Download pdf - Effect of regional trauma centralization on volume, injury severity and outcomes of injured patients admitted to trauma centres

Original article

Effect of regional trauma centralization on volume, injuryseverity and outcomes of injured patients admitted to traumacentres

D. Metcalfe1,2, O. Bouamra3, N. R. Parsons1, M.-O. Aletrari4, F. E. Lecky3,5 and M. L. Costa1

1Warwick Medical School, University of Warwick, Coventry, 2College of Medicine and Veterinary Medicine, University of Edinburgh, Edinburgh,3Trauma Audit and Research Network, University of Manchester, Salford, Manchester, 4Faculty of Medicine, Imperial College London, London, and5Emergency Medicine Research in Sheffield (EMRiS), School of Health and Related Research, University of Sheffield, Sheffield, UKCorrespondence to: Mr D. Metcalfe, Warwick Medical School, Clinical Sciences Building, University Hospital Coventry and Warwickshire,Coventry CV2 2DX, UK (e-mail: [email protected])

Background: Centralization of complex healthcare services into specialist high-volume centres isbelieved to improve outcomes. For injured patients, few studies have evaluated the centralization ofmajor trauma services. The aim of this study was to evaluate how a regional trauma network affectedtrends in admissions, case mix, and outcomes of injured patients.Methods: A retrospective before–after study was undertaken of severely injured patients attending fourhospitals that became major trauma centres (MTCs) in March 2012. Consecutive patients with majortrauma were identified from a national registry and divided into two groups according to injury beforeor after the launch of a new trauma network. The two cohorts were compared for differences in casemix, demand on hospital resources, and outcomes.Results: Patient volume increased from 442 to 1326 (200 per cent), operations from 349 to 1231 (253per cent), critical care bed-days from 1100 to 3704 (237 per cent), and total hospital bed-days from 7910to 22 772 (188 per cent). Patient age increased on MTC designation from 45·0 years before March 2012to 48·2 years afterwards (P = 0·021), as did the proportion of penetrating injuries (1·8 versus 4·1 per cent;P = 0·025). Injury severity fell as measured by median Injury Severity Score (16 versus 14) and RevisedTrauma Score (4·1 versus 7·8). Fewer patients required secondary transfer to a MTC from peripheralhospitals (19·9 versus 16·1 per cent; P = 0·100). There were no significant differences in total durationof hospital stay, critical care requirements or mortality. However, there was a significant increase, from55·5 to 62·3 per cent (P < 0·001), in the proportion of patients coded as having a ‘good recovery’ atdischarge after institution of the trauma network.Conclusion: MTC designation leads to an increased case volume with considerable implications foroperating theatre capacity and bed occupancy. Although no mortality benefit was demonstrated within6 months of establishing this trauma network, early detectable advantages included improved functionaloutcome at discharge.

Paper accepted 17 February 2014Published online in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.9498

Introduction

Major injuries are the third leading cause of lost disability-adjusted life-years worldwide1. A growing internationalevidence base2–11 suggests that outcomes for severelyinjured patients are optimal at dedicated trauma centres.The American College of Surgeons established anaccreditation programme for trauma centres in 1987, andmany studies from the USA2–4 favour management ofseverely injured patients at specialist institutions. Thesereports have shown that quality of care is better at

accredited trauma centres5, that patients benefit fromdirect triage to trauma centres over delayed transfer fromother hospitals6,7, and that overall mortality is lower attrauma centres8,9. However, it is unclear whether thesebenefits can be exported readily to other regions thatdiffer in terms of geography, healthcare economics andtrauma epidemiology. Although studies from Australia10,11

also suggest that trauma service centralization improvesoutcomes, a previous study12 of centralization in Englandfound no effect on mortality. No recent studies have

2014 BJS Society Ltd BJS 2014; 101: 959–964Published by John Wiley & Sons Ltd

960 D. Metcalfe, O. Bouamra, N. R. Parsons, M.-O. Aletrari, F. E. Lecky and M. L. Costa

described the effect of centralizing major trauma serviceswithin an English setting. However, in early 2012, a seriesof regional trauma networks were launched across England,each centred on dedicated major trauma centres (MTCs).

