Updating and recalibrating causal probabilistic models on a new target population.

OBJECTIVE: Very often the performance of a Bayesian Network (BN) is affected when applied to a new target population. This is mainly because of differences in population characteristics. External validation of the model performance on different populations is a standard approach to test model's generalisability. However, a good predictive performance is not enough to show that the model represents the unique population characteristics and can be adopted in the new environment. METHODS: In this paper, we present a methodology for updating and recalibrating developed BN models - both their structure and parameters - to better account for the characteristics of the target population. Attention has been given on incorporating expert knowledge and recalibrating latent variables, which are usually omitted from data-driven models. RESULTS: The method is successfully applied to a clinical case study about the prediction of trauma-induced coagulopathy, where a BN has already been developed for civilian trauma patients and now it is recalibrated on combat casualties. CONCLUSION: The methodology proposed in this study is important for developing credible models that can demonstrate a good predictive performance when applied to a target population. Another advantage of the proposed methodology is that it is not limited to data-driven techniques and shows how expert knowledge can also be used when updating and recalibrating the model.

Current knowledge and availability of machine learning across the spectrum of trauma science.

PURPOSE OF REVIEW: Recent technological advances have accelerated the use of Machine Learning in trauma science. This review provides an overview on the available evidence for research and patient care. The review aims to familiarize clinicians with this rapidly evolving field, offer perspectives, and identify existing and future challenges. RECENT FINDINGS: The available evidence predominantly focuses on retrospective algorithm construction to predict outcomes. Few studies have explored actionable outcomes, workflow integration, or the impact on patient care. Machine Learning and data science have the potential to simplify data capture and enhance counterfactual causal inference research from observational data to address complex issues. However, regulatory, legal, and ethical challenges associated with the use of Machine Learning in trauma care deserve particular attention. SUMMARY: Machine Learning holds promise for actionable decision support in trauma science, but rigorous proof-of-concept studies are urgently needed. Future research should assess workflow integration, human-machine interaction, and, most importantly, the impact on patient outcome. Machine Learning enhanced causal inference for observational data carries an enormous potential to change trauma research as complement to randomized studies. The scientific trauma community needs to engage with the existing challenges to drive progress in the field.

Understanding pre-hospital blood transfusion decision-making for injured patients: an interview study.

BACKGROUND: Blood transfusion for bleeding trauma patients is a promising pre-hospital intervention with potential to improve outcomes. However, it is not yet clear which patients may benefit from pre-hospital transfusions. The aim of this study was to enhance our understanding of how experienced pre-hospital clinicians make decisions regarding patient blood loss and the need for transfusion, and explore the factors that influence clinical decision-making. METHODS: Pre-hospital physicians, from two air ambulance sites in the south of England, were interviewed between December 2018 and January 2019. Participants were involved in teaching or publishing on the management of bleeding trauma patients and had at least 5 years of continuous and contemporary practice at consultant level. Interviews were semi-structured and explored how decisions were made and what made decisions difficult. A qualitative description approach was used with inductive thematic analysis to identify themes and subthemes related to blood transfusion decision-making in trauma. RESULTS: Ten pre-hospital physicians were interviewed and three themes were identified: recognition-primed analysis, uncertainty and imperfect decision analysis. The first theme describes how participants make decisions using selected cues, incorporating their experience and are influenced by external rules and group expectations. What made decisions difficult for the participants was encapsulated in the uncertainty theme. Uncertainty emerged regarding the patient's true underlying physiological state and the treatment effect of blood transfusion. The last theme focuses on the issues with decision-making itself. Participants demonstrated lapses in decision awareness, often incomplete decision evaluation and described challenges to effective learning due to incomplete patient outcome information. CONCLUSION: Pre-hospital clinicians make decisions about bleeding and transfusion which are recognition-primed and incorporate significant uncertainty. Decisions are influenced by experience and are subject to bias. Improved understanding of the decision-making processes provides a theoretical perspective of how decisions might be supported in the future.

Causes and associations with mortality in patients with pelvic ring injuries with haemorrhagic shock.

