ANNALS OF EMERGENCY MEDICINE JOURNAL CLUB

Speed Does Matter: Police “Scoop and Run” Transport ofCritical Trauma Victims

Answers to the May 2014 Journal Club Questions

Guest ContributorsSamuel J. Stratton, MD, MPH; Atilla Uner, MD, MPH

0196-0644/$-see front matterCopyright © 2014 by the American College of Emergency Physicians.http://dx.doi.org/10.1016/j.annemergmed.2014.06.021

Editor’s Note: You are reading the 39th installment of Annals ofEmergency Medicine Journal Club. This Journal Club refers to theBand et al1 article titled “Severity-Adjusted Mortality in TraumaPatients Transported by Police,” published in the May 2014edition. Information about Journal Club can be found at http://www.annemergmed.com/content/journalclub. Readers shouldrecognize that these are suggested answers. We hope they areaccurate; we know that they are not comprehensive. There aremany other points that could be made about these questions orabout the article in general. Questions are rated “novice” ( ),“intermediate” ( ), and “advanced” ( ) so that individualsplanning a journal club can assign the right question to the rightstudent. The “novice” rating does not imply that a novice should beable to spontaneously answer the question. “Novice” means weexpect that someone with little background should be able to do abit of reading, formulate an answer, and teach the material toothers. Intermediate and advanced questions also will likelyrequire some reading and research, and that reading will besufficiently difficult that some background in clinical epidemiologywill be helpful in understanding the reading and concepts. We areinterested in receiving feedback about this feature. Please e-mailjournalclub@acep.org with your comments.

DISCUSSION POINTS

1. Describe the study goal and researchers’ conclusions.What is the difference between nonmedical police transportand emergency medical services (EMS) transport?

2. A. For this retrospective study, patients with penetratingtrauma were divided into 2 groups: those transported bypolice and those transported by EMS. How mightselection bias affect study results?B. What are the characteristics of the emergencymanagement of penetrating trauma of the thorax thatsupport the study results? What aspects of theemergency management for penetrating trauma of thethorax contradict the logic of the study results? Whichcharacteristics of police transport could EMS emulateand which cannot be emulated?

3. A. The outcome measure for this study was inhospitalmortality. Is this outcome measure optimal? Are thereother outcome measures that would be of interest? Whatwould be the obstacles to studying them?

Volume 64, no. 4 : October 2014

B. The authors report that 4.7% of the initial studypopulation could not be included in the study becauseof missing information about transport mode (policeversus EMS). How could the data have been analyzed todetermine whether this 4.7% missing data biased studyresults?C. Are there other acute emergency conditions that maybenefit from incorporating rapid nonmedical transportinto a community emergency response system?

4. An adjustment of case mix among study subjects wasmade with the Charlson comorbidity index. Was thisan appropriate adjustment technique for victims ofpenetrating trauma? The Charlson index was modifiedto conduct the study; how may this affect the validityof the results?

ANSWER 1Q1. Describe the study goal and researchers’ conclusions. What is

the difference between nonmedical police transport and emergencymedical services (EMS) transport?

The study goal was to evaluate the association between modeof transport (police versus EMS) to the emergency department(ED) and survival among patients with proximal penetratingtrauma.1 In effect, the study compares mortality outcome forrapid transport of a trauma victim without medical intervention(police transport) versus standard out-of-hospital EMS on-scenecare before transport (EMS transport).

The difference between nonmedical police transport and EMStransport is best described in the following quote from therecently published article by Band et al1: “Although EMS followscitywide out-of-hospital protocols, no formal policy outlines howcare should be provided.by police.” It is further stated thatvictims transported by police “.are rendered no care..”

Police provided immediate transport, presumably withoutbasic medical care and without obtaining clearance for transportto a trauma center. Police officers apparently had no formalmedical training, but during law enforcement training are taughthow to best incapacitate an aggressor with their firearms and mayinfer that a victim hit in vulnerable body areas will be severelyinjured. Many police officers have a military background andwould have a soldier’s understanding of penetrating trauma

Annals of Emergency Medicine 417

Journal Club Stratton & Uner

severity. Band et al1 reported that patients transported by policewere on average more severely injured, suggesting that officersmight have some expertise in triaging more critical patients toimmediate police transport rather than waiting for EMS.

