RESEARCHA population-based study of thrombolysis for acute stroke in South Australia

James M Leyden, Woon K Chong, Tim Kleinig, Andrew Lee, John B Field and Jim JannesThe Medical Journal of Australia ISSN:0025-729X 7 February 2011 194 3 111-115The Medical Journal of Australia 2011www.mja.com.auResearch

Implementation of Thrombolysis in StrokeInternational Stroke Thrombolysis Register(SITS-ISTR) was established as a voluntaryregistry in over 700 centres around the worldto address this concern.3 The SITS-ISTR hadrecruited over 30 000 patients by February2010, including 704 Australian patients, and

admiurbapeopMainadmiguiddistaM

ABSTRACT

Objective: To report the rate of thrombolysis for treating acute stroke in South Australia from October 2007 to September 2009. We hypothesised that the rate of thrombolytic therapy would be related to distance from an acute stroke unit.Design, setting and patients: An observational, population-based, retrospective review of case notes and imaging, using multiple case-ascertainment methods. Patients

nistered a thrombolytic agent by any method for suspected ischaemic stroke in n, rural, public and private hospitals in SA (covering a population of 1.5 million le) were included. outcome measures: Absolute and relative contraindications for thrombolysis nistration in each case, according to the 2007 National Stroke Foundation elines; incidence of haemorrhage; and population thrombolysis rates according to nce from an acute stroke unit.

Results: A total of 158 cases of thrombolytic therapy for suspected acute ischaemic stroke were identified in 157 patients. Fifteen patients (10%) had symptomatic intracranial haemorrhage, of whom eight (5%) died within 3 months. Seven patients had symptomatic extracranial haemorrhage. Five patients (3%) received thrombolysis despite absolute contraindications. Patients living closer to stroke units were more likely to receive thrombolysis.Conclusions: Rates of symptomatic haemorrhage after thrombolysis were similar to those in voluntary registries. A large proportion of South Australians are currently missing out on acute stroke therapy as a result of poor access to acute stroke units in both urban and rural settings. It is estimated that fewer than 2% of ischaemic stroke

MJA 2011; 194: 111115

patients are administered thrombolysis in SA.hrmeadT ombolysis is a highly effective treat-nt for acute ischaemic stroke, whenministered within 4.5 hours accord-

ing to strict protocols.1,2 There have beenconcerns, however, that the results of clinicaltrials for stroke thrombolysis would be diffi-cult to replicate in routine practice. The Safe

demonstrated clinical outcomes superior tothose seen in clinical trials, including lowercomplication rates.4,5

The SITS registry may not reflect outcomesin the community. Centres registering caseson the SITS registry may have better out-comes. Although centres participating in theEuropean Safe Implementation of Thrombo-lysis in Stroke Monitoring Study (SITS-MOST) were externally audited, there is noguarantee that centres subsequently partici-pating in the SITS-ISTR report all cases.6

Stroke thrombolysis rates differ greatlybetween institutions and, to date, studies ofthrombolysis for acute stroke have not beenpopulation based. The proportion of patientswith acute stroke who receive thrombolysishas not been accurately estimated.7

We adapted previously developed princi-ples for population-based stroke research8

to identify all cases of acute ischaemic strokefor which thrombolysis was administeredfrom October 2007 to September 2009 inthe entire state of South Australia, covering apopulation of 1.5 million people. We reportthe rates and types of intentional and acci-dental protocol violations in thrombolysisadministration for stroke, and quantify thegeographic disparity in thrombolytic ther-apy for stroke in urban and rural SA. Wealso examined differences between SITS-registered and unregistered cases.

METHODS

A 2-year study period from 1 October 2007to 30 September 2009 was chosen to coin-cide with the opening of an acute stroke unit(a total of three acute stroke units wereoperating during the study period). Multiple

methods of case ascertainment were appliedacross urban, rural, public and private hos-pital systems to identify all cases of acutestroke for which thrombolysis was adminis-tered. SA is geographically isolated fromother states, so no patients with acute strokein SA are transferred interstate. Ethicsapproval was granted by the ethics commit-tees of the Queen Elizabeth Hospital, RoyalAdelaide Hospital and Flinders MedicalCentre.

