Injury, Int. J. Care Injured 42 (2011) 870–873

Contents lists available at ScienceDirect

Injury

journa l homepage: www.e lsevier .com/ locate / in jury

Preventability of trauma deaths in a Dutch Level-1 trauma centre

T.P. Saltzherr a,*, K.W. Wendt b, P. Nieboer b, M.W.N. Nijsten c, J.P. Valk d, J.S.K. Luitse a,K.J. Ponsen a, J.C. Goslings a

a Trauma Unit, Department of Surgery, Academic Medical Center, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlandsb Trauma Unit, Department of Surgery, University Medical Center Groningen, The Netherlandsc Department of Critical Care, University Medical Center Groningen, The Netherlandsd Department of Anesthesiology, University Medical Center Groningen, The Netherlands

A R T I C L E I N F O

Article history:

Accepted 7 April 2010

Keywords:

Trauma

Morbidity

Mortality

Preventability

Outcome and process assessment

Quality improvement

A B S T R A C T

Background: Monitoring the quality of trauma care is frequently done by analysing the preventability of

trauma deaths and errors during trauma care. In the Academic Medical Center trauma deaths are

discussed during a monthly Morbidity and Mortality meeting. In this study an external multidisciplinary

panel assessed the trauma deaths and errors in management of a Dutch Level-1 trauma centre for

(potential) preventability.

Methods: All patients who died during or after presentation in the trauma resuscitation room in a 2-year

period were eligible for review. All information on trauma evaluation and management was summarised

by an independent research fellow. An external multidisciplinary panel individually evaluated the cases

for preventability of death. Potential errors or mismanagements during the admission were classified for

type, phase and domain. Overall agreement on (potential) preventability was compared between the

external panel and the internal M&M consensus.

Results: Of the 62 evaluated trauma deaths one was judged as preventable and 17 were judged as

potentially preventable by the review panel. Overall agreement on preventability between the review

panel and the internal consensus was moderate (Kappa 0.51). The external panel judged one death as

preventable compared with three from the internal consensus. The interobserver agreement between

the external panel members was also moderate (Kappa 0.43). The panel judged 31 errors to have

occurred in the (potential) preventable death group and 23 errors in the non-preventable death group.

Such errors included choice or sequence of diagnostics, rewarming of hypothermic patients, and

correction of coagulopathies.

Conclusions: The preventable death rate in the present study was comparable to data in the available

literature. Compared to internal review, the external, multidisciplinary review did not find a higher

preventable death rate, although it provided several insights to optimise trauma care.

� 2010 Elsevier Ltd. All rights reserved.

Introduction

Trauma care is susceptible to medical errors or mismanage-ments due to the necessity of rapid diagnosis and treatment in asometimes stressful environment. To optimise trauma care and tominimise potential errors algorithms have been developed anddedicated trauma centres have originated.

For the continuous evaluation of the quality of trauma care,evaluating deaths and severely injured patients in whom potentialerrors in management occurred is a commonly applied method.Frequently, this analysis is performed using a classification inpreventability of trauma deaths.2,5,6,10,11,15–18

* Corresponding author. Tel.: +31 20 5666676; fax: +31 20 6914858.

E-mail address: t.p.saltzherr@amc.nl (T.P. Saltzherr).

0020–1383/$ – see front matter � 2010 Elsevier Ltd. All rights reserved.

doi:10.1016/j.injury.2010.04.007

The Academic Medical Center (AMC) is a Dutch Level-1 traumacentre with 32,000 emergency department visits and 700 traumateam notifications per year. Approximately one quarter of thesenotifications concern multitrauma patients. In the AMC all traumadeaths are internally discussed in a monthly quality-monitoringMorbidity and Mortality meeting (M&M) and classified based onconsensus. The review board consists of four trauma surgeons, thetrauma fellow, and the attending residents. All lethal trauma victimsare classified into one of four categories: non-preventable, poten-tially preventable or preventable death. In addition, potential errorsare registered for each phase of trauma care (i.e. pre-hospital,resuscitative, intervention-related and intensive care). The out-comes ofthe M&M meetingserveasa tool toreduce (potential)errorsand to improve the efficiency and effectiveness of our trauma care.