By contrast, paramedics have extensive formal medicaltraining and mandatory continuing medical education. There isno mention in the study by Band et al1 about what the localEMS protocols for proximal penetrating trauma required. EMSstandard practice usually includes airway management, assistingwith ventilations as needed, aggressive hemorrhage control, andother field and transport procedures, as illustrated in Table 1. Forthe study by Band et al,1 EMS-transported patients were alsolikely placed in complete spine immobilization because this wasstill recommended practice during the 2003 to 2007 enrollmentperiod. Although some interventions can be conducted en route,others must be conducted at the scene and require considerabletime. It is also not known whether there were on-scene delays fora decision about which trauma center to transport victims to whowere managed by EMS.

The overall assumption is that police transport was more rapidthan EMS transport both because the police did less and perhapsbecause they reached some victims more quickly than EMS, whowould have had to wait until the police declared the scene safe.These assumptions are not validated and it is not reported towhat extent EMS performed on-scene medical interventions.Scene times are not reported for either group.

The unadjusted data suggested that individuals transportedby police were more severely injured and had a higher mortalityrate. However, when adjusted for injury severity and patientdemographics, no important difference in mortality wasobserved. Post hoc subgroup analyses identified improvedsurvival for victims transported by police with gunshot wounds,with stab wounds, and when severely injured (defined as InjurySeverity Score [ISS] >15). These subgroup findings should beinterpreted with caution and the authors do not discuss why

Table 1. Scene activities: police versus EMS.*

Police Transport

1. Intuitive assessment with decision to transport2. Possible control of external hemorrhage3. Place in police vehicle and transport to hospital

1.2.3.4.

5.6.7.

8.9.

10.11.12.

*Illustration of typical on-scene actions for police and EMS.

418 Annals of Emergency Medicine

these subgroups show a potential survival benefit while othersubgroups (presumably of same age and sex, according toinformation provided) show the opposite (given that the overalleffect is neutral). Further study would be required to confirmthese subgroup effects.

ANSWER 2Q2.a For this retrospective study, patients with penetrating

trauma were divided into 2 groups: those transported by police andthose transported by EMS. How might selection bias affect studyresults?

Bias is an error in the design or conduct of a study thatproduces a spurious value for the association between exposureand outcome. Causal inference is possible only when anassociation is not confounded, meaning that both groupswould have had the same outcome had they received the sametreatment. In other words, in an unconfounded study bothpolice-transported and EMS-transported groups would have hadthe same outcomes if they had both been managed on scene andtransported by the same methods.

The study is confounded (biased) if the 2 groups would havehad different outcomes had they received identical on-scenetreatment and transport. Selection bias occurs when theplacement of the patient into a study group is associated withthe outcome.2 For example, in its extreme, if the most injured50% of patients are transported by police and the least injured50% by EMS, then results would be highly confounded favoringEMS transport, assuming that injury correlates with outcome.Although we cannot absolutely determine the degree of selectionbias (because we can’t have patients transported one way,measure an outcome, roll back time, and have them transportedthe other way), we can look for signs of confounding such asbetween-group differences in variables known to affect outcomeand, if necessary, attempt to adjust for confounding factors with

EMS Transport

Airway assessment: assisted ventilation as indicatedControl of external hemorrhageAssess and cover “sucking” chest woundsPrimary trauma assessment, including:a. Level of consciousnessb. Circulation: blood pressure, oximetryc. Disability assessment: spinal stabilizationOxygen therapy as indicatedNeedle decompression of tension pneumothorax as indicatedDefinitive airway management (supraglottic airway, intubation,cricothyroidotomy) as indicatedContact medical control for destinationDocument data for encounter (ongoing)Package, load, and transport to trauma centerIntravenous access with crystalloid infusion during transport as indicatedSecondary survey during transport

Volume 64, no. 4 : October 2014

Stratton & Uner Journal Club

multivariable analysis to determine the strength of association forthe factor with outcome. In the article by Band et al,1 Table 1shows that the police- and EMS-transported groups were similar,but that more severely injured victims, according to mean ISS,were transported by police. In their analysis, the authorsattempted to account for such differences by adjusting results fordemographics, injury type, and ISS.

