Predictors of mortality in adult trauma patients: the physiologic trauma score is equivalent to the Trauma and Injury Severity Score

Background:

Several statistical models (Trauma and Injury Severity Score [TRISS], New Injury Severity Score [NISS], and the International Classification of Disease, Ninth Revision-based Injury Severity Score [ICISS]) have been developed over the recent decades in an attempt to accurately predict outcomes in trauma patients. The anatomic portion of these models makes them difficult to use when performing a rapid initial trauma assessment. We sought to determine if a Physiologic Trauma Score, using the systemic inflammatory response syndrome (SIRS) score in combination with other commonly used indices, could accurately predict mortality in trauma.

Study Design:

Prospective data were analyzed in 9,539 trauma patients evaluated at a Level I Trauma Center over a 30-month period (January 1997 to July 1999). A SIRS score (1 to 4) was calculated on admission (1 point for each: temperature >38°C or <36°C, heart rate >90 beats per minute, respiratory rate >20 breaths per minute, neutrophil count > 12,000 or < 4,000. SIRS score, Injury Severity Score (ISS), Revised Trauma Score (RTS), TRISS, Glasgow Coma Score, age, gender, and race were used in logistic regression models to predict trauma patients’ risk of death. The area under the receiver-operating characteristic curves of sensitivity versus 1-specificity was used to assess the predictive ability of the models.

Results:

The study cohort of 9,539 trauma patients (of which 7,602 patients had complete data for trauma score calculations) had a mean ISS of 9 ± 9 (SD) and mean age of 37 ± 17 years. SIRS (SIRS score ≥ 2) was present in 2,165 of 7,602 patients (28.5%). In single-variable models, TRISS and ISS were most predictive of outcomes. A multiple-variable model, Physiologic Trauma Score combining SIRS score with Glasgow Coma Score and age (Hosmer-Lemenshow CHI-SQUARE = 4.74) was similar to TRISS and superior to ISS in predicting mortality. The addition of ISS to this model did not significantly improve its predictive ability.

Conclusions:

A new statistical model (Physiologic Trauma Score), including only physiologic variables (admission SIRS score combined with Glasgow Coma Score and age) and easily calculated at the patient bedside, accurately predicts mortality in trauma patients. The predictive ability of this model is comparable to other complex models that use both anatomic and physiologic data (TRISS, ISS, and ICISS).     

Prognostic value of trauma scores in pediatric patients with multiple injuries

BACKGROUND: For the quantification of multiple injuries in children, a range of different trauma scores are available, the actual prognostic value of which has, however, not so far been investigated and compared in a group of patients. METHODS: In 261 polytraumatized children and adolescents, 11 trauma scores (Abbreviated Injury Scale [AIS], Injury Severity Score [ISS], Glasgow Coma Scale [GCS], Acute Trauma Index [ATI], Shock Index [SI], Trauma Score [TS], Revised Trauma Score [RTS], Modified Injury Severity Score [MISS], Trauma and Injury Severity Score [TRISS]-Scan, Hannover Polytrauma Score [HPTS], and Pediatric Trauma Score [PTS]) were calculated, and their prognostic relevance in terms of survival, duration of intensive care treatment, hospital stay, and long-term outcome analyzed. RESULTS: With a specificity of 80%, physiologic scores (TS, RTS, GCS, ATI) showed a greater accuracy (79-86% vs. 73-79%) with regard to survival prediction than did the anatomic scores (AIS, HPTS, ISS, PTS); combined forms of these two types of score (TRISS-Scan, MISS) did not provide any additional information (76-80%). Overall, the TRISS-Scan was the score that showed the highest correlation with duration of treatment and long-term outcome. Trauma scores specially conceived for use with children (PTS, MISS) failed to show any superiority vis-à-vis trauma scores in general. CONCLUSION: With regard to prognostic quality and ease of use in the practical setting, TS and the TRISS-Scan are recommended for polytrauma in children and adolescents. Special pediatric scores are not necessary.     

