Model for predicting the injury severity score

Comparison and interpretability of machine learning models to predict severity of chest injury

Objective

Trauma quality improvement programs and registries improve care and outcomes for injured patients. Designated trauma centers calculate injury scores using dedicated trauma registrars; however, many injuries arrive at nontrauma centers, leaving a substantial amount of data uncaptured. We propose automated methods to identify severe chest injury using machine learning (ML) and natural language processing (NLP) methods from the electronic health record (EHR) for quality reporting.

Materials and Methods

A level I trauma center was queried for patients presenting after injury between 2014 and 2018. Prediction modeling was performed to classify severe chest injury using a reference dataset labeled by certified registrars. Clinical documents from trauma encounters were processed into concept unique identifiers for inputs to ML models: logistic regression with elastic net (EN) regularization, extreme gradient boosted (XGB) machines, and convolutional neural networks (CNN). The optimal model was identified by examining predictive and face validity metrics using global explanations.

Results

Of 8952 encounters, 542 (6.1%) had a severe chest injury. CNN and EN had the highest discrimination, with an area under the receiver operating characteristic curve of 0.93 and calibration slopes between 0.88 and 0.97. CNN had better performance across risk thresholds with fewer discordant cases. Examination of global explanations demonstrated the CNN model had better face validity, with top features including “contusion of lung” and “hemopneumothorax.”

Discussion

The CNN model featured optimal discrimination, calibration, and clinically relevant features selected.

Conclusion

NLP and ML methods to populate trauma registries for quality analyses are feasible.

Fibrin monomers and association with significant hemorrhage or mortality in severely injured trauma patients

BACKGROUND

Post-traumatic hemorrhage is still the leading cause of potentially preventable death in patients with severe trauma. Traumatic-induced coagulopathy has been described as a risk factor for significant hemorrhage and mortality in this population. Fibrin monomers (FMs) are a direct marker of thrombin action, and thus reflect coagulation activation. This study sought to determine the association of FMs levels at admission with significant hemorrhage and 28-day mortality after a severe trauma.

METHODS

We conducted a retrospective, observational study including all severe trauma patients admitted in a level-1 trauma center between January 2012 and December 2017. Patients with severe traumatic brain injury or previous anticoagulant / antiaggregant therapies were excluded. FMs measurements and standard coagulation test were taken at admission. Significant hemorrhage was defined as a hemorrhage requiring the transfusion of ≥ 4 Red Blood Cells units during the first 6 h. Multivariable analysis was applied to identify predictors of significant hemorrhage and a simple logistic regression analysis was applied to identify an association between FMs and 28-day mortality.

RESULTS

Overall, 299 patients were included. A total of 47 (16%) experienced a significant hemorrhage. The ROC curve demonstrated that FMs had a poor accuracy to predict the occurrence of significant hemorrhage with an AUC of 0.65 (0.57-0.74). The best threshold at 92.45 µg/ml had excellent sensitivity (87%) and negative predictive value (95%), but was not independently associated with significant hemorrhage (OR = 1.5; 95%CI (0.5-4.2)). The 28-day mortality rate was 5%. In simple logistic regression analysis, FMs values ≥109.5 µg/ml were significantly associated with 28-day mortality (unadjusted OR = 13.2; 95%CI (1.7-102)).

CONCLUSIONS

FMs levels at admission are not associated with the occurrence of a significant hemorrhage in patients with severe trauma. However, the excellent sensitivity and NPV of FMs could help to identify patients with a low risk of severe bleeding during hospital care. In addition, FMs levels ≥109.5 µg/ml might be predictive of 28-day mortality.

Risk factors for death of trauma patients admitted to an Intensive Care Unit

OBJECTIVE

To analyze the risk factors for death of trauma patients admitted to the intensive care unit (ICU).

METHOD

Retrospective cohort study with data from medical records of adults hospitalized for trauma in a general intensive care unit. We included patients 18 years of age and older and admitted for injuries. The variables were grouped into levels in a hierarchical manner. The distal level included sociodemographic variables, hospitalization, cause of trauma and comorbidities; the intermediate, the characteristics of trauma and prehospital care; the proximal, the variables of prognostic indices, intensive admission, procedures and complications. Multiple logistic regression analysis was performed.

RESULTS

The risk factors associated with death at the distal level were age 60 years or older and comorbidities; at intermediate level, severity of trauma and proximal level, severe circulatory complications, vasoactive drug use, mechanical ventilation, renal dysfunction, failure to perform blood culture on admission and Acute Physiology and Chronic Health Evaluation II.

