting with TBI and an indication for brain computed tomography. The CENTER-TBI core cohort consists of 4509 TBI patients available for analyses from 59 centers in 18 countries across Europe and Israel. The IMPACT validation cohort included 1173 patients with GCS ≤12, age ≥14, and 6-month Glasgow Outcome Scale-Extended (GOSE) available. The CRASH validation cohort contained 1742 patients with GCS ≤14, age ≥16, and 14-day mortality or 6-month GOSE available. Performance of the three IMPACT and two CRASH model variants was assessed with discrimination (area under the receiver operating characteristic curve; AUC) and calibration (comparison of observed vs. predicted outcome rates). For IMPACT, model discrimination was good, with AUCs ranging between 0.77 and 0.85 in 1173 patients and between 0.80 and 0.88 in the broader CRASH selection (n = 1742). For CRASH, AUCs ranged between 0.82 and 0.88 in 1742 patients and between 0.66 and 0.80 in the stricter IMPACT selection (n = 1173). Calibration of the IMPACT and CRASH models was generally moderate, with calibration-in-the-large and calibration slopes ranging between -2.02 and 0.61 and between 0.48 and 1.39, respectively. The IMPACT and CRASH models adequately identify patients at high risk for mortality or unfavorable outcome, which supports their use in research settings and for benchmarking in the context of quality-of-care assessment.

Prediction of in-hospital mortality in patients on mechanical ventilation post traumatic brain injury: machine learning approach

Background

The study aimed to introduce a machine learning model that predicts in-hospital mortality in patients on mechanical ventilation (MV) following moderate to severe traumatic brain injury (TBI).

Methods

A retrospective analysis was conducted for all adult patients who sustained TBI and were hospitalized at the trauma center from January 2014 to February 2019 with an abbreviated injury severity score for head region (HAIS) ≥ 3. We used the demographic characteristics, injuries and CT findings as predictors. Logistic regression (LR) and Artificial neural networks (ANN) were used to predict the in-hospital mortality. Accuracy, area under the receiver operating characteristics curve (AUROC), precision, negative predictive value (NPV), sensitivity, specificity and F-score were used to compare the models` performance.

Results

Across the study duration; 785 patients met the inclusion criteria (581 survived and 204 deceased). The two models (LR and ANN) achieved good performance with an accuracy over 80% and AUROC over 87%. However, when taking the other performance measures into account, LR achieved higher overall performance than the ANN with an accuracy and AUROC of 87% and 90.5%, respectively compared to 80.9% and 87.5%, respectively. Venous thromboembolism prophylaxis, severity of TBI as measured by abbreviated injury score, TBI diagnosis, the need for blood transfusion, heart rate upon admission to the emergency room and patient age were found to be the significant predictors of in-hospital mortality for TBI patients on MV.

Conclusions

Machine learning based LR achieved good predictive performance for the prognosis in mechanically ventilated TBI patients. This study presents an opportunity to integrate machine learning methods in the trauma registry to provide instant clinical decision-making support.

Systematic evaluation, verification and comparison of tuberculosis-related non-coding RNA diagnostic panels

We systematically summarized tuberculosis (TB)‐related non‐coding RNA (ncRNA) diagnostic panels, validated and compared panel performance. We searched TB‐related ncRNA panels in PubMed, OVID and Web of Science up to 28 February 2020, and available datasets in GEO, SRA and EBI ArrayExpress up to 1 March 2020. We rebuilt models and synthesized the results of each model in validation sets by bivariate mixed models. Specificity at 90% sensitivity, area under curve (AUC) and inconsistence index (I²) were calculated. NcRNA biofunctions were analysed. Nineteen models based on 18 ncRNA panels (miRNA, lncRNA, circRNA and snoRNA panels) and 18 datasets were included. Limited available datasets only allowed to evaluate miRNA panels further. Cui 2017 and Latorre 2015 exhibited specificity >70% at 90% sensitivity and AUC >80% in all validation sets. Cui 2017 showed higher specificity at 90% sensitivity (92%) and AUC (95%) and lower heterogeneity (I² = 0%) in ethological‐confirmation validation sets. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes analysis indicated that most ncRNAs in panels involved in immune cell activation, oxidative stress, and Wnt and MAPK signalling pathway. Cui 2017 outperformed other models in both all available and aetiological‐confirmed validation sets, meeting the criteria of target product profile of WHO. This work provided a basis for clinical choice of TB‐related ncRNA diagnostic panels to a certain extent.

Prediction of Global Functional Outcome and Post-Concussive Symptoms after Mild Traumatic Brain Injury: External Validation of Prognostic Models in the Collaborative European NeuroTrauma Effectiveness Research in Traumatic Brain Injury (CENTER-TBI) Study

The majority of traumatic brain injuries (TBIs) are categorized as mild, according to a baseline Glasgow Coma Scale (GCS) score of 13-15. Prognostic models that were developed to predict functional outcome and persistent post-concussive symptoms (PPCS) after mild TBI have rarely been externally validated. We aimed to externally validate models predicting 3-12-month Glasgow Outcome Scale Extended (GOSE) or PPCS in adults with mild TBI. We analyzed data from the Collaborative European NeuroTrauma Effectiveness Research in Traumatic Brain Injury (CENTER-TBI) project, which included 2862 adults with mild TBI, with 6-month GOSE available for 2374 and Rivermead Post-Concussion Symptoms Questionnaire (RPQ) results available for 1605 participants. Model performance was evaluated based on calibration (graphically and characterized by slope and intercept) and discrimination (C-index). We validated five published models for 6-month GOSE and three for 6-month PPCS scores. The models used different cutoffs for outcome and some included symptoms measured 2 weeks post-injury. Discriminative ability varied substantially (C-index between 0.58 and 0.79). The models developed in the Corticosteroid Randomisation After Significant Head Injury (CRASH) trial for prediction of GOSE <5 discriminated best (C-index 0.78 and 0.79), but were poorly calibrated. The best performing models for PPCS included 2-week symptoms (C-index 0.75 and 0.76). In conclusion, none of the prognostic models for early prediction of GOSE and PPCS has both good calibration and discrimination in persons with mild TBI. In future studies, prognostic models should be tailored to the population with mild TBI, predicting relevant end-points based on readily available predictors.

