Embed Size (px)
Citation preview
Journal of Safety Research 42 (2011) 345–350
Contents lists available at SciVerse ScienceDirect
Journal of Safety Research
j ourna l homepage: www.e lsev ie r .com/ locate / j s r
What is the potential of trauma registry data to be used for road traffic injurysurveillance and informing road safety policy?☆
Rebecca Mitchell a,b,⁎, Ann Williamson b, Kate Curtis a,c,d
a NSW Injury Risk Management Research Centre, University of New South Walesb Transport and Road Safety (TARS) Research, School of Aviation, University of New South Walesc St George Hospital¸ Sydney, Australiad Sydney Nursing School, University of Sydney
☆ The authors wish to thank the NSW Roads and Trafcess to TADS and the Centre for Epidemiology and Reseament for providing access to the Health Outcomes and(HOIST) to obtain data analysed in this study. The HOISTanalysis and reporting facility established and operated band Research, Population Health Division, NSW Departm⁎ Corresponding author at: Transport and Road Safe
Aviation, University of New South Wales, Sydney NSW9385 7555; fax: +61 2 9385 6637.
E-mail address: [email protected] (R. Mitchell
0022-4375/$ – see front matter © 2011 National Safetydoi:10.1016/j.jsr.2011.06.007
a b s t r a c t
a r t i c l e i n f oArticle history:
Received 11 March 2011Received in revised form 28 June 2011Accepted 29 June 2011Available online 17 September 2011Keywords:Road safetySurveillanceTraumaPolice reportsHospitalisationInjury
Introduction: Information from hospital trauma registries is increasingly being used to support injury surveil-lance efforts. This research examines the potential of using trauma registry data for road traffic injury surveil-lance for different types of road users in terms of both the information collected and how representativetrauma data are compared to two population-based road traffic injury data collections. Methods: The threedata collections were assessed against recommended variables to be collected for injury surveillance pur-poses and the representativeness of the distribution of road traffic-related injury data from the trauma reg-istry was compared to hospital admission and road traffic authority data collections. Results: Data from thetrauma registry was largely not representative of the distribution of age groups or activities compared tothe two population-based collections, but was representative for gender for some road user groups to atleast one population-based data collection. Conclusions: Trauma data could be used to supplement informa-tion from population-based data collections to inform road safety efforts. Impact on Industry: Road safety pol-icy makers should be aware of the potential and the limitations of using trauma registry data for road traffic
injury surveillance.© 2011 National Safety Council and Elsevier Ltd. All rights reserved.
1. Introduction
Around one-quarter of injuries experienced worldwide are roadtraffic-related (Peden, McGee, & Sharma, 2002) and in 2002 these in-juries accounted for an estimated 38.7 million disability adjusted lifeyears globally (World Health Organization [WHO], 2004). In Austra-lia, there are around 1,460 deaths (6.9 per 100,000; Department of In-frastructure Transport Regional Development and Local Government,2009) and 52,000 individuals hospitalized (251 per 100,000 popula-tion; Henley & Harrison, 2009) each year following a road traffic-re-lated injury. While Australia has the capacity to perform routineinjury mortality and hospitalized morbidity surveillance activities,road traffic injury surveillance capabilities across countries vary. Insome countries, a great deal of information may be available for
fic Authority for providing ac-rch at the NSW Health Depart-Information Statistical Toolkitsystem refers to a data access,y the Centre for Epidemiologyent of Health.
ty (TARS) Research, School of2052, Australia. Tel.: +61 2
).
Council and Elsevier Ltd. All rights
surveillance of road traffic injuries, such as the linked crash outcomesdata evaluation systems (CODES) in the United States (National High-way Traffic Safety Authority, 1996) used to examine road traffic-relat-ed injuries, while in other countries road traffic injury informationmay be obtained from hospital records, police reports, or newspaperclippings (Rahman, Andersson, & Svanstrom, 2000), which containonly limited information about the traffic incident and how itoccurred.
One type of data collection that is increasingly being used for inju-ry surveillance purposes are trauma registries (Forst, Hryhorczuk, &Jaros, 1999; Johnson & Moore, 1997; Kobusingye & Lett, 2000;Layde, Stueland, & Nordstrom, 1996; Pollock & McClain, 1989). Trau-ma registries primarily collect data on individuals who have been se-verely injured (Nwomeh, Lowell, Kable, Haley, & Ameh, 2006). Theypredominantly record information on medical care within hospitalsand are used to measure the quality of trauma care (Lloyd & Graitcer,1989). In some instances, trauma registries may also record informa-tion that would be useful for injury surveillance purposes (Pollock &McClain, 1989). A significant advantage of using trauma registrydata for injury surveillance is that the criterion for entry into the reg-istry is usually well-defined being based on a recognized injury mech-anism, physiologic parameters, or injury severity classification. Moreconventional data collections used for road traffic injury surveillanceoften have poorly defined entry criteria, for example the number of
reserved.
