ABSTRACT
Objectives
To examine the epidemiology of stroke and transient ischaemic attack (TIA), and quality of care provided to Aboriginal and Torres Strait Islander (‘Indigenous’) patients presenting to emergency medical services (EMSs) with stroke or TIA.
Design
Retrospective population‐based data linkage study.
Setting, Participants
Victoria, Australia. Patients aged ≥ 18 years with hospital‐confirmed stroke or TIA who presented to EMSs between 1 January 2015 and 30 June 2019 and were successfully linked with hospital or mortality datasets.
Main Outcome Measures
Thirty‐ and 90‐day mortality, crude and age‐standardised incidence rates, EMS quality of care.
Results
Among 32,163 patients, 287 (0.9%) were Indigenous. Indigenous patients were younger than non‐Indigenous patients (median age, 62 vs. 77 years, p < 0.001). Age‐standardised incidence of EMS attendance was higher within the Indigenous population (343 vs. 144 per 100,000 person‐years; incidence rate ratio, 2.383 [95% confidence interval (CI), 2.084–2.711]). Rates of call‐taker identification of stroke or TIA and hospital diagnoses were comparable; however, Indigenous patients were less likely to receive a stroke assessment from paramedics (166/285 [58.2%] vs. 21,271/31,605 [67.3%]; p = 0.001) and more frequently taken to stroke‐capable hospitals (257/275 [93.5%] vs. 27,371/30,647 [89.3%]; p = 0.027), but less frequently taken to endovascular thrombectomy‐capable hospitals (67/275 [24.4%] vs. 10,015/30,647 [32.7%]; p = 0.003). Adjusted mortality was similar between groups (30 days: hazard ratio, 0.814 [95% CI, 0.518–1.279]; 90 days: hazard ratio, 0.805 [95% CI, 0.534–1.214]). However, age‐standardised incidence of 90‐day mortality was higher within the Indigenous population (39 vs. 24 per 100,000 person‐years; incidence rate ratio, 1.621 [95% CI, 1.031–2.398]). Indigenous patients also more often recontacted EMSs for any reason within 90 days (subdistribution hazard ratio, 1.548 [95% CI, 1.194–2.007]).
Conclusions
The age‐standardised incidence of EMS attendance for stroke in the Indigenous population was more than double that for the non‐Indigenous population. Although there was no difference in stroke identification at the time of the emergency call, Indigenous patients were less likely to be evaluated by paramedics for stroke.
Keywords: epidemiology, Indigenous health, medical emergency services, stroke
Plain Language Summary
The Known: Aboriginal and Torres Strait Islander (‘Indigenous’) peoples have a roughly twofold greater stroke hospitalisation rate than non‐Indigenous patients. However, few studies have investigated variation in pre‐hospital stroke care between Indigenous and non‐Indigenous patients.
The New: In Victoria, Australia, the age‐standardised incidence of emergency medical service attendance for stroke within the Indigenous population was more than twice that for the non‐Indigenous population. Rates of stroke identification in emergency calls and stroke and transient ischaemic attack hospital diagnoses were comparable; however, Indigenous patients less frequently received a stroke assessment from paramedics. The age‐standardised incidence of 90‐day mortality was higher for Indigenous Australians.
The Implications: Opportunities exist to improve care for Indigenous patients with stroke.
1. Introduction
Stroke impacts around 46,000 Australians annually and carries a high burden of mortality and disability [1]. The burden of stroke is not equally distributed across the Australian population, with Aboriginal and Torres Strait Islander (‘Indigenous’) peoples having a hospitalisation rate roughly twofold greater than non‐Indigenous people, and a 2.5‐fold greater mortality rate [2]. In addition, Indigenous patients experience stroke, on average, 16 years earlier than non‐Indigenous Australians [2]. Despite this, little research has investigated whether Indigenous patients receive comparable care to non‐Indigenous patients, or experience disparities in stroke outcomes.
Existing research surrounding stroke care for Indigenous and non‐Indigenous Australians is conflicting, owing to heterogeneity in data definitions and identification of Indigenous status, as well as variation in geographic context. For example, a 2009 national audit suggested Indigenous patients are less likely to receive care according to optimal clinical guidelines or in a stroke unit [3]. Conversely, a New South Wales study (2020) reported similar treatment rates according to Indigenous status for patients admitted to a stroke unit [4]. Literature investigating post‐stroke outcomes and short‐term mortality in Indigenous and non‐Indigenous populations also conflict, with some studies reporting higher mortality in Indigenous populations and others reporting no difference [3, 4, 5, 6]. Importantly, to our knowledge, there is no literature exploring variation in prehospital care.