Comparing logistic models based on modified GCS motor component with other prognostic tools in prediction of mortality: results of study in 7226 trauma patients

A simple reproducible and sensitive prognostic trauma tool is still needed. In this article we have introduced modified GCS motor response (MGMR) and evaluated the performance of logistic models based on this variable. The records of 8452 trauma patients admitted to major hospitals of Tehran from 1999 to 2000 were analysed. 7226 records with known outcome were included in our study. Logistic models based on outcome (death versus survival) as a dependent variable and Injury Severity Score (ISS), Revised Trauma Score (RTS), Glasgow Coma Scale (GCS), GCS motor component (GMR) and MGMR (following command [=2], movement but not following [=1] command and without movement [=0]) were compared based on their accuracy and area under the Receiver Operating Characteristic (ROC) curve. The accuracy of the Trauma and Injury Severity Score (TRISS), RTS, GCS, GMR and MGMR models were almost the same. Considering both the area under the ROC curve and accuracy, the age included MGMR model was also comparable with other age included models (RTS+age, GCS+age, GMR+age). We concluded that although in some situations we need more sophisticated models, should our results be reproducible in other populations, MGMR (with or without age added) model may be of considerable practical value.     

Comparison of current injury scales for survival chance estimation: an evaluation comparing the predictive performance of the ISS, NISS, and AP scores in a Dutch local trauma registration

BACKGROUND: Prediction of survival chances for trauma patients is a basic requirement for evaluation of trauma care. The current methods are the Trauma and Injury Severity Score (TRISS) and A Severity Characterization of Trauma (ASCOT). Scales for scoring injury severity are part of these methods. This study compared three injury scales, the Injury Severity Score (ISS), the New ISS (NISS), and the Anatomic Profile (AP), in three otherwise identical predictive models. METHODS: Records of the Rotterdam Trauma Center were analyzed using logistic regression. The variables used in the models were age (as a linear variable), the corrected Revised Trauma Score (RTS), a denominator for blunt or penetrating trauma, and one of the three injury scales. The original TRISS and ASCOT models also were evaluated. The resulting models were compared in terms of their discriminative power, as indicated by the receiver-operator characteristic (ROC), and calibration (Hosmer-Lemeshow [HL]) for the entire range of injury severity. RESULTS: For this study, 1,102 patients, with an average ISS of 15, met the inclusion criteria. The TRISS and ASCOT models, using original coefficients, showed excellent discriminative power (ROC, 0.94 and 0.96, respectively), but insufficient fits (HL, p = 0.001 and p = 0.03, respectively). The three fitted models also had excellent discriminative abilities (ROC, 0.95, 0.97, and 0.96, respectively). The custom ISS model was unable to fit the entire range of survival chances sufficiently (p = 0.01). Models using the NISS and AP scales provided adequate fits to the actual survival chances of the population (HL, 0.32 and 0.12, respectively). CONCLUSIONS: The AP and NISS scores particularly both managed to outperform the ISS score in correctly predicting survival chances among a Dutch trauma population. Trauma registries stratifying injuries by the ISS score should evaluate the use of the NISS and AP scores.     

Impact of cirrhosis on outcomes in trauma

BACKGROUND: Cirrhosis as an independent predictor of poor outcomes in trauma patients was identified in 1990. We hypothesized that the degree of preinjury hepatic dysfunction is, by itself, an independent predictor of mortality. STUDY DESIGN: The trauma registry at our Level I trauma center was queried for all ICD-9 codes for liver disease from 1999 to 2003, and patients were categorized as having Child-Turcotte-Pugh (CTP) class A, B, or C cirrhosis. Data analyzed included age, mechanism of injury, Abbreviated Injury Score (AIS), Injury Severity Score (ISS), Glasgow Coma Score (GCS), hospital length of stay, ventilator days, procedures performed, transfusion of blood products, admission lactate, base deficit, and mortality. Trauma Related Injury Severity Score (TRISS) methodology was used to calculate the probability of survival. Outcomes data were analyzed, and statistical comparison was performed using group t-test. RESULTS: Of the 50 patients meeting study criteria, 31 had alcohol-related cirrhosis, 18 had a history of hepatitis C, and 1 had cryptogenic cirrhosis. Twenty (40%) met CTP A classification, 16 (32%) met CTP B criteria, and 14 (28%) had CTP class C cirrhosis. One death occurred in the CTP A and B groups. Comparison between the five survivors and nine nonsurvivors from CTP class C showed no statistical significance in terms of age, ISS, TRISS, or GCS. CONCLUSIONS: The mortality rate for class C cirrhotic patients posttrauma continues to be higher than that predicted by TRISS, although patients with less severe hepatic dysfunction do not appear to have significantly lower than predicted survival. The degree of hepatic dysfunction remains an independent predictor of mortality and CTP C criteria must be considered when determining outcomes for patients posttrauma.     