Table 1Information recommended for injury surveillance by the World Health Organization1
available in the SGPH trauma register, the NSW APDC, and TADS.
Variables2 SGPH APDC TADS
DemographicsGender ✓ ✓ ✓
Age group ✓ ✓ ✓
Indigenous origin X ✓ XCircumstances of the incidentActivity ✓ ✓ XPlace of occurrence ✓ ✓ ✓
Mechanism of injury ✓ ✓ ✓
Intent ✓ ✓ XDate and time of injury ✓ X ✓
External cause X ✓ XAlcohol a factor ✓ X ✓
Other drugs a factor ✓ X XCausal factors (eg. hazardous features, natural lighting,weather, speed, fatigue)
X X ✓
Preventive factors (eg. seat belt, airbag, helmet) ✓ X ✓
Injury outcomeNature of injury ✓ ✓ XInjury severity ✓ X X
1 World Health Organization. Injury Surveillance Guidelines. Geneva: WHO, 2001.2 Differences in classification schemes used to code place of occurrence, intent and
nature of injury and poorly coded data for preventive factors, alcohol and otherdrugs precluded comparison.
346 R. Mitchell et al. / Journal of Safety Research 42 (2011) 345–350
vehicle crash-related injuries reported in police crash data collectionsmay vary greatly with the severity of the vehicle crash and/or any as-sociated injuries (Lopez, Rosman, Jelinek, Wilkes, & Sprivulis, 2000).Trauma registries may therefore be a useful source of data for roadtraffic injury surveillance as they are likely to contain more homoge-neous data.
Unfortunately, however, little is known about the potential of a trau-ma registry to provide information for road traffic-related injury surveil-lance of different types of road users compared to other population-based data collections more commonly used for road traffic-relatedinjury surveillance. If trauma registry information is to be used forroad traffic injury surveillance, researchers and policy makers needto know how using information from a trauma registry will influencethe types of individuals and traffic incidents reported as this will af-fect the understanding of both the magnitude of the issue and whyroad traffic-related injuries occur.
In Australia, road traffic injury surveillance is routinely conductedusing information recorded by road safety authorities (usuallyobtained from police records; Roads and Traffic Authority NSW,2009) and/or from hospital admissions (Henley & Harrison, 2009),emergency department (ED) presentations (Watson & Ozanne-Smith, 2000), or a combination of these data collections (Boufous &Williamson, 2006; Cercarelli, Rosman, & Ryan, 1996; Lopez et al.,2000; Rosman, 2001; Rosman & Knuiman, 1994). The potential use
Table 2Number of road traffic-related injuries for the SGPH trauma register, the NSW APDC,and TADS, percent, NSW, 2002 to 2008.
SGPH APDC TADS
All crashes1 All crashes1 Traffic-only2 Traffic-only2
n % n % n % n %
Pedestrians 751 13.6 8,802 10.1 6,405 10.7 15,874 9.5Pedal cyclists 454 8.20 17,609 20.2 8,683 14.5 8,100 4.9Motor cyclists 659 11.9 24,290 27.9 12,489 20.8 15,239 9.1Car occupants3 3,671 66.3 36,367 41.8 32,423 54.0 127,835 76.5Total 5,535 100 87,068 100 60,000 100 167,048 100
1 MVCs that occurred either on a public (ie. traffic) or non-public (ie. non-traffic)roadway.
2 MVCs that occurred on a public roadway (ie. traffic).3 SGPH data in ‘car occupants’ refer to occupants of all vehicles.
of information from a trauma registry as an indicator of population-based road traffic injury surveillance has not been fully explored.This paper aims to examine the potential of a trauma registry forroad traffic injury surveillance for different types of road users interms of both the information collected that could be used for injurysurveillance purposes and how representative the distribution ofkey characteristics used for surveillance from trauma data are com-pared to two population-based road traffic injury data collections,such as hospital admission and road safety authority collections.
2. Methods and materials
A retrospective review of data from the St George Public Hospital(SGPH) trauma registry, the New South Wales (NSW) Admitted Pa-tients Data Collection (APDC), and the NSW Road and Traffic Author-ity's (RTA) Traffic Accident Database System (TADS) during January 1,2002 to December 31, 2008 was conducted. Each data collection wasassessed against the information recommended to be collected for in-jury surveillance purposes by the World Health Organization (WHO,2001) and representativeness of the distribution of road traffic-relat-ed injury data from the SGPH trauma registry for different types ofroad users was examined in terms of demographics and circum-stances of the injury incident compared to the APDC and TADS.