Pre-injury health status and excess mortality in persons with traumatic brain injury: A decade-long historical cohort study

An increasing number of patients are able to survive traumatic brain injuries (TBIs) with advanced resuscitation. However, the role of their pre-injury health status in mortality in the following years is not known. Here, we followed 77,088 consecutive patients (59% male) who survived the TBI event in Ontario, Canada for more than a decade, and examined the relationships between their pre-injury health status and mortality rates in excess to the expected mortality calculated using sex- and age-specific life tables. There were 5792 deaths over the studied period, 3163 (6.95%) deaths in male and 2629 (8.33%) in female patients. The average excess mortality rate over the follow-up period of 14 years was 1.81 (95% confidence interval = 1.76-1.86). Analyses of follow-up time windows showed different patterns for the average excess rate of mortality following TBI, with the greatest rates observed in year one after injury. Among identified pre-injury comorbidity factors, 33 were associated with excess mortality rates. These rates were comparable between sexes. Additional analyses in the validation dataset confirmed that these findings were unlikely a result of TBI misclassification or unmeasured confounding. Thus, detection and subsequent management of pre-injury health status should be an integral component of any strategy to reduce excess mortality in TBI patients. The complexity of pre-injury comorbidity calls for integration of multidisciplinary health services to meet TBI patients' needs and prevent adverse outcomes.

Reliability of the intracerebral hemorrhage score for predicting outcome in patients with intracerebral hemorrhage using oral anticoagulants

BACKGROUND AND PURPOSE

The intracerebral hemorrhage (ICH) score is the most widely used and validated prognostic model for estimating 30-day mortality in ICH. However, the score was developed and validated in an ICH population probably not using oral anticoagulants (OACs). The aim of this study was to determine the performance of the ICH score for predicting the 30-day mortality rate in the full range of ICH scores in patients using OACs.

METHODS

Data from admitted patients with ICH were collected retrospectively in two Dutch comprehensive stroke centers. The validity of the ICH score was evaluated by assessing both discrimination and calibration in OAC and OAC-naive patient groups.

RESULTS

A total of 1752 patients were included of which 462 (26%) patients were on OAC. The 30-day mortality was 54% for the OAC cohort and 34% for the OAC-naive cohort. The 30-day mortality was higher in the OAC cohort for ICH score 1 (33% vs. 12.5%; odds ratio, 3.4; 95% confidence intervals, 1.1-10.4) and ICH score 2 (53% vs. 26%; odds ratio, 3.2; 95% confidence intervals, 1.2-8.2) compared with the predicted mortality rate of the original ICH score. Overall, the discriminative ability of the ICH score was equally good in both cohorts (area under the curve 0.83 vs. 0.87, respectively).

CONCLUSIONS

The ICH score underestimated the 30-day mortality rate for lower ICH scores in OAC-ICH. When estimating the prognosis of ICH in patients using OAC, this underestimation of mortality must be taken into account.

Prediction of in-hospital mortality in patients with post traumatic brain injury using National Trauma Registry and Machine Learning Approach

BACKGROUND

The use of machine learning techniques to predict diseases outcomes has grown significantly in the last decade. Several studies prove that the machine learning predictive techniques outperform the classical multivariate techniques. We aimed to build a machine learning predictive model to predict the in-hospital mortality for patients who sustained Traumatic Brain Injury (TBI).

METHODS

Adult patients with TBI who were hospitalized in the level 1 trauma center in the period from January 2014 to February 2019 were included in this study. Patients' demographics, injury characteristics and CT findings were used as predictors. The predictive performance of Artificial Neural Networks (ANN) and Support Vector Machines (SVM) was evaluated in terms of accuracy, Area Under the Curve (AUC), sensitivity, precision, Negative Predictive Value (NPV), specificity and F-score.

RESULTS

A total of 1620 eligible patients were included in the study (1417 survival and 203 non-survivals). Both models achieved accuracy over 91% and AUC over 93%. SVM achieved the optimal performance with accuracy 95.6% and AUC 96%.

CONCLUSIONS

for prediction of mortality in patients with TBI, SVM outperformed the well-known classical models that utilized the conventional multivariate analytical techniques.

Prospective prediction of suicide attempts in community adolescents and young adults, using regression methods and machine learning

BACKGROUND

The use of machine learning (ML) algorithms to study suicidality has recently been recommended. Our aim was to explore whether ML approaches have the potential to improve the prediction of suicide attempt (SA) risk. Using the epidemiological multiwave prospective-longitudinal Early Developmental Stages of Psychopathology (EDSP) data set, we compared four algorithms-logistic regression, lasso, ridge, and random forest-in predicting a future SA in a community sample of adolescents and young adults.

METHODS

The EDSP Study prospectively assessed, over the course of 10 years, adolescents and young adults aged 14-24 years at baseline. Of 3021 subjects, 2797 were eligible for prospective analyses because they participated in at least one of the three follow-up assessments. Sixteen baseline predictors, all selected a priori from the literature, were used to predict follow-up SAs. Model performance was assessed using repeated nested 10-fold cross-validation. As the main measure of predictive performance we used the area under the curve (AUC).