The effect of patient age upon survival in pediatric trauma

Developmental changes in the anatomy and physiology of growing children are thought to improve the survivability of older children to significant injury. The effect of age upon survival, however, is poorly defined. Data for 4,615 patients less than 15 years old from a statewide trauma center registry were analyzed. Injury and survival were characterized by Abbreviated Injury Scale (AIS, 1985 revision), Injury Severity Score (ISS), Revised Trauma Score (RTS), and probability of survival [P(s)] and Z by TRISS. Patients were separated into age groups of 0 through 4, 5 through 9, and 10 through 14 years. The survival rate for patients with a maximum AIS 3 for any region was significantly higher in the 10-14-year age group. There were no significant differences in survival rates from head, thoracic, and abdominal injuries stratified by AIS among the three age groups. Survival rates for ISS cohorts were consistently lowest in the 0-4-year age group, but differences failed to reach significance. Survival for RTS and P(s) intervals were similar for all ages. The Z statistic reached significance for all children (Z = 4.717, W = 1.049), and for each group (Z = 2.203-3.029). Corresponding values of the W statistic suggest approximately one additional unexpected survivor per 100 admitted children when compared with the Major Trauma Outcome Study. Logistic regression for patients with all data required for TRISS showed no significant effect for any of the three age groups. We conclude that for this patient set, survival after childhood injury is independent of the age groups used in this study, after controlling for injury severity.     

Effect of alcohol on Glasgow Coma Scale in head-injured patients

OBJECTIVE: Almost 50% of traumatic brain-injured (TBI) patients are alcohol intoxicated. The Glasgow Coma Scale (GCS) is frequently used to direct diagnostic and therapeutic decisions in these patients. It is commonly assumed that alcohol intoxication reduces GCS, thus limiting its utility in intoxicated patients. The purpose of this study was to test the hypothesis that the presence of blood alcohol has a clinically significant impact on GCS in TBI patients. METHODS: The National Trauma Data Bank of the American College of Surgeons was queried (1994-2003). Patients 18 to 45 years of age with blunt injury mechanism, whose GCS in the emergency department, survival status, anatomic severity of TBI (Head Abbreviated Injury Score [AIS]), and blood alcohol testing status were known, were included. GCS of patients who tested positive for alcohol (n = 55,732) was compared with GCS of patients who tested negative (n = 53,197), stratified by head AIS. RESULTS: Groups were similar in age (31 +/- 8 vs. 30 +/- 8 years), Injury Severity Score (ISS; 12 +/- 11 vs. 12 +/- 11), systolic blood pressure in the ED (131 +/- 25 vs. 134 +/- 25 mm Hg), TRISS (Trauma Injury Severity Score; probability of survival (94% +/- 16% vs. 95% +/- 15%), and actual survival (96% vs. 96%). When stratified by anatomic severity of TBI, the presence of alcohol did not lower GCS by more than 1 point in any head AIS group (GCS in alcohol-positive vs. alcohol-negative patients; AIS 1 = 13.9 +/- 2.8 vs. 14.3 +/- 2.3; AIS 2 = 13.4 +/- 3.2 vs. 14.1 +/- 2.4; AIS 3 = 11.1 +/- 4.7 vs. 11.6 +/- 4.6; AIS 4 = 9.8 +/- 4.9 vs. 10.4 +/- 4.9; AIS 5 = 5.5 +/- 3.8 vs. 5.9 +/- 4.1, AIS 6: 3.4 +/- 1.1 vs. 3.8 +/- 2.8). CONCLUSION: Alcohol use does not result in a clinically significant reduction in GCS in trauma patients. Attributing low GCS to alcohol intoxication in TBI patients may delay necessary diagnostic and therapeutic interventions.     