Case ascertainment

In the urban public health system, caseascertainment was performed under twobroad headings: searching electronic datarepositories and collating data from otherprospective clinical registries. SAs uniquecomprehensive public hospital electronicdata repository was queried. All dischargesummaries over the study period for anypatient with a diagnosis or past history ofstroke were queried for the case-insensitive

word stems thrombol, tpa, alteplaseand tenecteplase. As all patients whoreceived thrombolysis for stroke should ide-ally have subsequent brain imaging, the textreports of all brain scans in public hospitalswere also queried for the above word stems.All public hospital encounters, regardless ofdiagnosis, were queried for the above wordstems in the procedure field. Pharmacyrecords in each public hospital were exam-ined for any vials of thrombolytic drugsdispensed against a patients name. Twopublic hospitals were keeping prospectiveregistries of stroke thrombolysis from Octo-ber 2007, and a prospective population-based study of stroke was being performedin a subset of the study population (140 000people) for the final 3 months of the studyperiod. Data from these were collated.

There were no private hospital centreswith protocols for thrombolytic therapyoperating in SA. However, two private hos-pitals in Adelaide had emergency depart-ments during the study period and both hadJA Volume 194 Number 3 7 February 2011 111

RESEARCHa policy to direct acute stroke patients toother institutions. To identify any cases ofpatients who had received thrombolysiswhile in a private hospital emergencydepartment or while admitted as a privatehospital inpatient for another reason, wecontacted every neurologist covering privatehospital inpatients and hospital coordina-tors at each private hospital.

Each major regional centre was contactedto determine whether any patients hadreceived thrombolysis for stroke. In addi-tion, rural general practitioners, who mayrun isolated remote hospitals, were con-tacted during education sessions providedby the GP networks.

Case reviewCase notes were examined by one of us(J M L) for time of symptom onset, time ofarrival at hospital, whether the patient hadbeen transferred from another hospital andtime of thrombolysis administration.

Symptomatic intracranial haemorrhagewas defined as any recorded haemorrhagewith altered neurological findings (theNational Institute of Neurological Disordersand Stroke [NINDS] definition9). TheNINDS definition was the most suitable fora retrospective audit of case notes and islikely to overestimate symptomatic haemor-rhage compared with other definitions.4 Allimaging was reviewed by one of us (W K C)

and haemorrhage was classified using theEuropean Cooperative Acute Stroke Study(ECASS) grading system.10 Finally, the deathregister was examined for ascertained cases.

The accepted protocol for the administra-tion of thrombolysis was that described inthe 2007 National Stroke Foundation guide-lines.11 Following the publication of resultsfrom the ECASS-III trial in September 2008,the allowable time window for administra-tion was extended from 3 hours to 4.5hours.1 Evidence of exclusion criteria in casenotes was recorded. If the notes mentionedthat the treating medical team was aware ofa particular protocol violation but decidedto administer the medication anyway, theviolation was regarded as intentional. Thereported final diagnosis for the administra-tion of thrombolysis was also recorded.

Population demographicsStatistics regarding the population of SAwere obtained from the Australian Bureau ofStatistics 2006 census. Numbers of admis-sions of patients with stroke in urban, rural,public and private hospitals were obtainedfrom Medicare casemix funding databetween June 2006 and June 2008. Thedistance between postcodes for each patientand the nearest stroke unit was calculatedfrom longitude and latitude coordinates,obtained from Google Maps, using thehaversine formula.

Statistical analysisProportions were analysed using the Fisherexact test for counts with two-tailed P val-ues. Differences in arrival-to-treatment timeand symptom-to-treatment time betweenthose patients registered with SITS andthose not registered were compared usingthe Wilcoxon rank sum test. Computationwas done in the R statistical language.

RESULTS

Over the 2-year study period, 158 instancesof thrombolytic therapy for suspected acuteischaemic stroke in SA were identified in157 patients; 84 of these patients were maleand 73 were female. The median age ofpatients was 76 years (interquartile range,7179 years), and 43 patients (27%) wereaged over 80 years. One patient was anIndigenous Australian.

Eighteen of the 157 patients who receivedthrombolysis (11%) initially presented tohospitals that do not have stroke units five received thrombolysis after transfer to ahospital with a stroke unit, and 13 (8%)received thrombolysis in hospitals withoutstroke units. More than half of the admis-sions for stroke in SA recorded by Medicarefrom June 2006 to June 2008 were tohospitals without acute stroke units.

