INTRODUCTION:Tranexamic acid (TXA) is an antifibrinolytic drug that has been demonstrated to reduce head injury-related mortality when given within 2 h of injury in patients with traumatic brain injury and intracranial hemorrhage. It is usually administered via intravenous (IV) access, which can be difficult to obtain in prehospital and austere settings. Intraosseous (IO) access is fast and offers an alternative when IV access proves challenging; however, TXA administration via IO access has never been studied in humans. We sought to determine if the total drug exposure of TXA given in the prehospital setting in patients with moderate or severe brain injury differs based on route of administration. METHODS:We performed a retrospective analysis of prospectively collected data from the prehospital TXA for traumatic brain injury trial (NCT01990768). Participants who received TXA via IO administration were compared to those who received TXA via IV administration and stratified by renal function category based on the Kidney Disease Improving Global Outcomes criteria. The area under the plasma drug concentration-time curve (AUC) was calculated using the trapezoidal rule (Phoenix WinNonlin 8.3, Certara, Princeton NJ) to obtain total drug exposure. The inverse variance method was used to combine observations within strata and calculate mean differences. RESULTS:Of the 966 participants enrolled in the trial, 345 participants received a 2-g TXA prehospital bolus (11 IO, 334 IV); 312 participants received a 1-g TXA prehospital bolus followed by a 1-g TXA infusion in-hospital over 8 h (13 IO, 299 IV). After exclusion because of missing data and extreme estimated AUC, 233 IV and eight IO participants in the 2-g bolus arm and 152 IV and eight IO participants in the 1-g bolus 1-g infusion arm remained. Participants did not differ by age, sex, race, ethnicity, body mass index, serum creatinine, estimated glomerular filtration rate, or clot lysis at 30 min on thromboelastography. No difference in the mean AUCs were observed between IV and IO for either the 2-g bolus group (-2.6 μ g/mL/h [IO] compared to IV, 95% confidence interval: -28.4 to 23.3 μ g/mL/h) or the 1-g bolus/1-g infusion group (-13.0 μ g/mL/h [IO] compared to IV, 95% confidence interval: -236.2 to 210.3 μ g/mL/h). CONCLUSIONS:These preliminary data suggest that the administration of TXA via IO and IV routes may result in similar total drug exposure. Further studies incorporating larger numbers with clinical outcomes are needed to confirm this finding.

Disparities in breast cancer mortality persist despite improvements in screening and therapeutic options. Understanding the impact of social determinants of health on disparate breast cancer outcomes is challenging due to heterogeneity of prior assessments. We examined the association between social vulnerability and breast cancer stage at diagnosis and mortality using a standardized measure of population risk for external stressors on health. Using institutional cancer registry data, female patients aged 18 or older diagnosed with breast cancer between 2012 and 2019 were assigned a 2018 Social Vulnerability Index (SVI) rank based upon home address census tract. We used multinomial logistic regression and Cox proportional hazards model to examine the relationships between SVI and breast cancer stage at diagnosis and all-cause mortality. Covariates included age and, when assessing mortality, cancer stage, comorbidities, body mass index, insurance type, and treatment regimen. A total of 3,499 women with a median age of 59 (IQR 48–69) were included. 60

Abstract Background Optimizing resuscitation to reduce inflammation and organ dysfunction following human trauma-associated hemorrhagic shock is a major clinical hurdle. This is limited by the short duration of pre-clinical studies and the sparsity of early data in the clinical setting. Methods We sought to bridge this gap by linking preclinical data in a porcine model with clinical data from patients from the Prospective, Observational, Multicenter, Major Trauma Transfusion (PROMMTT) study via a three-compartment ordinary differential equation model of inflammation and coagulation. Results The mathematical model accurately predicts physiologic, inflammatory, and laboratory measures in both the porcine model and patients, as well as the outcome and time of death in the PROMMTT cohort. Model simulation suggests that resuscitation with plasma and red blood cells outperformed resuscitation with crystalloid or plasma alone, and that earlier plasma resuscitation reduced injury severity and increased survival time. Conclusions This workflow may serve as a translational bridge from pre-clinical to clinical studies in trauma-associated hemorrhagic shock and other complex disease settings.

BACKGROUND Brain specific biomarkers such as glial fibrillary acidic protein (GFAP), ubiquitin C-terminal hydrolase L1 (UCH-L1), and microtubule-associated protein-2 (MAP-2) have been identified as tools for diagnosis in traumatic brain injury (TBI). Tranexamic acid (TXA) has been shown to decrease mortality in patients with intracranial hemorrhage (ICH). The effect of TXA on these biomarkers is unknown. We investigated whether TXA affects levels of GFAP, UCH-L1, and MAP-2, and whether biomarker levels are associated with mortality in patients receiving TXA.METHODS Patients enrolled in the prehospital TXA for TBI trial had GFAP, UCHL-1 and MAP-2 levels drawn at 0 hour and 24 hours postinjury (n = 422). Patients with ICH from blunt trauma with a GCS <13 and SBP >90 were randomized to placebo, 2 g TXA bolus, or 1 g bolus +1 g/8 hours TXA infusion. Associations of TXA and 24-hour biomarker change were assessed with multivariate linear regression. Association of biomarkers with 28-day mortality was assessed with multivariate logistic regression. All models were controlled for age, GCS, ISS, and AIS head.RESULTS Administration of TXA was not associated with a change in biomarkers over 24 hours postinjury. Changes in biomarker levels were associated with AIS head and age. On admission, higher GFAP (odds ratio [OR], 1.75; confidence interval [CI], 1.31-2.38; p < 0.001) was associated with increased 28-day mortality. At 24 hours postinjury, higher levels of GFAP (OR, 2.09; CI, 1.37-3.30; p < 0.001 and UCHL-1 (OR, 2.98; CI, 1.77-5.25; p < 0.001) were associated with mortality. A change in UCH levels from 0 hour to 24 hours postinjury was also associated with increased mortality (OR, 1.68; CI, 1.15-2.49; p < 0.01).CONCLUSION Administration of TXA does not impact change in GFAP, UCHL-1, or MAP-2 during the first 24 hours after blunt TBI with ICH. Higher levels of GFAP and UCH early after injury may help identify patients at high risk for 28-day mortality.LEVEL OF EVIDENCE Therapeutic/Care Management; Level III.