RESULTS

The mean AUCs of the four predictive models, logistic regression, lasso, ridge, and random forest, were 0.828, 0.826, 0.829, and 0.824, respectively.

CONCLUSIONS

Based on our comparison, each algorithm performed equally well in distinguishing between a future SA case and a non-SA case in community adolescents and young adults. When choosing an algorithm, different considerations, however, such as ease of implementation, might in some instances lead to one algorithm being prioritized over another. Further research and replication studies are required in this regard.

Admission Perfusion CT for Classifying Early In-Hospital Mortality of Patients With Severe Traumatic Brain Injury: A Pilot Study

OBJECTIVE. The purposes of this study were to assess the feasibility and safety of perfusion CT of patients with severe traumatic brain injury (TBI) at hospital admission and to examine whether early in-hospital mortality could be characterized with perfusion CT (PCT). The hypothesis was that PCT can be used to characterize brain death, when present, in patients with severe TBI at hospital admission. SUBJECTS AND METHODS. In this prospective cohort pilot study, PCT was performed on patients with severe TBI at first imaging workup at hospital admission. PCT images were processed at the end of the study and assessed for features of brain death. The PCT features were then compared with the clinical outcome of in-hospital mortality. RESULTS. A total of 19 patients (13 men [68.4%]; six women [31.6%]; mean age, 36.4 years; median, 27.5 years) had a mean hospital stay longer than 1 month. No complications of PCT were found. In the first 48 hours after admission, four patients (21%) died. Admission PCT changes suggesting brainstem death were sensitive (75%) and specific (100%) and had high positive (100%) and negative (93.75%) predictive value for correct classification early in-hospital mortality. CONCLUSION. Admission PCT of patients with severe TBI was feasible and safe. Admission PCT findings helped in correctly classifying early in-hospital mortality in the first 48 hours of hospital admission.

IMPACT and CRASH prognostic models for traumatic brain injury: external validation in a South-American cohort

OBJECTIVE

To develop a robust prognostic model, the more diverse the settings in which the system is tested and found to be accurate, the more likely it will be generalisable to untested settings. This study aimed to externally validate the International Mission for Prognosis and Clinical Trials in Traumatic Brain Injury (IMPACT) and Corticosteroid Randomization after Significant Head Injury (CRASH) models for low-income and middle-income countries using a dataset of patients with severe traumatic brain injury (TBI) from the Benchmark Evidence from South American Trials: Treatment of Intracranial Pressure study and a simultaneously conducted observational study.

METHOD

A total of 550 patients with severe TBI were enrolled in the study, and 466 of those were included in the analysis. Patient admission characteristics were extracted to predict unfavourable outcome (Glasgow Outcome Scale: GOS<3) and mortality (GOS 1) at 14 days or 6 months.

RESULTS

There were 48% of the participants who had unfavourable outcome at 6 months and these included 38% who had died. The area under the receiver operating characteristic curve (AUC) values were 0.683-0.775 and 0.640-0.731 for the IMPACT and CRASH models respectively. The IMPACT CT model had the highest AUC for predicting unfavourable outcomes, and the IMPACT Lab model had the best discrimination for predicting 6-month mortality. The discrimination for both the IMPACT and CRASH models improved with increasing complexity of the models. Calibration revealed that there were disagreement between observed and predicted outcomes in the IMPACT and CRASH models.

CONCLUSION

The overall performance of all IMPACT and CRASH models was adequate when used to predict outcomes in the dataset. However, some disagreement in calibration suggests the necessity for updating prognostic models to maintain currency and generalisability.

Outcome Prediction after Traumatic Brain Injury: Comparison of the Performance of Routinely Used Severity Scores and Multivariable Prognostic Models

Objectives:

Prognosis of outcome after traumatic brain injury (TBI) is important in the assessment of quality of care and can help improve treatment and outcome. The aim of this study was to compare the prognostic value of relatively simple injury severity scores between each other and against a gold standard model – the IMPACT-extended (IMP-E) multivariable prognostic model.

Materials and Methods:

For this study, 866 patients with moderate/severe TBI from Austria were analyzed. The prognostic performances of the Glasgow coma scale (GCS), GCS motor (GCSM) score, abbreviated injury scale for the head region, Marshall computed tomographic (CT) classification, and Rotterdam CT score were compared side-by-side and against the IMP-E score. The area under the receiver operating characteristics curve (AUC) and Nagelkerke's R² were used to assess the prognostic performance. Outcomes at the Intensive Care Unit, at hospital discharge, and at 6 months (mortality and unfavorable outcome) were used as end-points.

Results:

Comparing AUCs and R²s of the same model across four outcomes, only little variation was apparent. A similar pattern is observed when comparing the models between each other: Variation of AUCs <±0.09 and R²s by up to ±0.17 points suggest that all scores perform similarly in predicting outcomes at various points (AUCs: 0.65–0.77; R²s: 0.09–0.27). All scores performed significantly worse than the IMP-E model (with AUC > 0.83 and R² > 0.42 for all outcomes): AUCs were worse by 0.10–0.22 (P < 0.05) and R²s were worse by 0.22–0.39 points.

Conclusions:

All tested simple scores can provide reasonably valid prognosis. However, it is confirmed that well-developed multivariable prognostic models outperform these scores significantly and should be used for prognosis in patients after TBI wherever possible.