There is another form of selection bias that could affect studyresults. Only patients transported to Level I and II trauma centerswere included as subjects. If either police or EMS brought somemoribund patients to the nearest facility rather than a traumacenter, these patients would not be in the study and these badoutcomes (assuming the patients died) would not be accountedfor when study results were rendered. Although the study usesacceptable techniques for accounting for selection bias, there isno guarantee that adjustments were adequate and did notintroduce additional biases. Further independent observationalstudies or, ideally, a randomized trial is needed to confirm theresults obtained by Band et al.1

Q2.b What are the characteristics of the emergency managementof penetrating trauma of the thorax that support the study results?What aspects of the emergency management for penetrating traumaof the thorax contradict the logic of the study results? Whichcharacteristics of police transport could EMS emulate and whichcannot be emulated?

Few penetrating thoracic injuries require field interventionand most often require the resources available at a trauma center.Delays in the field can result in physiologic deterioration from

Table 2. EMS actions to stabilize thoracic trauma victims.

Potential Acute On-Scene PenetratingThoracic Trauma Conditions

1. Airway obstruction or occlusion

2. Spinal cord or unstable spinal column injury3. Poor ventilatory effort, respiratory failure

4. Tension pneumothorax5. Open chest wound with risk of tension pneumothorax6. Hypoxia7. Uncontrolled external hemorrhage

8. Hypovolemic and other shock states

9. Potential hypothermia10. Pain in a hemodynamically stable victim11. Cardiac rhythm disturbances of significance

The items listed are common actions that EMS personnel may use to stabilize a thoracicsuspected pericardial tamponade and Focused Assessment with Sonography for Trauma (

Volume 64, no. 4 : October 2014

progressing shock and hypoxia, with subsequent poor outcome.Some penetrating thoracic injuries may stabilize in the field withimmediate lifesaving interventions, such as tension pneumothorax(needle or tube thoracotomy). But most often, there is no standardtreatment in the field other than rapid transport to a traumacenter. Other traditional EMS interventions have provedcontroversial (routine spinal immobilization) or have beenquestioned (aggressive intravenous fluid resuscitation). Table 2provides a summary of out-of-hospital procedures that maybenefit victims of penetrating thoracic trauma.

With the exception of thoracostomy and ventilatory support,few EMS thoracic injury interventions are truly required withinminutes and can be effectively provided at the trauma centerwithin the short transport times of an urban EMS system. Ifpolice transport is faster than EMS transport, some conditions,such as tension pneumothorax or hypotension caused byhemothorax, may not become clinically evident before arrival to atrauma center. On the other hand, some patients may dieunnecessarily during police transport because of lack of on-scenemedical intervention. The study does not provide information onhow often this happened or for which injuries a poor outcomeoccurred.

The study by Band et al1 is not limited to thoracic injuries.Other groups of patients whom on-scene medical intervention(hemorrhage control, airway control, or intravenous fluidresuscitation) may have benefited are those with exsanguinatingproximal extremity injuries, penetrating neck injuries, or vascularand solid organ abdominal injuries.

Potential On-Scene EMS Actions to StabilizeVictims of Penetrating Thoracic Trauma

1. Establish patent airway as immediately requireda. Manual airway maneuversb. Airway adjuncts (oral pharyngeal, nasal pharyngeal)c. Supraglottic airway devicesd. Intubatione. Cricothyroid procedure

2. Selective spinal immobilization3. Assist ventilation

a. Bag-valve-maskb. Bag-valve-mask to airway devicec. Autoventilating devices (when intubated)

4. Needle or tube decompression of suspected tension pneumothorax5. Semiocclusive dressing for open chest wound(s)6. Monitor oxygen saturation, provide supplemental oxygen7. Control external hemorrhage

a. Direct pressureb. Compressive dressingc. Tourniquet

8. Fluid resuscitation of shock states (preferred during transportwithout delay on scene)

9. Monitor for and measures to avoid hypothermia10. Administer pain medication as appropriate11. Cardiac rhythm monitoring and treatment as appropriate

trauma victim in the field. Some EMS systems also provide for pericardiocentesis ofFAST) assessment with field ultrasonography.