Introduction Tranexamic acid (TXA) administered within 2 h of injury reduces mortality in traumatic brain injury (TBI) with intracranial hemorrhage. TXA also reduces the seizure threshold in a dose-dependent manner. We examined whether a 2-g bolus of prehospital TXA administered in moderate or severe TBI is associated with seizure activity within 72 h of injury. Methods Patients from the prehospital TXA for TBI trial with Glasgow Coma Scale < 13, blunt head injury, and time-of-seizure data were included in this analysis. The original trial randomized patients with suspected TBI to placebo, 1-g TXA bolus + 1-g 8-h TXA infusion, or 2-g TXA bolus within 2 h of injury. In this secondary analysis, multivariable logistic regression was performed to examine the association of treatment group with seizure incidence. The model controlled for age, Glasgow Coma Scale, Injury Severity Score, intracranial hemorrhage, Abbreviated Injury Scale-head, and home antiseizure medication use. Results Of the 786 patients who met the inclusion criteria, 19 had seizures within 72 h (five in placebo, two in 1-g bolus/1-g infusion, and 12 in 2-g bolus). The 2-g TXA bolus was not associated with increased seizures compared to placebo (odds ratio 0.41, 95% confidence interval 0.12-1.18, P = 0.12). Home antiseizure medication use was associated with increased seizures (odds ratio 15.95, 95% confidence interval 3.79-60.57, P < 0.001). Conclusions A prehospital 2-g TXA bolus in moderate or severe TBI was not associated with increased seizure activity during the first 72 h after injury; however, limited power, limited use of continuous electroencephalography, and unavailable seizure prophylaxis data highlight the need for further study.

