Surgical prioritization based on decision model outcomes is not sensitive to differences between the health-related quality of life values estimates of physicians and citizens.

PURPOSE: Decision models can be used to support allocation of scarce surgical resources. These models incorporate health-related quality of life (HRQoL) values that can be determined using physician panels. The predominant opinion is that one should use values obtained from citizens. We investigated whether physicians give different HRQoL values to citizens and evaluate whether such differences impact decision model outcomes. METHODS: A two-round Delphi study was conducted. Citizens estimated HRQoL of pre- and post-operative health states for ten surgeries using a visual analogue scale. These values were compared using Bland-Altman analysis with HRQoL values previously obtained from physicians. Impact on decision model outcomes was evaluated by calculating the correlation between the rankings of surgeries established using the physicians' and the citizens' values. RESULTS: A total of 71 citizens estimated HRQoL. Citizens' values on the VAS scale were - 0.07 points (95% CI - 0.12 to - 0.01) lower than the physicians' values. The correlation between the rankings of surgeries based on citizens' and physicians' values was 0.96 (p < 0.001). CONCLUSION: Physicians put higher values on health states than citizens. However, these differences only result in switches between adjacent entries in the ranking. It would seem that HRQoL values obtained from physicians are adequate to inform decision models during crises.

Minimizing population health loss due to scarcity in OR capacity: validation of quality of life input.

OBJECTIVES: A previously developed decision model to prioritize surgical procedures in times of scarce surgical capacity used quality of life (QoL) primarily derived from experts in one center. These estimates are key input of the model, and might be more context-dependent than the other input parameters (age, survival). The aim of this study was to validate our model by replicating these QoL estimates. METHODS: The original study estimated QoL of patients in need of commonly performed procedures in live expert-panel meetings. This study replicated this procedure using a web-based Delphi approach in a different hospital. The new QoL scores were compared with the original scores using mixed effects linear regression. The ranking of surgical procedures based on combined QoL values from the validation and original study was compared to the ranking based solely on the original QoL values. RESULTS: The overall mean difference in QoL estimates between the validation study and the original study was - 0.11 (95% CI: -0.12 - -0.10). The model output (DALY/month delay) based on QoL data from both studies was similar to the model output based on the original data only: The Spearman's correlation coefficient between the ranking of all procedures before and after including the new QoL estimates was 0.988. DISCUSSION: Even though the new QoL estimates were systematically lower than the values from the original study, the ranking for urgency based on health loss per unit of time delay of procedures was consistent. This underscores the robustness and generalizability of the decision model for prioritization of surgical procedures.

Minimising population health loss in times of scarce surgical capacity: a modelling study for surgical procedures performed in nonacademic hospitals.

BACKGROUND: The burden of the COVID-19 pandemic resulted in a reduction of available health care capacity for regular care. To guide prioritisation of semielective surgery in times of scarcity, we previously developed a decision model to quantify the expected health loss due to delay of surgery, in an academic hospital setting. The aim of this study is to validate our decision model in a nonacademic setting and include additional elective surgical procedures. METHODS: In this study, we used the previously published three-state cohort state-transition model, to evaluate the health effects of surgery postponement for 28 surgical procedures commonly performed in nonacademic hospitals. Scientific literature and national registries yielded nearly all input parameters, except for the quality of life (QoL) estimates which were obtained from experts using the Delphi method. Two expert panels, one from a single nonacademic hospital and one from different nonacademic hospitals in the Netherlands, were invited to estimate QoL weights. We compared estimated model results (disability adjusted life years (DALY)/month of surgical delay) based on the QoL estimates from the two panels by calculating the mean difference and the correlation between the ranks of the different surgical procedures. The eventual model was based on the combined QoL estimates from both panels. RESULTS: Pacemaker implantation was associated with the most DALY/month of surgical delay (0.054 DALY/month, 95% CI: 0.025-0.103) and hemithyreoidectomy with the least DALY/month (0.006 DALY/month, 95% CI: 0.002-0.009). The overall mean difference of QoL estimates between the two panels was 0.005 (95% CI -0.014-0.004). The correlation between ranks was 0.983 (p < 0.001). CONCLUSIONS: Our study provides an overview of incurred health loss due to surgical delay for surgeries frequently performed in nonacademic hospitals. The quality of life estimates currently used in our model are robust and validate towards a different group of experts. These results enrich our earlier published results on academic surgeries and contribute to prioritising a more complete set of surgeries.