2.1. St George public hospital trauma registry
The SGPH is a 600-bed acute care tertiary referral facility. Since1987, the SGPH has served as the major trauma referral hospital forthe south eastern area of Sydney, Australia, a geographic zone of ap-proximately 1.5 million inhabitants. The SGPH ED is the fourth busiestin NSW, averaging over 59,000 patient visits per year since 2004. Ofthese, an average of nearly 1,000 trauma patients are admitted annu-ally, 250 of which have an Injury Severity Score greater than 15(ISSN15).
Data are collected on all trauma presentations and stored in a pur-pose built data registry, maintained since 1991. This information isobtained from a number of sources, including ambulance runningsheets, patient medical records, clinical ward rounds, and patient in-terviews. Information collected include: demographic details, injuryseverity, injury description and body region, total hospital, and inten-sive care length of stay. Road traffic-related trauma presentations inthe SGPH trauma registry were identified if the mechanism recordedindicated that the incident involved either a motor vehicle crash or amotorcycle, pedal cycle, or pedestrian collision.
2.2. NSW admitted patients data collection
NSW hospitalization data include information on inpatient sepa-rations from NSW public and private hospitals, private day proce-dures, and public psychiatric hospitals. Included are data onepisodes of care in hospital that end with the discharge, transfer, ordeath of the patient, or when the service category for the admittedpatient changes. Information collected includes patient demo-graphics, circumstances of the incident, diagnoses, and clinical proce-dures. The hospitalization data were coded using the InternationalClassification of Disease, 10th Revision, Australian Modified (ICD-10-AM; National Centre for Classification in Health, 2004). The dataalso include hospitalizations of NSW residents that occur in anotherstate or territory in Australia. At the time of writing, these datawere not available for post-July 2007 or all of 2008, but were estimat-ed to comprise 911 road traffic-related hospitalizations, based on theaverage of the past four years.
Road traffic-related hospitalizations of NSW residents in the APDCwere identified using the following criteria: (a) the hospitalizationwas for a patient who was a resident of NSW; (b) the principal diag-nosis was coded as ‘injury, poisoning and certain other consequences
Table 3Distribution of road traffic-related injuries by type and gender for the SGPH traumaregister, the NSW APDC, and TADS, percent, NSW, 2002 to 2008.
Gender SGPH APDC TADS
All crashes1 All crashes1,2 Traffic-only3 Traffic-only3
PedestriansMale 59.8 61.4 61.2 56.5Female 40.2 38.6 38.8 43.5Pedal cyclistsMale 85.7 81.8 82.3 84.8Female 14.3 18.2 17.8 15.2Motor cyclistsMale 90.3 92.0 91.6 88.9Female 9.7 8.1 8.4 11.1Car occupants4
Male 53.0 51.9 51.5 44.5Female 47.0 48.1 48.5 55.5
1 MVCs that occurred either on a public (ie. traffic) or non-public (ie. non-traffic)roadway.
2 Unknown genders of pedal and motor cyclists in the APDC were removed for anal-ysis due to low cell sizes.
3 MVCs that occurred on a public roadway (ie. traffic).4 SGPH data in ‘car occupants’ refer to occupants of all vehicles.
347R. Mitchell et al. / Journal of Safety Research 42 (2011) 345–350
of external causes’ (i.e., ICD-10-AM: range S00-T98); and (c) an exter-nal cause code indicated either car occupants (i.e., ICD-10-AM: rangeV40-V49); a motorcycle (i.e., ICD-10-AM: range V20-V29) or pedal
Table 4Distribution of road traffic-related injuries by type and age group for the SGPH traumaregister, the NSW APDC, and TADS, percent, NSW, 2002 to 2008.
Age group SGPH APDC TADS
All crashes1 All crashes1 Traffic-only2 Traffic-only2
Pedestrians0-14 19.6 18.7 16.6 16.415-19 10.3 9.2 10.0 11.320-24 9.2 9.8 10.7 11.825-34 13.0 14.4 14.6 15.735-44 8.9 11.7 11.8 11.945-54 9.7 9.7 10.0 10.055-64 6.5 8.0 8.0 7.865+ 22.8 18.4 18.3 15.2Pedal cyclists0-14 20.3 43.8 33.7 16.315-19 14.1 12.7 11.6 11.720-24 6.6 6.0 6.6 9.625-34 18.9 12.8 16.0 24.035-44 20.5 10.8 13.8 19.645-54 11.5 7.7 10.2 10.855-64 4.6 4.0 5.1 5.065+ 3.5 2.4 3.1 3.0Motor cyclists0-14 1.4 13.9 6.1 1.115-19 10.3 16.0 12.4 8.720-24 19.1 16.2 17.3 17.725-34 30.0 23.4 26.6 29.635-44 22.0 16.8 19.6 21.245-54 11.4 8.8 11.6 14.655-64 4.7 3.5 4.8 5.765+ 1.1 1.5 1.7 1.4Car occupants3
0-14 7.5 5.7 5.3 5.315-19 12.3 15.0 15.1 14.020-24 14.8 13.9 14.3 15.125-34 18.4 17.6 18.0 19.935-44 14.4 14.2 14.5 15.945-54 12.0 11.7 11.9 12.855-64 8.9 8.2 8.2 8.265+ 11.6 13.8 12.6 8.9
1 MVCs that occurred either on a public (ie. traffic) or non-public (ie. non-traffic)roadway.
2 MVCs that occurred on a public roadway (ie. traffic).3 SGPH data in ‘car occupants’ refer to occupants of all vehicles.
cycle (i.e., ICD-10-AM: range V10-V19) crash, or a pedestrian (i.e.,ICD-10-AM: range V00-V09).