Uncontrolled hemorrhage continues to be the most common cause of preventable deaths both for combat1, 2 and civilian3-5 injuries. Injury outcomes have improved with the establishment of trauma systems6 and damage control resuscitation (DCR) clinical practice guidelines (CPG).7-9 Additionally, implementing massive transfusion protocols (MTP) and maintaining a high ratio of plasma and platelets to red blood cells is associated with improved survival for critically injured patients.7, 10 However, early identification of patients with occult massive hemorrhage in time to prevent progressing hemorrhage, repay the oxygen debt, and halt the "bloody vicious cycle" of death during a team-based resuscitation across multiple phases of care proves difficult even under the best circumstances.11-14 Though challenges in DCR exist, several promising solutions may improve hemorrhagic shock recognition, optimize MTP triggers for early blood product mobilization, refine metrics used for optimal DCR practice, and improve adherence to guidelines. Clinical decision support systems (CDSS) represent a range of tools for handling and presenting information to clinical care teams in a way that improves diagnosis and treatment.15 For example, real-time DCR decision support systems can help trauma teams track blood products and adjuncts across multiple phases of patient care.16 Nudges change the architecture of critical choices leading to altered behavior in a predictable way without forbidding options or altering incentives.17 Today, the use of behavioral insights and nudge theory is common practice and used in fields like economics, politics, marketing, and even health care.18-20 With proper optimization for clinical workflow, nudge interventions have significant potential to improve both patient outcomes and the delivery of health care.18-21 Lastly, machine learning algorithms excel in real-time responsiveness and the iterative process helps the system adapt to new patterns often missed by the human eye. Machine learning has the potential to help us identify patients with occult bleeding early and navigate the intricacies and challenges of treating patients with traumatic injuries.22 This manuscript outlines the importance of DCR, identifies barriers in early identification of patients requiring massive transfusion and adherence to DCR CPGs, and highlights promising solutions for addressing these barriers so care teams can consistently execute optimized resuscitations. As with many other concepts in medicine, DCR emerged from innovations in military resuscitation practices and describes a strategic approach to the resuscitation of critically ill trauma patients.23, 24 DCR is based on the understanding that early and aggressive correction and prevention of metabolic derangements can mitigate against further hemorrhage and mortality (Table 1). Initial DCR protocols focused on permissive hypotension and early transfusion of blood products.24 However, the DCR concept evolved to include a more balanced approach to trauma resuscitation with a focus on rapid hemorrhage control, early and balanced blood product transfusion, and prevention of acidosis and hypothermia. Minimize blood loss with early hemorrhage control measures Transfuse blood components that optimize hemostasis Activate MTP Obtain functional laboratory measures of coagulation to refine ongoing resuscitation Give pharmacologic adjuncts to safely promote hemostasis The initial stages of resuscitation may consist of limited volumes of crystalloid fluid to expand plasma volume while awaiting blood products; however, large amounts of crystalloid fluid can lead to acidosis, hemodilution, and decreased oxygen delivery. Thus crystalloid resuscitation ultimately impedes hemostasis and fails to resolve the accumulating oxygen debt.25 Whole blood represents the optimal resuscitation fluid, providing the patient with physiologic ratios of coagulation factors, functioning platelets, and oxygen-carrying hemoglobin.26 In the absence of whole blood, a balanced, 1:1:1 ratio of packed red blood cells (PRBCs), plasma, and platelets should be started early, a practice supported by a large prospective cohort study and a randomized clinical trial.27, 28 However, this ratio-based treatment is not equivalent to whole blood for many reasons; for example, whole blood contains increased concentration of cellular components, lower amounts of anticoagulants and additives, increased oxygen carrying capacity, single-donor exposure, preserved platelet function, and other potential unmeasured benefits.29-31 Additionally, whole blood carries logistical advantages over administering multiple component units. At the same time, avoidance of over-resuscitation is essential to DCR success.32, 33 The "Lethal Diamond of Trauma" consists of hypothermia, acidosis, coagulopathy, and hypocalcemia.34, 35 The interplay among these four factors is complex and multifactorial, and each should be rapidly identified and corrected. For example, acidosis prevents adequate regulation of vascular tone and cardiac function leading to impaired circulatory function and hypothermia. Both acidosis and hypothermia contribute to coagulopathy by numerous mechanisms, such as impaired thrombin production and accelerated fibrinogen consumption.36-38 Acute hypocalcemia after injury, mediated by direct tissue damage and the chelating effect of blood product preservatives, is associated with prehospital hypotension, increased blood product administration, and acute traumatic coagulopathy (INR >1.5) and is more predictive of mortality than base deficit.39-42 Several studies identified ionized calcium (iCa) as an independent predictor for mortality.34 As such, iCa levels could serve as a substantial predictor for mortality. Current approaches to mitigating this lethal diamond include the empiric use of adjuncts such as calcium and tranexamic acid (TXA).43, 44 The optimal dosing strategy for calcium is not fully elucidated; however, the Joint Trauma Systems Damage Control Clinical Practice Guideline recommends first administering 1 g of calcium chloride or 3 g of calcium gluconate empirically before or with the first unit of blood and then repeating administration every four blood products.9 CRASH-2 demonstrated that early TXA (1 g over 10 min, followed by 1 g over 8 h), administered within 3 h of injury, was associated with a lower rate of early mortality.44 Recently, administering a single 2 g bolus of TXA has gained increased attention, and a systematic review and meta-analysis suggest that a single high-dose bolus may be associated with decreased transfusion requirements without increased complications.45 The care of trauma patients begins at the scene of the injury and prehospital providers play a crucial role in thwarting the ensuing lethal diamond.46 While prehospital blood product resuscitation has been associated with improved survival, the STAAMP trial found that prehospital TXA administration did not improve mortality at 30 days.47-50 Additionally, a recent trial of prehospital TXA showed no improvement in favorable functional outcomes at 6 months.51 In the CRYOSTAT-2 trial patients received three pools of cryoprecipitate within 3 hours of injury in addition to standard care; the investigators found no difference in all-cause mortality but also no difference in adverse events.52 Airway management in patients with hemorrhagic shock also requires a thoughtful and measured approach. While many of these patients will have compromised airways, due to either altered mental status or a contaminated airway, rapid sequence induction and positive pressure ventilation (PPV) can be perilous in a hypovolemic patient. Post-intubation, hypotension can result from several mechanisms. PPV reduces venous return due to increased intrathoracid pressure while loss of vasomotor tone results in vasodilation with associated decreased afterload and venous return.53, 54 Induction medications further suppress the release of catecholamines needed to maintain adequate perfusion.55 Definitive airway management with intubation should be delayed until after resuscitation has been initiated to prevent hemodynamic