Is a full team required for emergency management of pediatric trauma?

Pediatric trauma centers often do not meet the guidelines requiring a trauma team as recommended by the American Academy of Pediatrics (AAP). We reviewed our experience with a team consisting of a pediatric emergency physician, resident, nurse, and respiratory therapist. The surgical and pediatric critical care residents and staff were available within 5 minutes. We conducted a retrospective chart review of 146 patients (aged 8.1 +/- 4.8 years) between 1987 and 1989, with Injury Severity Scores (ISS) greater than or equal to 16 or admitted to the pediatric critical care unit. The time of presentation, surgical services consulted, and the nature of the injury were obtained from chart review. The Pediatric Trauma Score (PTS), the Revised Trauma Score (RTS), the Injury Severity Score (ISS), Glasgow Coma Scale (GCS) score, and Pediatric Risk of Mortality (PRISM) were used to determine the severity of insult and physiologic derangement on admission. The Modified Injury Severity Score (MISS) was determined and the Delta score for Disability Assessment was assigned at discharge. The Delta score was also determined at 3-month intervals up to one year. The probability of survival (Ps) was calculated, using the ISS and RTS. The Z statistic for this group of patients was then determined, using the Major Trauma Outcome Study (MTOS) methodology. The percentages of patients who were normal, disabled, and dead were 61%, 31.5%, and 7.5%, respectively, at 6 months follow-up. Eleven deaths were expected based on PRISM and TRISS analysis. Our mortality and morbidity figures were comparable with those of centers with teams based on AAP guidelines.(ABSTRACT TRUNCATED AT 250 WORDS)     

Prediction of outcomes in trauma: anatomic or physiologic parameters?

BACKGROUND: Prediction of outcomes after injury has traditionally incorporated measures of injury severity, but recent studies suggest that including physiologic and shock measures can improve accuracy of anatomic-based models. A recent single-institution study described a mortality predictive equation [f(x) = 3.48 - .22 (GCS) - .08 (BE) + .08 (Tx) + .05 (ISS) + .04 (Age)], where GSC is Glasgow Coma Score, BE is base excess, Tx is transfusion requirement, and ISS is Injury Severity Score, which had 63% sensitivity, 94% specificity, (receiver operating characteristic [ROC] 0.96), but did not provide comparative data for other models. We have previously documented that the Physiologic Trauma Score, including only physiologic variables (systemic inflammatory response syndrome, Glasgow Coma Score, age) also accurately predicts mortality in trauma. The objective of this study was to compare the predictive abilities of these statistical models in trauma outcomes. METHODS: Area under the ROC curve of sensitivity versus 1-specificity was used to assess predictive ability and measured discrimination of the models. RESULTS: The study cohort consisted of 15,534 trauma patients (80% blunt mechanism) admitted to a Level I trauma center over a 3-year period (mean age 37 +/- 18 years; mean Injury Severity Score 10 +/- 10; mortality 4%). Sensitivity of the new predictive model was 45%, specificity was 96%, ROC was 0.91, validating this new trauma outcomes model in our institution. This was comparable with area under the ROC for Revised Trauma Score (ROC 0.88), Trauma and Injury Severity Score (ROC 0.97), and Physiologic Trauma Score (ROC 0.95), but superior compared with admission Glasgow Coma Score (ROC 0.79), Injury Severity Score (ROC 0.79), and age (ROC 0.60). CONCLUSIONS: The predictive ability of this new model is superior to anatomic-based models such as Injury Severity Score, but comparable with other physiologic-based models such as Revised Trauma Score, Physiologic Trauma Score and Trauma, and Injury Severity Score.     