Prognosis in Moderate and Severe Traumatic Brain Injury: A Systematic Review of Contemporary Models and Validation Studies

Outcome prognostication in traumatic brain injury (TBI) is important but challenging due to heterogeneity of the disease. The aim of this systematic review is to present the current state-of-the-art on prognostic models for outcome after moderate and severe TBI and evidence on their validity. We searched for studies reporting on the development, validation or extension of prognostic models for functional outcome after TBI with Glasgow Coma Scale (GCS) ≤12 published between 2006-2018. Studies with patients age ≥14 years and evaluating a multi-variable prognostic model based on admission characteristics were included. Model discrimination was expressed with the area under the receiver operating characteristic curve (AUC), and model calibration with calibration slope and intercept. We included 58 studies describing 67 different prognostic models, comprising the development of 42 models, 149 external validations of 31 models, and 12 model extensions. The most common predictors were GCS (motor) score (n = 55), age (n = 54), and pupillary reactivity (n = 48). Model discrimination varied substantially between studies. The International Mission for Prognosis and Analysis of Clinical Trials (IMPACT) and Corticoid Randomisation After Significant Head injury (CRASH) models were developed on the largest cohorts (8509 and 10,008 patients, respectively) and were most often externally validated (n = 91), yielding AUCs ranging between 0.65-0.90 and 0.66-1.00, respectively. Model calibration was reported with a calibration intercept and slope for seven models in 53 validations, and was highly variable. In conclusion, the discriminatory validity of the IMPACT and CRASH prognostic models is supported across a range of settings. The variation in calibration, reflecting heterogeneity in reliability of predictions, motivates continuous validation and updating if clinical implementation is pursued.

Assessment of heterogeneity in an individual participant data meta-analysis of prediction models: An overview and illustration

Clinical prediction models aim to provide estimates of absolute risk for a diagnostic or prognostic endpoint. Such models may be derived from data from various studies in the context of a meta‐analysis. We describe and propose approaches for assessing heterogeneity in predictor effects and predictions arising from models based on data from different sources. These methods are illustrated in a case study with patients suffering from traumatic brain injury, where we aim to predict 6‐month mortality based on individual patient data using meta‐analytic techniques (15 studies, n = 11 022 patients). The insights into various aspects of heterogeneity are important to develop better models and understand problems with the transportability of absolute risk predictions.

The use of rigorous methods was strongly warranted among prognostic prediction models for obstetric care

OBJECTIVE

The objective of the study was to examine methodological characteristics about the design and conduct in prognostic prediction models used for obstetric care.

STUDY DESIGN AND SETTING

We searched PubMed for studies on prognostic prediction models for obstetric care, published in top general medicine or major specialty journals between January 2011 and February 2018. Teams of method-trained investigators independently screened titles and abstracts and collected data using a prespecified, pilot-tested, structured questionnaire.

RESULTS

In total, 91 studies were eligible, of which two were published in top general medicine journals, 20 (22.0%) involved an epidemiologist or statistician, 18 (19.4%) published study protocols, 53 (58.2%) did not include any model validation, 20 (22.0%) did not clearly state the intended timing of use, 23 (25.3%) had no eligibility criteria, 15 (16.5%) did not use clear criteria for ascertaining outcome, and 69 (75.82%) did not apply blinding to outcome assessment. Among those models, 11 (12.1%) included participants fewer than 200 events, 41 (48.8%) had fewer than 100 events, and 19 (24.7%) had fewer than 10 events per variable.

CONCLUSION

The prognostic prediction models have important limitations in design and conduct. Substantial efforts are needed to strengthen the production of reliable prognostic prediction models for obstetric care.

Why Does Brain Trauma Research Fail?

Traumatic brain injury (TBI) represents a major health care problem and a significant social and economic issue worldwide. Considering the generalized failure in introducing effective drugs and clinical protocols, there is an urgent need for efficient treatment modalities, able to improve devastating posttraumatic morbidity and mortality. In this work, the status of brain trauma research is analyzed in all its aspects, including basic and translational science and clinical trials. Implicit and explicit challenges to different lines of research are discussed and clinical trial structures and outcomes are scrutinized, along with possible explanations for systematic therapeutic failures and their implications for future development of drug and clinical trials. Despite significant advances in basic and clinical research in recent years, no specific therapeutic protocols for TBI have been shown to be effective. New potential therapeutic targets have been identified, following a better understanding of pathophysiologic mechanisms underlying TBI, although with disappointing results. Several reasons can be pinpointed at different levels, from inaccurate animal models of disease to faulty preclinical and clinical trials, with poor design and subjective outcome measures. Distinct strategies can be delineated to overcome specific shortcomings of research studies. Identifying and contextualizing the failures that have dominated TBI research is mandatory. This review analyzes current approaches and discusses possible strategies for improving outcomes.

External validation of a modified trauma and injury severity score model in major trauma injury

BACKGROUND

The establishment of an accurate prognostic model in major trauma patients is important mainly because this group of patients will benefit the most. Clinical prediction models must be validated internally and externally on a regular basis to ensure the prediction is accurate and current. This study aims to externally validate two prediction models, the Trauma and Injury Severity Score model developed using the Major Trauma Outcome Study in North America (MTOS-TRISS model), and the NTrD-TRISS model, which is a refined MTOS-TRISS model with coefficients derived from the Malaysian National Trauma Database (NTrD), by regarding mortality as the outcome measurement.

METHOD

This retrospective study included patients with major trauma injuries reported to a trauma centre of Hospital Sultanah Aminah over a 6-year period from 2011 and 2017. Model validation was examined using the measures of discrimination and calibration. Discrimination was assessed using the area under the receiver operating characteristic curve (AUC) and 95% confidence interval (CI). The Hosmer-Lemeshow (H-L) goodness-of-fit test was used to examine calibration capabilities. The predictive validity of both MTOS-TRISS and NTrD-TRISS models were further evaluated by incorporating parameters such as the New Injury Severity Scale and the Injury Severity Score.