collapse.56, 57 Recent research highlights the importance of addressing hypovolemia prior to performing rapid sequence intubation (RSI) and will likely be reflected in future editions of ATLS.57, 58 To ensure ultimate patient survival, damage control surgery (DCS) must accompany DCR. In this synchronized approach, surgical hemostasis staunches ongoing hemorrhage while DCR mitigates or reverses trauma-induced coagulopathy. Definitive surgical repair or reconstruction then commences after all metabolic derangements of hemorrhage are corrected.59, 60 Bogert and colleagues astutely noted that DCS is truly a component of DCR, despite its origins being decades before the advent of DCR.23 The entire process of DCS occurs in conjunction with resuscitation practices guided by DCR concepts. Both systems are complex and require coordination among a team of healthcare providers along the entire continuum of care for the injured trauma patient. Modern-day DCR CPGs, including the American College of Surgeons (ACS) Committee on Trauma (COT) Massive Transfusion in Trauma Guidelines,7-9 include MTP with indications for activation, transfusion service processes, blood product ratios, adjunctive medications, termination of MTPs, and performance improvement (PI) metrics while considering multi-disciplinary viewpoints. The most commonly accepted DCR CPGs outside of the ACS include DCR in patients with severe traumatic hemorrhage: A practice management guideline from the Eastern Association for the Surgery of Trauma7 and the Joint Trauma System Damage Control Resuscitation Clinical Practice Guideline.9 The 10th edition of the Advanced Trauma Life Support guidelines acknowledges that massive infusion of crystalloid is associated with higher mortality rates and advocates for earlier blood product transfusion.61 To best approximate the blood lost by an exsanguinating trauma victim, blood product ratios should target a high ratio of plasma and platelets to PRBCs.7 AB and A plasma are considered universal donors for plasma, while type O is universal for PRBCs. An alternative transfusion strategy that delivers a maximally balanced resuscitation involves the transfusion of Low Titer O Whole Blood (LTOWB) with platelet functionality.9 Preparing coolers with a predetermined 1:1:1 ratio27, 28 of Plasma: Platelets: PRBCS (6 units of plasma, 1 unit of apheresis platelets, and 6 units of PRBCs) can mitigate significant ratio imbalances and time spent within ratio imbalances.7, 10 DCR CPGs recommend protocolized ratios and adjunctive medications to minimize the lethal diamond.9 Empiric TXA should be infused within 3 hours after traumatic insult, while calcium infusion should occur after the first unit of blood and for every four units of blood product after that with a repletion goal of normocalcemia (1.2 mmol/L as mentioned in the Joint Trauma System (JTS) CPG).7 The infusion of recombinant human-activated factor VIIa and the use of hydroxyethyl starch (Hextend, Hespan) as a resuscitative fluid are not endorsed by current DCR CPGs.7-9 Prothrombin complex concentrate (PCC) effectively reverses anticoagulant medications as per its Food and Drug Administration cleared use and is also occasionally used off-label to manage trauma-induced coagulopathy that proves refractory to platelets, cryoprecipitate, or fibrinogen and TXA. Early use of PCC for acutely bleeding trauma patients is the subject of a recently completed study and an ongoing large randomized controlled trial (ClinicalTrials.gov identifier NCT05568888).62, 63 In the PROCOAG trial, a double-blind, randomized, placebo-controlled multi-center superiority trial, administration of 4F-PCC showed no significant reduction in 24-hour blood product consumption for patients with traumatic injuries at risk for massive transfusion.63 The target systolic blood pressure during DCR is 100 mmHg (110 mmHg for traumatic brain injuries) to balance resuscitation while avoiding the disruption of unstable, early hemostatic clots by having higher blood pressure.9 The use of mechanical hemostatic adjuncts and DCR tools (tourniquets, resuscitative endovascular occlusion of the aorta, direct peritoneal resuscitation) used during DCR lies outside the scope of this review. Early identification of the high-risk patient has several benefits for the team and the patient.64 First, the patient can be more quickly and accurately triaged to the appropriate level of care. The team can better prepare with a focus on clear communication that highlights the patient's circumstances. For example, anticipating the need for a massive transfusion provides a clinically valuable advantage by enabling early and effective communication with the blood bank. Finally, the team can initiate treatment with the optimal level of aggressiveness, without over- or underutilizing valuable resources. By accurately predicting the requirement for a significant volume of blood products, healthcare providers can allocate resources more efficiently, ensuring an adequate supply is readily available. Taken together, these improvements will translate into better patient outcomes. Identification of the critically ill and coagulopathic trauma patient is beneficial, but challenging with clinicians often relying on gestalt and experience to recognize patients at risk for decompensation and death.65 Rapid detection of hemorrhagic shock is a crucial step in the early initiation of DCR. Physical exams and vital signs are typically the only data available immediately upon arrival to the Emergency Department (ED). Compensatory mechanisms, however, can hide signs of severe shock until they tire and fail. Shock Index and the Assessment of Blood Consumption (ABC) score attempt to quantify the degree of hemorrhagic shock and the possible need for massive transfusion. Conventional lab tests generally take too much time to guide immediate treatment and individual tests have specific inherent flaws. Viscoelastic hemostatic assays (VHAs) have shown potential to guide resuscitation and transfusion of blood products and adjuncts66; additionally, VHA-guided resuscitation generally utilizes fewer blood products than empiric transfusion.67 Real-time PI is the process of monitoring, analyzing, and optimizing processes in real-time.68, 69 This process is ongoing and iterative, with the goal of improving patient care and outcomes. Traditional PI approaches such as "Plan-Do-Check-Act" or the JTS PI pillars of conduct, support, inform, and consult can be used as a model for how to review, analyze, and implement PI actions as well as future standards of practice.69 However, these approaches use an asynchronous approach to PI that takes weeks, months, or even years to appreciate any positive impact. We propose that for time-sensitive, complex clinical situations like hemorrhagic shock, a real-time feedback loop represents a more logical and ultimately more effective approach. Though we know that early identification of patients with severe hemorrhage is a critically important step in the initiation of DCR, robust compensatory physiologic mechanisms make this exceedingly difficult to achieve consistently. Keeping track of and administering blood products and adjuncts appropriately across multiple phases of care while coordinating a team-based resuscitation proves difficult.16 As such, adherence to DCR principles remains surprisingly low.10, 27, 28 In light of these myriad challenges, using CDSS, nudges, and machine learning as part of a real-time DCR PI process may improve adherence to DCR best practices leading to optimal outcomes from team-based resuscitations. One key aspect of real-time PI in healthcare is decision support. Broadly, the advantages of CDSS include reduction of clinician errors, increased adherence to clinical management guidelines, and diagnostic support.11 They have been developed for diverse applications,15 and work by leveraging technology, data analytics, and monitoring to provide clinicians with real-time evidence-based recommendations at the point of care.70, 71 CDSS also enables clinical teams to identify and address deviations from guidelines, either in real-time or in the form of an after-action