Bayesian Scoring Systems for Military Pelvic and Perineal Blast Injuries

MILITARY MEDICINE, 181, 5:127, 2016

Bayesian Scoring Systems for Military Pelvic and Perineal Blast Injuries: Is it Time to Take a New Approach? Somayyeh Mossadegh, BM, MRCS*; Shan He, BEng, MSc, PhD†; Paul Parker, FIMC, FRCS (Orth)‡

INTRODUCTION Trauma scoring systems are used by the United Kingdom (U.K.) Defence Medical Services, in order to provide accurate risk assessments for injured service personnel. This is seen as essential research and investigation as part of the quality assurance process, in order to maintain and improve clinical practice.1 As these scoring systems are well known internationally, it allows comparison of clinical performances between institutions and countries. An in-depth analysis of pelvic and perineal blast injuries in U.K. military service personnel over an 8-year period of the war in Afghanistan provided the evidence for our study injury pattern and created a didactic module for the Military Operational Surgical Training course at the Royal College of Surgeon of England.2 Review of actual injuries sustained by this cohort of patients identified that Injury Severity Score (ISS) did not correlate with the severity of pelvic and perineal injuries,3 a finding that is well known amongst the military medical profession and one that carries a high mortality (Fig. 1).2,4–6 Recent publications by Penn-Barwell et al7 have examined the temporal changes in injury patterns in U.K.

*Queen Mary University of London, Blizard Institute, Barts and The London School of Medicine and Dentistry, The Blizard Building, 4 Newark Street, London E1 2AT, United Kingdom. †School of Computer Science Centre for Systems Biology, University of Birmingham, Edgbaston, Birmingham B15 2TT, United Kingdom. ‡Royal Centre for Defence Medicine, New Queen Elizabeth Hospital Birmingham, Mindelsohn Way, Edgbaston, Birmingham B15 2GW, United Kingdom. This work was presented at the Military Health Systems Research Symposium, Fort Lauderdale, FL, August 19, 2014. This presentation was awarded the Silver Prize Winner at the Young Investigators Competition. The views expressed in this scientific article are that of the authors and not necessarily the views of the United Kingdom’s Ministry of Defense. doi: 10.7205/MILMED-D-15-00171

MILITARY MEDICINE, Vol. 181, May Supplement 2016

military casualties over the last decade, and have shown a year-on-year improvement in survival. A simple approach to better identify the true severity of these casualties has already been published. A cumulative scoring system, which calculated the total injury burden of every tissue/area affected in the pelvic and perineal region.3 Significantly better correlation indicated that further research was necessary to develop this scoring system into a valid model worthy of use in current trauma systems. This article outlines the preliminary steps in developing a mortality prediction algorithm using Naïve Bayesian (NB) analysis specifically in a population of combat trauma patients exposed to an improvised explosive devices (IED) blast injury. We hypothesize that using a nonlogistic regression based algorithm, we were able to produce a more accurate scoring system with highly significant implications for future development on a global scale. METHODS The cohort of casualties in this article was initially analyzed in two previous publications.3,8 The U.K. Joint Theatre Trauma Registry (JTTR) was used to identify all U.K. military service personnel who had sustained a perineal injury over an 8-year period (January 2003–December 2010). Coalition military were excluded due to nonavailability of follow-up. To ensure complete data capture, anal and genitourinary injuries were also specified in search terms. Anatomical injuries affecting the pelvis and perineal region were firstly identified using abbreviated injury scale (AIS) codes and then corroborated using free-text descriptions to confirm that they matched the AIS Military (2005) descriptions.24 Although this was time consuming, it was necessary as frequent input errors were identified and corrected. Physiological data were also available relating to observations at various times, from point of wounding (Role 1 127

Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_5/127/4209476 by guest on 02 January 2019

ABSTRACT Background: Various injury severity scores exist for trauma; it is known that they do not correlate accurately to military injuries. A promising anatomical scoring system for blast pelvic and perineal injury led to the development of an improved scoring system using machine-learning techniques. Methods: An unbiased genetic algorithm selected optimal anatomical and physiological parameters from 118 military cases. A Naïve Bayesian model was built using the proposed parameters to predict the probability of survival. Ten-fold cross validation was employed to evaluate its performance. Results: Our model significantly out-performed Injury Severity Score (ISS), Trauma ISS, New ISS, and the Revised Trauma Score in virtually all areas; positive predictive value 0.8941, specificity 0.9027, accuracy 0.9056, and area under curve 0.9059. A two-sample t test showed that the predictive performance of the proposed scoring system was significantly better than the other systems ( p < 0.001). Conclusion: With limited resources and the simplest of Bayesian methodologies, we have demonstrated that the Naïve Bayesian model performed significantly better in virtually all areas assessed by current scoring systems used for trauma. This is encouraging and highlights that more can be done to improve trauma systems not only for our military injured, but also for civilian trauma victims.