BACKGROUND: High energy pelvic ring injuries are associated with significant morbidity and mortality and can be accompanied by haemorrhagic shock following associated vascular injury. This study evaluated the causes and predictors of mortality in haemodynamically unstable pelvic fractures. METHODS: This retrospective observational study at a Major Trauma Centre reviewed 938 consecutive adult patients (≥ 18yrs) with pelvic ring injuries between December 2014 and November 2018. Patients with features of haemorrhagic shock were included, defined as: arrival Systolic BP < 90 mmHg, Base Deficit ≥ 6.0 mmol/l, or transfusion of ≥ 4 units of packed red blood cells within 24 h. RESULTS: Of the 102 patients included, all sustained injuries from high energy trauma, and 47.1% underwent a haemorrhage control intervention (Resuscitative Endovascular Balloon Occlusion of the Aorta-REBOA, Interventional Radiology-IR, or Laparotomy). These were more often required following vertical shear injuries (OR 10.7, p = 0.036). Overall, 33 patients (32.4%) died; 16 due to a head injury, and only 2 directly from acute pelvic exsanguination (6.1%). Multivariable logistic regression demonstrated that increasing age, Injury Severity Score, Abbreviated Injury Scale (AIS) Head ≥ 3 and open pelvic fracture were all independent predictors of mortality, and IR was associated with reduced mortality. Lateral Compression III (LC3) injuries were associated with mortality due to multiple organ dysfunction syndrome (MODS). CONCLUSION: Haemodynamically unstable patients with pelvic ring injuries have a high mortality rate, but death is usually attributed to other injuries or later complications, and not from acute exsanguination. This reflects improvements in resuscitative care, transfusion protocols, and haemorrhage control techniques.

Impact of Ischemia Duration on Lower Limb Salvage in Combat Casualties.

INTRODUCTION: The 6-hour threshold to revascularization of an ischemic limb is ubiquitous in the trauma literature, however, contemporary evidence suggests that this threshold should be less. This study aims to characterize the relationship between the duration of limb ischemia and successful limb salvage following lower extremity arterial trauma. METHODS: This is a cohort study of the United States and UK military service members injured while serving in Iraq or Afghanistan between 2003 and 2013. Consecutive patients who sustained iliac, femoral, or popliteal artery injuries, and underwent surgery to attempt revascularization, were included. The association between limb outcome and the duration of limb ischemia was assessed using the Kaplan-Meier method. RESULTS: One hundred twenty-two patients (129 limbs) who sustained iliac (2.3%), femoral (56.6%), and popliteal (41.1%) arterial injuries were included. Overall, 87 limbs (67.4%) were successfully salvaged. The probability of limb salvage was 86.0% when ischemia was ≤1 hour; 68.3% when between 1 and 3 hours; 56.3% when between 3 and 6 hours; and 6.7% when >6 hours ( P <0.0001). Shock more than doubled the risk of failed limb salvage [hazard ratio=2.42 (95% confidence interval: 1.27-4.62)]. CONCLUSIONS: Limb salvage is critically dependent on the duration of ischemia with a 10% reduction in the probability of successful limb salvage for every hour delay to revascularization. The presence of shock significantly worsens this relationship. Military trauma systems should prioritize rapid hemorrhage control and early limb revascularization within 1 hour of injury.

Early Identification of Trauma-induced Coagulopathy: Development and Validation of a Multivariable Risk Prediction Model.

OBJECTIVE: The aim of this study was to develop and validate a risk prediction tool for trauma-induced coagulopathy (TIC), to support early therapeutic decision-making. BACKGROUND: TIC exacerbates hemorrhage and is associated with higher morbidity and mortality. Early and aggressive treatment of TIC improves outcome. However, injured patients that develop TIC can be difficult to identify, which may compromise effective treatment. METHODS: A Bayesian Network (BN) prediction model was developed using domain knowledge of the causal mechanisms of TIC, and trained using data from 600 patients recruited into the Activation of Coagulation and Inflammation in Trauma (ACIT) study. Performance (discrimination, calibration, and accuracy) was tested using 10-fold cross-validation and externally validated on data from new patients recruited at 3 trauma centers. RESULTS: Rates of TIC in the derivation and validation cohorts were 11.8% and 11.0%, respectively. Patients who developed TIC were significantly more likely to die (54.0% vs 5.5%, P < 0.0001), require a massive blood transfusion (43.5% vs 1.1%, P < 0.0001), or require damage control surgery (55.8% vs 3.4%, P < 0.0001), than those with normal coagulation. In the development dataset, the 14-predictor BN accurately predicted this high-risk patient group: area under the receiver operating characteristic curve (AUROC) 0.93, calibration slope (CS) 0.96, brier score (BS) 0.06, and brier skill score (BSS) 0.40. The model maintained excellent performance in the validation population: AUROC 0.95, CS 1.22, BS 0.05, and BSS 0.46. CONCLUSIONS: A BN (http://www.traumamodels.com) can accurately predict the risk of TIC in an individual patient from standard admission clinical variables. This information may support early, accurate, and efficient activation of hemostatic resuscitation protocols.