review. In clinical settings where decisions are often made under constraints of time and uncertainty, such as the Intensive Care Unit (ICU) and ED, CDSS have supported clinicians' management of acutely ill and critically injured patients.11, 72 In the context of DCR, an iterative development and human factors testing approach resulted in a clinically usable CDSS capable of prompting activation of MTP, tracking and prompting blood product and adjunct administration, viewing viscoelastic testing, and after-action review (Figure 1).16 Device-level feedback highlighted the device's ease of use, and users overall had a positive impression. In a prospective pilot study, use of a DCR CDSS with a real-time PI process for massively transfused patients resulted in more time spent in target ratios of plasma and platelets to PRBCs compared to both controls at the same institution and PROMMTT patients. This initial pilot study highlights the application of CDSS for DCR; however, multi-center validation is warranted.73, 74 Nudge theory was initially applied in the field of behavioral economics, but has since been used in many other fields, health care included.18-20 For example, in health care, nudges have been described in relation to both behavioral intervention and ordering practices.75-77 Some academic and medical instititutions have created Nudge Units or embedded behavioral design teams within the health care system.19, 20 Such Nudge Units have helped clinicians and researchers utilize nudge theory to improve the delivery of care. Nudges can take many forms and produce varying degrees of behavioral impact (Figure 2), with information framing exerting lighter influence and guiding choices through defaults exerting stronger influence.18-20 While nudges at the bottom of the ladder passively influence members of the care team (i.e., an email reminding clinicians of existing guidelines or a poster with information placed strategically in a Trauma Bay or an Operating Room (OR)), nudges at the top of the ladder more directly influence clinicians at the time of decision making (i.e., enabling active choice between two options of medications or changing the default selection for a medication in the electronic medical record). The most effective nudges tend to be the more assertive ones, often limiting the set of choices or changing default options.18 That being said, the more assertive nudges are not always feasible given the existing workflows in clinical spaces. For that reason, it is important to evaluate how best to optimize the nudge for the environment.20 In a pre-post study of severely injured patients in an urban Level I trauma center, a nudge providing information on calcium-specific guidelines during blood product resuscitation was a simple solution and barriers to implementation were minimal.78 In this study, a sign indicating that 1 g calcium chloride was to be administered empirically following the administration of the fourth blood product was posted in the trauma bay, OR, and ICU. While the simplest solution in this case did not significantly improve adherence to the institutional guidelines, it was the most feasible to implement. Given competing priorities during massive transfusion, a more aggressive nudge like an automated default order for calcium in accordance with the institutional guidelines might have been more impactful.78 For example, modeling after the INPUT trial, additional strategies might be explored for improving adherence using the electronic health record.79 First, a default order for calcium can be programed following the scanning and administration of the fourth blood product during the resuscitation. Alternatively, the accountable justification strategy might be implemented. This would require the clinician to provide a justification for not administering calcium after the fourth product before being able to order or administer any other products (i.e., blood products, adjuncts). Researchers and clinicians have employed the concept of machine learning to recognize patterns among often heterogeneous and noisy datasets in the setting of both pediatric and adult critical care.80-82 Models have been used to predict the onset of sepsis, readmission to the ICU, and volume responsiveness.81, 83-86 Machine learning algorithms have even predicted mortality and length of stay in ICU patients with high accuracy.87 Patients at risk for hemorrhagic shock often present with a heterogeneous range of compensatory mechanisms leading to varying degrees of physiologic compensation. Traditional scoring systems like the ABC and Trauma-Associated Severe Hemorrhage (TASH) serve as effective tools for predicting the need for MTP.88, 89 However, these systems simplify variables for ease of clinical use—a practical yet limiting approach. This simplification can miss nuanced relationships between critical variables for accurate prediction. Machine learning offers a transformative approach to address these exact limitations. Machine learning, a subset of artificial intelligence, employs algorithms to autonomously analyze data and make predictions. Leveraging the power of granular data, machine learning algorithms can identify complex relationships often overlooked by traditional scoring systems. These algorithms have been shown to outperform established methods like the ABC score, even when operating on a very limited set of variables.22, 90 Beyond their predictive accuracy, machine learning algorithms excel in adaptability and real-time responsiveness. These models can seamlessly integrate into existing clinical workflows, providing continuously updated predictions as new data and lab results become available.91, 92 Furthermore, such analysis can reveal previously unexpected relationships between input variables and clinical outcomes. For example, a recent machine learning algorithm designed to predict MT demonstrated a surprising relationship between small perturbations in serum glucose in admission labs and the need for MT (Figure 3).22 The ability to exploit these subtle relationships and the dynamic nature of machine learning algorithms serve to equip trauma providers with the most current and accurate predictive data, facilitating timely and appropriate interventions while limiting unnecessary usage of limited resources. In addition, as more data becomes available for training, the algorithms can refine their predictive models to enhance precision and accuracy. This iterative learning process allows the system to adapt to new patterns and presentations, making it an optimal tool for navigating the intricate, time-critical challenges inherent to treating the critically injured trauma patients. Early identification of individuals requiring massive transfusion remains challenging despite existing guidelines and best practices. However, emerging technologies such as decision support systems, nudges, and machine learning may prove useful. Real-time PI decision support systems have demonstrated potential benefits with favorable reviews by end users.16 The ability to track blood product ratios, time spent in high ratio targets during the resuscitation, time to MTP, and time to adjuncts is a way to combat challenges preventing adherence to DCR best practice guidelines and optimized resuscitations. The use of nudges in acute resuscitations warrants further exploration as well. The implementation of a high-impact nudge, like a default order in the electronic medical record, may also improve adherence to guidelines by clinical care teams. Lastly, machine learning models should be further explored to both help identify patients with severe hemorrhage and prompt MTP.22 Though challenges in the early identification of patients and optimal DCR practice have persisted over the years, there are promising solutions on the rise to help improve patient outcomes. This work was supported by a Measey Fund Scholarship from the Department of Surgery, Perelman School of Medicine at the University of Pennsylvania (DS), and by the US Army Medical Research and Materiel Command under contract number W81XWH-18-C-0163 (DS, JWC). The authors have no disclosures related to this work.