The main purpose of this study was to analyse all trauma deathsfrom a Dutch Level-1 trauma centre for (potential) preventability

Table 1Demographic data.

Non-preventable Potentially preventable

n 44 17

Age 48 (9–88) 53 (19–92)

Gender (male) 64% 71%

ISS 32 (9–75) 36 (16–75)

Blunt mechanism 68% 88%

GCS �8 91% 53%

Causes of death

CNS injury 55% 29%

Haemorrhage 18% 35%

MOF 2% 18%

Drowning 18% 6%

Myocardial infarction/fail. 5% 0%

Respiratory fail. 2% 6%

Pulmonary embol. 0% 6%

CNS = central nervous system; fail. = failure; embol. = embolism.

Table 2Preventable death classification as judged by the internal consensus and external

review.

External review Internal consensus

Non-preventable Potentially

preventable

Preventable Total

Non-preventable 39 4 1 44

Potentially preventable 5 10 2 17

Preventable 0 1 0 1

Total 44 15 3 62

T.P. Saltzherr et al. / Injury, Int. J. Care Injured 42 (2011) 870–873 871

by an external panel. Second, by comparing the external paneljudgement with the internal consensus we assessed if our internalevaluation system functions adequately.

Methods

All trauma patients who died during initial admission afterpresentation to the trauma resuscitating room (TR) from 1-1-2006until 31-12-2007 were eligible for review. Patients in whomresuscitation was stopped before arriving in the TR werepronounced dead on arrival and were excluded for this study.All patients in whom information about initial trauma resuscita-tion or during admission was incomplete were also excluded.During the study period our TR was equipped with a sliding 4-sliceCT scanner (SOMATOM Sensation 4, Siemens) and conventionalradiography equipment.13

The selected population was individually peer reviewed by anexternal, independent, multidisciplinary peer review panel.Participation of the external panel was based on goodwill andthere was no legislative or personal obligation. There was no directrelationship to the authors’ institution. The review panel consistedof two experienced trauma surgeons, one intensivist and onetrauma anesthesiologist, all working in another Dutch Level-1trauma centre, the University Medical Center Groningen (UMCG)and all with 10–20 years of experience. For the review all availablepatient data was summarised and anonymised by an independentresearch fellow. Patient information was abstracted from pre-hospital and emergency department forms, radiology, operationand pathology records and the in-hospital electronic registrationsystem. Data was anonymised for both patients’ data and for dataon their care givers. Abstracts contained all valuable informationon relevant pre-hospital findings, initial physical examination andvital parameters, first and subsequent laboratory measurements,initial diagnostic examinations and primary interventions andtreatment. If no information was provided on these measurementsor findings, the outcomes could be considered normal by thereview panel.

Before the start of the external review process threerandomly chosen example cases were judged adequatelysummarised for review by one of the members of the externalreview panel. For these cases the member did not require anyadditional information on trauma characteristics, evaluation andtreatment methods and findings to classify the preventabilityand potential errors.

All patients who were not unanimously classified by theindividual reviewers were discussed again during two interactivemeetings where the independent research fellow was present withall detailed data on the different cases. If required, additionalinformation on trauma care could then be provided. In thesepatients the final classification of preventability was reached bygeneral consensus between the four panel members. Prior to thereview the panel members all received written information on howto classify deaths and errors. Classification was based on thecriteria from the American College of Surgeons.1 The errors ormismanagements that have been identified were classifiedaccording to type, phase and domain in which they occurred.Potential pre-hospital errors were not assessed systematicallybecause information on pre-hospital trauma care was notsufficient in all patients.

Kappa statistics were performed for comparison of the externalreview and the outcomes of the previously described internalM&M meetings as well as for the interobserver agreement betweenthe individual panel members. Continuous variables weredescribed as mean with ranges and categorical variables weredescribed as percentages. Analyses were performed using SPSSBase for Windows, version 15.0.1 (SPSS, Inc., Chicago, IL).