Case ascertainmentThe 158 cases of stroke for which thrombo-lysis was administered were ascertained assummarised in Box 1. A total of 8914admissions to urban public hospitals in SAthat related to any patient coded with anyform of stroke or transient ischaemic attackwere identified, for which 8201 dischargesummaries were available (92%). From thedischarge summaries it was determined thatstroke was the principle reason for, oroccurred during, 4106 admissions, and 916of these admissions were rehabilitationadmissions or interhospital transfers. Of theremaining 3190 admissions, 2485 hospitaladmissions represented acute presentationswith ischaemic stroke. Ischaemic strokeoccurred while the patient was in hospital in76 of the admissions. Thrombolytic therapyfor stroke was identified in 145 hospitaladmissions in 144 patients from dischargesummaries (one patient received thrombo-lysis twice during the study period).

Of 72 082 brain scans (computed tomo-graphy and magnetic resonance imaging)performed in Adelaide public hospitals dur-ing the study period, 136 instances ofthrombolysis in 135 patients were identi-

1 Ascertainment of cases of acute stroke in South Australia for which thrombolysis was administered, October 2007 to September 2009*

*Only three cases were ascertained by one method alone.

Electronic data repository: 155 cases

Urban public health system: 155 cases

Discharge summaries8914 admissions; 8201 discharge summaries available; 2485 acute

ischaemic strokes identified145 cases

Imaging reports72 082 brain scans (text reports

queried for thrombolysis keywords)136 cases

Safe Implementation of Thrombolysis in Stroke

registry72 cases

Pharmacy recordsDispensing records for

thrombolytic drugs28 cases

Hospital encounters131 064 encounters

(procedure list queried for thrombolysis keywords)

50 cases

Population-based stroke study

8 cases

Emergency departments4 cases

Private neurologists4 cases

Prospective registries: 72 cases

Private hospital system: 4 cases

Thrombolysis for acute ischaemic stroke

confirmed from case notes

158 cases in 157 patients

Rural health system: 4 cases

Rural hospitals3 cases

Rural general practitioners2 cases112 MJA Volume 194 Number 3 7 February 2011

RESEARCHfied. Of 131 064 public hospital encounters,50 cases were identified where thrombolysisfor stroke had been reported as a procedureduring the study period. From pharmacydispensing records for thrombolytic drugs,28 cases were identified (thrombolytic drugswere frequently dispensed without patientidentifiers).

Seventy-two patients had been recruitedinto the SITS registry, all of whom werealso ascertained by other means. Withinthe two hospitals participating in the SITSregistry, 15 eligible cases were identified inpatients who had not been enrolled intothe SITS registry. The prospective popula-tion-based stroke study detected eightpatients, all of whom were also ascertainedby other means.

Private hospital thrombolysis. Four patientsreceived thrombolysis in the private sector;each was identified by both the treatingneurologist and the hospital coordinator.Medicare recorded 782 private hospitaladmissions for stroke from June 2006 toJune 2008.

Patients residing outside Adelaide. Elevenpatients residing outside the urban region ofAdelaide received thrombolysis over thestudy period, of whom two received throm-bolysis while visiting Adelaide and five weretaken directly to urban hospitals. The popu-lation outside Adelaide comprises 26% ofthe total South Australian population(394 000 of 1.5 million) but comprised only7% of all stroke thrombolysis cases

(P < 0.001). Over the 2-year period fromJune 2006 to June 2008, Medicare recorded997 stroke admissions in hospitals outsideAdelaide, out of a total of 3521 for SA(28%), suggesting that stroke incidence is atleast as high in these areas as it is in urbanAdelaide.

Case reviewSymptom-to-treatment times. Symptomonset time could not be determined for 10cases. For the remaining patients, themedian time from symptom onset to hospitalarrival was 55 minutes. In 21 cases, throm-bolysis was administered after a stroke thatoccurred in hospital. In 11 cases, the hospi-tal arrival time or time of thrombolytic ther-apy could not be identified from the notes.

For the remaining patients, the median timefrom arrival to thrombolysis was 85 minutes,which was significantly greater (P < 0.001)than the 60 minutes recommended by theAmerican Heart Association.12

Protocol violation and misdiagnoses. In fivecases, thrombolytic therapy was adminis-tered when absolute contraindicationsexisted (Box 2). In one of these cases thepatient had a fatal intracerebral haemor-rhage after an asymptomatic subacute inf-arction was missed in the initial radiologicalreview, and in three the treating physicianwas aware of the contraindication. In 53cases, thrombolytic therapy was adminis-tered when at least one relative contraindi-cation existed (Box 2). In all these cases thepatient was aged over 80 years.

In four cases, the diagnosis was notstroke. One of these diagnoses was Bellspalsy with psychogenic features, one washemiplegic migraine, and two cases weredeemed psychogenic presentations.