This study therefore aimed to investigate the epidemiology of stroke and transient ischaemic attack (TIA), and quality of care provided to Indigenous and non‐Indigenous patients with hospital‐confirmed stroke or TIA attended by emergency medical services (EMSs), in Victoria, Australia.
2. Methods
2.1. Positionality Statement
This study was conceived by an Aboriginal researcher (L.J.B.). The research team included two Aboriginal clinician researchers (A.D.S., M.C.) with expertise in stroke and prehospital care. Aboriginal liaison officers from hospitals represented in the dataset formed an Indigenous advisory group, providing guidance on study design, community needs and priorities. Aboriginal members of the research team (L.J.B., A.D.S., M.C.), which included members of the Kulin Nation from which data were collected, participated across all stages of research, ensuring the priorities identified by Aboriginal liaison officers were centred across the project.
2.2. Study Design
This was a retrospective observational data linkage study of adult patients (≥ 18 years old) with confirmed stroke or TIA within Victoria, Australia, between 1 January 2015 and 30 June 2019. We excluded cases of trauma, interhospital transfers and patients with missing Indigenous status. Repeat stroke and TIA presentations within 90 days of the index event were excluded and instead coded as a 90‐day outcome. The inclusive ‘ever‐Indigenous’ approach was taken, whereby patients who identified as Indigenous once within the dataset were classified as Indigenous throughout the entire analysis [7, 8]. This approach ensures that the Indigenous population is not under‐represented [9]. Ethics approval for this Aboriginal and Torres Strait Islander research was obtained from Monash University Human Research Ethics Committee (No. 39764). This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines (Table S1).
2.3. Study Setting
Victoria spans 227,500 km2 and has a population of 6.7 million inhabitants. Ambulance Victoria is the statewide EMS provider and is two‐tiered, including Mobile Intensive Care Ambulance paramedics and Advanced Life Support paramedics. Emergency Triple Zero (000) calls are triaged using the universal Medical Priority Dispatch System (MPDS), which uses structured questioning to arrive at the patient's primary complaint. Most call types receive an emergency ambulance response. However, predefined low‐acuity call types are referred for secondary triage by an experienced clinician, which may result in emergency ambulance or non‐emergency patient transport service dispatch, referral to an alternative service (including the emergency department [ED]) or provision of self‐care advice.
Upon arrival at the patient, paramedics follow clinical practice guidelines. If paramedics suspect a stroke or TIA, the Melbourne Ambulance Stroke Screen (MASS) is recommended to assess symptoms including facial droop, grip weakness, slurred speech and abnormal blood glucose levels. If stroke symptoms are present, patients are transported to the nearest thrombolysis or endovascular thrombectomy (EVT)‐capable hospital [10]. Within rural and regional areas where stroke‐specific care is limited, 19 hospitals have access to the Victorian Stroke Telemedicine (VST) service, which allows regional and remote hospitals to connect virtually with neurologists in metropolitan areas, providing lower access hospitals with greater capacity to provide care [11].
2.4. Data Collection and Sources
A linked dataset was utilised, which included data from: (1) prehospital electronic patient care records; (2) the Victorian Emergency Minimum Dataset, which captures demographic, administrative and clinical data for ED presentations at Victorian public hospitals; (3) the Victorian Admitted Episodes Dataset, which includes information on the causes, effects and nature of illness for all admissions to public and private hospitals; and (4) the Victorian Death Index, which collects information on the date and cause of death. The data linkage process has been described previously [12]. For patients attended by paramedics but not transported to hospital, ED presentations and hospital admissions occurring up to 48 h after calling Triple Zero (000) were linked with the prehospital record. We also obtained population estimates from the Australian Bureau of Statistics to calculate crude and age‐standardised incidence rates.