2.3. NSW traffic accident database system
The TADS collection includes information on all motor-vehiclecrashes (MVCs)where a personwas unintentionally fatally or non-fatallyinjured or at least one motor vehicle was towed away and the incidentoccurred on a public road in NSW. Information recorded in TADS in-cludes: information describing the crash and conditions at the incidentsite; information regarding the traffic unit or vehicle and the vehicle con-troller; and information describing any casualties resulting from thecrash. Data were obtained from TADS for records where: (a) the ‘degreeof casualty’ variable indicated that the person had been non-fatally in-jured (i.e., ‘2’); and (b) the type of road-related incident was identifiedusing the appropriate codes from the ‘traffic unit group’ variable to iden-tify pedestrians (i.e., ‘11’), pedal cyclists (i.e., ‘9’), motorcyclists (i.e., ‘8’),and car occupants (i.e., ‘1’).
2.4. Data management and analysis
Hospitalizations from the APDC relating to transfers or statisticaldischarges were excluded from the APDC in order to attempt to partlyeliminate ‘multiple counts.’ These exclusions refer to transfers be-tween hospitals or changes in the service category (e.g., a changefrom acute to rehabilitation) for a patient during one episode ofcare in a single facility.
Analysis was performed using SAS version 9.1 (SAS Institute,2003). An assessment of the homogeneity of the distribution of twoindividual demographics (i.e., gender and age group) between theSGPH trauma data and data from TADS and the APDC and activity atthe time of the incident between the SGPH trauma data and theAPDC for each type of road traffic incident was conducted using achi-square test (Armitage, Berry, & Matthews, 2002). Unknown ageand genders were removed from the analysis. As TADS should onlyrecord crashes that occurred on public roads (i.e., ‘traffic-only’crashes), road traffic crashes in the APDC that were identified as‘non-traffic’ were also identified in each Table but not reported inthe analyses.
3. Results
All three data collections recorded information on the age andgender of the injured individual, but only the APDC recorded informa-tion on an individual's indigenous status. The trauma registryrecorded 80% of the information recommended by the WHO on cir-cumstances of the incident, while TADS and the APDC recorded 60%and 50% of this information, respectively. The trauma registry wasalso the only data collection to record information on both the natureand the severity of the injury (Table 1).
During 2002 to 2008 there were 5,535 road traffic-related presen-tations in the SGPH trauma registry, 87,068 hospital admissions in theAPDC for all road traffic-related injuries of NSW residents and 60,000hospitalizations where the road traffic incident occurred on a publicroad, and 167,048 entries of road traffic-related injuries recorded inthe TADS collection. Car occupants made up the largest proportionof road traffic-related injuries in all data collections. There wereover twice the proportion of motorcyclist crash injuries reported inthe APDC compared to both the trauma registry and TADS. However,there was only a slightly higher proportion of motorcyclist injuries inthe trauma registry compared to TADS. There were a higher propor-tion of incidents involving pedestrians in the trauma registry com-pared to both the APDC and TADS collections and there was a lowerproportion of pedal cyclist injuries reported in the TADS collectioncompared to the trauma registry or all hospitalizations (Table 2).
Table 5Distribution of road traffic-related injuries by type and activity for the SGPH traumaregister, the NSW APDC, and TADS, percent, NSW, 2002 to 2008.
Activity at time of incident SGPH APDC
All crashes All crashes
PedestriansSports activity 0.4 1.5Leisure activity 39.0 0.7Working for income 3.6 4.0Other types of work 47.0 1.2Vital activities (eg. resting, sleeping, eating) 0.1 0.8Other specified activities 7.4 14.6Unspecified activities - 76.9Not known 2.5 0.3Pedal cyclistsSports activity 9.7 27.2Leisure activity 62.6 5.0Working for income 1.5 1.6Other types of work 22.2 0.2Vital activities (eg. resting, sleeping, eating) - 0.2Other specified activities 0.7 14.9Unspecified activities - 50.7Not known 3.3 0.3Motor cyclistsSports activity 1.5 18.8Leisure activity 42.3 2.9Working for income 5.6 3.6Other types of work 45.5 0.3Vital activities (eg. resting, sleeping, eating) 0.2 0.1Other specified activities 0.5 12.0Unspecified activities - 61.4Not known 4.4 0.9Car occupants2
Sports activity 0.1 0.3Leisure activity 27.6 0.3Working for income 5.8 3.4Other types of work 57.5 0.3Vital activities (eg. resting, sleeping, eating) 0.1 0.5Other specified activities 5.3 13.9Unspecified activities - 80.6Not known 3.7 0.9
1MVCs that occurred either on a public (ie. traffic) or non-public (ie. non-traffic)roadway.2SGPH data in ‘car occupants’ refer to occupants of all vehicles.