Effectiveness of intubation devices in patients with cervical spine immobilisation: a systematic review and network meta-analysis.

BACKGROUND: Cervical spine immobilisation increases the difficulty of tracheal intubation. Many intubation devices have been evaluated in this setting, but their relative performance remains uncertain. METHODS: MEDLINE, EMBASE, and the Cochrane Library were searched to identify randomised trials comparing two or more intubation devices in adults with cervical spine immobilisation. After critical appraisal, a random-effects network meta-analysis was used to pool and compare device performance. The primary outcome was the probability of first-attempt intubation success (first-pass success). For relative performance, the Macintosh direct laryngoscopy blade was chosen as the reference device. RESULTS: We included 80 trials (8039 subjects) comparing 26 devices. Compared with the Macintosh, McGrath™ (odds ratio [OR]=11.5; 95% credible interval [CrI] 3.19-46.20), C-MAC D Blade™ (OR=7.44; 95% CrI, 1.06-52.50), Airtraq™ (OR=5.43; 95% CrI, 2.15-14.2), King Vision™ (OR=4.54; 95% CrI, 1.28-16.30), and C-MAC™ (OR=4.20; 95% CrI=1.28-15.10) had a greater probability of first-pass success. This was also true for the GlideScope™ when a tube guide was used (OR=3.54; 95% CrI, 1.05-12.50). Only the Airway Scope™ had a better probability of first-pass success compared with the Macintosh when manual-in-line stabilisation (MILS) was used as the immobilisation technique (OR=7.98; 95% CrI, 1.06-73.00). CONCLUSIONS: For intubation performed with cervical immobilisation, seven devices had a better probability of first-pass success compared with the Macintosh. However, more studies using MILS (rather than a cervical collar or other alternative) are needed, which more accurately represent clinical practice. CLINICAL TRIAL REGISTRATION: PROSPERO 2019 CRD42019158067 (https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=158067).

Predicting the Outcome of Limb Revascularization in Patients With Lower-extremity Arterial Trauma: Development and External Validation of a Supervised Machine-learning Algorithm to Support Surgical Decisions.

OBJECTIVES: Estimating the likely success of limb revascularization in patients with lower-extremity arterial trauma is central to decisions between attempting limb salvage and amputation. However, the projected outcome is often unclear at the time these decisions need to be made, making them difficult and threatening sound judgement. The objective of this study was to develop and validate a prediction model that can quantify an individual patient's risk of failed revascularization. METHODS: A BN prognostic model was developed using domain knowledge and data from the US joint trauma system. Performance (discrimination, calibration, and accuracy) was tested using ten-fold cross validation and externally validated on data from the UK Joint Theatre Trauma Registry. BN performance was compared to the mangled extremity severity score. RESULTS: Rates of amputation performed because of nonviable limb tissue were 12.2% and 19.6% in the US joint trauma system (n = 508) and UK Joint Theatre Trauma Registry (n = 51) populations respectively. A 10-predictor BN accurately predicted failed revascularization: area under the receiver operating characteristic curve (AUROC) 0.95, calibration slope 1.96, Brier score (BS) 0.05, and Brier skill score 0.50. The model maintained excellent performance in an external validation population: AUROC 0.97, calibration slope 1.72, Brier score 0.08, Brier skill score 0.58, and had significantly better performance than mangled extremity severity score at predicting the need for amputation [AUROC 0.95 (0.92-0.98) vs 0.74 (0.67-0.80); P < 0.0001]. CONCLUSIONS: A BN (https://www.traumamodels.com) can accurately predict the outcome of limb revascularization at the time of initial wound evaluation. This information may complement clinical judgement, support rational and shared treatment decisions, and establish sensible treatment expectations.

Diagnostic performance of EUS in non-jaundiced patients with an incidental finding of double duct sign on cross-sectional imaging: A systematic review and meta-analysis.