Results

During the study period 72 deceased trauma patients wereeligible for external review. Four elderly patients with hip fracturesdied of non-trauma related causes of death and were excludedbecause of incorrect trauma team notifications. Six other patientshad incomplete information on initial resuscitation and hospitaltreatment and were excluded for review. In total, 62 patients weresummarised for external peer review. Overall, mean age was 50years (9–92), 66% were male, 79% had sustained blunt trauma, 13%were drowning victims and 8% had a penetrating traumamechanism. The mean injury severity score (ISS) was 33 (9–75).The most frequent cause of death was central nervous system(CNS) injury (47%), followed by a combination of cardio-respiratory arrest (27%), uncontrollable bleeding (19%) andmulti-organ failure (7%).

The external reviewers judged one death as preventable (2%)and 17 deaths as potentially preventable (27%). Most deaths wereconsidered potentially preventable when the reviewers wouldhave chosen for other therapeutic and diagnostic pathways (e.g.more aggressive ICU treatment, blood transfusions and rewarmingwith extracorporeal circulation, earlier operative interventionsinstead of extra radiodiagnostics). The one death that wasconsidered preventable was a patient who died as a result ofintracranial haemorrhage after a fall from the stairs. Earlier andcorrect diagnosis of the injury in combination with earliercorrection of haemostatic disorders could have prevented thepatient’s death in the opinion of the external reviewers. Thedemographic data for both non-preventable and potentiallypreventable deaths according to the external judgement is shownin Table 1.

The death that was judged preventable according to the panelwas considered potentially preventable. However, in contrast tothe external reviewers the internal consensus judged three deathsas preventable instead of potentially or not preventable. The

Table 3Preventability agreement between external reviewers (Kappa).

Reviewers 1 2 3 4

1 X X X X

2 0.59 X X X

3 0.23 0.27 X X

4 0.50 0.66 0.33 X

Fig. 1. Classification per type of error for (potentially) preventable and non-

preventable deaths.

T.P. Saltzherr et al. / Injury, Int. J. Care Injured 42 (2011) 870–873872

overall agreement between the internal consensus and theexternal reviewers is shown in Table 2. The overall kappa was0.51 which can be considered moderate according to the method ofLandis and Koch.8 The Kappa between panel members is presentedin Table 3. Overall, the Kappa between panel members was 0.43which also can be considered moderate.

In total, there were 31 errors for the potential and preventabledeath group and 23 errors for the non-preventable death group.The classification for type of error is presented in Fig. 1. Mostfrequently mentioned errors were a delay in treatment due towrong choice or order of diagnostics, not adequate rewarming ofhypothermic patients and correction of coagulopathies. Within the(potential) preventable deaths 5 errors occurred in the investiga-tion phase, 20 in the resuscitation phase, 3 in the operative phaseand 3 were classified as other. These numbers for errors within thenon-preventable death group were 15 in the resuscitation phase, 5in the investigation phase and 3 were classified as other. Overall,38 errors occurred in the trauma resuscitation room, 10 in the ICU,4 in the operation room and 2 on the general ward.

Discussion

The introduction of trauma systems has improved trauma careconsiderably.4,14 For the continuous quality and performanceimprovement of trauma care the assessment of potentiallypreventable deaths and errors is an important method. The primarypurpose of this study was to assess the in-hospital trauma deaths byan external, multidisciplinary review panel and classify them forpreventability. One of the 62 deaths was judgedpreventable(2%) and17 as potentially preventable deaths (27%). Compared with theliterature the numbers in the present study are lower but a validcomparison is difficult.3,6–8,11,12,15,17,18 The main reason for thisdifficult comparison is that several different methods for reviewingare used, as was also noted by MacKenzie et al.9 In some studies theTRISS score was applied for calculating the probability of survival onwhich the preventability classification was based. In other studies a

peer review panel or a combination of both is used for classifying thepreventability. In studies that used a review panel the size andcomposition of the panels differed (i.e. uni- versus multidisciplinaryand/or independency). Second, whether the judgement is based onclinical experiences or if standardised criteria were used also variedwidely. Another problem for comparison is that in most studiesdifferent inclusion criteria were used (i.e. in- or exclusion of pre-hospital deaths or certain trauma mechanisms and injuries). Thiswide variation in study methods hampers the comparison of moststudies.