This article offers an analysis of the implicit conception of translation at work in Malebranche's Entretien d'un philosophe chr & eacute;tien, et d'un philosophe chinois (1708).

Hoefer, Lea E MD; Castro, Helen J Ms; Polcari, Ann MD; Keegan, Grace E BA, BS; Plackett, Timothy P DO, FACS; Benjamin, Andrew J MD, MS; Zakrison, Tanya Liv MD MPH FACS; Cone, Jennifer T MD, FACS Author Information

Objective:. Develop a novel machine learning (ML) model to rapidly identify trauma patients with severe hemorrhage at risk of early mortality. Background:. The critical administration threshold (CAT, 3 or more units of red blood cells in a 60-minute period) indicates severe hemorrhage and predicts mortality, whereas early identification of such patients improves survival. Methods:. Patients from the PRospective, Observational, Multicenter, Major Trauma Transfusion and Pragmatic, Randomized Optimal Platelet, and Plasma Ratio studies were identified as either CAT+ or CAT−. Candidate variables were separated into 4 tiers based on the anticipated time of availability during the patient’s assessment. ML models were created with the stepwise addition of variables and compared with the baseline performance of the assessment of blood consumption (ABC) score for CAT+ prediction using a cross-validated training set and a hold-out validation test set. Results:. Of 1245 PRospective, Observational, Multicenter, Major Trauma Transfusion and 680 Pragmatic, Randomized Optimal Platelet and Plasma Ratio study patients, 1312 were included in this analysis, including 862 CAT+ and 450 CAT−. A CatBoost gradient-boosted decision tree model performed best. Using only variables available prehospital or on initial assessment (Tier 1), the ML model performed superior to the ABC score in predicting CAT+ patients [area under the receiver-operator curve (AUC = 0.71 vs 0.62)]. Model discrimination increased with the addition of Tier 2 (AUC = 0.75), Tier 3 (AUC = 0.77), and Tier 4 (AUC = 0.81) variables. Conclusions:. A dynamic ML model reliably identified CAT+ trauma patients with data available within minutes of trauma center arrival, and the quality of the prediction improved as more patient-level data became available. Such an approach can optimize the accuracy and timeliness of massive transfusion protocol activation.

BACKGROUND:Firearm violence is now endemic to certain US neighborhoods. Understanding factors that impact a neighborhood's susceptibility to firearm violence is crucial for prevention. Using a nationally standardized measure to characterize community-level firearm violence risk has not been broadly studied but could enhance prevention efforts. Thus, we sought to examine the association between firearm violence and the social, structural, and geospatial determinants of health, as defined by the Social Vulnerability Index (SVI). STUDY DESIGN:In this cross-sectional study, we merged 2018 SVI data on census tract with shooting incidents between 2015 and 2021 from Baltimore, Chicago, Los Angeles, New York City, and Philadelphia. We used negative binomial regression to associate the SVI with shooting incidents per 1,000 people in a census tract. Moran's I statistics and spatial lag models were used for geospatial analysis. RESULTS:We evaluated 71,296 shooting incidents across 4,415 census tracts. Fifty-five percent of shootings occurred in 9.4% of census tracts. In all cities combined, a decile rise in SVI resulted in a 37% increase in shooting incidents (p < 0.001). A similar relationship existed in each city: 30% increase in Baltimore (p < 0.001), 50% in Chicago (p < 0.001), 28% in Los Angeles (p < 0.001), 34% in New York City (p < 0.001), and 41% in Philadelphia (p < 0.001). Shootings were highly clustered within the most vulnerable neighborhoods. CONCLUSIONS:In 5 major US cities, firearm violence was concentrated in neighborhoods with high social vulnerability. A tool such as the SVI could be used to inform prevention efforts by directing resources to communities most in need and identifying factors on which to focus these programs and policies.