BACKGROUND/OBJECTIVES: Dilatation of the pancreatic duct and common bile duct, known as double duct sign (DDS), suggests sinister pathology at the periampullary region. Non-jaundiced patients with incidental DDS and otherwise normal cross-sectional imaging present a diagnostic dilemma to the multidisciplinary team (MDT). The primary aim of this review was to assess the diagnostic yield of endoscopic ultrasound (EUS) in detecting causal pathology in this patient population. METHODS: A systematic literature search (Medline, EMBASE, Google Scholar, Cochrane database and PROSPERO) was performed to identify original studies that reported EUS findings in patients with incidental DDS. Primary outcome was detection of a periampullary tumour. Secondary outcome was detection of benign causal pathology. Meta-analysis was used to calculate an absolute measure (pooled proportion) of pathology detection. RESULTS: Four studies (177 patients) were included. EUS detection rate for a periampullary tumour was 5% (95% CI, 0-10%) including both adenocarcinomas and adenomas. EUS detection rate for benign causal pathology was 22% (95% CI, 10-34%), the most common being chronic pancreatitis CONCLUSION: Non-jaundiced patients with incidental DDS on cross-sectional imaging have a 5% risk of a periampullary tumour that can be detected by EUS.

Management and outcome of 597 wartime penetrating lower extremity arterial injuries from an international military cohort.

OBJECTIVE: Vascular injury is a leading cause of death and disability in military and civilian settings. Most wartime and an increasing amount of civilian vascular trauma arises from penetrating mechanisms of injury due to gunshot or explosion. The objective of this study was to provide a comprehensive examination of penetrating lower extremity arterial injury and to characterize long-term limb salvage and differences related to mechanisms of injury. METHODS: The military trauma registries of the United States and the United Kingdom were analyzed to identify service members who sustained penetrating lower limb arterial injury (2001-2014). Treatment and limb salvage data were studied and comparisons made of patients whose penetrating vascular trauma arose from explosion (group 1) vs gunshot (group 2). Standardized statistical testing was used, with Bonferroni corrections for multiple comparisons. RESULTS: The cohort consisted of 568 combat casualties (mean age, 25.2 years) with 597 injuries (explosion, n = 416; gunshot, n = 181). Group 1 had higher Injury Severity Score (P < .05) and Mangled Extremity Severity Score (P < .0001), required more blood transfusion (P < .05), and had more tibial (P < .01) and popliteal (P < .05) arterial injuries; group 2 had more profunda femoris injuries (P < .05). Initial surgical management for the whole cohort included vein interposition graft (33%), ligation (31%), primary repair with or without patch angioplasty (16%), temporary vascular shunting (15%), and primary amputation (6%). No difference in patency of arterial reconstruction was found between group 1 and group 2, although group 1 had a higher incidence of primary (13% vs 2%; P < .05) and secondary (19% vs 9%; P < .05) amputation. Similarly, longer term freedom from amputation was lower for group 1 than for group 2 (68% vs 89% at 5.5 years; Cox hazard ratio, 0.30; P < .0001), as was physical functioning (36-Item Short Form Health Survey data; mean, 39.80 vs 43.20; P < .05). CONCLUSIONS: The majority of wartime lower extremity arterial injuries result from an explosive mechanism that preferentially affects the tibial vasculature and results in poorer long-term limb salvage compared with those injured with firearms. The mortality associated with immediate limb salvage attempts is low, and delayed amputations occur weeks later, affording the patient involvement in the decision-making and rehabilitation planning. We recommend assertive attempts at vascular repair and limb salvage for service members injured by explosive and gunshot mechanisms.

Trauma-associated acute kidney injury.

PURPOSE OF REVIEW: A summary of recent research into the epidemiology, cause, management and outcomes of trauma-associated acute kidney injury (AKI). There is an increasing focus on subtypes of AKI to better target clinical management and future research. RECENT FINDINGS: AKI associated with trauma occurs in 20-24% of patients admitted to ICU. On the basis of creatinine and/or urine output, AKI occurs in the first few days of traumatic illness. Although various associations have been identified, shock and high-volume blood transfusion are the most consistent risks for development of trauma-associated AKI. Short-term outcomes appear worse for patients with AKI, but extent of longer term kidney function recovery remains unknown. Recent research in the general critical care population is beginning to better inform AKI management; however, currently, preventive and supportive strategies remain the mainstay of AKI management after trauma. SUMMARY: Well-designed, prospective research is required to better understand the phenotype, pathophysiology and recovery trajectory of trauma-associated AKI. Only then can potentially unique therapeutic targets be developed for this common subtype of AKI.