348 R. Mitchell et al. / Journal of Safety Research 42 (2011) 345–350
3.1. Gender
As shown in Table 3, there were significant differences in the gen-der distribution evident for pedal cyclists between the trauma regis-try and the APDC (χ²=4.5, df=1, p=0.03) and for car occupantsbetween the trauma registry and TADS (χ²=102.1, df=1,pb .0001). There was a greater proportion of female pedal cyclists inthe APDC than in the trauma registry and the proportion of femalecar occupants was higher in TADS compared to the trauma registry.There were no significant differences in the distribution of genderfor pedestrians between the trauma registry and the APDC (p=0.4)or TADS (p=0.07), for pedal cyclists between the trauma registryand TADS (p=0.6), for motorcyclists between the trauma registryand the APDC (p=0.1) and TADS (p=0.3), or for car occupants be-tween the trauma registry and the APDC (p=0.2).
3.2. Age group
The age group distribution showed significant differences be-tween the trauma registry and both the APDC and TADS for pedes-trians (χ²=15.8, df=7, p=0.03 and χ²=46.0, df=7, pb .0001,respectively), pedal cyclists (χ²=119.9, df=7, pb .0001 andχ²=15.9, df=7, p=0.03, respectively), car occupants (χ²=52.5,df=7, pb .0001 and χ²=83.3, df=7, pb .0001, respectively), andmotorcyclists between the trauma registry and the APDC(χ²=121.0, df=7, pb .0001). There were no significant differences
in the distribution of age groups between the trauma registry andTADS for motorcyclists (p=0.2; Table 4).
3.3. Activity
The distribution of activities between the trauma registry and theAPDC showed significant differences for pedestrians (χ²=6,417.8,df=7, pb .0001), pedal cyclists (χ²=5,812.0, df=7, pb .0001), mo-torcyclists (χ²=12,082.8, df=7, pb .0001), and car occupants(χ²=32,167.3, df=7, pb .0001) (Table 5).
4. Discussion
Road traffic-related injuries require accurate information from in-jury surveillance activities in order to effectively target road safetyprevention efforts. Injury information from a trauma registry was ex-amined to assess its potential, in terms of available information andrepresentativeness of the distribution of key characteristics used forinjury surveillance to determine the capacity of trauma data to actas a type of sentinel indicator to assist with road traffic-related injurysurveillance activities for different road user groups.
The SGPH trauma registry collected information on the majority ofvariables recommended by theWHO for injury surveillance, while thehospital admission collection did not record information related topreventive factors or to the precursors of the crash, such as alcohol,drugs, and other causal factors. The data collection maintained bythe road safety authority did not collect information on injury out-comes or some aspects of the circumstances of the incident, such asthe type of activity conducted or intent. As these data collections allprimarily collect information for other purposes, such as administra-tive needs, they were not designed specifically for injury surveillancecapabilities. However, of the three data collections, the SGPH traumaregistry was the better equipped, in terms of information collected,for injury surveillance purposes. Yet, while able to be collected inthe trauma registry, for some variables, such as alcohol and drugsand the identification of preventive factors, the information waslargely not recorded.
It would have been ideal to examine the representativeness of thedistribution of all the key injury surveillance variables in Table 1 forroad traffic-related injuries. However, this was not possible due tothe use of different classification systems used to classify the re-sponses for many variables, including place of occurrence, intent,and nature of injury, between the data collections examined and bymissing data for some variables, such as use of alcohol and drugsand the identification of preventive factors in the trauma registry.Only the distribution of the two individual demographics (i.e., ageand gender) were able to be assessed for all three data collectionsand the activity at the time of the incident for the trauma registryand the APDC.
Not surprisingly there were differences in the number of road traf-fic-related injuries identified in each of the three data collections aseach collects data from differently defined populations. For example,road traffic injuries recorded in the trauma registry only includecrashes that occurred in the SGPH catchment area or those individ-uals who were transferred to SGPH due to the severity of their inju-ries. In addition, pre-hospital trauma triage protocol criteria requiremore severely injured individuals to be conveyed to the nearestmajor trauma facility, with the injury mechanism criteria during thetime period of this study specifically stating that motor-vehiclecrashes N60kph, major deformation of vehicle/airbag deployment,fatal injury in same vehicle, patient ejected from vehicle, cyclist or pe-destrian hit by vehicleN30 kph, falls from N3 metres, and injuries tomultiple body regions should be taken immediately to a major trau-ma facility (Ambulance Service of New South Wales, 2008).