BACKGROUND Firearm-related injury in children is a public health crisis. The Social Vulnerability Index (SVI) identifies communities at risk for adverse effects due to natural or human-caused crises. We sought to determine if SVI was associated with pediatric firearm-related injury and thus could assist in prevention planning. METHODS The Centers for Disease Control and Prevention's 2018 SVI data were merged on census tract with 2015 to 2022 open-access shooting incident data in children 19 years or younger from Baltimore, Chicago, Los Angeles, New York City, and Philadelphia. Regression analyses were performed to uncover associations between firearm violence, SVI, SVI themes, and social factors at the census tract level. RESULTS Of 11,654 shooting incidents involving children, 52% occurred in just 6.7% of census tracts, which were on average in the highest quartile of SVI. A decile increase in SVI was associated with a 45% increase in pediatric firearm-related injury in all cities combined (incidence rate ratio, 1.45; 95% confidence interval, 1.41–1.49; p < 0.001). A similar relationship was found in each city: 30% in Baltimore, 51% in Chicago, 29% in Los Angeles, 37% in New York City, and 35% in Philadelphia (all p < 0.001). Socioeconomic status and household composition were SVI themes positively associated with shootings in children, as well as the social factors below poverty, lacking a high school diploma, civilian with a disability, single-parent household, minority, and no vehicle access. Living in areas with multi-unit structures, populations 17 years or younger, and speaking English less than well were negatively associated. CONCLUSION Geospatial disparities exist in pediatric firearm-related injury and are significantly associated with neighborhood vulnerability. We demonstrate a strong association between SVI and pediatric shooting incidents in multiple major US cities. Social Vulnerability Index can help identify social and structural factors, as well as geographic areas, to assist in developing meaningful and targeted intervention and prevention efforts. LEVEL OF EVIDENCE Prognostic and Epidemiological; Level III.

BACKGROUND Firearm violence in the United States is a public health crisis, but accessing accurate firearm assault data to inform prevention strategies is a challenge. Vulnerability indices have been used in other fields to better characterize and identify at-risk populations during crises, but no tool currently exists to predict where rates of firearm violence are highest. We sought to develop and validate a novel machine-learning algorithm, the Firearm Violence Vulnerability Index (FVVI), to forecast community risk for shooting incidents, fill data gaps, and enhance prevention efforts. METHODS Open-access 2015 to 2022 fatal and nonfatal shooting incident data from Baltimore, Boston, Chicago, Cincinnati, Los Angeles, New York City, Philadelphia, and Rochester were merged on census tract with 30 population characteristics derived from the 2020 American Community Survey. The data set was split into training (80%) and validation (20%) sets; Chicago data were withheld for an unseen test set. XGBoost, a decision tree-based machine-learning algorithm, was used to construct the FVVI model, which predicts shooting incident rates within urban census tracts. RESULTS A total of 64,909 shooting incidents in 3,962 census tracts were used to build the model; 14,898 shooting incidents in 766 census tracts were in the test set. Historical third grade math scores and having a parent jailed during childhood were population characteristics exhibiting the greatest impact on FVVI’s decision making. The model had strong predictive power in the test set, with a goodness of fit (D 2) of 0.77. CONCLUSION The Firearm Violence Vulnerability Index accurately predicts firearm violence in urban communities at a granular geographic level based solely on population characteristics. The Firearm Violence Vulnerability Index can fill gaps in currently available firearm violence data while helping to geographically target and identify social or environmental areas of focus for prevention programs. Dissemination of this standardized risk tool could also enhance firearm violence research and resource allocation. LEVEL OF EVIDENCE Prognostic and Epidemiological; Level IV.

BACKGROUND:Duodenal leak is a feared complication of repair, and innovative complex repairs with adjunctive measures (CRAM) were developed to decrease both leak occurrence and severity when leaks occur. Data on the association of CRAM and duodenal leak are sparse, and its impact on duodenal leak outcomes is nonexistent. We hypothesized that primary repair alone (PRA) would be associated with decreased duodenal leak rates; however, CRAM would be associated with improved recovery and outcomes when leaks do occur. METHODS:A retrospective, multicenter analysis from 35 Level 1 trauma centers included patients older than 14 years with operative, traumatic duodenal injuries (January 2010 to December 2020). The study sample compared duodenal operative repair strategy: PRA versus CRAM (any repair plus pyloric exclusion, gastrojejunostomy, triple tube drainage, duodenectomy). RESULTS:The sample (N = 861) was primarily young (33 years) men (84%) with penetrating injuries (77%); 523 underwent PRA and 338 underwent CRAM. Complex repairs with adjunctive measures were more critically injured than PRA and had higher leak rates (CRAM 21% vs. PRA 8%, p < 0.001). Adverse outcomes were more common after CRAM with more interventional radiology drains, prolonged nothing by mouth and length of stay, greater mortality, and more readmissions than PRA (all p < 0.05). Importantly, CRAM had no positive impact on leak recovery; there was no difference in number of operations, drain duration, nothing by mouth duration, need for interventional radiology drainage, hospital length of stay, or mortality between PRA leak versus CRAM leak patients (all p > 0.05). Furthermore, CRAM leaks had longer antibiotic duration, more gastrointestinal complications, and longer duration until leak resolution (all p < 0.05). Primary repair alone was associated with 60% lower odds of leak, whereas injury grades II to IV, damage control, and body mass index had higher odds of leak (all p < 0.05). There were no leaks among patients with grades IV and V injuries repaired by PRA. CONCLUSION:Complex repairs with adjunctive measures did not prevent duodenal leaks and, moreover, did not reduce adverse sequelae when leaks did occur. Our results suggest that CRAM is not a protective operative duodenal repair strategy, and PRA should be pursued for all injury grades when feasible. LEVEL OF EVIDENCE:Therapeutic/Care Management; Level IV.