Prehospital emergency anaesthesia: an updated survey of UK practice with emphasis on the role of standardisation and checklists.

INTRODUCTION: Prehospital emergency anaesthesia (PHEA or 'prehospital rapid sequence intubation') is a high-risk procedure. Standard operating procedures (SOPs) and checklists within healthcare systems have been demonstrated to reduce human error and improve patient safety. We aimed to describe the current practice of PHEA in the UK, determine the use of checklists for PHEA and describe the content, format and layout of any such checklists currently used in the UK. METHOD: A survey of UK prehospital teams was conducted to establish the incidence and conduct of PHEA practice. Results were grouped into systems delivering a high volume of PHEA per year (>50 PHEAs) and low volume (≤50 PHEAs per annum). Standard and 'crash' (immediate) induction checklists were reviewed for length, content and layout. RESULTS: 59 UK physician-led prehospital services were identified of which 43 (74%) participated. Thirty services (70%) provide PHEA and perform approximately 1629 PHEAs annually. Ten 'high volume' services deliver 84% of PHEAs per year with PHEA being performed on a median of 11% of active missions. The most common indication for PHEA was trauma. 25 of the 30 services (83%) used a PHEA checklist prior to induction of anaesthesia and 24 (80%) had an SOP for the procedure. 19 (76%) of the 'standard' checklists and 5 (50%) of the 'crash' induction checklists used were analysed. On average, standard checklists contained 169 (range: 52-286) words and 41 (range: 28-70) individual checks. The style and language complexity varied significantly between different checklists. CONCLUSION: PHEA is now performed commonly in the UK. The use of checklists for PHEA is relatively common among prehospital systems delivering this intervention. Care must be taken to limit checklist length and to use simple, unambiguous language in order to maximise the safety of this high-risk intervention.

A systematic review and meta-analysis comparing mortality in pre-hospital tracheal intubation to emergency department intubation in trauma patients.

BACKGROUND: Pre-hospital endotracheal intubation is frequently used for trauma patients in many emergency medical systems. Despite a wide range of publications in the field, it is debated whether the intervention is associated with a favourable outcome, when compared to more conservative airway measures. METHODS: A systematic literature search was conducted to identify interventional and observational studies where the mortality rates of adult trauma patients undergoing pre-hospital endotracheal intubation were compared to those undergoing emergency department intubation. RESULTS: Twenty-one studies examining 35,838 patients were included. The median mortality rate in patients undergoing pre-hospital intubation was 48% (range 8-94%), compared to 29% (range 6-67%) in patients undergoing intubation in the emergency department. Odds ratios were in favour of emergency department intubation both in crude and adjusted mortality, with 2.56 (95% CI: 2.06, 3.18) and 2.59 (95% CI: 1.97, 3.39), respectively. The overall quality of evidence is very low. Twelve of the twenty-one studies found a significantly higher mortality rate after pre-hospital intubation, seven found no significant differences, one found a positive effect, and for one study an analysis of the mortality rate was beyond the scope of the article. CONCLUSIONS: The rationale for wide and unspecific indications for pre-hospital intubation seems to lack support in the literature, despite several publications involving a relatively large number of patients. Pre-hospital intubation is a complex intervention where guidelines and research findings should be approached cautiously. The association between pre-hospital intubation and a higher mortality rate does not necessarily contradict the importance of the intervention, but it does call for a thorough investigation by clinicians and researchers into possible causes for this finding.

Resuscitative endovascular balloon occlusion of the aorta: promise, practice, and progress?

PURPOSE OF REVIEW: Resuscitative endovascular balloon occlusion of the aorta (REBOA) is a minimally invasive damage control procedure for life-threatening abdominal or pelvic haemorrhage. The purpose of this review is to summarize the current understanding and experience with REBOA, outline potential future applications of this technology, and highlight priority areas for further research. RECENT FINDINGS: REBOA is a feasible method of achieving temporary aortic occlusion and can be performed rapidly, with a high degree of success, in the emergency setting (including at the scene of injury) by appropriately trained clinicians. The procedure supports central perfusion, controls noncompressible haemorrhage, and may improve survival in certain profoundly shocked patient groups; but is also associated with significant risks, including ischaemic tissue damage and procedural complications. Evolutions of this strategy are being explored, with promising proof-of-concept studies in the fields of partial aortic occlusion and the combination of REBOA with extracorporeal support. SUMMARY: Noncompressible torso haemorrhage is the leading cause of preventable trauma deaths. The majority of these deaths occur soon after injury, often before any opportunity for definitive haemorrhage control. For a meaningful reduction in trauma mortality, novel methods of rapid haemorrhage control are required.