Comparative effectiveness of intracranial hypertension management guided by ventricular versus intraparenchymal pressure monitoring: a CENTER-TBI study.

OBJECTIVE: To compare outcomes between patients with primary external ventricular device (EVD)-driven treatment of intracranial hypertension and those with primary intraparenchymal monitor (IP)-driven treatment. METHODS: The CENTER-TBI study is a prospective, multicenter, longitudinal observational cohort study that enrolled patients of all TBI severities from 62 participating centers (mainly level I trauma centers) across Europe between 2015 and 2017. Functional outcome was assessed at 6 months and a year. We used multivariable adjusted instrumental variable (IV) analysis with "center" as instrument and logistic regression with covariate adjustment to determine the effect estimate of EVD on 6-month functional outcome. RESULTS: A total of 878 patients of all TBI severities with an indication for intracranial pressure (ICP) monitoring were included in the present study, of whom 739 (84%) patients had an IP monitor and 139 (16%) an EVD. Patients included were predominantly male (74% in the IP monitor and 76% in the EVD group), with a median age of 46 years in the IP group and 48 in the EVD group. Six-month GOS-E was similar between IP and EVD patients (adjusted odds ratio (aOR) and 95% confidence interval [CI] OR 0.74 and 95% CI [0.36-1.52], adjusted IV analysis). The length of intensive care unit stay was greater in the EVD group than in the IP group (adjusted rate ratio [95% CI] 1.70 [1.34-2.12], IV analysis). One hundred eighty-seven of the 739 patients in the IP group (25%) required an EVD due to refractory ICPs. CONCLUSION: We found no major differences in outcomes of patients with TBI when comparing EVD-guided and IP monitor-guided ICP management. In our cohort, a quarter of patients that initially received an IP monitor required an EVD later for ICP control. The prevalence of complications was higher in the EVD group. PROTOCOL: The core study is registered with ClinicalTrials.gov , number NCT02210221, and the Resource Identification Portal (RRID: SCR_015582).

Modern Learning from Big Data in Critical Care: Primum Non Nocere.

Large and complex data sets are increasingly available for research in critical care. To analyze these data, researchers use techniques commonly referred to as statistical learning or machine learning (ML). The latter is known for large successes in the field of diagnostics, for example, by identification of radiological anomalies. In other research areas, such as clustering and prediction studies, there is more discussion regarding the benefit and efficiency of ML techniques compared with statistical learning. In this viewpoint, we aim to explain commonly used statistical learning and ML techniques and provide guidance for responsible use in the case of clustering and prediction questions in critical care. Clustering studies have been increasingly popular in critical care research, aiming to inform how patients can be characterized, classified, or treated differently. An important challenge for clustering studies is to ensure and assess generalizability. This limits the application of findings in these studies toward individual patients. In the case of predictive questions, there is much discussion as to what algorithm should be used to most accurately predict outcome. Aspects that determine usefulness of ML, compared with statistical techniques, include the volume of the data, the dimensionality of the preferred model, and the extent of missing data. There are areas in which modern ML methods may be preferred. However, efforts should be made to implement statistical frameworks (e.g., for dealing with missing data or measurement error, both omnipresent in clinical data) in ML methods. To conclude, there are important opportunities but also pitfalls to consider when performing clustering or predictive studies with ML techniques. We advocate careful valuation of new data-driven findings. More interaction is needed between the engineer mindset of experts in ML methods, the insight in bias of epidemiologists, and the probabilistic thinking of statisticians to extract as much information and knowledge from data as possible, while avoiding harm.

Performance of the Hull Salford Cambridge Decision Rule (HSC DR) for early discharge of patients with findings on CT scan of the brain: a CENTER-TBI validation study.