Validity of applying adult TRISS analysis to injured children

Injury severity measures are becoming increasingly important for quality assurance and injury research. TRISS analysis, which uses the Revised Trauma Score (RTS) and Injury Severity Score (ISS) to predict survival, is an effective tool for comparing outcomes between trauma centers. It has been argued that blunt trauma outcome differs between children and adults, yet the Major Trauma Outcome Study (MTOS) adult data base (ages 15-54 years) regression weights have been used by others to calculate TRISS scores for injured children. This study appears to be the first to perform TRISS analysis on groups of children and adults treated by the same surgeons using the same treatment protocols to assess the validity of applying "adult" TRISS analysis to children. The charts of 346 consecutive children (ages 0-14) and 346 random adults (ages 15-54) admitted to a regional trauma center for isolated blunt trauma over a 30-month period were reviewed for demographics, mechanism of injury, RTS, ISS, and survival. Statistical evaluation included TRISS survival analysis and calculation of the Z statistic. The median ISS was 10 for both children and adults. The Z statistics for children and adults were similar (1.85 and 1.81). Analysis demonstrated the groups to be statistically identical with a nonsignificant trend toward improved survival compared with the MTOS baseline group. These data support the use of existing TRISS analysis for evaluation of pediatric trauma care.     

Direct transport to the operating room for resuscitation of trauma patients

Two hundred forty trauma patients were transported directly from the scene to a specially designed operating room (OR) for resuscitation, bypassing the Emergency Department (ED). Triage criteria included a systolic BP less than or equal to 80 mm Hg, penetrating torso trauma, multiple long-bone fractures, major limb amputation, extensive soft-tissue wounds, severe maxillofacial hemorrhage, and witnessed arrest (WA). The mechanism of injury, transport mode, age, sex, admitting Revised Trauma Score (RTS), Injury Severity Score (ISS), Abbreviated Injury Scale (AIS), operative procedures, and outcome were recorded. Utilizing the current weights from the Major Trauma Outcome Study, the predicted survival (TRISS) of the total group and of several subgroups was compared to the observed survival. The mean ISS was 29.3. The survival rate for the total group was 70.4%. For the 58.7% who required major operative intervention, the mean time of OR arrival to anesthesia induction was 8.5 minutes. Non-arrested, hypotensive blunt trauma victims requiring therapeutic laparotomy had a higher than predicted survival observed survival = 0.75 versus average TRISS = 0.55; p less than 0.0002) and therefore appeared to benefit from this technique. Patients suffering witnessed arrest in the field did not benefit.     

Validation of international trauma scoring systems in urban trauma centres in India

IntroductionIn the Lower-Middle Income Country setting, we validate trauma severity scoring systems, namely Injury Severity Score (ISS), New Injury Severity Scale (NISS) score, the Kampala Trauma Score (KTS), Revised Trauma Score (RTS) score and the TRauma Injury Severity Score (TRISS) using Indian trauma patients.Patients and methodsFrom 1 September 2013 to 28 February 2015, we conducted a prospective multi-centre observational cohort study of trauma patients in four Indian university hospitals, in three megacities, Kolkata, Mumbai and Delhi. All adult patients presenting to the casualty department with a history of injury and who were admitted to inpatient care were included. The primary outcome was in-hospital mortality within 30-days of admission. The sensitivity and specificity of each score to predict inpatient mortality within 30 days was assessed by the areas under the receiver operating characteristic curve (AUC). Model fit for the performance of individual scoring systems was accomplished by using the Akaike Information criterion (AIC).ResultsIn a registry of 8791 adult trauma patients, we had a cohort of 7197 patients eligible for the study. 4091 (56.8%)patients had all five scores available and was the sample for a complete case analysis. Over a 30-day period, the scores (AUC) was TRISS (0.82), RTS (0.81), KTS (0.74), NISS (0.65) and ISS (0.62). RTS was the most parsimonious model with the lowest AIC score. Considering overall mortality, both physiologic scores (RTS, KTS) had better discrimination and goodness-of-fit than ISS or NISS. The ability of all Injury scores to predict early mortality (24 h) was better than late mortality (30 day).ConclusionOn-admission physiological scores outperformed the more expensive anatomy-based ISS and NISS. The retrospective nature of ISS and TRISS score calculations and incomplete imaging in LMICs precludes its use in the casualty department of LMICs. They will remain useful for outcome comparison across trauma centres. Physiological scores like the RTS and KTS will be the practical score to use in casualty departments in the urban Indian setting, to predict early trauma mortality and improve triage.     