Bayesian Scoring Systems for Military Pelvic and Perineal Blast Injuries

[R1]), to the small forward field medical facility (Role 2 [R2]), to Camp Bastion, a large well-equipped field hospital (Role 3 [R3]), and finally repatriation to the U.K. (Role 4, [R4]) medical facility at Queen Elizabeth Hospital, Birmingham. The physiological variables included the following; systolic and diastolic blood pressure (BP), heart rate, respiratory rate, oxygen saturations, temperature, and white cell count (Table I). ISS and New ISS (NISS) were calculated for all cases due to complete data capture; however, 23 cases had missing physiological variables, which meant that the Trauma ISS (TRISS) and Revised Trauma Score (RTS), which rely on these variables, could not be calculated; therefore, only 95 cases had TRISS and RTS values. The standard coeffiTABLE I.

All Variables Considered, Variables Selected, and Their Coefficients All Variables

Anatomical

Physiological

Pelvis Penis Scrotum Anus Testes Rectum Urethra Perineum Hip/Buttocks BP Systolic, R3, R2, R1 BP Diastolic, R3, R2, R1 Pulse R3, R2, R1 Respiratory Rate R3, R2, R1 SpO2 R3, R2, R1 Temperature R3 WCC

Variables Selected Pelvis Penis Scrotum Ano-Rectum Testes

Coefficient −132.781 −93.22 −63.540 −54.486 53.492

BP Systolic R3

14.872

BP Diastolic R3

−22.233

Pulse R3

7.581

BP, blood pressure; R3, Role 3, R2, Role 2; R1, Role 1; SpO2, oxygen saturation; WCC, white cell count; ISS – Injury Severity Score; NISS, New ISS; , TRISS, Trauma ISS, RTS – Revised Trauma Score.

128

RESULTS The JTTR contained 4,808 coalition casualties, of which 2,204 were U.K. Military and 118 (5.4% of 2,204) were identified as having sustained a perineal injury. The mean age of the cohort was 25, case fatality rate was 47%, and there were 62 (53%) survivors. The demographics of this cohort of patients as well as the distribution of ISS are summarized in Table III.8 The overall case fatality rate for all U.K. military IED related casualties was 28% (all mechanisms was 25%). Using the definition of an unexpected survivor having a TRISS of less than 50% in a surviving casualty, there were 9 unexpected survivors (17%). Twenty three patients did not have TRISS assigned because of index parameters not being recorded, 13 from this group survived. Genitourinary injuries were identified in 85 (72%) of 118 patients. There was a significantly lower rate of mortality in patients with perineal injuries alone compared to MILITARY MEDICINE, Vol. 181, May Supplement 2016

Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_5/127/4209476 by guest on 02 January 2019

FIGURE 1. Bilateral traumatic amputation (often high transfemoral) with devastating pelvic and perineal injuries—the signature injury from the war in Afghanistan (copyright permission from Elsevier).