RESULTS

Total patients of 3788 (3434 blunt and 354 penetrating injuries) with average age of 37 years (standard deviation of 16 years) were included in this study. All MTOS-TRISS and NTrD-TRISS models examined in this study showed adequate discriminative ability with AUCs ranged from 0.86 to 0.89 for patients with blunt trauma mechanism and 0.89 to 0.99 for patients with penetrating trauma mechanism. The H-L goodness-of-fit test indicated the NTrD-TRISS model calibrated as good as the MTOS-TRISS model for patients with blunt trauma mechanism.

CONCLUSION

For patients with blunt trauma mechanism, both the MTOS-TRISS and NTrD-TRISS models showed good discrimination and calibration performances. Discrimination performance for the NTrD-TRISS model was revealed to be as good as the MTOS-TRISS model specifically for patients with penetrating trauma mechanism. Overall, this validation study has ascertained the discrimination and calibration performances of the NTrD-TRISS model to be as good as the MTOS-TRISS model particularly for patients with blunt trauma mechanism.

Acute and Subacute Outcome Predictors in Moderate and Severe Traumatic Brain Injury: A Retrospective Monocentric Study

BACKGROUND

Prognostic factors affecting outcome of traumatic brain injury (TBI), despite their importance, are still under discussion. The purpose of this study was to describe risk factors of in-hospital mortality and outcome at 1 year in a homogeneously treated population of patients with moderate/severe TBI.

METHODS

A total of 193 consecutive patients with moderate or severe TBI (Glasgow Coma Scale [GCS] score 13-3, including patients with initial GCS score of 13 at high risk for subsequent neurologic deterioration), admitted to the intensive care unit, were retrospectively analyzed. In-hospital mortality and unfavorable outcome at 1 year, based on a Glasgow Outcome Scale-Extended score ≤4, were considered as primary and secondary outcomes.

RESULTS

At 1 year, unfavorable outcome occurred in 47.2%, including an in-hospital mortality of 19.7%. Increasing age, GCS motor score <3, coagulation disorders, and intracranial hypertension were acute risk factors of in-hospital mortality. In the 155 remaining survivors, Oxford Handicap Scale (OHS), posttraumatic cerebral infarction, cerebrospinal fluid disturbances, and length of intensive care unit stay were associated with unfavorable outcome at 1 year, in univariate analysis. A cutoff OHS score ≥3 discriminated the probability of an unfavorable outcome (area under the curve, 0.87; P < 0.001; specificity, 74%; sensitivity, 84%). Combining the effect of acute and subacute variables in a multivariate analysis, increasing age and OHS score were independent predictors of outcome.

CONCLUSIONS

The results of this retrospective study confirmed age as the main acute risk factor and identified OHS as new potential subacute predictor of unfavorable outcome in moderate and severe TBI.

Untapped potential of multicenter studies: a review of cardiovascular risk prediction models revealed inappropriate analyses and wide variation in reporting

BACKGROUND

Clinical prediction models are often constructed using multicenter databases. Such a data structure poses additional challenges for statistical analysis (clustered data) but offers opportunities for model generalizability to a broad range of centers. The purpose of this study was to describe properties, analysis, and reporting of multicenter studies in the Tufts PACE Clinical Prediction Model Registry and to illustrate consequences of common design and analyses choices.

METHODS

Fifty randomly selected studies that are included in the Tufts registry as multicenter and published after 2000 underwent full-text screening. Simulated examples illustrate some key concepts relevant to multicenter prediction research.

RESULTS

Multicenter studies differed widely in the number of participating centers (range 2 to 5473). Thirty-nine of 50 studies ignored the multicenter nature of data in the statistical analysis. In the others, clustering was resolved by developing the model on only one center, using mixed effects or stratified regression, or by using center-level characteristics as predictors. Twenty-three of 50 studies did not describe the clinical settings or type of centers from which data was obtained. Four of 50 studies discussed neither generalizability nor external validity of the developed model.

CONCLUSIONS

Regression methods and validation strategies tailored to multicenter studies are underutilized. Reporting on generalizability and potential external validity of the model lacks transparency. Hence, multicenter prediction research has untapped potential.

REGISTRATION

This review was not registered.

A Multidimensional Approach to Post-concussion Symptoms in Mild Traumatic Brain Injury

Mild traumatic brain injury (mTBI) presents a substantial burden to patients, families, and health care systems. Whereas, recovery can be expected in the majority of patients, a subset continues to report persisting somatic, cognitive, emotional, and/or behavioral problems, generally referred to as post-concussion syndrome (PCS). However, this term has been the subject of debate since the mechanisms underlying post-concussion symptoms and the role of pre- and post-injury-related factors are still poorly understood. We review current evidence and controversies concerning the use of the terms post-concussion symptoms vs. syndrome, its diagnosis, etiology, prevalence, assessment, and treatment in both adults and children. Prevalence rates of post-concussion symptoms vary between 11 and 82%, depending on diagnostic criteria, population and timing of assessment. Post-concussion symptoms are dependent on complex interactions between somatic, psychological, and social factors. Progress in understanding has been hampered by inconsistent classification and variable assessment procedures. There are substantial limitations in research to date, resulting in gaps in our understanding, leading to uncertainty regarding epidemiology, etiology, prognosis, and treatment. Future directions including the identification of potential mechanisms, new imaging techniques, comprehensive, multidisciplinary assessment and treatment options are discussed. Treatment of post-concussion symptoms is highly variable, and primarily directed at symptom relief, rather than at modifying the underlying pathology. Longitudinal studies applying standardized assessment strategies, diagnoses, and evidence-based interventions are required in adult and pediatric mTBI populations to optimize recovery and reduce the substantial socio-economic burden of post-concussion symptoms.