A Comparison of the Kampala Trauma Score (KTS) with the Revised Trauma Score (RTS), Injury Severity Score (ISS) and the TRISS Method in a Ugandan Trauma Registry

Abstract Background: The public health significance of injuries that occur in developing countries is now recognized. In 1996, as part of the injury surveillance registry in Kampala, Uganda, a new score, the Kampala Trauma Score (KTS) was instituted. The KTS, developed in light of the limited resource base of sub-Saharan Africa, is a simplified composite of the Revised Trauma Score (RTS) and the Injury Severity Score (ISS) and closely resembles the Trauma Score and Injury Severity Score (TRISS). Patients and Methods: The KTS was applied retrospectively to a cohort of prospectively accrued urban trauma patients with the RTS, ISS and TRISS calculated. Using ROC (receiver operating characteristics) analysis, logistic regression models and sensitivity and specificity cutoff analysis, the KTS was compared to these three scores. Results: Using logistic regression models and areas under the ROC curve, the RTS proved a more robust predictor of death at 2 weeks in comparison to the KTS. However, differences in screening performance were marginal (areas under the ROC curves were 87% for the RTS and 84% for the KTS) with statistical significance only reached for an improved specificity (67% vs. 47%; p < 0.001), at a fixed sensitivity of 90%. In addition, the KTS predicted hospitalization at 2 weeks more accurately. Conclusion: The KTS statistically performs comparably to the RTS and ISS alone as well as to the TRISS but has the added advantage of utility. Therefore, the KTS has potential as a triage tool in resource-poor and similar health care settings.     

Exploring injury severity measures and in-hospital mortality: A multi-hospital study in Kenya

IntroductionLow- and middle-income countries (LMICs) have a disproportionately high burden of injuries. Most injury severity measures were developed in high-income settings and there have been limited studies on their application and validity in low-resource settings. In this study, we compared the performance of seven injury severity measures: estimated Injury Severity Score (eISS), Glasgow Coma Score (GCS), Mechanism, GCS, Age, Pressure score (MGAP), GCS, Age, Pressure score (GAP), Revised Trauma Score (RTS), Trauma and Injury Severity Score (TRISS) and Kampala Trauma Score (KTS), in predicting in-hospital mortality in a multi-hospital cohort of adult patients in Kenya.MethodsThis study was performed using data from trauma registries implemented in four public hospitals in Kenya. Estimated ISS, MGAP, GAP, RTS, TRISS and KTS were computed according to algorithms described in the literature. All seven measures were compared for discrimination by computing area under curve (AUC) for the receiver operating characteristics (ROC), model fit information using Akaike information criterion (AIC), and model calibration curves. Sensitivity analysis was conducted to include all trauma patients during the study period who had missing information on any of the injury severity measure(s) through multiple imputations.ResultsA total of 16,548 patients were included in the study. Complete data analysis included 14,762 (90.2%) patients for the seven injury severity measures. TRISS (complete case AUC: 0.889, 95% CI: 0.866–0.907) and KTS (complete case AUC: 0.873, 95% CI: 0.852–0.892) demonstrated similarly better discrimination measured by AUC on in-hospital deaths overall in both complete case analysis and multiple imputations. Estimated ISS had lower AUC (0.764, 95% CI: 0.736–0.787) than some injury severity measures. Calibration plots showed eISS and RTS had lower calibration than models from other injury severity measures.ConclusionsThis multi-hospital study in Kenya found statistical significant higher performance of KTS and TRISS than other injury severity measures. The KTS, is however, an easier score to compute as compared to the TRISS and has stable good performance across several hospital settings and robust to missing values. It is therefore a practical and robust option for use in low-resource settings, and is applicable to settings similar to Kenya.     