2.5. Definitions
Hospital‐confirmed stroke or TIA was defined as a primary, secondary or tertiary International Classification of Diseases, tenth revision, Australian modification (ICD‐10‐AM) diagnosis code of stroke (I61–I64) or TIA (G459). Sex was defined as the patient's sex at birth. Accessibility/Remoteness Index of Australia scores were used to categorise patient residential and scene addresses based on geographical accessibility to services [13]. Patients in outer regional, remote and very remote areas of Victoria were grouped into one cohort. Socio‐economic status was reported using the Index of Relative Socio‐economic Disadvantage, which groups people and households within an area into deciles based on their relative disadvantage [14]. For analysis, deciles were further collapsed into quintiles. ‘Stroke identified by paramedics’ describes cases where paramedics recorded a final diagnosis of stroke, TIA, subarachnoid haemorrhage or intracranial haemorrhage on their patient care record. ‘Stroke identified by call‐taker’ included cases allocated as Protocol 28 (Stroke) MPDS code during the emergency telephone triage. Pain scores were self‐reported on a scale of 0 (no pain) to 10 (worst imaginable pain). Vital signs were categorised based on Ambulance Victoria's clinical practice guidelines [15]. Receiving hospitals were defined as stroke‐capable if they had a stroke unit and/or the ability to administer thrombolysis, and EVT‐capable if they had the capacity to perform EVT [16]. Subsequent interhospital transfers were not examined, therefore transport time was defined as the duration of transport to the original receiving ED. Days to death, repeat stroke and recontact with the EMS were calculated from the date of the initial emergency call.
2.6. Data Analysis
All analyses were performed using Stata 17 statistical software (StataCorp). Crude, age‐specific and age‐standardised incidence and mortality rates per 100,000 person‐years were calculated using 2016 Indigenous and non‐Indigenous population estimates from the Australian Bureau of Statistics, with the 2001 Victorian population used as the reference year [17]. Incidence rates were compared using incidence rate ratios (IRRs) with 95% confidence intervals (CIs). Categorical data are presented as frequencies and proportions and compared using the chi‐squared test. Continuous data are presented as medians and interquartile ranges (IQRs) with comparisons made using the Mann–Whitney U test.
Multivariable survival analyses were performed to examine differences in 30‐ and 90‐day mortality, repeat stroke presentation and recontact with the EMS for any reason. For mortality, parametric Weibull models were used based on examination of the baseline hazard function, allowing for a flexible representation of time‐varying risk. Results are presented as hazard ratios (HRs) and 95% CIs. For repeat stroke and recontact, Fine–Gray competing risks regression models were used, with death within 30 or 90 days modelled as a competing risk. Results are presented as subdistribution hazard ratios (sHRs) and 95% CIs. All models were adjusted for Indigenous status, patient age, sex, hospital diagnosis (stroke or TIA), scene remoteness, patient socio‐economic status quintile and comorbidities including previous stroke or TIA, atrial fibrillation, chronic obstructive pulmonary disease and liver disease. Sensitivity analyses were performed in which Indigenous status was classified according to individual hospital admission records, rather than using the ever‐Indigenous approach. For the mortality models, additional sensitivity analyses were performed using multiple imputation with 20 chained equations to account for missingness in the comorbidity and socio‐economic status variables. Rubin's rules were used to pool results. An age‐adjusted logistic regression analysis was also conducted to understand differences in performance of the MASS assessment. The result of this analysis is presented as an adjusted odds ratio with 95% CI. p values of < 0.05 were considered statistically significant.
3. Results
After exclusions, 32,163 confirmed stroke or TIA patients were included (Figure 1). Of these patients, 287 (0.9%) identified as Indigenous. Of the 31,876 non‐Indigenous patients, 27,142 (85.1%) had a hospital diagnosis of stroke and 4734 (14.9%) had a hospital diagnosis of TIA. Among Indigenous patients, 246 (85.7%) were diagnosed with stroke, and 41 (14.3%) with TIA.
FIGURE 1.
Patient selection. EMS, emergency medical services, ICD, international classification of diseases; TIA, transient ischaemic attack.
3.1. Incidence
The crude incidence of EMS attendance for stroke was higher in the Indigenous population than the non‐Indigenous population (220 vs. 164 per 100,000 person‐years; IRR, 1.346 [95% CI, 1.194–1.512], Figure 2A, Table S2). After age standardisation, incidence in the Indigenous population was more than twice that in the non‐Indigenous population (343 vs. 144 per 100,000 person‐years; IRR, 2.383 [95% CI, 2.084–2.711]). Age‐specific incidence across 10‐year age categories is presented in Figure 2B and was consistently higher within the Indigenous population.
FIGURE 2.
Crude, age‐standardised and age‐specific incidence per 100,000 person‐years within Indigenous and non‐Indigenous populations.