The aim of the present study was to analyse data from asingle regional trauma network to evaluate the early impactof MTC designation on case mix, hospital resources andshort-term patient outcomes.

Methods

An observational before–after study was conductedusing registry data from all four hospitals in a singleregion (West Midlands, UK) that became MTCs inMarch 2012. The West Midlands is a region of 2·7million people; its designated MTCs are UniversityHospital Birmingham, University Hospital Coventry andWarwickshire, University Hospital of North Staffordshireand Birmingham Children’s Hospital, which serves as theregional paediatric MTC.

Data source and case selection

All patients with major trauma presenting within a 200-dayperiod before and after launch of the trauma networkon 26 March 2012 were included in the study. Datawere extracted from nationally collected Trauma Auditand Research Network (TARN) submissions. The studyinclusion criteria are therefore identical to those used toidentify cases that should be reported to TARN: all injuredpatients regardless of age who were inpatients for 72 h ormore, or were admitted to a high-dependency area, or diedafter reaching hospital, and who sustained a severe injury asdefined in the TARN procedures manual (such as a femoralfracture excluding femoral neck fracture in patients aged65 years or more)13.

Patient outcomes recorded within TARN include deathand functional recovery at discharge, which is coded usingthe Glasgow Outcome Scale. This is a five-point scale thatincludes death, persistent vegetative state, severe disability,moderate disability, and good recovery14. Patients withinTARN are recorded as having a ‘good recovery’ if theywere discharged home to live independently, with no newarrangements for domiciliary care.

The comprehensiveness of TARN is estimated byreporting the number of submissions from each hospitalas a proportion of patients with a primary diagnosisof International Classification of Diseases tenth revision(ICD-10) code range S00–T75 contained within HospitalEpisode Statistics. These values for the four MTCs in thepresent study ranged from 67 to 88 per cent in 2012.

Statistical analysis

All statistical analyses were performed using GraphPadPrism 6 (GraphPad Software, San Diego, California,USA). Continuous variables were compared between 200-day periods before and after launch of the trauma networkusing unpaired t tests for normally distributed data andMann–Whitney U tests for non-normally distributed data.Categorical variables were compared using the χ2 test withYates’ continuity correction.

Revised Trauma Score (RTS) values, which lie alonga physiological severity scale from zero (most injured)to 12 (least injured)15, were calculated using individualcomponents recorded on arrival at hospital: GlasgowComa Scale (GCS) score, systolic blood pressure andrespiratory rate.

Outcome data are presented both for all injured patientsand for the most severely injured patients (defined as InjurySeverity Score (ISS) of 15 or above). Specific outcomemeasures such as hospital and critical care length of stayexcluded patients who had died, to avoid these measuresbeing downwardly biased by the inclusion of patients whodied at an early stage.

Adjusted mortality statistics (W and W s) were calculatedfor the two groups. Initial sample sizes were 442 and 1326respectively, but patients transferred to other centres wereexcluded as their final outcome was unknown. The Wstatistic16 represents the excess number of survivors per100 cases and is defined as: W = 100 × [(observed no. ofsurvivors − expected no. of survivors)/no. of patients]. W sis a further standardization of the W statistic that is usefulwhen hospitals with different case mix are compared16. Inthis case, W is most valuable because it allows comparisonover time of the same group of hospitals.

Expected survival was calculated using the sum ofsurvival probability obtained from the risk-adjusted modelused in TARN17. Missing GCS scores in the emergencydepartment were replaced by the prehospital GCS score, ifrecorded, or imputed using a locally determined multipleimputation technique. Statistical significance was set for Pvalues of less than 0·050.