An artificial neural network as a model for prediction of survival in trauma patients: validation for a regional trauma area

BACKGROUND: To develop and validate an artificial neural network (ANN) for predicting survival of trauma patients based on standard prehospital variables, emergency room admission variables, and Injury Severity Score (ISS) using data derived from a regional area trauma system, and to compare this model with known trauma scoring systems. PATIENT POPULATION: The study was composed of 10,609 patients admitted to 24 hospitals comprising a seven-county suburban/rural trauma region adjacent to a major metropolitan area. The data was generated as part of the New York State trauma registry. Study period was from January 1993 through December 1996 (1993-1994: 5,168 patients; 1995: 2,768 patients; 1996: 2,673 patients). METHODS: A standard feed-forward back-propagation neural network was developed using Glasgow Coma Scale, systolic blood pressure, heart rate, respiratory rate, temperature, hematocrit, age, sex, intubation status, ICD-9-CM Injury E-code, and ISS as input variables. The network had a single layer of hidden nodes. Initial network development of the model was performed on the 1993-1994 data. Subsequent models were generated using the 1993, 1994, and 1995 data. The model was tested first on the 1995 and then on the 1996 data. The ANN model was tested against Trauma and Injury Severity Score (TRISS) and ISS using the receiver operator characteristic (ROC) area under the curve [ROC-A(z)], Lemeshow-Hosmer C-statistic, and calibration curves. RESULTS: The ANN showed good clustering of the data, with good separation of nonsurvivors and survivors. The ROCA(z) was 0.912 for the ANN, 0.895 for TRISS, and 0.766 for ISS. The ANN exceeded TRISS with respect to calibration (Lemeshow-Hosmer C-statistic: 7.4 for ANN; 17.1 for TRISS). The prediction of survivors was good for both models. The ANN exceeded TRISS in nonsurvivor prediction. CONCLUSION: An ANN developed for trauma patients using prehospital, emergency room admission data, and ISS gave good prediction of survival. It was accurate and had excellent calibration. This study expands our previous results developed at a single Level I trauma center and shows that an ANN model for predicting trauma deaths can be applied across hospitals with good results     

Predicting inpatient mortality for pediatric trauma patients with blunt injuries: a better alternative

PURPOSE: The aim of this study was to identify significant independent predictors of inpatient mortality rates for pediatric victims of blunt trauma and to develop a formula for predicting the probability of inpatient mortality for these patients. METHODS: Emergency department and inpatient data from 2,923 pediatric victims of blunt injury in the New York State Trauma Registry in 1994 and 1995 were used to explore the relationship between patient risk factors and mortality rate. A stepwise logistic regression model with P<.05 was developed using survival status asthe dependent variable. Independent variables included are elements of the Pediatric Trauma Score (PTS), additional elements from the Revised Trauma Score (RTS), the motor response and eye opening components of the Glasgow Coma Scale (GCS), age-specific systolic blood pressure, the AVPU score, and 2 measures of anatomic injury severity (the Injury Severity Score [ISS] and the International Classification of Disease, Ninth Revision-based Injury Severity Score [ICISS]). RESULTS: The only significant independent predictors of severity that emerged were the ICISS, no motor response (best motor response = 1) from the GCS, and the unresponsive component from the AVPU score. The statistical model exhibited an excellent fit (C statistic = .964). The specificity associated with the prediction of inpatient mortality rate based on the presence of 1 or more of these risk factors was .926, and the sensitivity was .944. CONCLUSION: The best independent predictors of inpatient mortality rate for pediatric trauma patients with blunt injuries include variables not specifically contained in the PTS or the RTS: ICISS, no motor response (best motor response = 1) from the GCS, and the unresponsive component of the AVPU score.     

Harborview assessment for risk of mortality: an improved measure of injury severity on the basis of ICD-9-CM