BACKGROUND: There is international variation in hospital admission practices for patients with mild traumatic brain injury (TBI) and injuries on CT scan. Only a small proportion of patients require neurosurgical intervention, while many guidelines recommend routine admission of all patients. We aim to validate the Hull Salford Cambridge Decision Rule (HSC DR) and the Brain Injury Guidelines (BIG) criteria to select low-risk patients for discharge from the emergency department. METHOD: A cohort from 18 countries of Glasgow Coma Scale 13-15 patients with injuries on CT imaging was identified from the multicentre Collaborative European NeuroTrauma Effectiveness Research in TBI (CENTER-TBI) Study (conducted from 2014 to 2017) for secondary analysis. A composite outcome measure encompassing need for ongoing hospital admission was used, including seizure activity, death, intubation, neurosurgical intervention and neurological deterioration. We assessed the performance of our previously derived prognostic model, the HSC DR and the BIG criteria at predicting deterioration in this validation cohort. RESULTS: Among 1047 patients meeting the inclusion criteria, 267 (26%) deteriorated. Our prognostic model achieved a C-statistic of 0.81 (95% CI: 0.78 to 0.84). The HSC DR achieved a sensitivity of 100% (95% CI: 97% to 100%) and specificity of only 4.7% (95% CI: 3.3% to 6.5%) for deterioration. Using the BIG criteria for discharge from the ED achieved a higher specificity (13.3%, 95% CI: 10.9% to 16.1%) and lower sensitivity (94.6%, 95% CI: 90.5% to 97%), with 12/105 patients recommended for discharge subsequently deteriorating, compared with 0/34 with the HSC DR. CONCLUSION: Our decision rule would have allowed 3.5% of patients to be discharged, none of whom would have deteriorated. Use of the BIG criteria may select patients for discharge who have too high a risk of subsequent deterioration to be used clinically. Further validation and implementation studies are required to support use in clinical practice.

Minimizing Population Health Loss in Times of Scarce Surgical Capacity During the Coronavirus Disease 2019 Crisis and Beyond: A Modeling Study.

OBJECTIVES: Coronavirus disease 2019 has put unprecedented pressure on healthcare systems worldwide, leading to a reduction of the available healthcare capacity. Our objective was to develop a decision model to estimate the impact of postponing semielective surgical procedures on health, to support prioritization of care from a utilitarian perspective. METHODS: A cohort state-transition model was developed and applied to 43 semielective nonpediatric surgical procedures commonly performed in academic hospitals. Scenarios of delaying surgery from 2 weeks were compared with delaying up to 1 year and no surgery at all. Model parameters were based on registries, scientific literature, and the World Health Organization Global Burden of Disease study. For each surgical procedure, the model estimated the average expected disability-adjusted life-years (DALYs) per month of delay. RESULTS: Given the best available evidence, the 2 surgical procedures associated with most DALYs owing to delay were bypass surgery for Fontaine III/IV peripheral arterial disease (0.23 DALY/month, 95% confidence interval [CI]: 0.13-0.36) and transaortic valve implantation (0.15 DALY/month, 95% CI: 0.09-0.24). The 2 surgical procedures with the least DALYs were placing a shunt for dialysis (0.01, 95% CI: 0.005-0.01) and thyroid carcinoma resection (0.01, 95% CI: 0.01-0.02). CONCLUSION: Expected health loss owing to surgical delay can be objectively calculated with our decision model based on best available evidence, which can guide prioritization of surgical procedures to minimize population health loss in times of scarcity. The model results should be placed in the context of different ethical perspectives and combined with capacity management tools to facilitate large-scale implementation.

Prehospital Management of Traumatic Brain Injury across Europe: A CENTER-TBI Study.