Results

The total number of patients with major injury increasedsubstantially from 442 (mean 2·2 per day) to 1326 (mean6·6 per day) following the launch of the trauma network;this represents a 200 per cent increase. Mean age increasedsignificantly from 45·0 (95 per cent confidence interval(c.i.) 42·7 to 47·3) to 48·2 (46·9 to 49·5) years (P = 0·021,t test). The proportion of penetrating injuries increasedfrom 1·8 to 4·1 per cent (P = 0·025).

2014 BJS Society Ltd www.bjs.co.uk BJS 2014; 101: 959–964Published by John Wiley & Sons Ltd

Effect of regional trauma centralization on outcomes 961

Falls from height were the leading cause of severe injuryin both groups, although the proportion did not changesignificantly in the two time periods (49·8 versus 48·5 percent; P = 0·641). The proportion of injuries caused by roadtraffic collisions was similar (36·7 versus 36·8 per cent;P = 0·955), as was the proportion of intentional injuries(12·9 versus 10·7 per cent; P = 0·972).

The proportion of patients transferred from otherhospitals decreased (19·9 versus 16·1 per cent), althoughthis change was not statistically significant (P = 0·100).Similarly, there was no significant difference in theproportion transported by air ambulance (21·8 versus 23·5per cent; P = 0·516).

Median ISS fell from 16 to 14, and the proportion ofseverely injured patients (ISS 15 or more) fell from 52·3to 48·1 per cent (P = 0·131). The median RTS increasedfrom 4·1 to 7·8, indicating a reduction in physiologicaldisturbance among patients in the later group. Theproportions of patients with tachycardia (heart rate 100beats per min or more) (28·9 versus 28·1 per cent; P = 0·783)and hypotension (systolic blood pressure 100 mmHg orless) (5·3 versus 6·0 per cent; P = 0·630) did not change.

The overwhelming majority had a GCS score of 15 onarrival, and the proportion of those scoring 8 or less did notincrease significantly (8·5 versus 10·0 per cent; P = 0·475).There were, however, ten patients with an ISS of 59 ormore in the later group whereas no such patients wereidentified before institution of the trauma network. Thedistribution of ISS is shown in Fig. 1.

Resource demands

The proportion of patients requiring surgery increasednon-significantly from 50·2 to 55·0 per cent (P = 0·083)and the mean number of operations per patient remainedconstant (1·6 versus 1·7; P = 0·286, Mann–Whitney Utest). In keeping with the increased volume of patients, theabsolute number of operations increased by 253 per cent,from 349 (1·7 per day) to 1231 (6·2 per day).

The proportion of patients requiring hospital admissionwas unchanged (95·7 versus 96·5 per cent; P = 0·469).However, the total number of hospital bed-days increasedby 188 per cent, from 7910 to 22 772 after launch of thetrauma network. The proportion of patients admitted to a

15

30Inju

ry S

ever

ity S

core

45

60

75

0Before After

Fig. 1 Scatter plot showing the distribution of the Injury Severity Score in injured patients before and after institution of the traumanetwork. Each point represents a severely injured patient reported to the Trauma Audit and Research Network. The central bardenotes the group median and the error bars show the range



critical care area did not change (29·0 versus 31·4 per cent;P = 0·341), although total bed occupancy increased by 237per cent, from 1100 to 3704 days.

Outcomes

Outcomes in the two groups for all injured patients andfor the severely injured (ISS 15 or above) are shown inTables 1 and 2 respectively. Critical care length of stay wascalculated only for patients admitted to a high-dependencyarea during their admission.

Duration of hospital and critical care stay did not differbetween the groups. However, the proportion of patientscoded as having a good recovery at discharge increasedsignificantly, from 55·5 to 62·3 per cent (P < 0·001).Although the proportion of patients with an ISS of 15 orabove coded as having a good recovery appeared to increaseby an even greater amount, from 44·3 to 55·5 per cent, thisdifference was not statistically significant (P = 0·821).