External Validation of Hematoma Expansion Scores in Spontaneous Intracerebral Hemorrhage in an Asian Patient Cohort

BACKGROUND

Hematoma expansion (HE) occurs in approximately one-third of patients with intracerebral hemorrhage (ICH) and is known to be a strong predictor of neurological deterioration as well as poor functional outcome. This study aims to externally validate three risk prediction models of HE (PREDICT, 9-point, and BRAIN scores) in an Asian population.

METHODS

A prospective cohort of 123 spontaneous ICH patients admitted to a tertiary hospital (certified stroke center) in Singapore was recruited. Logistic recalibrations were performed to obtain updated calibration slopes and intercepts for all models. The discrimination (c-statistic), calibration (Hosmer-Lemeshow test, le Cessie-van Houwelingen-Copas-Hosmer test, Akaike information criterion), overall performance (Brier score, R²), and clinical usefulness (decision curve analysis) of the risk prediction models were examined.

RESULTS

Overall, the recalibrated PREDICT performed best among the three models in our study cohort based on the novel matrix comprising of Akaike information criterion and c-statistic. The PREDICT model had the highest R² (0.26) and lowest Brier score (0.14). Decision curve analyses showed that recalibrated PREDICT was more clinically useful than 9-point and BRAIN models over the greatest range of threshold probabilities. The two scores (PREDICT and 9-point) which incorporated computed tomography (CT) angiography spot sign outperformed the one without (BRAIN).

CONCLUSIONS

To our knowledge, this is the first study to validate HE scores, namely PREDICT, 9-Point and BRAIN, in a multi-ethnic Asian ICH patient population. The PREDICT score was the best performing model in our study cohort, based on the performance metrics employed in this study. Our findings also showed support for CT angiography spot sign as a predictor of outcome after ICH. Although the models assessed are sufficient for risk stratification, the discrimination and calibration are at best moderate and could be improved.

A Machine Learning Framework for Automatic and Continuous MMN Detection With Preliminary Results for Coma Outcome Prediction

Mismatch negativity (MMN) is a component of the event-related potential (ERP) that is elicited through an odd-ball paradigm. The existence of the MMN in a coma patient has a good correlation with coma emergence; however, this component can be difficult to detect. Previously, MMN detection was based on visual inspection of the averaged ERPs by a skilled clinician, a process that is expensive and not always feasible in practice. In this paper, we propose a practical machine learning (ML) based approach for detection of MMN component, thus, improving the accuracy of prediction of emergence from coma. Furthermore, the method can operate on an automatic and continuous basis thus alleviating the need for clinician involvement. The proposed method is capable of the MMN detection over intervals as short as two minutes. This finer time resolution enables identification of waxing and waning cycles of a conscious state. An auditory odd-ball paradigm was applied to 22 healthy subjects and 2 coma patients. A coma patient is tested by measuring the similarity of the patient's ERP responses with the aggregate healthy responses. Because the training process for measuring similarity requires only healthy subjects, the complexity and practicality of training procedure of the proposed method are greatly improved relative to training on coma patients directly. Since there are only two coma patients involved with this study, the results are reported on a very preliminary basis. Preliminary results indicate we can detect the MMN component with an accuracy of 92.7% on healthy subjects. The method successfully predicted emergence in both coma patients when conventional methods failed. The proposed method for collecting training data using exclusively healthy subjects is a novel approach that may prove useful in future, unrelated studies where ML methods are used.

Prediction of Persistent Post-Concussion Symptoms after Mild Traumatic Brain Injury

Persistent post-concussion symptoms (PPCS) occur frequently after mild traumatic brain injury (mTBI). The identification of patients at risk for poor outcome remains challenging because valid prediction models are missing. The objectives of the current study were to assess the quality and clinical value of prediction models for PPCS and to develop a new model based on the synthesis of existing models and addition of complaints at the emergency department (ED). Patients with mTBI (Glasgow Coma Scale score 13-15) were recruited prospectively from three Dutch level I trauma centers between 2013 and 2015 in the UPFRONT study. PPCS were assessed using the Head Injury Severity Checklist at six months post-injury. Two prediction models (Stulemeijer 2008; Cnossen 2017) were examined for calibration and discrimination. The final model comprised variables of existing models with the addition of headache, nausea/vomiting, and neck pain at ED, using logistic regression and bootstrap validation. Overall, 591 patients (mean age 51years, 41% female) were included; PPCS developed in 241 (41%). Existing models performed poorly at external validation (area under the curve [AUC]: 0.57-0.64). The newly developed model included female sex (odds ratio [OR] 1.48, 95% confidence interval [CI] [1.01-2.18]), neck pain (OR 2.58, [1.39-4.78]), two-week post-concussion symptoms (OR 4.89, [3.19-7.49]) and two-week post-traumatic stress (OR 2.98, [1.88-4.73]) as significant predictors. Discrimination of this model was adequate (AUC after bootstrap validation: 0.75). Existing prediction models for PPCS perform poorly. A new model performs reasonably with predictive factors already discernible at ED warranting further external validation. Prediction research in mTBI should be improved by standardizing definitions and data collection and by using sound methodology.

A Decision Support System for Diagnostics and Treatment Planning in Traumatic Brain Injury

Traumatic brain injury (TBI) occurs when an external force causes functional or structural alterations in the brain. Clinical characteristics of TBI vary greatly from patient to patient, and a large amount of data is gathered during various phases of clinical care in these patients. It is hard for clinicians to efficiently integrate and interpret all of these data and plan interventions in a timely manner. This paper describes the technical architecture and functionality of a web-based decision support system (DSS), which not only provides advanced support for visualizing complex TBI data but also predicts a possible outcome by using a state-of-the-art Disease State Index machine-learning algorithm. The DSS is developed by using a three-layered architecture and by employing modern programming principles, software design patterns, and using robust technologies (C#, ASP.NET MVC, HTML5, JavaScript, Entity Framework, etc.). The DSS is comprised of a patient overview module, a disease-state prediction module, and an imaging module. After deploying it on a web-server, the DSS was made available to two hospitals in U.K. and Finland. Afterwards, we conducted a validation study to evaluate its usability in clinical settings. Initial results of the study indicate that especially less experience clinicians may benefit from this type of decision support software tool.