BACKGROUND: Prehospital care for traumatic brain injury (TBI) is important to prevent secondary brain injury. We aim to compare prehospital care systems within Europe and investigate the association of system characteristics with the stability of patients at hospital arrival. METHODS: We studied TBI patients who were transported to CENTER-TBI centers, a pan-European, prospective TBI cohort study, by emergency medical services between 2014 and 2017. The association of demographic factors, injury severity, situational factors, and interventions associated with on-scene time was assessed using linear regression. We used mixed effects models to investigate the case mix adjusted variation between countries in prehospital times and interventions. The case mix adjusted impact of on-scene time and interventions on hypoxia (oxygen saturation <90%) and hypotension (systolic blood pressure <100mmHg) at hospital arrival was analyzed with logistic regression. RESULTS: Among 3878 patients, the greatest driver of longer on-scene time was intubation (+8.3 min, 95% CI: 5.6-11.1). Secondary referral was associated with shorter on-scene time (-5.0 min 95% CI: -6.2- -3.8). Between countries, there was a large variation in response (range: 12-25 min), on-scene (range: 16-36 min) and travel time (range: 15-32 min) and in prehospital interventions. These variations were not explained by patient factors such as conscious level or severity of injury (expected OR between countries: 1.8 for intubation, 1.8 for IV fluids, 2.0 for helicopter). On-scene time was not associated with the regional EMS policy (p= 0.58). Hypotension and/or hypoxia were seen in 180 (6%) and 97 (3%) patients in the overall cohort and in 13% and 7% of patients with severe TBI (GCS <8). The largest association with secondary insults at hospital arrival was with major extracranial injury: the OR was 3.6 (95% CI: 2.6-5.0) for hypotension and 4.4 (95% CI: 2.9-6.7) for hypoxia. DISCUSSION: Hypoxia and hypotension continue to occur in patients who suffer a TBI, and remain relatively common in severe TBI. Substantial variation in prehospital care exists for patients after TBI in Europe, which is only partially explained by patient factors.

Identifying trauma patients with benefit from direct transportation to Level-1 trauma centers.

BACKGROUND: Prehospital triage protocols typically try to select patients with Injury Severity Score (ISS) above 15 for direct transportation to a Level-1 trauma center. However, ISS does not necessarily discriminate between patients who benefit from immediate care at Level-1 trauma centers. The aim of this study was to assess which patients benefit from direct transportation to Level-1 trauma centers. METHODS: We used the American National Trauma Data Bank (NTDB), a retrospective observational cohort. All adult patients (ISS > 3) between 2015 and 2016 were included. Patients who were self-presenting or had isolated limb injury were excluded. We used logistic regression to assess the association of direct transportation to Level-1 trauma centers with in-hospital mortality adjusted for clinically relevant confounders. We used this model to define benefit as predicted probability of mortality associated with transportation to a non-Level-1 trauma center minus predicted probability associated with transportation to a Level-1 trauma center. We used a threshold of 1% as absolute benefit. Potential interaction terms with transportation to Level-1 trauma centers were included in a penalized logistic regression model to study which patients benefit. RESULTS: We included 388,845 trauma patients from 232 Level-1 centers and 429 Level-2/3 centers. A small beneficial effect was found for direct transportation to Level-1 trauma centers (adjusted Odds Ratio: 0.96, 95% Confidence Interval: 0.92-0.99) which disappeared when comparing Level-1 and 2 versus Level-3 trauma centers. In the risk approach, predicted benefit ranged between 0 and 1%. When allowing for interactions, 7% of the patients (n = 27,753) had more than 1% absolute benefit from direct transportation to Level-1 trauma centers. These patients had higher AIS Head and Thorax scores, lower GCS and lower SBP. A quarter of the patients with ISS > 15 were predicted to benefit from transportation to Level-1 centers (n = 26,522, 22%). CONCLUSIONS: Benefit of transportation to a Level-1 trauma centers is quite heterogeneous across patients and the difference between Level-1 and Level-2 trauma centers is small. In particular, patients with head injury and signs of shock may benefit from care in a Level-1 trauma center. Future prehospital triage models should incorporate more complete risk profiles.

Neurological outcome after extracorporeal cardiopulmonary resuscitation for in-hospital cardiac arrest: a systematic review and meta-analysis.

BACKGROUND: In-hospital cardiac arrest (IHCA) is a major adverse event with a high mortality rate if not treated appropriately. Extracorporeal cardiopulmonary resuscitation (ECPR), as adjunct to conventional cardiopulmonary resuscitation (CCPR), is a promising technique for IHCA treatment. Evidence pertaining to neurological outcomes after ECPR is still scarce. METHODS: We performed a comprehensive systematic search of all studies up to December 20, 2019. Our primary outcome was neurological outcome after ECPR at any moment after hospital discharge, defined by the Cerebral Performance Category (CPC) score. A score of 1 or 2 was defined as favourable outcome. Our secondary outcome was post-discharge mortality. A fixed-effects meta-analysis was performed. RESULTS: Our search yielded 1215 results, of which 19 studies were included in this systematic review. The average survival rate was 30% (95% CI 28-33%, I2 = 0%, p = 0.24). In the surviving patients, the pooled percentage of favourable neurological outcome was 84% (95% CI 80-88%, I2 = 24%, p = 0.90). CONCLUSION: ECPR as treatment for in-hospital cardiac arrest is associated with a large proportion of patients with good neurological outcome. The large proportion of favourable outcome could potentially be explained by the selection of patients for treatment using ECPR. Moreover, survival is higher than described in the conventional CPR literature. As indications for ECPR might extend to older or more fragile patient populations in the future, research should focus on increasing survival, while maintaining optimal neurological outcome.