cients used for both the U.K. and U.S. JTTR are still based on the 1980’s blunt and penetrating injuries. The same standard weighting factors were also used for the RTS. Ultimately, for blast injuries, the most predominant injury would decide if a blunt or penetrating coefficient would be used. If intubated, a casualty would have been given a Glasgow Coma Scale of 3, respiratory rate was recorded as 0 if being ventilated and the rate achieved if breathing spontaneously through the tube. All anatomical injuries in the pelvis and perineal region (Table II) as well as the physiological parameters were put through a process of optimization using an unbiased genetic algorithm to select the most important variables using the Waikato Environment for Knowledge Analysis.8 This is a relatively easy data-mining tool used by many trauma registries. A genetic algorithm is a search heuristic that mimics the process of natural selection. The algorithm first generates a population of random solutions (e.g., a number of physiological measures as features). Then the algorithm will execute selection, e.g., selecting those good solutions (sets of features) that can generate better classification results. Genetic operators, e.g., mutation (changing the features randomly) and crossover (combining two sets of features) will then be applied to create a new generation of solutions. The algorithm will then perform selection and the iterations continue until some stopping criteria are met.9 This process is represented diagrammatically in Figure 2. The algorithm selected 3 variables from the physiological measures: “BP systolic R3,” “BP diastolic R3,” and “pulse R3.” We also included all pelvic and perineal measurements: “pelvis,” “penis,” “testes,” “scrotum,” and “anorectum.” The pelvis measurements referred to severity of pelvic fracture sustained. This scale was derived from the Young and Burgess classification.3,10,11 These variables were then used to build a Naïve Bayes Classifier model. All the algorithms were programmed using MATLAB (Mathworks, Natick, Massachusetts). The average performance of our system was evaluated using 30 runs of 10-fold cross validation and this compared against the performance of ISS, NISS, TRISS, and RTS.

Bayesian Scoring Systems for Military Pelvic and Perineal Blast Injuries TABLE II.

Anatomical Components of Cumulative Pelvic and Perineal Trauma Scoring System

Soft Tissue Components

Pelvic Fracture Scores

Severity Score

Severity Score

Description

Penis Scrotum Testes Urethra Perineum Anus Rectum Hip/Buttocks

1–5 1–5 1–5 1–5 1–4 1–4 1–5 1–3

0 1 2 3 4 5

No Pelvic Fracture Stable Fracture/Fracture Not Involving Ring (Iliac Wing/Coccyx/Distal Sacrum) Pelvic Ring Fracture—Posterior Arch Intact (APC1/LC1/Rami/Ischium) Pelvic Ring Fracture—Posterior Ring Fracture Partially/Vertically Stable (APC2/LC2) Substantial Deformation and Displacement with Associated Vascular Disruption (APC3/CMI) Bilateral Substantial Deformation and Displacement with Vascular Disruption or Traumatic Hemipelvectomy (VS/SIJ Disruption) Hemicorporectomy. Type 6 Injuries are Theoretically Meant to be Unsurvivable

6

Severity score based on abbreviated injury scale (AIS); APC, anterior posterior compression; CMI, combined mechanism injury; LC, lateral compression; SIJ, sacroiliac joint; VS, vertical sheer.

FIGURE 2.

Diagrammatic representation of the whole data analysis process.

those with a coexisting pelvic fracture (p < 0.001). There were 8 patients who had bilateral testicular loss with a mortality rate of 25%.8 Of all the inputted variables, five anatomical and three physiological variables had the most significant effect on outcome as the injury severity increased. Anatomical variables were “penis,” “testes,” “scrotum,” “ano-rectum,” and “pelvis” and physiological variables were “BP systolic R3,” “BP diastolic R3,” and “pulse R3.” Testicular injury had a positive coefficient, implying a positive correlation to survival; whereas the other four areas had a negative coefficient, with the pelvis having the highest negative value. As the severity of the testicular injury increased, the likelihood of the patient having a better outcome improved, whereas the other areas had a negative correlation, which is expected—as the injury gets worse, so too does the likelihood of survival. Using the NB analysis, we demonstrated that there was a statistically significant improvement across the whole range of outcomes when comparing the novel score to current scoring systems (Table IV). DISCUSSION Numerous scoring systems have been developed for trauma over the past 40 years. These are based on anatomical or physiological descriptors or some combination of the two.1,12–17 Over time there have been major improvements and modificaMILITARY MEDICINE, Vol. 181, May Supplement 2016