The crude mortality rate for all injured patients fellfrom 7·9 to 6·3 per cent following the launch of thetrauma network, although this change was not significant(P = 0·298). Similarly, the mortality rate in the subgroupof patients with an ISS of 15 or above fell from 12·6

Table 1 Outcomes for all injured patients before and afterinstitution of the trauma network

Beforecentralization

Aftercentralization P†

Duration of hospitalstay (days)*

9·0 (16·0, 21·9) 11·0 (16·6, 18·8) 0·068

Duration of criticalcare stay (days)*

5·0 (7·2, 11·4) 5·0 (8·1, 10·5) 0·306

‘Good recovery’ atdischarge

172 of 310 (55·5) 308 of 494 (62·3) < 0·001‡

Crude mortality 35 of 443 (7·9) 84 of 1325 (6·3) 0·298‡

Values in parentheses are percentages unless indicated otherwise; *valuesare median (95 per cent confidence interval). †Mann–Whitney U test,except ‡χ2 test.

Table 2 Outcomes for patients with an Injury Severity Score of15 or above before and after institution of the trauma network

Beforecentralization

Aftercentralization P†

Duration of hospitalstay (days)*

12·0 (20·4, 28·8) 14·0 (20·3, 24·2) 0·599

Duration of criticalcare stay (days)*

6·0 (7·9, 12·7) 6·0 (9·4, 12·3) 0·181

‘Good recovery’ atdischarge

90 of 203 (44·3) 254 of 458 (55·5) 0·821‡

Crude mortality 29 of 230 (12·6) 65 of 639 (10·2) 0·385‡

Values in parentheses are percentages unless indicated otherwise; *valuesare median (95 per cent confidence interval). †Mann–Whitney U testtest, except ‡χ2 test.

to 10·2 per cent (P = 0·385). The W statistic increasedfrom 1·80 (95 per cent c.i. −0·44 to 4·05) to 3·73 (2·44to 5·02), indicating that following MTC designation1·93 lives were saved per 100 patients beyond thatexpected, although the overlapping confidence intervalspoint to no statistically significant difference. The W sstatistic also increased, from 1·33 (−0·49 to 3·15) to2·56 (1·54 to 3·58).

Discussion

This study describes a longitudinal evaluation of traumaservice centralization and demonstrates an increasedvolume of severely injured patients with predictableimplications for operating theatre capacity, hospital bed,and critical care bed occupancy in trauma centres. Areduction in overall injury severity (ISS) and physiologicaldisturbance (RTS) was noted, perhaps due to expandedambulance triage criteria, for example based on mechanismof injury. A greater proportion of penetrating injuries areseen in trauma centres, as well as a group of criticallyinjured patients who might previously have been triageddirectly to the nearest resuscitation facility.

Following their designation as MTCs, the four hospitalscontributing to this analysis shouldered an additional14 862 hospital bed-days and 2604 critical care bed-days ina single 6-month period. Fewer patients required secondarytransfer from other hospitals following the launch ofthe trauma network, although this reduction was notstatistically significant.

Possible advantages of prehospital triage to a MTCinclude avoiding complex transfer arrangements, delayedoperative intervention (such as wound closure) and reducedtotal length of stay. Evidence from the USA6,7,18,19 suggeststhat outcomes are improved for patients triaged directly toan appropriate facility compared with those transferred ata later time.

In this analysis, no statistically significant improvementswere identified for hospital length of stay, critical carelength of stay, or mortality. Importantly, one institutionreported atypical mortality data to TARN: no traumadeaths after becoming a MTC. The mortality data shouldtherefore be interpreted cautiously, although the datafrom this hospital seem unlikely to have influenced theoverall finding of no statistically significant improvementin mortality rate. There was a highly significant (P < 0·001)improvement in the proportion of patients dischargedwith a coded outcome of good recovery compared againstall other outcomes (moderate disability, severe disability,persistent vegetative state and death). Improvement in thesubgroup of patients with an ISS of 15 or more would havebeen consistent with studies showing that such patients


Effect of regional trauma centralization on outcomes 963

gain most from treatment at a trauma centre18,20, althoughthis finding was not statistically significant.