The overall agreement between the results of our monthlyM&M meeting and the external review panel was consideredmoderate. The external panel’s judgement seemed more favour-able than the internal consensus. Consequently, we think that weshould take measures to improve interobserver and inter-teamagreement (i.e. by using standardised definitions and classifica-tions). After such measures further, more symmetrical multilateralpeer review involving more trauma centres may be useful.

Although the expert panel in this study consisted of specialistsfrom only three specialties it did show the need for a multidisci-plinary review panel since there was a moderate agreementbetween panel members. Especially the anaesthesiologist andintensive care physician had some criticism about several potentialerrors, such as methods of rewarming and coagulopathy correc-tions, which were not the primary objections of the other panelmembers. Because in our monthly M&M meeting we lack themultidisciplinary structure this is something to consider in thefuture to optimise our M&M meeting.

There were two patients with an ISS of 75 which the externalpanel classified as potential preventable. This seems odd because anISS of 75 indicates unsurvivable injuries. The explanation for this isthat we did not provide the ISS scores in the summary to thereviewers but rather a detailed description of the sustained injuries.One patient had a complete avulsion of the right hepatic lobe withlarge vessel disruptions. The second patient had a traumatic aorticrupture with haemorrhage not confined to the mediastinum.Therefore, the judgement of the reviewers was based on theirspecific knowledge on the prognosis and survivability of theseinjuries. If we had used other methods for classification, such as theTRISS methodology, these cases would have been classified as non-preventable. With our methodology the panel members’ reason forclassifying these deaths as potentially preventable was in both casesthe choice to perform a CT scan before going to the OR.

The largest proportion of errors that was classified by the reviewpanel was due to a delay in treatment or error in management. Mostof the suggested errors in both categories were based on the wrongchoice or order of diagnostic imaging. A possible explanation is thatin the AMC there is an infrastructure that includes a sliding,multislice CT scanner in the TR. This enables early and fast CTscanning in severely injured but haemodynamically stable patients.Although in some cases the internal consensus also concluded thatthe reason and timing for CT was debatable, we consider most ofthese suggested errors by the external panel as the result of theirdifferent perspective as for example the external panel’s traumacentre does not possess the advanced rapid CT facilities. This couldalso explain the relatively high frequency of suggested errorsoccurring in the TR during resuscitation and diagnostics.

Other frequent remarks concerned the methods of rewarming ofhypothermic patients. The external panel suggested in several casesof drowning victims the use of rewarming with extracorporealcirculation. In our centre the cardiothoracic team is alwaysconsulted when this may be an option. In general they are reluctantto start this treatment based on interpretation of the availableliterature. It should be noted that the indications for initiation ofextracorporeal circulatory warming among other alternatives havelong been debated. Another issue was that more aggressive

T.P. Saltzherr et al. / Injury, Int. J. Care Injured 42 (2011) 870–873 873

correction of coagulopathy was advised in some patients. A potentialexplanation could be that only summaries on the amount of usedblood products and plasma derivatives were provided to theexternal panel. As a result the review panel could have had the falseimpression that we did not correct coagulopathies adequately.

A limitation of the study is the limited constitution of theexternal panel in terms of specialties represented. It could beargued that we should have formed an external panel consisting ofmore specialists including neurosurgeons, radiologists, emergencyphysicians, emergency nurses or pre-hospital care givers. Thismight have improved the value of the insights to further improvethe outcome of trauma care in our institution.