Tracheal intubation in traumatic brain injury: a multicentre prospective observational study.

BACKGROUND: We aimed to study the associations between pre- and in-hospital tracheal intubation and outcomes in traumatic brain injury (TBI), and whether the association varied according to injury severity. METHODS: Data from the international prospective pan-European cohort study, Collaborative European NeuroTrauma Effectiveness Research for TBI (CENTER-TBI), were used (n=4509). For prehospital intubation, we excluded self-presenters. For in-hospital intubation, patients whose tracheas were intubated on-scene were excluded. The association between intubation and outcome was analysed with ordinal regression with adjustment for the International Mission for Prognosis and Analysis of Clinical Trials in TBI variables and extracranial injury. We assessed whether the effect of intubation varied by injury severity by testing the added value of an interaction term with likelihood ratio tests. RESULTS: In the prehospital analysis, 890/3736 (24%) patients had their tracheas intubated at scene. In the in-hospital analysis, 460/2930 (16%) patients had their tracheas intubated in the emergency department. There was no adjusted overall effect on functional outcome of prehospital intubation (odds ratio=1.01; 95% confidence interval, 0.79-1.28; P=0.96), and the adjusted overall effect of in-hospital intubation was not significant (odds ratio=0.86; 95% confidence interval, 0.65-1.13; P=0.28). However, prehospital intubation was associated with better functional outcome in patients with higher thorax and abdominal Abbreviated Injury Scale scores (P=0.009 and P=0.02, respectively), whereas in-hospital intubation was associated with better outcome in patients with lower Glasgow Coma Scale scores (P=0.01): in-hospital intubation was associated with better functional outcome in patients with Glasgow Coma Scale scores of 10 or lower. CONCLUSION: The benefits and harms of tracheal intubation should be carefully evaluated in patients with TBI to optimise benefit. This study suggests that extracranial injury should influence the decision in the prehospital setting, and level of consciousness in the in-hospital setting. CLINICAL TRIAL REGISTRATION: NCT02210221.

Prevalence of post-concussion-like symptoms in the general population in Italy, The Netherlands and the United Kingdom.

Objectives: To evaluate the frequency of post-concussion symptoms and prevalence and risk factors of post-concussion syndrome (PCS) in the general population, investigate the association between the Rivermead Post-Concussion Symptoms Questionnaire (RPQ) and self-perceived health, and evaluate differences between three European countries. Methods: A web-based survey including the RPQ and EQ-5D was conducted among representative samples in three European countries. Results: A total of 11,759 respondents completed the questionnaire. The most frequently reported symptom was fatigue (49.9%). Almost half (45.1%) of the respondents were classified as having PCS considering rating score 2 (three RPQ items with score ≥ 2) as a cut-off. Chronic health complaints were found as a significant risk factor for PCS. All items of the RPQ were positively correlated with the EQ-5D and the strongest positive correlation (0.633, p<0.001) was between RPQ item 'feeling depressed or tearful' and EQ-5D domain 'anxiety/depression'. Conclusions: We found a high frequency of post-concussion-like symptoms and PCS in the general population, indicating that these symptoms are not specific for patients with traumatic brain injury (TBI), and PCS is not a unique syndrome after TBI. Therefore, the use of post-concussion symptoms and PCS as outcome following mild TBI should be interpreted with caution.

How to Study the Brain While Anesthetizing It?! A Scoping Review on Running Neuroanesthesiologic Studies and Trials That Include Neurosurgical Patients.