Although the West Midlands trauma network is servedby two ambulance organizations, both implemented similarmajor trauma triage tools on 26 March 2012. After this date,patients were triaged directly to MTCs based on vital signs(such as systolic blood pressure below 90 mmHg), anatomyof injury (for example, penetrating chest trauma) and mech-anism of injury (for instance, ejection from motor vehicle).It is therefore likely that a substantial proportion of severelyinjured patients were taken directly to MTCs after openingof the trauma network. However, there are no reliable datathat describe the outcomes of patients taken to other hos-pitals in the earlier group, before the launch of the traumanetwork. Similarly, there is a possibility of undertriage, inwhich case some severely injured patients might have beentreated at non-MTC hospitals and so missed from the latergroup. The proportion of patients transferred secondarilyto MTCs from other hospitals (16·1 per cent) providessome indication of the number of patients triaged else-where but ultimately requiring care at a specialist traumacentre. A definitive evaluation of trauma network impact onpatient outcomes would, however, require comprehensivedata from all hospitals in the study region.

Although English trauma networks are already showingprocedural improvements21, the present study found nosignificant mortality benefit. It may, however, be tooearly to detect improved mortality outcomes. Although aprevious study22 reported dramatic improvements withinmonths of trauma centre designation, others4,23,24 havesuggested that the full benefits become apparent only givensufficient time for maturation of staff experience, triageprotocols and structural changes. For example, one USstudy4 reported that it took up to 3 years before mortalityoutcomes in new trauma centres were comparable tothose of established institutions in the same region. Aretrospective analysis23 of all road traffic collision deathsin the USA found that mortality began to decline only10 years after implemention of a new trauma system.Similarly, a before–after study from the USA24 reportedno mortality improvement soon after the launch of atrauma network, but a 35 per cent reduction in mortalityafter 2 years. It is therefore possible that future work willshow improved patient outcomes after maturation of thenew trauma network system.

A broad consensus from studies generated outsideEurope2–11 is that trauma service centralization improvespatient outcomes. However, the only previously conductedtrauma network evaluation in the UK12 reported littleimprovement in terms of patient mortality. The authors ofthis study proposed that international data might not apply

to the UK owing to differences in trauma epidemiology,such as greater trauma volume and proportion ofpenetrating injuries in the USA12. Although Englishtrauma service centralization may ultimately improveoutcomes, this model should be evaluated continually toensure that services are organized to optimize outcomesfor severely injured patients.

Acknowledgements

This work was prepared for and supported by theEdinburgh Surgical Sciences Qualification programme,led jointly by the University of Edinburgh and the RoyalCollege of Surgeons of Edinburgh. TARN is funded bysubscriptions from participating hospitals.Disclosure: The authors declare no conflict of interest.

References

1 World Health Organization. Disability Adjusted Life YearsEstimates, 2000–2011. Global Summary Estimates.http://www.who.int/healthinfo/en [accessed 2 January 2014].

2 Demetriades D, Martin M, Salim A, Rhee P, Brown C,Chan L. The effect of trauma center designation and traumavolume on outcome in specific severe injuries. Ann Surg2005; 242: 512–517.

3 Mackenzie EJ, Rivara FP, Jurkovich GJ, Nathens AB,Egleston BL, Salkever DS et al. The impact oftrauma-center care on functional outcomes following majorlower-limb trauma. J Bone Joint Surg Am 2008; 90: 101–109.

4 Pracht EE, Tepas JJ III, Langland-Orban B, Simpson L,Pieper P, Flint LM. Do pediatric patients with trauma inFlorida have reduced mortality rates when treated indesignated trauma centers? J Pediatr Surg 2008; 43:212–221.

5 Draaisma JM, de Haan AF, Goris RJ. Preventable traumadeaths in the Netherlands – a prospective multicenter study.J Trauma 1989; 29: 1552–1557.