Haemorrhage after thrombolysis. All haem-orrhages occurred within 36 hours ofthrombolysis. Fifteen patients (10%) hadsymptomatic intracranial haemorrhage. Thisdid not differ significantly from the 6.4%rate recorded in the NINDS trial (P = 0.30).9

Haemorrhage rates apparent on imaging,defined using the ECASS classification, areshown in Box 3.

Case fatality. Eight patients with sympto-matic haemorrhage died within 3 months(5%). In five of these patients (3%), theintracranial haemorrhage was remote to thesite of the stroke. Seven additional patientshad symptomatic extracranial haemorrhage,of whom one had a myocardial infarct anddied as a consequence. Three months afterthrombolysis, 34 patients (22%) had diedfrom any cause.

2 Protocol violations according to the 2007 National Stroke Foundation guidelines11

CT = computed tomography. * Fatal intracerebral haemorrhage. Both patients had bleeding at surgical site, and both survived. Seven patients had intracerebral haemorrhages, of whom five died.

No. of patients

Intended by

treating doctors

Symptomatic haemorrhage

Absolute contraindications

Stroke symptoms on waking (time of onset unknown) 1 Yes No

International normalised ratio of 1.9 before administration 1 No No

Reported seizure of unaffected side 1 Yes No

Retrospective radiological review showed subacute ischaemic stroke on contralateral side

1 No Yes*

No imaging performed before administration in rural setting 1 Yes No

Relative contraindications

National Institutes of Health Stroke Scale score > 22 1 Yes No

Large established cerebral infarction on preliminary CT scan 1 No No

Recent major surgery 2 Yes Yes

Patient aged over 80 years 53 Yes 10

3 Occurance of haemorrhage within 36 hours of thrombolysis based on imaging, according to the European Cooperative Acute Stroke Study grading system10

Classification No.

Parenchymal haematoma type 2 (PH2): a dense haematoma in >30% of the infarcted area with substantial space-occupying effect, or any haemorrhagic lesion outside the infarcted area)

21

Parenchymal haematoma type 1 (PH1): a haematoma in 30% of infarcted area with some slight space-occupying effect

9

Haemorrhagic infarction type 2 (HI2): confluent petechiae within the infarcted area but without space-occupying effect

7

Haemorrhagic infarction type 1 (HI1): small petechiae along the margins of the infarct 2

No haemorrhage visible 116

Imaging not performed or imaging not available 3MJA Volume 194 Number 3 7 February 2011 113

RESEARCHDistance from acute stroke unit. For eachpostcode, the number of patients whoreceived thrombolysis per 100 000 popula-tion was plotted against the distance fromthe nearest acute stroke unit (Box 4), and astrong inverse relationship was observedbetween these variables. There was no rela-tionship between average or median ageand distance from the nearest acute strokeunit.

Difference between SITS-registered andunregistered cases. Seventy-two of the 158cases were entered in the SITS registry (Box5). There was no significant difference indemographics or bleeding complicationsbetween those entered and not entered intothe SITS registry. Cases not entered hadlonger arrival-to-treatment times (P < 0.01)and longer symptom-to-treatment times,including cases in which the stroke occurredin hospital (P < 0.001). This analysisexcluded patients treated outside urbanAdelaide.

Proportion of stroke admissions thatinvolved thrombolysis. During the studyperiod, 4% (155/4106) of acute hospitaladmissions for stroke in urban public hospi-tals involved administration of thromboly-sis. Eight patients received thrombolysisoutside the urban public hospital system.Medicare estimates from 2006 to 2008 sug-gest that over 50% of acute stroke admis-sions in SA occur outside the urban publichospital system, hence the proportion ofacute strokes in SA for which thrombolysisis administered is less than 2%. However,for patients residing within 5 km of an acutestroke unit, the thrombolysis rate for strokewas just over 10% (65 patients in 647strokes).