This scoping review addresses the challenges of neuroanesthesiologic research: the population, the methods/treatment/exposure, and the outcome/results. These challenges are put into the context of a future research agenda for peri-/intraoperative anesthetic management, neurocritical care, and applied neurosciences. Finally, the opportunities of adaptive trial design in neuroanesthesiologic research are discussed.

Development and internal validation of a clinical prediction model using machine learning algorithms for 90 day and 2 year mortality in femoral neck fracture patients aged 65 years or above.

PURPOSE: Preoperative prediction of mortality in femoral neck fracture patients aged 65 years or above may be valuable in the treatment decision-making. A preoperative clinical prediction model can aid surgeons and patients in the shared decision-making process, and optimize care for elderly femoral neck fracture patients. This study aimed to develop and internally validate a clinical prediction model using machine learning (ML) algorithms for 90 day and 2 year mortality in femoral neck fracture patients aged 65 years or above. METHODS: A retrospective cohort study at two trauma level I centers and three (non-level I) community hospitals was conducted to identify patients undergoing surgical fixation for a femoral neck fracture. Five different ML algorithms were developed and internally validated and assessed by discrimination, calibration, Brier score and decision curve analysis. RESULTS: In total, 2478 patients were included with 90 day and 2 year mortality rates of 9.1% (n = 225) and 23.5% (n = 582) respectively. The models included patient characteristics, comorbidities and laboratory values. The stochastic gradient boosting algorithm had the best performance for 90 day mortality prediction, with good discrimination (c-statistic = 0.74), calibration (intercept = - 0.05, slope = 1.11) and Brier score (0.078). The elastic-net penalized logistic regression algorithm had the best performance for 2 year mortality prediction, with good discrimination (c-statistic = 0.70), calibration (intercept = - 0.03, slope = 0.89) and Brier score (0.16). The models were incorporated into a freely available web-based application, including individual patient explanations for interpretation of the model to understand the reasoning how the model made a certain prediction: https://sorg-apps.shinyapps.io/hipfracturemortality/ CONCLUSIONS: The clinical prediction models show promise in estimating mortality prediction in elderly femoral neck fracture patients. External and prospective validation of the models may improve surgeon ability when faced with the treatment decision-making. LEVEL OF EVIDENCE: Prognostic Level II.

Feasibility of Machine Learning and Logistic Regression Algorithms to Predict Outcome in Orthopaedic Trauma Surgery.

BACKGROUND: Statistical models using machine learning (ML) have the potential for more accurate estimates of the probability of binary events than logistic regression. The present study used existing data sets from large musculoskeletal trauma trials to address the following study questions: (1) Do ML models produce better probability estimates than logistic regression models? (2) Are ML models influenced by different variables than logistic regression models? METHODS: We created ML and logistic regression models that estimated the probability of a specific fracture (posterior malleolar involvement in distal spiral tibial shaft and ankle fractures, scaphoid fracture, and distal radial fracture) or adverse event (subsequent surgery [after distal biceps repair or tibial shaft fracture], surgical site infection, and postoperative delirium) using 9 data sets from published musculoskeletal trauma studies. Each data set was split into training (80%) and test (20%) subsets. Fivefold cross-validation of the training set was used to develop the ML models. The best-performing model was then assessed in the independent testing data. Performance was assessed by (1) discrimination (c-statistic), (2) calibration (slope and intercept), and (3) overall performance (Brier score). RESULTS: The mean c-statistic was 0.01 higher for the logistic regression models compared with the best ML models for each data set (range, -0.01 to 0.06). There were fewer variables strongly associated with variation in the ML models, and many were dissimilar from those in the logistic regression models. CONCLUSIONS: The observation that ML models produce probability estimates comparable with logistic regression models for binary events in musculoskeletal trauma suggests that their benefit may be limited in this context.

Missing Data in Prediction Research: A Five-Step Approach for Multiple Imputation, Illustrated in the CENTER-TBI Study.