Annals of Emergency Medicine 419

Journal Club Stratton & Uner

EMS patient management in the field is regulated byprotocols and procedures. Extensive scrutiny is applied byagencies to each case, and paramedics face professionalconsequences if they do not comply with all applicable rules andregulations. This includes recording medical response data,which is required for state and national databases. In most EMSsystems, individual paramedic judgment such as omitting spinalprecautions to decrease transport delays is discouraged regardlessof patient outcomes. Following protocols takes significant time,and scrutiny is also applied to EMS by receiving hospitals(“What, no IV?”), leading to complaints to EMS superiors,whereas police officers likely face fewer such pressures. It appearsthat Philadelphia police officers were not restricted by protocolsand regulations and were free to load a victim into a policevehicle and transport immediately.

EMS can learn from the police transport model described inthe study by Band et al1 by hastening EMS procedures. Forexample, omitting routine spinal precautions for patients withoutneurologic deficit is being implemented in many EMS systems.Emphasizing short on-scene time (5 minutes or less) andperforming interventions and communicating with the receivingtrauma center while en route are other ways for EMS to emulatepolice procedures that may have been effective in this study.Receiving hospitals and regulatory agencies can help by being lesscritical of omitted actions that do not directly affect patientoutcome.

Although police officers have the training and equipment toenter hostile areas, EMS providers are not equipped to safelyenter areas of ongoing violence and will not always be able toextricate victims as quickly as the police. One way to bridge thisgap is to form tactical EMS teams that can extricate patients froma tactical environment and hand them off to EMS transportcrews in a safe zone. Police can learn from EMS by providingsimple and quick medical procedures such as hemorrhagecontrol, and the fact that Philadelphia is issuing tourniquets to allpolice officers supports this concept.

ANSWER 3Q3.a The outcome measure for this study was in-hospital

mortality. Is this outcome measure optimal? Are there other outcomemeasures that would be of interest? What would be the obstacles tostudying them?

Mortality is a common outcome measure in many studies thatevaluate serious acute trauma. It is a binary outcome measurethat is easily understood and measured. As a dichotomousvariable, mortality lends itself to unambiguous accounting andstraightforward statistical analysis and interpretation. On theother hand, mortality alone provides an incomplete picture of thehealth status of a population and fails to provide informationabout nonfatal aspects of trauma outcomes such as chronicmental and physical disabilities. Although mortality was notinfrequent in the study by Band et al,1 disability is likely evenmore prevalent and, consequently, may produce a greater overallhealth burden. For example, there may have been considerable

420 Annals of Emergency Medicine

variation in the level of neurologic survival characteristics amongthe 2 study groups that was unaccounted for in this study.Anoxic brain injury caused by prolonged lack of ventilation couldhave been a significant clinical outcome for either group: forpolice-transported patients because of lack of airway control orfor EMS-transported patients because of prolonged “down time”before arrival and institution of airway management. Short- andlong-term disability may be as important to measure as mortalitywhen trauma outcome is considered.

Measuring disability is more difficult than counting deaths. Incardiac resuscitation research, a 4-level scale of function, theCerebral Performance Category, is commonly used.3 This scoreprovides a gross assessment of an individual’s ability to return todaily, self-sustained living activity. A more sophisticated measureis Years Lived with Disability, which is commonly used in healthimpact assessment research.4 Disability outcomes are usuallybased on validated scoring systems. Analysis of this type of datarequires more advanced study methods and statistical techniquesthan are required for analysis of mortality data. These measuresare typically conducted at some time distant from the injury andhospitalization and therefore require a substantially moreexpensive study, given that patients have to be followed, located,and reassessed.

Q3.b The authors report that 4.7% of the initial studypopulation could not be included in the study because of missinginformation about transport mode (police versus EMS). How couldthe data have been analyzed to determine whether this 4.7% missingdata biased study results?

The authors were required to exclude 4.7% of the initial studypopulation because there was no record on how the patient wastransported to the trauma center. It is customary that patientstransported by EMS come with EMS field records, which aremade part of the medical record. It can be assumed that whenpatients are transported by police, such field care documentationis less likely to exist in the patient’s medical record and thusthe trauma registry, and therefore it is assumed that a greaterproportion of the 4.7% without a record on mode of transportwere transported by police rather than by EMS.