Unlike the trauma registry, both TADS and the APDC are popula-tion-based collections but the APDC only includes injuries serious
349R. Mitchell et al. / Journal of Safety Research 42 (2011) 345–350
enough to require hospitalization, whereas the TADS collection caninclude injury-related crashes reported to police, regardless of sever-ity. The numbers of cases reflect these differences. Nevertheless, itmight be expected that similar proportions of different types of roadusers would be represented in each of the data collections, especiallythe trauma (SGPH) and hospitalization (APDC) collections. Interest-ingly, all data collections had roughly similar proportions of pedes-trians, and car occupants were always the largest percentage ofcases, but the data collections differed considerably on the percent-ages of pedal cyclists and motorcyclists.
The discrepancies in the number of road traffic crashes recordedby road user group between the data collections can be explainedby a number of differences in inclusion criteria. The TADS collectiononly describes crashes that occurred on public roads (i.e., trafficareas), whereas the trauma registry and the APDC both contained in-formation on crashes that occurred in either traffic or non-trafficareas, such as bush or farmland. For example, the APDC recordedaround twice the number of injuries of pedal cyclists and one and ahalf times the number of injuries to motorcyclists compared toTADS. The lower number of both pedal cyclist and motorcyclist inju-ries in TADS is likely to be partially due to these crashes occurringin non-traffic areas and thus not being captured in TADS. This wasalso common in other comparisons of police and hospital data collec-tions (Amoros, Martin, & Laumon, 2006; Langley, Dow, Stephenson, &Kypri, 2003). However, the number of pedal cyclist crashes reportedto police and included in TADS are similar to the traffic-only pedal cy-clist hospital admissions in the APDC, which suggests that most of thepedal cyclists that are involved in collisions on a public roadway areadmitted to hospital and are reported to the police. Some of theremaining discrepancy between the number of road traffic incidentrecords are likely to be due to differences in vehicle definitions,with TADS including vans and utilities as cars whereas the APDC in-clude occupants of these vehicles in separate ICD-10-AM codes (i.e.,V50-V59 pick-up truck or van), and in the severity of the injury expe-rienced. Only road traffic-related injuries severe enough to requirehospital presentation and/or admission are recorded in the SGPHtrauma registry and the APDC, while, in theory, all unintentionalroad traffic-related injuries that occurred on a public road that arereported to police are recorded in TADS.
For the most part, the distribution of gender by road user groupwas similar in each collection, except for pedal cyclists between thetrauma data and the APDC and for car occupants in the trauma dataand TADS. In both instances, a higher proportion of males were ob-served in the trauma registry. It is difficult to explain exactly whythis gender difference for these two road injury mechanisms occurs,but could be due to female drivers being more likely to report vehiclecrashes to police than males and thus have their crash informationrecorded in TADS, and vehicle crashes that involve males might in-volve more severe injuries and thus males would be more likely tobe transported directly to a major trauma facility compared to fe-males. Males are known to have a high hospitalization rate fortransport-related injuries (Bradley & Harrison, 2008) and it mightbe expected that this rate would be stable across different road injurymechanisms.
The age group distributions in the trauma registry for all road usergroups were not representative of the age group distribution in theAPDC and TADS, except for motorcycle crashes between the traumaregistry and TADS. The significant differences between the data col-lections suggest a systematic bias between the collections for age,with a higher proportion of both older (for pedestrians and pedal cy-clists) and younger (for pedestrians and car occupants) age groupsrepresented in the trauma data. Possible reasons for the higher pro-portion of older individuals represented in the trauma registrycould be because pedestrian collisions involving individuals over65 years are more likely to involve serious injuries and these individ-uals are more likely to be conveyed to a major trauma center (Oxley,
Corben, Fildes, & O'Hare, 2004). In addition, nearly 30% of major trau-ma admissions to SGPH are of individuals aged over 55 years, withthe SGPH having one of the highest severe pedestrian injury rates inNSW, peaking for individuals in the 65–74 age group. Younger agegroups may be more likely to be represented in the trauma data, asthese age groups may be more likely to be severely injured (NSW In-stitute of Trauma and Injury Management, 2008). There may also bedifferences in admission practices for different age groups, withyounger (b15 years) and older individuals (65+ years) more likelyto be taken to a major trauma hospital following an traumatic eventas per pre-hospital trauma protocol guidelines (Ambulance Serviceof New South Wales, 2008).
The activity conducted at the time of incident as is was not repre-sentative for any road user group for the SGPH trauma data and theAPDC. Differences in coding strategies appear to largely account forthe non-representativeness of activity data between these two collec-tions. It appears that ‘leisure activities’ and ‘other types of work’ weremore likely to be used as ‘dump codes’ in the trauma registry, whilein the APDC, ‘other and unspecified’ activities were more likely to beused to indicate unspecified activities and/or ‘dump codes’ (AustralianInstitute of Health and Welfare, 2009).