This cohort study examines the association of tranexamic acid administration with intracranial hemorrhage type, neurologic outcomes, and mortality in patients with traumatic brain injury.

INTRODUCTION: The Social Vulnerability Index (SVI), composed of 4 social factors, is used to predict communities most at risk for negative consequences during natural disasters and public health crises. Violence, especially with a firearm, is one such crisis previously shown to behave like a transmissible disease. We hypothesized that SVI could predict community risk for violence. METHODS: The CDC’s 2018 SVI data were merged by census tract with violent crimes (assault, battery, and homicide) from the Chicago Data Portal (Figure). Cubic regression was used to show correlations between crime rate, overall SVI, and each SVI social factor; univariate and multivariate negative binomial regression was performed to calculate risk ratios.SVI, Social Vulnerability Index.RESULTS: SVI predicted risk of violent crime in Chicago. Negative binomial regression found that each percentile increase in SVI resulted in a 7 times increase in violent crime (RR 7.09, 95% CI 6.06 to 8.29, p < 0.001). The factors most strongly associated with violence were socioeconomic status (RR 3.63, 95% CI 2.76 to 4.78, p < 0.001), household composition/disability (RR 2.59, 95% CI 2.09 to 3.22, p < 0.001) and housing type/transportation (RR 1.41, 95% CI 1.60 to 2.25, p < 0.001). Minority status/language was associated with a decreased risk of violent crime (RR 0.34, 95% CI 0.26 to 0.42, p < 0.001). CONCLUSION: This is the first study to demonstrate SVI may predict community risk for violence. SVI can identify geospatial areas of focus for violence intervention and prevention. Academic researchers and trauma surgeons should use SVI when demonstrating disparities related to violent injury risk and outcomes, because it encompasses the effect of social, economic, and built environments at a granular level that is standardized across the US.

INTRODUCTION: The Pediatric Shock Index (PSI) uses age-based thresholds to define shock in pediatric trauma. One group showed that PSI outperforms Shock Index Pediatric Age-Adjusted (SIPA) in predicting adverse outcomes. Glasgow Coma Score (GCS) can also forecast trauma risk. We hypothesized that merging PSI with GCS would improve PSI’s predictive value and outperform other pediatric shock indices. METHODS: A cohort of patients ≤18 years old from the 2010 to 2018 Trauma Quality Programs Participant Use File (TQP-PUF) files was split into training and test sets (70/30). The training set analysis optimized age-adjusted cutoffs for shock using PSI+GCS = HR/SBP/GCS by maximizing the Youden Index (Fig. 1A). The test set evaluated positive (PPV) and negative (NPV) predictive values of PSI+GCS, PSI alone, SIPA, and rSIG (reverse shock index * GCS) for mortality, pediatric intensive care unit (PICU) admission, and need for blood transfusion.RESULTS: PSI+GCS outperformed the other measures of shock in predicting adverse outcomes. PSI+GCS had the highest PPV for death (9.65% vs 5.1% PSI, 3.5% SIPA, 2.8% rSIG), PICU admission (63.9% vs 39% PSI, 32.6% SIPA, 31.9% rSIG), ventilator use (47.5% vs 19.4% PSI, 14.6% SIPA, 15.2% rSIG), and early transfusion (15.9% vs 11.6% PSI, 7.9% SIPA, 5.2% rSIG; Fig. 1B). NPV was overall similar, although PSI+GCS did best (99.76% vs 99.27% PSI, 99.28% SIPA, 99.75% rSIG). CONCLUSION: PSI+GCS improves the predictive value of adverse pediatric trauma outcomes compared with PSI alone and outperforms other proposed shock indices in predicting PICU admission, ventilator use, and need for blood transfusion. PSI+GCS has potential use in prehospital and PICU triage, and as a research tool.

Blunt cardiac injury (BCI) encompasses a wide range of presentations, some of them severe in nature. Therefore, appropriate screening is necessary so as not to miss a clinically significant BCI. Electrocardiogram (ECG) has consistently been shown to be the most sensitive test for ruling out BCI; however, it is not sufficient in isolation. ECG combined with troponin has been shown to be nearly 100% sensitive, and both should routinely be checked in patients who present with blunt thoracic trauma. Transthoracic echocardiogram (TTE) is poorly sensitive, but highly specific and provides insight into cardiac function. Therefore, a TTE should be performed in any patient with BCI and persistent arrhythmia or shock state, as it may be useful in guiding management. Although initial studies suggest axial imaging may be sensitive and specific, they do not currently serve a role in the critically ill trauma patient given their significant limitations, especially in terms of the need for transport of patients to less monitored settings. Given the rarity of BCI, there are no specific treatments. Therefore, the mainstay of BCI management is supportive care and usual management of any clinically significant manifestations.