If the transport mode were primarily by police for the missing4.7% of the study population, the outcome results could havebeen more or less favorable, depending on mortality rates for thisportion of the population. These missing cases present risk forselection bias because they represent lost data related to exposure(mode of transport) and outcome (mortality) that could havechanged the statistics for one or both study groups. The handlingof missing data is a critical study design component, especiallyin retrospective cohort studies. Readers are directed to theNovember 2008 and March 2010 Annals of Emergency MedicineJournal Clubs Answers for a detailed discussion on thesetopics.5,6

Given that the between-group difference in this study is small,if the mortality rate in the 4.7% of patients with unknowntransport type differed greatly from that observed in the 95.3%who were analyzed, then inclusion of these patients could changethe study results. In a study that shows slight or no difference in

Volume 64, no. 4 : October 2014

Stratton & Uner Journal Club

outcomes, this number could bias the results. Ideally, the authorscould have performed and reported a sensitivity analysis todetermine how much results could be skewed by the missingdata. This could be accomplished by reanalysis of study resultsunder the 2 extreme situations that (1) all 4.7% of patients lived,and (2) that they died and by assuming that every patient in the4.7% with unknown transport status was transported by police,and, conversely, all of these patients were transported by EMS.This would put boundaries on the results that are moremeaningful than confidence intervals (CIs) because they take intoconsideration nonrandom error, whereas CIs assume that there isonly random error (aka bias).

Q3.c Are there other acute emergency conditions that may benefitfrom incorporating rapid nonmedical transport into a communityemergency response system?

A common practice in some US EMS systems is for police,when first on scene, to transport small children who are victimsof near drowning with continuing basic pediatriccardiopulmonary resuscitation (CPR) while in transit. There areinsufficient data to determine whether this practice is associatedwith improved outcomes because of infrequency of occurrenceand need for organized multicenter study design. Although mostwill accept that EMS transport of pediatric near-drowningvictims is preferred because of ability to apply oxygen and likelybetter CPR performance and airway management, there is logicfor immediate transport from the scene by police with CPR inprogress so that the child is more quickly delivered to theadvanced care available in an ED.

Although not rapid transport by police, use of automaticexternal defibrillators is an analogous situation to rapid policetransport of victims with penetrating trauma.7 Police use ofautomatic external defibrillators when they are first to arrive to ascene of cardiac arrest has support throughout most US EMSsystems.

ANSWER 4Q4. An adjustment of case mix among study subjects was made

with the Charlson comorbidity index. Was this an appropriateadjustment technique for victims of penetrating trauma? TheCharlson index was modified to conduct the study; how may thisaffect the validity of the results?

The Charlson comorbidity index is a method to adjust datafor 19 comorbid conditions according to their association withmortality. The index can be calculated from InternationalClassification of Diseases, Ninth Revision, Clinical Modification(ICD-9-CM) data and is therefore commonly used to adjuststudies of medical conditions.8 To apply the Charlson index,ICD-9-CM coded data (in the study by Band et al,1 presumablyICD-9-CM coded data in the Trauma Registry) are used toestimate and rank underlying risk of mortality related to medicalcomorbidities. The Charlson index includes medical conditionssuch as myocardial infarction, congestive heart failure, andperipheral vascular disease and was derived with an index

Volume 64, no. 4 : October 2014

population of medical patients.9 A potential concern for use ofthe Charlson index to adjust for case mix in the study by Bandet al1 is that the study population was young (average range 27.7to 30.6 years) and likely did not have the comorbidities that areincluded in the Charlson index.

The Charlson index has been validated for medical patients.8

The use of the index for adjusting trauma-related data has beenquestioned, and it has been suggested that a similar trauma-focused scoring system, the Mortality Risk Score for Trauma(MoRT), is more appropriate for trauma case-mixadjustment.9,10 Although the MoRT case-mix adjustment indexhas been validated for trauma victims, it relies on chronicconditions such as severe liver disease, myocardial infarction, anddementia that would be of low prevalence in the population withpenetrating trauma described for the study by Band et al.1

Furthermore, it has not been shown that the MoRT method is abetter predictive tool than the Charlson method for trauma case-mix adjustment.10 In summary, the Charlson index and MoRTmethod for mortality case-mix adjustment of a relatively youngpopulation with penetrating trauma likely have little adjustmenteffect because the factors used by either method wouldinfrequently be present.