BACKGROUND: There have been several attempts to develop a scoring system that can accurately reflect the severity of a trauma patient's injuries, particularly with respect to the effect of the injury on survival. Current methodologies require unreliable physiologic data for the assignment of a survival probability and fail to account for the potential synergism of different injury combinations. The purpose of this study was to develop a scoring system to better estimate probability of mortality on the basis of information that is readily available from the hospital discharge sheet and does not rely on physiologic data. METHODS: Records from the trauma registry from an urban Level I trauma center were analyzed using logistic regression. Included in the regression were Internation Classification of Diseases-9th Rev (ICD-9CM) codes for anatomic injury, mechanism, intent, and preexisting medical conditions, as well as age. Two-way interaction terms for several combinations of injuries were also included in the regression model. The resulting Harborview Assessment for Risk of Mortality (HARM) score was then applied to an independent test data set and compared with Trauma and Injury Severity Score (TRISS) probability of survival and ICD-9-CM Injury Severity Score (ICISS) for ability to predict mortality using the area under the receiver operator characteristic curve. RESULTS: The HARM score was based on analysis of 16,042 records (design set). When applied to an independent validation set of 15,957 records, the area under the receiver operator characteristic curve (AUC) for HARM was 0.9592. This represented significantly better discrimination than both TRISS probability of survival (AUC = 0.9473, p = 0.005) and ICISS (AUC = 0.9402, p = 0.001). HARM also had a better calibration (Hosmer-Lemeshow statistic [HL] = 19.74) than TRISS (HL = 55.71) and ICISS (HL = 709.19). Physiologic data were incomplete for 6,124 records (38%) of the validation set; TRISS could not be calculated at all for these records. CONCLUSION: The HARM score is an effective tool for predicting probability of in-hospital mortality for trauma patients. It outperforms both the TRISS and ICD9-CM Injury Severity Score (ICISS) methodologies with respect to both discrimination and calibration, using information that is readily available from hospital discharge coding, and without requiring emergency department physiologic data.     

Comparison of trauma assessment scores and their use in prediction of infection and death.

Current trauma assessment scores do not include an assessment of immune competence and have not been designed to predict late death from or risk of infection. We have compared the use of the Outcome Predictive Score (OPS) with other standard scales to predict clinical outcome after trauma. The OPS combines the Injury Severity Score (ISS) corrected for age (%LD50), degree of bacterial contamination, and monocyte HLA-DR antigen expression on hospital admission. The OPS was compared to the ISS, %LD50, Revised Trauma Score (RTS), Combined Trauma Score-ISS (TRISS), and Anatomical Index (AI). Sixty-one seriously ill patients were studied. Patient outcome was defined as uneventful recovery (n = 18), major infection (n = 27), and death (13 of 16 deaths resulted from infection). The assessment scores were compared for their use in prediction of these outcomes, as well as their ability to distinguish patients with good outcome from those patients who developed major infection or died, and to differentiate survival from death. Only the OPS was able to significantly segregate all five outcome groups (p less than 0.05). Although the age-adjusted ISS distinguished between survival and death (p less than 0.05), only OPS consistently distinguished between good outcome and sepsis/death (p less than 0.05), and therefore best identified the patients who developed infection. AI, RTS, and TRISS had little predictive value.     

Validation of the International Classification of Diseases 10th Edition-based Injury Severity Score (ICISS)

OBJECTIVE: To compare the predictive power of International Classification of Diseases 10th Edition (ICD-10)-based International Classification of Diseases 9th Edition-based Injury Severity Score (ICISS) with Trauma and Injury Severity Score (TRISS) and ICD-9CM-based ICISS in the injury severity measure. METHODS: ICD-10 version of survival risk ratios was derived from 47,750 trauma patients from 35 emergency centers for 1 year. The predictive power of TRISS, the ICD-9CM-based ICISS and ICD-10-based ICISS were compared in a group of 367 severely injured patients admitted to two university hospitals. The predictive power was compared by using the measures of discrimination (disparity, sensitivity, specificity, misclassification rates, and receiver operating characteristic curve analysis) and calibration (Hosmer-Lemeshow goodness-of-fit statistics), all calculated by logistic regression procedure. RESULTS: ICD-10-based ICISS showed a lower performance than TRISS and ICD-9CM-based ICISS. When age and Revised Trauma Score were incorporated into the survival probability model, however, ICD-10-based ICISS full model showed a similar predictive power compared with TRISS and ICD-9CM-based ICISS full model. ICD-10-based ICISS had some disadvantages in predicting outcomes among patients with intracranial injuries. However, such weakness was largely compensated by incorporating age and Revised Trauma Score in the model. CONCLUSION: The ICISS methodology can be extended to ICD-10 horizon as a standard injury severity measure in the place of TRISS, especially when age and Revised Trauma Score were incorporated in the model. For patients with intracranial injuries, the predictive power of ICD-10-based ICISS was relatively low because of differences in the classifying system between ICD-10 and ICD-9CM.