Predicting Long-Term Global Outcome after Traumatic Brain Injury: Development of a Practical Prognostic Tool Using the Traumatic Brain Injury Model Systems National Database

For patients surviving serious traumatic brain injury (TBI), families and other stakeholders often desire information on long-term functional prognosis, but accurate and easy-to-use clinical tools are lacking. We aimed to build utilitarian decision trees from commonly collected clinical variables to predict Glasgow Outcome Scale (GOS) functional levels at 1, 2, and 5 years after moderate-to-severe closed TBI. Flexible classification tree statistical modeling was used on prospectively collected data from the TBI-Model Systems (TBIMS) inception cohort study. Enrollments occurred at 17 designated, or previously designated, TBIMS inpatient rehabilitation facilities. Analysis included all participants with nonpenetrating TBI injured between January 1997 and January 2017. Sample sizes were 10,125 (year-1), 8,821 (year-2), and 6,165 (year-5) after cross-sectional exclusions (death, vegetative state, insufficient post-injury time, and unavailable outcome). In our final models, post-traumatic amnesia (PTA) duration consistently dominated branching hierarchy and was the lone injury characteristic significantly contributing to GOS predictability. Lower-order variables that added predictability were age, pre-morbid education, productivity, and occupational category. Generally, patient outcomes improved with shorter PTA, younger age, greater pre-morbid productivity, and higher pre-morbid vocational or educational achievement. Across all prognostic groups, the best and worst good recovery rates were 65.7% and 10.9%, respectively, and the best and worst severe disability rates were 3.9% and 64.1%. Predictability in test data sets ranged from C-statistic of 0.691 (year-1; confidence interval [CI], 0.675, 0.711) to 0.731 (year-2; CI, 0.724, 0.738). In conclusion, we developed a clinically useful tool to provide prognostic information on long-term functional outcomes for adult survivors of moderate and severe closed TBI. Predictive accuracy for GOS level was demonstrated in an independent test sample. Length of PTA, a clinical marker of injury severity, was by far the most critical outcome determinant.

Complement cascade on severe traumatic brain injury patients at the chronic unconscious stage: implication for pathogenesis

BACKGROUND

Patients who awake from severely traumatic brain injury (TBI) may remain unconscious for many years. Although behavioral assessment and functional imaging are currently used as diagnostic tools, the molecular basis underlying chronic condition has yet to be explored.

METHOD

Plasma samples were obtained at 3 time points (1, 3 and 6 months) from 18 patients with chronic disorders of consciousness who survived severe TBI, and 6 healthy volunteers. A coupled isobaric tag for relative and absolute quantitation (iTRAQ)-based proteomics approach was used to screen differentially expressed proteins (DEPs) between patients and controls. Potential molecular mechanisms were further discussed through bioinformatics analyses.

RESULT

In total, 300 plasma proteins <1% false discovery rates were identified and 32 proteins were consistently altered between patients and controls. Biological pathway analysis revealed that the DEPs were predominantly involved in complement cascade.

CONCLUSIONS

This study discussed potential mechanisms of complement cascade underlying chronic stage in severe TBI.

Simplifying the use of prognostic information in traumatic brain injury. Part 2: Graphical presentation of probabilities

OBJECTIVE Clinical features such as those included in the Glasgow Coma Scale (GCS) score, pupil reactivity, and patient age, as well as CT findings, have clear established relationships with patient outcomes due to neurotrauma. Nevertheless, predictions made from combining these features in probabilistic models have not found a role in clinical practice. In this study, the authors aimed to develop a method of displaying probabilities graphically that would be simple and easy to use, thus improving the usefulness of prognostic information in neurotrauma. This work builds on a companion paper describing the GCS-Pupils score (GCS-P) as a tool for assessing the clinical severity of neurotrauma. METHODS Information about early GCS score, pupil response, patient age, CT findings, late outcome according to the Glasgow Outcome Scale, and mortality were obtained at the individual adult patient level from the CRASH (Corticosteroid Randomisation After Significant Head Injury; n = 9045) and IMPACT (International Mission for Prognosis and Clinical Trials in TBI; n = 6855) databases. These data were combined into a pooled data set for the main analysis. Logistic regression was first used to model the combined association between the GCS-P and patient age and outcome, following which CT findings were added to the models. The proportion of variability in outcomes "explained" by each model was assessed using Nagelkerke's R². RESULTS The authors observed that patient age and GCS-P have an additive effect on outcome. The probability of mortality 6 months after neurotrauma is greater with increasing age, and for all age groups the probability of death is greater with decreasing GCS-P. Conversely, the probability of favorable recovery becomes lower with increasing age and lessens with decreasing GCS-P. The effect of combining the GCS-P with patient age was substantially more informative than the GCS-P, age, GCS score, or pupil reactivity alone. Two-dimensional charts were produced displaying outcome probabilities, as percentages, for 5-year increments in age between 15 and 85 years, and for GCS-Ps ranging from 1 to 15; it is readily seen that the movement toward combinations at the top right of the charts reflects a decreasing likelihood of mortality and an increasing likelihood of favorable outcome. Analysis of CT findings showed that differences in outcome are very similar between patients with or without a hematoma, absent cisterns, or subarachnoid hemorrhage. Taken in combination, there is a gradation in risk that aligns with increasing numbers of any of these abnormalities. This information provides added value over age and GCS-P alone, supporting a simple extension of the earlier prognostic charts by stratifying the original charts in the following 3 CT groupings: none, only 1, and 2 or more CT abnormalities. CONCLUSIONS The important prognostic features in neurotrauma can be brought together to display graphically their combined effects on risks of death or on prospects for independent recovery. This approach can support decision making and improve communication of risk among health care professionals, patients, and their relatives. These charts will not replace clinical judgment, but they will reduce the risk of influences from biases.