In medical research, missing data is common. In acute diseases, such as traumatic brain injury (TBI), even well-conducted prospective studies may suffer from missing data in baseline characteristics and outcomes. Statistical models may simply drop patients with any missing values, potentially leaving a selected subset of the original cohort. Imputation is widely accepted by methodologists as an appropriate way to deal with missing data. We aim to provide practical guidance on handling missing data for prediction modeling. We hereto propose a five-step approach, centered around single and multiple imputation: 1) explore the missing data patterns; 2) choose a method of imputation; 3) perform imputation; 4) assess diagnostics of the imputation; and 5) analyze the imputed data sets. We illustrate these five steps with the estimation and validation of the IMPACT (International Mission on Prognosis and Analysis of Clinical Trials in Traumatic Brain Injury) prognostic model in 1375 patients from the CENTER-TBI database, included in 53 centers across 17 countries, with moderate or severe TBI in the prospective European CENTER-TBI study. Future prediction modeling studies in acute diseases may benefit from following the suggested five steps for optimal statistical analysis and interpretation, after maximal effort has been made to minimize missing data.

Primary versus early secondary referral to a specialized neurotrauma center in patients with moderate/severe traumatic brain injury: a CENTER TBI study.

BACKGROUND: Prehospital care for patients with traumatic brain injury (TBI) varies with some emergency medical systems recommending direct transport of patients with moderate to severe TBI to hospitals with specialist neurotrauma care (SNCs). The aim of this study is to assess variation in levels of early secondary referral within European SNCs and to compare the outcomes of directly admitted and secondarily transferred patients. METHODS: Patients with moderate and severe TBI (Glasgow Coma Scale < 13) from the prospective European CENTER-TBI study were included in this study. All participating hospitals were specialist neuroscience centers. First, adjusted between-country differences were analysed using random effects logistic regression where early secondary referral was the dependent variable, and a random intercept for country was included. Second, the adjusted effect of early secondary referral on survival to hospital discharge and functional outcome [6 months Glasgow Outcome Scale Extended (GOSE)] was estimated using logistic and ordinal mixed effects models, respectively. RESULTS: A total of 1347 moderate/severe TBI patients from 53 SNCs in 18 European countries were included. Of these 1347 patients, 195 (14.5%) were admitted after early secondary referral. Secondarily referred moderate/severe TBI patients presented more often with a CT abnormality: mass lesion (52% vs. 34%), midline shift (54% vs. 36%) and acute subdural hematoma (77% vs. 65%). After adjusting for case-mix, there was a large European variation in early secondary referral, with a median OR of 1.69 between countries. Early secondary referral was not associated with functional outcome (adjusted OR 1.07, 95% CI 0.78-1.69), nor with survival at discharge (1.05, 0.58-1.90). CONCLUSIONS: Across Europe, substantial practice variation exists in the proportion of secondarily referred TBI patients at SNCs that is not explained by case mix. Within SNCs early secondary referral does not seem to impact functional outcome and survival after stabilisation in a non-specialised hospital. Future research should identify which patients with TBI truly benefit from direct transportation.

Machine learning algorithms performed no better than regression models for prognostication in traumatic brain injury.

OBJECTIVE: We aimed to explore the added value of common machine learning (ML) algorithms for prediction of outcome for moderate and severe traumatic brain injury. STUDY DESIGN AND SETTING: We performed logistic regression (LR), lasso regression, and ridge regression with key baseline predictors in the IMPACT-II database (15 studies, n = 11,022). ML algorithms included support vector machines, random forests, gradient boosting machines, and artificial neural networks and were trained using the same predictors. To assess generalizability of predictions, we performed internal, internal-external, and external validation on the recent CENTER-TBI study (patients with Glasgow Coma Scale <13, n = 1,554). Both calibration (calibration slope/intercept) and discrimination (area under the curve) was quantified. RESULTS: In the IMPACT-II database, 3,332/11,022 (30%) died and 5,233(48%) had unfavorable outcome (Glasgow Outcome Scale less than 4). In the CENTER-TBI study, 348/1,554(29%) died and 651(54%) had unfavorable outcome. Discrimination and calibration varied widely between the studies and less so between the studied algorithms. The mean area under the curve was 0.82 for mortality and 0.77 for unfavorable outcomes in the CENTER-TBI study. CONCLUSION: ML algorithms may not outperform traditional regression approaches in a low-dimensional setting for outcome prediction after moderate or severe traumatic brain injury. Similar to regression-based prediction models, ML algorithms should be rigorously validated to ensure applicability to new populations.