A second limitation is the relative high amount of inadequate orincomplete registration of pre-hospital information and initialvital parameters within our trauma deaths. This was the reasonthat we did not assess any pre-hospital errors but also why weexcluded six patients. This deficit in registration is probably theresult of the sometimes stressful or hectic situation during initialtrauma care. In the future, special attention will be paid to preventthis deficiency.

Conclusion

The preventable death rate of trauma patients in the presentstudy was comparable to data available in the literature. Althoughexternal review does not seem imperative to improve our currentsystem of quality monitoring, multidisciplinary reviewing of ourtrauma care did offer valuable insights to optimise the trauma carein our institution. Most errors were due to a delay in treatment orerrors in management. Documentation in lethally injured traumapatients should be improved.

Conflict of interest

The authors hereby declare that there is no conflict of interest.

References

[1] American College of Surgeons Committee on Trauma: Resources for OptimalCare of the Injured Patient: 1993. Chicago: American College of Surgeons; 1993.

[2] Cayten CG, Stahl WM, Agarwal N, Murphy JG. Analyses of preventable deathsby mechanism of injury among 13,500 trauma admissions. Ann Surg1991;214:510–20.

[3] Danne P, Brazenor G, Cade R, et al. The major trauma management study: ananalysis of the efficacy of current trauma care. Aust NZ J Surg 1998;68:50–7.

[4] Demetriades D, Martin M, Salim A, et al. The effect of trauma center designa-tion and trauma volume on outcome in specific severe injuries. Ann Surg2005;242:512–7.

[5] Gruen RL, Jurkovich GJ, McIntyre LK, et al. Patterns of errors contributing totrauma mortality: lessons learned from 2,594 deaths. Ann Surg 2006;244:371–80.

[6] Iau PT, Ong CL, Chan ST. Preventable trauma deaths in Singapore. Aust NZ JSurg 1998;68:820–5.

[7] Ivatury RR, Guilford K, Malhotra AK, et al. Patient safety in trauma: maximalimpact management errors at a level I trauma center. J Trauma 2008;64:265–70.

[8] Landis JR, Koch GG. The measurement of observer agreement for categoricaldata. Biometrics 1977;33:159–74.

[9] MacKenzie EJ. Review of evidence regarding trauma system effectivenessresulting from panel studies. J Trauma 1999;47:S34–41.

[10] MacLeod JB, Cohn SM, Johnson EW, McKenney MG. Trauma deaths in the firsthour: are they all unsalvageable injuries? Am J Surg 2007;193:195–9.

[11] Maio RF, Burney RE, Gregor MA, Baranski MG. A study of preventable traumamortality in rural Michigan. J Trauma 1996;41:83–90.

[12] McDermott FT, Cordner SM, Tremayne AB. Road traffic fatalities in Victoria,Australia and changes to the trauma care system. Br J Surg 2001;88:1099–104.

[13] Saltzherr TP, Fung Kon Jin PP, Bakker FC, et al. An evaluation of a Shockroomlocated CT scanner: a randomized study of early assessment by CT scanning intrauma patients in the bi-located trauma center North-West Netherlands(REACT trial). BMC Emerg Med 2008;8:10.

[14] Sampalis JS, Denis R, Lavoie A, et al. Trauma care regionalization: a process-outcome evaluation. J Trauma 1999;46:565–79.

[15] Teixeira PG, Inaba K, Hadjizacharia P, et al. Preventable or potentially pre-ventable mortality at a mature trauma center. J Trauma 2007;63:1338–46.

[16] Teixeira PG, Inaba K, Salim A, et al. Preventable morbidity at a mature traumacenter. Arch Surg 2009;144:536–41.

[17] Tien HC, Spencer F, Tremblay LN, et al. Preventable deaths from hemorrhage ata level I Canadian trauma center. J Trauma 2007;62:142–6.

[18] Zafarghandi MR, Modaghegh MH, Roudsari BS. Preventable trauma death inTehran: an estimate of trauma care quality in teaching hospitals. J Trauma2003;55:459–65.