Significant modification of traditional rapid sequence induction improves safety and effectiveness of pre-hospital trauma anaesthesia.

INTRODUCTION: Rapid Sequence Induction of anaesthesia (RSI) is the recommended method to facilitate emergency tracheal intubation in trauma patients. In emergency situations, a simple and standardised RSI protocol may improve the safety and effectiveness of the procedure. A crucial component of developing a standardised protocol is the selection of induction agents. The aim of this study is to compare the safety and effectiveness of a traditional RSI protocol using etomidate and suxamethonium with a modified RSI protocol using fentanyl, ketamine and rocuronium. METHODS: We performed a comparative cohort study of major trauma patients undergoing pre-hospital RSI by a physician-led Helicopter Emergency Medical Service. Group 1 underwent RSI using etomidate and suxamethonium and Group 2 underwent RSI using fentanyl, ketamine and rocuronium. Apart from the induction agents, the RSI protocol was identical in both groups. Outcomes measured included laryngoscopy view, intubation success, haemodynamic response to laryngoscopy and tracheal intubation, and mortality. RESULTS: Compared to Group 1 (n = 116), Group 2 RSI (n = 145) produced significantly better laryngoscopy views (p = 0.013) and resulted in significantly higher first-pass intubation success (95% versus 100%; p = 0.007). A hypertensive response to laryngoscopy and tracheal intubation was less frequent following Group 2 RSI (79% versus 37%; p < 0.0001). A hypotensive response was uncommon in both groups (1% versus 6%; p = 0.05). Only one patient in each group developed true hypotension (SBP < 90 mmHg) on induction. CONCLUSIONS: In a comparative, cohort study, pre-hospital RSI using fentanyl, ketamine and rocuronium produced superior intubating conditions and a more favourable haemodynamic response to laryngoscopy and tracheal intubation. An RSI protocol using fixed ratios of these agents delivers effective pre-hospital trauma anaesthesia.

Not just data: a method for improving prediction with knowledge.

Many medical conditions are only indirectly observed through symptoms and tests. Developing predictive models for such conditions is challenging since they can be thought of as 'latent' variables. They are not present in the data and often get confused with measurements. As a result, building a model that fits data well is not the same as making a prediction that is useful for decision makers. In this paper, we present a methodology for developing Bayesian network (BN) models that predict and reason with latent variables, using a combination of expert knowledge and available data. The method is illustrated by a case study into the prediction of acute traumatic coagulopathy (ATC), a disorder of blood clotting that significantly increases the risk of death following traumatic injuries. There are several measurements for ATC and previous models have predicted one of these measurements instead of the state of ATC itself. Our case study illustrates the advantages of models that distinguish between an underlying latent condition and its measurements, and of a continuing dialogue between the modeller and the domain experts as the model is developed using knowledge as well as data.

Combining data and meta-analysis to build Bayesian networks for clinical decision support.

Complex clinical decisions require the decision maker to evaluate multiple factors that may interact with each other. Many clinical studies, however, report 'univariate' relations between a single factor and outcome. Such univariate statistics are often insufficient to provide useful support for complex clinical decisions even when they are pooled using meta-analysis. More useful decision support could be provided by evidence-based models that take the interaction between factors into account. In this paper, we propose a method of integrating the univariate results of a meta-analysis with a clinical dataset and expert knowledge to construct multivariate Bayesian network (BN) models. The technique reduces the size of the dataset needed to learn the parameters of a model of a given complexity. Supplementing the data with the meta-analysis results avoids the need to either simplify the model - ignoring some complexities of the problem - or to gather more data. The method is illustrated by a clinical case study into the prediction of the viability of severely injured lower extremities. The case study illustrates the advantages of integrating combined evidence into BN development: the BN developed using our method outperformed four different data-driven structure learning methods, and a well-known scoring model (MESS) in this domain.