DISCUSSIONThis study was possible due to unique clini-cal data collection methods and cooperativehospital relationships. Case ascertainmentwas comprehensive and validated againstcoexisting registers and a population strokestudy. Although the numbers are small,there was no difference in haemorrhage ratesbetween the 72 cases registered in the SITStrial and the 86 which were not. Patientsregistered in the SITS trial received treat-ment about 30 minutes sooner, excludingrural cases. With larger numbers, such adifference in time to treatment could resultin different outcomes, suggesting that out-comes for stroke thrombolysis in the SITSregistry may be better than those in thebroader community.13

The rate of symptomatic intracranialhaemorrhage was not significantly differentfrom the rate reported for the NINDS study(10% v 6.4%). A third of patients receivedthrombolysis despite contraindicationsdefined by the National Stroke Foundationguidelines.11 In a Finnish study of morethan 1000 stroke patients who were treatedwith thrombolysis, half were treated outsideguidelines suggesting that most contra-indications may be unnecessary, or at leastrelative contraindications.14 Haemorrhagerates apparent on imaging were similar tothose reported from previous studies.10

In a 2007 prospective hospital-basedstudy of 259 stroke patients admitted to atertiary hospital in Adelaide, it was esti-mated that 16% of stroke patients werepotentially eligible for thrombolysis andarrived within the treatment time for throm-bolysis.15 A much larger proportion wouldhave been eligible had they arrived sooner.Currently, many stroke patients who wouldbenefit from thrombolysis do not receive it.Similar data have been reported throughoutthe Western world. In the United States, it isestimated from Medicare data alone thatfewer than 3% of stroke patients admitted tohospital receive thrombolysis.16 In theUnited Kingdom, a national audit showedthat 1.4% of stroke patients were treatedwith thrombolysis, but some individual cen-tres achieved much higher thrombolysisrates.7

Distance is an obstacle to thrombolysis forstroke in SA, even within metropolitanAdelaide. In our study, residing more thanfive kilometres from an acute stroke unitdiminished the chance of receiving throm-bolytic therapy this distance is only amatter of minutes by car in Adelaide. At adistance of more than about five kilometresfrom an acute stroke unit, a patient is muchmore likely to be taken to a closer alternativehospital. Despite over half of stroke admis-sions occurring at hospitals without strokeunits, transferred patients represented only3% of those who received thrombolysis.Also, although age is a strong risk factor forstroke,17 there was no relationship betweenage and distance from acute stroke unit.

The barriers to thrombolysis for stroke inrural areas, including lack of rapid imaging

4 Number of patients who received thrombolysis for acute stroke per 100 000 population according to distance from an acute stroke unit*

*The dashed line represents locally weighted regression applied to individual postcode data.

00

5

10

15

20

5 10Distance from an acute stroke unit (km)

Num

ber

of p

atie

nts

who

rece

ived

thr

omb

oly

sis

for

acut

e st

roke

per

100

000

po

pul

atio

n

15 20 50

0 to

RESEARCHand medical expertise, are formidable; yet asmall number of stroke patients do receivethrombolysis in the rural health system. Ruraldoctors come under pressure to provide treat-ment options to their patients. As unsup-ported administration of thrombolysis forstroke is not recommended in Australianguidelines,18 Australia must develop networksthat link rural doctors with stroke physicians.

The shortfalls in stroke service delivery inSA are currently being addressed by theStroke Clinical Network, established by SAHealth in 2009. Offering thrombolysis to alleligible patients is a main objective. Strate-gies include paramedic stroke identification,patient redirection to stroke units, acutestroke management protocols, and ruralstroke service networks. Unified hospitalcoding of thrombolysis is being imple-mented to monitor the success of thesestrategies.

Our results show that a large proportionof patients in SA are missing out on acutestroke therapy as a result of poor access toacute stroke units in urban and rural set-tings. Australia must face the challenge ofdelivering stroke thrombolysis safely andfrequently.

ACKNOWLEDGEMENTSWe thank Lizzie Dodd, Stroke Nurse, Queen Eliza-beth Hospital, for help in reviewing SITS cases;Grant Emmerson, Manager, Analysis and Report-ing, Information Communication and TechnologyDelivery Services, SA Health, for help in develop-ing searches within public hospital data repositor-ies; and Clarabelle Pham and Jonathan Karnon,Risk Adjusted Cost-Effectiveness (RAC-E) team,University of Adelaide, for providing Medicaredata.

COMPETING INTERESTSNone identified.

AUTHOR DETAILSJames M Leyden, MB BS, FRACP, Neurologist1

Woon K Chong, MB BS, FRANZCR, Neuroradiologist1

Tim Kleinig, MB BS(Hons), FRACP, PhD, Neurologist2

Andrew Lee, MB BS, MPH, FRACP, Director of Stroke Medicine3

John B Field, PhD, AStat, Statistical Consultant4

Jim Jannes, BM BS, FRACP, PhD, Head, Stroke Unit1

1 Queen Elizabeth Hospital, Adelaide, SA.2 Royal Adelaide Hospital, Adelaide, SA.3 Comprehensive Stroke Centre, Flinders

Medical Centre, Adelaide, SA.4 Faculty of Health Sciences, University of

Adelaide, Adelaide, SA.Correspondence: james.leyden@health.sa.gov.au

REFERENCES1 Hacke W, Donnan G, Fieschi C, et al. Associa-

tion of outcome with early stroke treatment:pooled analysis of ATLANTIS, ECASS, andNINDS rt-PA stroke trials. Lancet 2004; 363:768-774.