As well as the differences in case enumeration described abovethat account for some of the differences between the data collectionsand the use of different classification systems that precluded the ex-amination of representativeness of the distribution for some vari-ables, a further limitation of this examination of the potential oftrauma data for road traffic injury surveillance was that in the traumadata from SGPH car occupant data included occupants of all motor ve-hicles. As drivers of other vehicles, such as trucks, could not be ex-cluded from the trauma data, this resulted in more vehicleoccupants being observed to be working (5.8%) compared to hospitaladmissions data (3.4%) that only included car occupants. It is also pos-sible that some discrepancy exists between the APDC and TADS as theAPDC includes injuries involving all NSW residents, regardless of inwhat Australian state or territory the incident occurred, whereas in-formation in TADS describes road crashes that occurred in NSW thatwere reported to NSW police. Therefore, it is possible that crashesrecorded in TADS involve injuries of non-NSW residents that oc-curred on public roads in NSW.
Data from trauma registries is able to provide additional informationregarding road traffic crashes that is not recorded in other data collec-tions, such as detailed information regardingmedical treatment and in-jury severity. Within NSW, the move to a common state wide traumacollection system “the Collector,” that requires entry of ICD-10 codes,may see less variation in the data collected across trauma centers. Inthe future, by linking trauma data using available identifying variables,such as name, address, date of birth, date of injury/ admission, to infor-mation obtained from population-based collections, a proportion of se-vere road traffic-related injuries in NSWwill have detailed informationavailable across the injury continuum (i.e., from the circumstances ofthe crash to injuries incurred through to treatment received, and insome instances, through to long-term trauma outcomes from follow-up post-discharge surveys). This type of data linkage, using multipleroad safety data collections, has been performed in the United Stateswith CODES to bring together comprehensive data regarding events oc-curring before the crash, during the crash, and after the crash to assist inidentifying the sequence of events leading up to, and subsequent to acrash, including care provided in the medical system.
5. Conclusion
While the trauma registry recorded themajority of variables recom-mend for injury surveillance purposes, data from the trauma registrywas largely not representative of the distribution of population-basedcollections in NSW in terms of age group or activity performed, butwas similar in terms of gender. For this reason, trauma registries cannot
350 R. Mitchell et al. / Journal of Safety Research 42 (2011) 345–350
really be used alone as a sentinel indicator for road injury surveillance.However, information from trauma registries could be used in conjunc-tion with information from population-based data collections to en-hance road injury surveillance. Data linkage studies are likely toprovide additional information on individuals captured in trauma regis-tries and the circumstances of their injury and injury outcome.
Acknowledgements
R Mitchell was supported by an ARC-linkage post-doctoral fellow-ship (LP0990057). A Williamson is supported by an NHMRC SeniorResearch Fellowship.
References
Ambulance Service of New South Wales (2008). Pre-hospital Management of MajorTrauma. Protocol T1. Trauma Triage Tool. Major Trauma Criteria (MIST). Rozelle:Rozelle NSW: Author.
Amoros, E., Martin, J., & Laumon, B. (2006). Under-reporting of road crash casualties inFrance. Accident Analysis and Prevention, 38, 627–635.
Armitage, P., Berry, G., & Matthews, J. (2002). Statistical Methods in Medical Research(Fourth Edition ed.). Cornwell: Blackwell Science.
Australian Institute of Health andWelfare (2009). Australian hospital statistics 2007–08.Canberra: Author.
Boufous, S., & Williamson, A. (2006). Work-related traffic crashes: a record linkagestudy. Accident Analysis & Prevention, 38(1), 14–21.
Bradley, C., & Harrison, J. (2008). Hospital separations due to injury and poisoning, Aus-tralia 2004–05 (No. INJCAT 117). Adelaide: AIHW.
Cercarelli, L., Rosman, D., & Ryan, G. (1996). Comparison of accident and emergencywith police road injury data. The journal of trauma injury, infection, and criticalcare, 40(5), 805–809.
Department of Infrastructure Transport Regional Development and Local Government(2009). Road Deaths Australia 2008 Statistical Summary. Canberra: Commonwealthof Australia.
Forst, L. S., Hryhorczuk, D., & Jaros, M. (1999). A state trauma registry as a tool for oc-cupational injury surveillance. Journal of Occupational & Environmental Medicine,41(6), 514–520.
Henley, G., & Harrison, J. (2009). Serious injury due to land transport accidents, Australia2006–07. Canberra: Australian Institute of Health and Welfare.
Johnson, M. S., & Moore, M. (1997). Alaska's statewide trauma registry: a useful sur-veillance tool for injury prevention planning and evaluation. Journal of PublicHealth Management & Practice, 3(6), 1–7.