Prognostic value of coagulation tests for in-hospital mortality in patients with traumatic brain injury

BACKGROUND

Coagulopathy is commonly observed after traumatic brain injury (TBI). However, it is not known whether using the standard independent predictors in conjunction with coagulation tests would improve their prognostic value. We determined the incidence of TBI-associated coagulopathy in patients with isolated TBI (iTBI), evaluated the prognostic value of coagulation tests for in-hospital mortality, and tested their predictive power for in-hospital mortality in patients with iTBI.

METHODS

We conducted a retrospective, observational database study on 2319 consecutive patients with iTBI who attended the Huashan Hospital Department of the Neurosurgery Neurotrauma Center at Fudan University in China between December 2004 and June 2015. Two models based on the admission characteristics were developed: model A included predictors such as age, Glasgow Coma Scale (GCS) score, pupil reactivity, type of injury, and hemoglobin and glucose levels, while model B included the predictors from model A as well as coagulation test results. A total of 1643 patients enrolled between December 2004 and December 2011 were used to derive the prognostic models, and 676 patients enrolled between January 2012 and June 2015 were used to validate the models.

RESULTS

Overall, 18.6% (n = 432) of the patients developed coagulopathy after iTBI. The prevalence of acute traumatic coagulopathy is associated with the severity of brain injury. The percentage of platelet count <100 × 109/L, international normalized ratio (INR) > 1.25, the prothrombin time (PT) > 14 s, activated partial thromboplastin time (APTT) > 36 s, D-dimer >5 mg/L and fibrinogen (FIB) < 1.5 g/L was also closely related to the severity of brain injury, significance being found among three groups. Age, pupillary reactivity, GCS score, epidural hematoma (EDH), and glucose levels were independent prognostic factors for in-hospital mortality in model A, whereas age, pupillary reactivity, GCS score, EDH, glucose levels, INR >1.25, and APTT >36 s exhibited strong prognostic effects in model B. Discrimination and calibration were good for the development group in both prediction models. However, the external validation test showed that calibration was better in model B than in model A for patients from the validation population (Hosmer-Lemeshow test, p = 0.152 vs. p = 0.046, respectively).

CONCLUSIONS

Coagulation tests can improve the predictive power of the standard model for in-hospital mortality after TBI.

Poor performance of clinical prediction models: the harm of commonly applied methods

OBJECTIVE

To evaluate limitations of common statistical modeling approaches in deriving clinical prediction models and explore alternative strategies.

STUDY DESIGN AND SETTING

A previously published model predicted the likelihood of having a mutation in germline DNA mismatch repair genes at the time of diagnosis of colorectal cancer. This model was based on a cohort where 38 mutations were found among 870 participants, with validation in an independent cohort with 35 mutations. The modeling strategy included stepwise selection of predictors from a pool of over 37 candidate predictors and dichotomization of continuous predictors. We simulated this strategy in small subsets of a large contemporary cohort (2,051 mutations among 19,866 participants) and made comparisons to other modeling approaches. All models were evaluated according to bias and discriminative ability (concordance index, c) in independent data.

RESULTS

We found over 50% bias for five of six originally selected predictors, unstable model specification, and poor performance at validation (median c = 0.74). A small validation sample hampered stable assessment of performance. Model prespecification based on external knowledge and using continuous predictors led to better performance (c = 0.836 and c = 0.852 with 38 and 2,051 events respectively).

CONCLUSION

Prediction models perform poorly if based on small numbers of events and developed with common but suboptimal statistical approaches. Alternative modeling strategies to best exploit available predictive information need wider implementation, with collaborative research to increase sample sizes.

New astroglial injury-defined biomarkers for neurotrauma assessment

Traumatic brain injury (TBI) is an expanding public health epidemic with pathophysiology that is difficult to diagnose and thus treat. TBI biomarkers should assess patients across severities and reveal pathophysiology, but currently, their kinetics and specificity are unclear. No single ideal TBI biomarker exists. We identified new candidates from a TBI CSF proteome by selecting trauma-released, astrocyte-enriched proteins including aldolase C (ALDOC), its 38kD breakdown product (BDP), brain lipid binding protein (BLBP), astrocytic phosphoprotein (PEA15), glutamine synthetase (GS) and new 18-25kD-GFAP-BDPs. Their levels increased over four orders of magnitude in severe TBI CSF. First post-injury week, ALDOC levels were markedly high and stable. Short-lived BLBP and PEA15 related to injury progression. ALDOC, BLBP and PEA15 appeared hyper-acutely and were similarly robust in severe and mild TBI blood; 25kD-GFAP-BDP appeared overnight after TBI and was rarely present after mild TBI. Using a human culture trauma model, we investigated biomarker kinetics. Wounded (mechanoporated) astrocytes released ALDOC, BLBP and PEA15 acutely. Delayed cell death corresponded with GFAP release and proteolysis into small GFAP-BDPs. Associating biomarkers with cellular injury stages produced astroglial injury-defined (AID) biomarkers that facilitate TBI assessment, as neurological deficits are rooted not only in death of CNS cells, but also in their functional compromise.