2 Hacke W, Kaste M, Bluhmki E, et al. Thromboly-sis with alteplase 3 to 4.5 hours after acuteischemic stroke. N Engl J Med 2008; 359: 1317-1329.

3 Kaste M, Thomassen L, Grond M, et al. Throm-bolysis for acute ischemic stroke: a consensusstatement of the 3rd Karolinska Stroke Update,October 3031, 2000. Stroke 2001; 32: 2717-2718.

4 Wardlaw JM, Murray V, Berge E, Del Zoppo GJ.Thrombolysis for acute ischaemic stroke.Cochrane Database Syst Rev 2009; (4):CD000213.

5 Wahlgren N, Ahmed N, Davalos A, et al.Thrombolysis with alteplase 3-4.5 h after acuteischaemic stroke (SITS-ISTR): an observationalstudy. Lancet 2008; 372: 1303-1309.

6 Leung TW, Wong KS. Thrombolysis withalteplase for acute ischemic stroke: safe andeffective outside the 3-hour time window? NatClin Pract Neurol 2009; 5: 70-71.

7 Sudlow C, Warlow C. Getting the priorities rightfor stroke care. BMJ 2009; 338: b2083.

8 Sudlow CL, Warlow CP. Comparing stroke inci-dence worldwide: what makes studies compa-rable? Stroke 1996; 27: 550-558.

9 The National Institute of Neurological Disor-ders and Stroke rt-PA Stroke Study Group.Tissue plasminogen activator for acuteischemic stroke. N Engl J Med 1995; 333: 1581-1587.

10 Fiorelli M, Bastianello S, von Kummer R, et al.Hemorrhagic transformation within 36 hours ofa cerebral infarct: relationships with early clini-cal deterioration and 3-month outcome in theEuropean Cooperative Acute Stroke Study I(ECASS I) cohort. Stroke 1999; 30: 2280-2284.

11 National Stroke Foundation. Clinical guidelinesfor acute stroke management. Melbourne: NSF,2007.

12 Fonarow GC, Reeves MJ, Smith EE, et al. Char-acteristics, performance measures, and in-hos-pital outcomes of the first one million strokeand transient ischemic attack admissions in GetWith The Guidelines-Stroke. Circ CardiovascQual Outcomes 2010; 3: 291-302.

13 Lees KR, Bluhmki E, von Kummer R, et al. Timeto treatment with intravenous alteplase andoutcome in stroke: an updated pooled analysisof ECASS, ATLANTIS, NINDS, and EPITHETtrials. Lancet 2010; 375: 1695-1703.

14 Meretoja A, Putaala J, Tatlisumak T, et al. Off-label thrombolysis is not associated with pooroutcome in patients with stroke. Stroke 2010;41: 1450-1458.

15 Kleinig TJ, Kimber TE, Thompson PD. Strokeprevention and stroke thrombolysis: quantify-ing the potential benefits of best practice ther-apies. Med J Aust 2009; 190: 678-682.

16 Kleindorfer D, Xu Y, Moomaw CJ, et al. USgeographic distribution of rt-PA utilization byhospital for acute ischemic stroke. Stroke 2009;40: 3580-3584.

17 Feigin VL, Lawes CM, Bennett DA, AndersonCS. Stroke epidemiology: a review of popula-tion-based studies of incidence, prevalence,and case-fatality in the late 20th century. LancetNeurol 2003; 2: 43-53.

18 Ad Hoc Committee representing the NationalStroke Foundation and the Stroke Society ofAustralasia. The implementation of intravenoustissue plasminogen activator in acute ischaemicstroke a scientific position statement fromthe National Stroke Foundation and the StrokeSociety of Australasia. Intern Med J 2009; 39:317-324.

(Received 14 Jul 2010, accepted 12 Oct 2010) MJA Volume 194 Number 3 7 February 2011 115

Case ascertainmentCase reviewPopulation demographicsStatistical analysisCase ascertainmentCase review