Kobusingye, O. C., & Lett, R. R. (2000). Hospital-based trauma registries in Uganda. Jour-nal of Trauma-Injury Infection & Critical Care, 48(3), 498–502.
Langley, J., Dow, N., Stephenson, S., & Kypri, K. (2003). Missing cyclists. Injury Preven-tion, 9(4), 376–379.
Layde, P.M., Stueland, D. T., & Nordstrom, D. L. (1996). Representativeness of trauma centerregistries for farm injury surveillance. Accident Analysis & Prevention, 28(5), 581–586.
Lloyd, L. E., & Graitcer, P. L. (1989). The potential for using a trauma registry for injurysurveillance and prevention. American Journal of Preventive Medicine, 5(1), 34–37.
Lopez, D., Rosman, D., Jelinek, G., Wilkes, G., & Sprivulis, P. (2000). Complementing po-lice road-crash records with trauma registry data - an inital evaluation. AccidentAnalysis & Prevention, 32, 771–777.
National Centre for Classification in Health (2004). ICD-10-AM (Fourth ed.). Sydney:Commonwealth Department of Health.
National Highway Traffic Safety Authority (1996). The Crash Outcome Data EvaluationSystem (CODES): Technical Report. Washington DC: National Highway Traffic SafetyAuthority.
NSW Institute of Trauma and Injury Management (2008). The NSW Trauma RegistryProfile of Serious to Critical Injuries 2006. Sydney: Author.
Nwomeh, B., Lowell, W., Kable, R., Haley, K., & Ameh, E. (2006). History and develop-ment of trauma registry: lessons from developed to developing countries. WorldJournal of Emergency Surgery, 1(32).
Oxley, J., Corben, B., Fildes, B., & O'Hare, M. (2004). Older vulnerable road users - mea-sures to reduce crash and injury risk (No. 218). Melbourne: Monash University Acci-dent Research Centre.
Peden, M., McGee, K., & Sharma, G. (2002). The Injury Chart Book: a Graphical Overviewof the Global Burden of Injuries. Geneva: World Health Organization.
Pollock, D. A., & McClain, P. W. (1989). Trauma registries. Current status and futureprospects. JAMA, 262(16), 2280–2283.
Rahman, F., Andersson, R., & Svanstrom, L. (2000). Potential of using existing injury in-formation for injury surveillance at the local level in developing countries: experi-ences from Bangladesh. Public Health, 114(2), 133–136.
Roads and Traffic Authority NSW (2009). Road Traffic Crashes in New South Wales. Sta-tistcal statement for the year ended 31 December 2008. Sydney: Roads and TrafficAuthority, NSW.
Rosman, D. (2001). The Western Australian road injury database (1987–1996): tenyears of linked police, hospital and death records of road crashes and injuries. Ac-cident Analysis & Prevention, 33, 81–88.
Rosman, D., & Knuiman, M. (1994). A comparison of hospital and police road injurydata. Accident Analysis & Prevention, 26(2), 215–222.
SAS Institute (2003). SAS: statistical software, version 9.1. Cary, North Carolina: Author.Watson, W. L., & Ozanne-Smith, J. (2000). Injury surveillance in Victoria, Australia: de-
veloping comprehensive injury incidence estimates. Accident Analysis & Prevention,32(2), 277–286.
World Health Organization [WHO] (2001). Injury Surveillance Guidelines. Geneva:Author.
World Health Organization [WHO] (2004). The World Health Report 2004. Geneva:Author.
ReMse
becca Mitchell is a research fellow at the School of Aviation and the NSW Injury Riskanagement Research Centre at the University of New South Wales. Her primary re-arch interests include injury surveillance methods, data quality, evaluative tool de-
velopment, and epidemiological and evaluation studies. Previously, she worked as aSenior Policy Analyst in the Injury Prevention and Policy Branch of the NSW Depart-ment of Health and as a Senior Officer in the Epidemiology Unit at the National Occu-pational Health and Safety Commission.
AnnWilliamson is the Professor of Aviation Safety at the School of Aviation at the Uni-versity of New South Wales. She has a significant track record of research in safety es-pecially in the areas of workplace and road safety. Previously, she established andworked at the NSW Injury Risk Management Research Centre and also at the NationalInstitute of Occupational Health and Safety where she was Principal Research Scientistand Head of the Human Factors and Ergonomics Unit.
Kate Curtis is a Clinical Associate Professor at the Sydney Nursing School at the Univer-sity of Sydney. She is an internationally respected trauma nurse and researcher with aspecial interest in trauma models of care, trauma nursing and financial aspects of trau-ma systems. She also works as a Trauma Clinical Nursing Consultant at St George Hos-pital. Previously, she worked in aged care, general medical/surgical, mental health andemergency nursing.