tions to these scoring systems. However, the simple fact remains that these scoring systems using anatomical descriptors are based on the AIS, originally founded in the early 1970s, by the U.S. automotive industry.18 This scale was developed to measure the severity of (frequently blunt) vehicular trauma at that time. It became useful in measuring other traumatic injuries from different mechanisms of injury, and since its inception there have been several revisions. The most recent revision is AIS 2005 update 2008.19–23 Some modifications were made for military trauma (AIS 2005Military) using expert military surgeons to account for multiple injury etiology from high-energy weapons and explosives, but these did not envisage any triple amputee pelvi-perineally injured IED survivors.24,25 In light of this fact, a Military Injury Scoring Summit was convened in 2008 at the U.S. Army Institute of Surgical Research in San Antonio, Texas. It comprised of a panel of military and civilian experts and resulted in the Military Combat Injury Scale and the Military Functional Incapacity Scale, which was introduced into the literature in 2013.26 This may be reproducible and relatively simple to use, however, with further advances in technology and cross-disciplinary collaborations between computer sciences and medicine, it may already be outdated. The algorithm for our NB model was created using MATLAB (Mathworks)—this clearly is not a program used by clinicians on a daily basis; however, this is where cross-disciplinary collaboration is paramount. Yet et al27 have demonstrated that the development of a Bayesian Network (BN) utilizes not only the raw data available, but also integrates domain expertise to develop and refine a prediction model. Once a model has been created, the next step is to create an interface that is user friendly and readily available. Naturally, there are many steps before a usable interface is created; however, it is possible and as an example, a coagulopathy prediction tool is available online with a fully working BN.27–29 A NB analysis is a form of machine learning, where NB classifiers represent a family of simple probabilistic classifiers based on applying Bayes’ theorem using strong (naïve) independence assumptions between the features.30,31 A Naïve 129

Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_5/127/4209476 by guest on 02 January 2019

Structure

Bayesian Scoring Systems for Military Pelvic and Perineal Blast Injuries TABLE III.

Age ISS (n = 118) NISS (n = 118) TRISS (n = 95) RTS (n = 95)

ALL

Perineal Injury Alone

Perineal Injury and Pelvic Fracture

p

118 56 (48) 35 (67) 21 (37) 47 62 83

56 (47) 11 (20) 8 (23) 3 (14) 20 45 48 (58)

62 (53) 45 (73) 27 (77) 18 (86) 73 17 35 (42)

NS 0.0001 0.0001 0.0001 0.0001 0.0001 NS

Mean

Median (IQR)

25 41.03 53.14 43.51 7.95

25 38 57 28.04 7.84

(21–29) (29–57) (36–75) (0.36–96.69) (0.11–92.88)

CFR, case fatality rate; DOW, died of wounds; KIA, killed in action; NS, not significant; ISS, Injury Severity Score; NISS, New ISS; TRISS, Trauma ISS; p, significance.

TABLE IV.

Naïve Bayesian Analysis of Novel Score vs. Current Trauma Scoring Systems

Experiment

Precision

Recall

Specificity

Accuracy

AUC

p Value

ISS NISS TRISS RTS Novel Scoring System

0.858 0.891 0.784 0.788 0.894

0.800 0.857 0.954 0.929 0.909

0.879 0.905 0.761 0.774 0.903

0.842 0.882 0.853 0.848 0.906

0.844 0.880 0.859 0.851 0.906

<0.001 <0.001 <0.001 <0.001

ISS, Injury Severity Score; NISS, New ISS; , TRISS, Trauma ISS, RTS, Revised Trauma Score; AUC, area under curve (calculated by 30 runs of 10-fold cross validation); p, significance.

Bayes classifier assigns a new observation to the most probable class, assuming the features are conditionally independent given the class value. Bayesian updating is especially important in analyzing sequences of data inference, and has found application in a range of fields including science, engineering, philosophy, medicine, and law.30,31 Some might argue that maybe all that is needed is a new blast TRISS coefficient; however, we would argue that this is frequently based on retrospective data and requires the information to be in a linear distribution. Bayesian algorithms are nonlinear and can use prospective data to learn and develop with more information, including subject matter experts. This concept is currently used to create consensus statements; however, it has not been incorporated into trauma scoring models. Regarding the seemingly contradictory positive effect of testicular loss, recent studies have shown that there is indeed a sex-related difference in outcome in trauma patients with males being more susceptible to multiple organ failure, sepsis, and mortality after trauma.32 Female sex was associated with improved organ function following traumatic injury and hemorrhagic shock, and in particular, in the reproductive age groups (16–44 years).33,34 Testicular injury was the only anatomical area with a positive coefficient in our study, thus corroborating the notion that there may be a protective mechanism as the degree of testicular injury worsened. 130