6 Haas B, Stukel TA, Gomez D, Zagorski B, De Mestral C,Sharma SV et al. The mortality benefit of direct traumacenter transport in a regional trauma system: apopulation-based analysis. J Trauma Acute Care Surg 2012;72: 1510–1515.

7 Nirula R, Maier R, Moore E, Sperry J, Gentilello L. Scoopand run to the trauma center or stay and play at the localhospital: hospital transfer’s effect on mortality. J Trauma2010; 69: 595–599.

8 Garwe T, Cowan LD, Neas B, Cathey T, Danford BC,Greenawalt P. Survival benefit of transfer to tertiary traumacenters for major trauma patients initially presenting tonontertiary trauma centers. Acad Emerg Med 2010; 17:1223–1232.

9 Meisler R, Thomsen AB, Abildstrøm H, Guldstad N, BorgeP, Rasmussen SW et al. Triage and mortality in 2875



consecutive trauma patients. Acta Anaesthesiol Scand 2010;54: 218–223.

10 Gabbe BJ, Biostat GD, Lecky FE, Bouamra O, WoodfordM, Jenks T et al. The effect of an organized trauma systemon mortality in major trauma involving serious head injury: acomparison of the United Kingdom and Victoria, Australia.Ann Surg 2011; 253: 138–143.

11 Cameron PA, Gabbe BJ, Cooper DJ, Walker T, Judson R,McNeil J. A statewide system of trauma care in Victoria:effect on patient survival. Med J Aust 2008; 189: 546–550.

12 Nicholl J, Turner J. Effectiveness of a regional traumasystem in reducing mortality from major trauma: before andafter study. BMJ 1997; 315: 1349–1354.

13 Trauma Audit and Research Network. Procedures Manual.TARN: Salford, 2012.

14 Jennett B, Bond M. Assessment of outcome after severebrain damage. Lancet 1975; 1: 480–484.

15 Champion HR, Sacco WJ, Copes WS, Gann DS,Gennarelli TA, Flanagan ME. A revision of the TraumaScore. J Trauma 1989; 29: 623–629.

16 Hollis S, Yates DW, Woodford M, Foster P. Standardizedcomparison of performance indicators in trauma: a newapproach to case-mix variation. J Trauma 1995; 38:763–766.

17 Bouamra O, Wrotchford A, Hollis S, Vail A, Woodford M,Lecky F. A new approach to outcome prediction in trauma: acomparison with the TRISS model. J Trauma 2006; 61:701–710.

18 Sampalis JS, Denis R, Frechette P, Brown R, Fleiszer D,Mulder D. Direct transport to tertiary trauma centers versustransfer from lower level facilities: impact on mortality andmorbidity among patients with major trauma. J Trauma1997; 43: 288–295.

19 Barringer ML, Thomason MH, Kilgo P, Spallone L.Improving outcomes in a regional trauma system: impact ofa level III trauma center. Am J Surg 2006; 192: 685–689.

20 Rivara FP, Koepsell TD, Wang J, Nathens A, Jurkovich GA,Mackenzie EJ et al. Outcomes of trauma patients aftertransfer to a level I trauma center. J Trauma 2008; 64:1594–1599.

21 Stammers J, Williams D, Hunter J, Vesely M, Nielsen D.The impact of trauma centre designation on open tibialfracture management. Ann R Coll Surg 2013; 95: 184–187.

22 Shackford SR, Hollingworth-Fridlund P, Cooper GF,Eastman AB. The effect of regionalization upon the qualityof trauma care as assessed by concurrent audit before andafter institution of a trauma system: a preliminary report. JTrauma 1986; 26: 812–820.

23 Nathens AB, Jurkovich GJ, Cummings P, Rivara FP, MaierRV. The effect of organized systems of trauma care onmotor vehicle crash mortality. JAMA 2000; 283:1990–1994.

24 Mullins RJ, Veum-Stone J, Helfand M, Zimmer-GembeckM, Hedges JR, Southard PA et al. Outcome of hospitalizedinjured patients after institution of a trauma system in anurban area. JAMA 1994; 271: 1919–1924.