The Charlson comorbidity index has been evaluated incomparison with the Trauma and Injury Severity Score for blunttrauma outcomes, and both scoring systems were found to haveequal predictability for inhospital death.9 Similar studies for theuse of the Charlson index and penetrating trauma have not beenpublished, to our knowledge.

A further concern about the use of the Charlson Index as arisk-adjustment tool in the study by Band et al1 is the authors’decision to use a modification of the index that excluded 4 of 19conditions present in the complete index. Excluding these 4conditions might have affected the validity of the index. Thearticle does not report how frequently the 4 excluded conditionsoccurred within the study group, so it is difficult to tell whetherbias was introduced.

The validity of case-mix adjustment methods for this studyis important because the unadjusted data showed that policetransported more severely injured victims and that those victimshad a higher mortality (odds ratio [OR] 1.18; 95% CI 1.00 to1.39). However, the multivariable analysis that adjusted forTrauma and Injury Severity Score, modified Charlson index, age,and sex produced an OR of 0.78 (95% CI 0.6 to 1.0). Asdiscussed in question 2a, it is impossible to know whether thisOR represents a true association of transport method andinhospital death or whether it is a biased estimate.

Section editors: Tyler W. Barrett, MD, MSCI; David L. Schriger,MD, MPH

Author affiliations: From the UCLA Fielding School of Public Health,Los Angeles, CA (Stratton); and the David Geffen School ofMedicine at UCLA, Los Angeles, CA (Stratton, Uner).

Annals of Emergency Medicine 421

Journal Club Stratton & Uner

REFERENCES1. Band RA, Salhi RA, HolenaDN, et al. Severity-adjustedmortality in trauma

patients transported by police. Ann Emerg Med. 2014;63:608-614.2. Aschengrau A, Seage GR. Selection bias. In Essentials of Epidemiology

in Public Health. Jones & Bartlett (eds): Boston, MA; 2008:263-270.3. Phelps R, Dumas F, Maynard C, et al. Cerebral Performance Category

and long-term prognosis following out-of-hospital cardiac arrest. CritCare Med. 2013;41:1252-1257.

4. Vos T, Flaxman AD, Naghavi M, et al. Years Lived with Disability (YLDs)for 1160 sequelae of 289 diseases and injuries 1990-2010: asystematic analysis for the Global Burden of Disease Study 2010.Lancet. 2012;380:2163-2169.

5. Reynolds TA, Schriger DL, Barrett TW, et al. Empiric antibiotic therapyfor sepsis patients: monotherapy with b-lactam or b-lactam plus anaminoglycoside? Ann Emerg Med. 2009;32:536-543.

422 Annals of Emergency Medicine

6. Barrett TW, Brywczynski JJ, Schriger DL. Is the golden hourtarnished? registries and multivariable regression. Answers to theMarch 2010 Journal Club Questions. Ann Emerg Med.2010;56:188-200.

7. Husain S, Eisenberg M. Police AED programs: a systematic review andmeta-analysis. Resuscitation. 2013;84:1184-1191.

8. D’Hoore W, Bouckaert A, Tilquin C. Practical considerations on the useof the Charlson comorbidity index with administrative data bases.J Clin Epidemiol. 1996;49:1429-1433.

9. Gabbe BJ, Magtengaard K, Hannaford AP, et al. Is the Charlson indexuseful for predicting trauma outcomes? Acad Emerg Med. 2005;12:318-321.

10. Thompson HJ, Rivara FP, Nathens A, et al. Development and validationof the mortality risk for trauma comorbidity index. Ann Surg. 2010;252:370-375.

Get the Latest Articles WithoutWaiting for the Print Edition –

Annals Articles In Press

Annals offers its readers an advancelook at articles in press. This papershows how your colleagues aretreating a difficult disease in achallenging patient population.

Visit Articles in Press onwww.annemergmed.com.

Volume 64, no. 4 : October 2014