Further evidence is required before the use of sex steroids or testosterone blockade as a therapeutic intervention in critically ill trauma patients can be advocated. In statistics, logistic regression is a method of analyzing multiple independent variables in a dataset in order to determine the outcome, which is a binary measurement. It aims to identify the best fitting, yet biologically reasonable, model to describe the relationship between the two outcomes and the set of independent variables. Coefficients are generated to predict a logit (log odds) transformation of the probability of presence of the characteristic of interest.32 Unfortunately, it does not account for missing data. Bayesian methodology does, and this analysis was used for our study. Finally, we believe that this data is encouraging and highlights an exciting prospect for the future of improving trauma systems, especially as the NB is the simplest of all the Bayesian models and is purely data driven. As more information is given to the model and developments change, the model will learn from the accumulating data and will allow a more meaningful comparison between different trauma systems, theatres, and time periods. Future work will be directed toward using nonlinear trauma models such as BN to aid clinical governance as well as develop mortality prediction tools looking at total injury burden, which may lead to creating a more accurate global trauma scoring system. MILITARY MEDICINE, Vol. 181, May Supplement 2016

Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_5/127/4209476 by guest on 02 January 2019

Total Casualties, n Mortality, n (%) KIA, n (%) DOW, n (%) CFR, % Survivors, n Exclude KIA, n (%)

Demographics and Distribution of Injury Severity Scores

Bayesian Scoring Systems for Military Pelvic and Perineal Blast Injuries

REFERENCES 1. Russell RJ, Hodgetts TJ, McLeod J, et al: The role of trauma scoring in developing trauma clinical governance in the Defence Medical Services. Philos Trans R Soc Lond B Biol Sci 2011; 366(1562): 171–91. 2. Mossadegh S, Tai N, Midwinter M, Parker P: Improvised explosive device related pelvi-perineal trauma: anatomic injuries and surgical management. J Trauma Acute Care Surg 2012; 73(2 Suppl 1): S24–31. 3. Mossadegh S, Midwinter M, Parker P: Developing a cumulative anatomic scoring system for military perineal and pelvic blast injuries. J R Army Med Corps 2013; 159(Suppl 1): i40–4. 4. Jansen JO, Thomas GO, Adams SA, et al: Early management of proximal traumatic lower extremity amputation and pelvic injury caused by improvised explosive devices (IEDs). Injury 2012; 43(7): 976–9. 5. Ramasamy A, Evans S, Kendrew JM, Cooper J: The open blast pelvis: the significant burden of management. J Bone Joint Surg Br 2012; 94(6): 829–35. 6. Cross AM, Davis C, Penn-Barwell J, Taylor DM, De Mello W, Matthews JJ: The incidence of pelvic fractures with traumatic lower limb amputation in modern warfare due to improvised explosive devices. J R Nav Med Serv 2014; 100(2): 152–6. 7. Penn-Barwell JG, Roberts SAG, Midwinter MJ, Bishop JRB. Improved survival in UK combat casualties from Iraq and Afghanistan. J Trauma Acute Care Surg 2015; 78(5): 1014–20. 8. Hall M, Frank E, Holmes G, Pfahringer B, Reutemann P, Witten IH: The WEKA Data Mining Software: An Update. SIGKDD Explorations. ACM SIGKDD Explorations Newsletter 2009; 11(1): 10–8. Available at http://www.cs.waikato.ac.nz/~eibe/pubs/weka_update.pdf; accessed February 12, 2014. 9. Holland J: Adaptation in Natural and Artificial Systems. Cambridge, MA, MIT Press, 1992. 10. Young JW, Burgess AR, Brumback RJ, Poka A: Lateral compression fractures of the pelvis: the importance of plain radiographs in the diagnosis and surgical management. Skeletal Radiol 1986; 15(2): 103–9. 11. Dalal SA, Burgess AR, Siegel JH, et al: Pelvic fracture in multiple trauma: classification by mechanism is key to pattern of organ injury, resuscitative requirements, and outcome. J Trauma 1989; 29(7): 981–1000.