CONCLUSION

The identified factors are useful to compose a clinical profile and to plan intensive care to avoid complications and deaths of traumatized patients.

The significance of the levels of fibrin/fibrinogen degradation products for predicting trauma severity

Background:

On initial treatment in the emergency room, trauma patients should be assessed using simple clinical indicators that can be measured quickly.

Objectives:

The purpose of this study is to investigate the relationship between the injury severity score and blood test parameters measured on emergency room arrival in trauma patients.

Methods:

Trauma patients transferred to Gunma University Hospital between May 2013 and April 2014 were evaluated in this prospective, observational study. Blood samples were collected immediately on their arrival at our emergency room and their hematocrit, platelet, international normalized ratio of prothrombin time, activated partial thromboplastin time, fibrin/fibrinogen degradation products, and D-dimer were measured. We evaluated the correlations between the injury severity score and those biomarkers, and examined whether the correlation varied according to the injury severity score value. We also evaluated the correlations between the biomarkers and the abbreviated injury scale values of six regions.

Results:

We analyzed 371 patients. Fibrin/fibrinogen degradation products and D-dimer showed the greatest coefficients of correlation with injury severity score (0.556 and 0.543, respectively). The area under the curve of the receiver operating characteristic was larger in patients with injury severity score ⩾ 9 than in those with injury severity score ⩾ 4; however, patients with injury severity score ⩾ 9 or ⩾16 showed no significant differences. The area under the curve of fibrin/fibrinogen degradation products was larger than that of D-dimer at all injury severity score values. The chest abbreviated injury scale had the strongest relationship with fibrin/fibrinogen degradation products.

Conclusion:

Fibrin/fibrinogen degradation products and D-dimer were positively correlated with injury severity score, and the relationships varied according to trauma severity. Chest trauma contributed most strongly to fibrin/fibrinogen degradation product elevation.

Geriatric Trauma Protocol

Study design for a quality improvement project. Objective was to implement a geriatric trauma protocol (GTP) based on American College of Surgeons recommendations to improve patient outcomes. Geriatric trauma patients comprise a vulnerable and high-risk trauma population, and must be treated with specific protocols that take into account physiological, psychosocial, environmental, and pharmacological needs. A growing body of research and organizations such as the American College of Surgeons and the Eastern Association for the Surgery of Trauma recommend that a specific trauma protocol for geriatric adults must be utilized in hospitals and trauma centers. A retrospective chart review was conducted to assess geriatric patient outcomes prior to GTP implementation. Surgical residents then received training on the GTP, including performing additional diagnostics, referrals, and discussing goals of care early in treatment. The GTP was then implemented for 8 weeks and monitored to determine its effects on patient outcomes. The training for surgical residents in the GTP yielded a 9.2% increase in provider knowledge. The results of the GTP showed a reduced length of stay and increased geriatric consultations. More patients received a full evaluation by the trauma team, contributing the reduced length of stay. The use of a GTP shows promise in being able to improve patient outcomes, including morbidity and mortality. The principles of the GTP can be applied in all clinical settings, especially emergency rooms, which are on the frontlines of initial evaluations. In order to improve health care delivery to an aging population, organizations and clinicians should adopt a specialized GTP into their practices.

FDP/fibrinogen ratio reflects the requirement of packed red blood cell transfusion in patients with blunt trauma

PURPOSE

To find factors that predict the requirement of packed red blood cells (pRBC) transfusion in patients with blunt trauma on arrival at the hospital.

METHODS

We conducted blood tests in trauma patients whose trauma severity was suspected as being 3 and over in the Abbreviated Injury Scale. Patients were divided into the blood transfusion (BT) and control groups according to the requirement of pRBC transfusion within 24h after arrival.

RESULTS

We analyzed 347 patients (BT group, n=14; control group, n=333). On univariate analysis, there were significant differences in Glasgow Coma Scale (GCS), rate of positive FAST (focused assessment with sonography for trauma) finding, hematocrit, international normalized ratio of prothrombin time, activated partial thromboplastin time, fibrinogen (Fib), and level of fibrin degradation products (FDP). On multivariable analysis, positive FAST finding, GCS, Fib, and FDP influenced the requirement of pRBC transfusion. In the area under the receiver operating characteristic curve analysis, Fib and FDP were markers that predicted the requirement of pRBC transfusion. The FDP/Fib ratio had a better correlation with the requirement of pRBC transfusion than FDP or Fib.

CONCLUSIONS

The FDP/Fib ratio can be easily measured and may be a predictor of the need for pRBC transfusion.