MILITARY MEDICINE, Vol. 181, May Supplement 2016

12. Champion HR, Copes WS, Sacco WJ, et al: A new characterization of injury severity. J Trauma 1990; 30(5): 539–45. 13. Ali T, Shepherd JP: The measurement of injury severity. Br J Oral Maxillofac Surg 1994; 32(1): 13–8. 14. Champion HR, Copes WS, Sacco WJ, et al: Improved predictions from a severity characterization of trauma (ASCOT) over trauma and injury severity score (TRISS): results of an independent evaluation. J Trauma Acute Care Surg 1996; 40(1): 42–8. 15. Osler T, Baker SP, Long W: A modification of the injury severity score that both improves accuracy and simplifies scoring. J Trauma 1997; 43(6): 922–5. 16. Schluter PJ: The trauma and injury severity score (TRISS) revised. Injury 2011; 42(1): 90–6. 17. Kim YJ: Injury severity scoring systems: a review of application to practice. Nurs Crit Care 2012; 17(3): 138–50. 18. Rating the severity of tissue damage. I. The abbreviated scale. JAMA 1971; 215(2): 277–80. 19. States JD: The Abbreviated Injury Scale: AIS-80. Barrington, IL, American Association for Automotive Medicine, 1980. 20. Gennarelli TA, Committee on Injury Scaling: Abbreviated Injury Scale. University of Chicago (1985). Arlington Heights, IL, American Association for Automotive Medicine, 1985. 21. Gennarelli TA: Abbreviated Injury Scale (1990). Des Plaines, IL, Association for the Advancement of Automotive Medicine, 1990. 22. Gennarelli TA, Wodzin E: The Abbreviated Injury Scale 2005. Barrington, IL, Association for the Advancement of Automotive Medicine (AAAM), 2005. 23. Gennarelli TA, Wodzin E: The Abbreviated Injury Scale 2005: Update 2008. Barrington, IL, Association for the Advancement of Automotive Medicine, 2008. 24. Gennarelli TA, Wodzin E: AIS 2005: a contemporary injury scale. Injury 2006; 37(12): 1083–91. 25. Champion HR, Holcomb JB, Lawnick MM, et al: Improved characterization of combat injury. J Trauma Acute Care Surg 2010; 68(5): 1139–50. 26. Lawnick MM, Champion HR, Gennarelli T, et al: Combat injury coding: a review and reconfiguration. J Trauma Acute Care Surg 2013; 75(4): 573–81. 27. Yet B, Perkins Z, Tai N, Marsh W. Explicit evidence for prognostic bayesian network models. Stud Health Technol Inform 2014; 205: 53–7. 28. Yet B, Perkins Z, Fenton N, Tai N, Marsh W. Not just data: a method for improving prediction with knowledge. J Biomed Inform 2014; 48: 28–37. 29. Acute Traumatic Coagulopathy Bayesian Network Evidence Browser. Available at http://valinor.agena.co.uk:8080/atcbn/atcbn.html; accessed June 3, 2015. 30. Bishop CM: Pattern Recognition and Machine Learning, Vol. 1. New York, NY, Springer, 2006. 31. Hand DJ, Yu K: Idiot’s Bayes—Not so stupid after all? Int Stat Rev 2001; 69(3): 385–99. 32. Pampel FC: Logistic Regression: A Primer (Quantitative Applications in the Social Sciences). Thousand Oaks, CA, Sage Publications, 2000. 33. Trentzsch H, Lefering R, Nienaber U, Kraft R, Faist E, Piltz S: The role of biological sex in severely traumatized patients on outcomes: a matched-pair analysis. Ann Surg 2015; 261(4): 774–80. 34. Trentzsch H, Nienaber U, Behnke M, Lefering R, Piltz S: Female sex protects from organ failure and sepsis after major trauma haemorrhage. Injury 2014; 45(Suppl 3): S20–8.

131

Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_5/127/4209476 by guest on 02 January 2019

SUMMARY This article describes a novel concept for developing a scoring system that correlates better to injuries sustained by U.K. military personnel from IEDs. It has not taken into account any other injuries at present and is no doubt a weak study, which we fully acknowledge, yet it still outperforms current scoring systems. Its progression from a simple cumulative scoring system3 to a NB mathematical model, independent of any formal funding or working groups, is testament to the authors’ determination that this is the future of trauma scoring systems. When used for the purpose that they were designed for—moderate speed automotive crashes, the AIS is an excellent tool. However, newer concepts such as en route airway care, prehospital hemostatic blood transfusion, damage control resuscitation and surgery, and surgeon and institutional trauma volume now means that an ISS of 75 does not always mean death. As treatments evolve—so must scoring systems.