3.2. Baseline Characteristics
Table 1 shows the baseline characteristics of the study population. Indigenous patients were younger (median age, 62 years [IQR, 49–72 years] vs. 77 years [IQR, 66–85 years]; p < 0.001) and more often resided in areas of lower socioeconomic status. Most patients resided in metropolitan areas, but Indigenous patients had a higher proportion of residents in inner regional, outer regional and remote areas. Indigenous patients had an increased history of liver disease and more frequently presented with a rapid heart rate. Non‐Indigenous patients had an increased history of atrial fibrillation.
TABLE 1.
Baseline characteristics of Indigenous and non‐Indigenous patients.
| Overall | Non‐Indigenous | Indigenous | p | |
|---|---|---|---|---|
| Total | 32,163 | 31,876 | 287 | |
| Age, median (IQR) | 77 (66–85) | 77 (66–85) | 62 (49–72) | < 0.001 |
| Females, n (%) a | 15,893 (49.9) | 15,743 (49.9) | 150 (52.8) | 0.328 |
| Missing | Not provided | 329 (1.0%) | < 5 | — |
| Patient residential location, n (%) | ||||
| Metropolitan a | 23,005 (73.5) | 22,851 (73.6) | 154 (55.8) | < 0.001 |
| Inner regional a | 6776 (21.6) | 6684 (21.5) | 92 (33.3) | < 0.001 |
| Outer regional/remote a | 1527 (4.9) | 1497 (4.8) | 30 (10.9) | < 0.001 |
| Missing | 855 (2.7) | 844 (2.6) | 11 (3.8) | — |
| Socio‐economic status, n (%) | ||||
| Quintile 1 (lowest) a | 6899 (24.0) | 6797 (23.9) | 102 (40.5) | < 0.001 |
| Quintile 2 a | 5893 (20.5) | 5838 (20.5) | 55 (21.8) | 0.612 |
| Quintile 3 a | 5815 (20.3) | 5771 (20.3) | 44 (17.5) | 0.265 |
| Quintile 4 a | 5626 (19.6) | 5594 (19.7) | 32 (12.7) | 0.006 |
| Quintile 5 (highest) a | 4455 (15.5) | 4436 (15.6) | 19 (7.5) | < 0.001 |
| Missing | 3475 (10.8) | 3440 (10.8) | 35 (12.2) | — |
| Patient event location, n (%) | ||||
| Metropolitan a | 23,555 (73.8) | 23,398 (74.0) | 157 (55.5) | < 0.001 |
| Inner regional a | 6830 (21.4) | 6736 (21.3) | 94 (33.2) | < 0.001 |
| Outer regional/remote a | 1521 (4.8) | 1489 (4.7) | 32 (11.3) | < 0.001 |
| Missing | Not provided | 253 (0.8) | < 5 | — |
| Pre‐existing comorbidities, n (%) | ||||
| Stroke/TIA a | 9315 (30.7) | 9219 (30.6) | 96 (36.1) | 0.054 |
| Myocardial infarction a | 4538 (14.9) | 4497 (14.9) | 41 (15.4) | 0.826 |
| Renal diseases a | 1073 (3.5) | 1064 (3.5) | 9 (3.4) | 0.895 |
| COPD a | 2135 (7.0) | 2106 (7.0) | 29 (10.9) | 0.013 |
| Liver disease a | 192 (0.6) | 184 (0.6) | 8 (3.0) | < 0.001 |
| Diabetes a | 6839 (22.5) | 6772 (22.5) | 67 (25.2) | 0.294 |
| Dementia a | 2400 (7.9) | 2385 (7.9) | 15 (5.6) | 0.170 |
| Atrial fibrillation a | 4867 (16.0) | 4853 (16.1) | 14 (5.3) | < 0.001 |
| Missing | 1781 (5.5) | 1760 (5.5) | 21 (7.3) | — |
| Initial EMS vital signs, n (%) | ||||
| Impaired conscious state (GCS score < 15) a | 12,828 (40.9) | 12,717 (40.9) | 111 (39.8) | 0.701 |
| Missing | 806 (2.5) | 798 (2.5) | 8 (2.8) | — |
| Fever (≥ 38°C) a | 722 (2.4) | 714 (2.4) | 8 (3.0) | 0.534 |
| Missing | 2038 (6.3) | 2020 (6.3) | 18 (6.3) | — |
| Any pain a | 4972 (16.4) | 4917 (16.4) | 55 (20.3) | 0.084 |
| Missing | 1895 (5.9) | 1879 (5.9) | 16 (5.6) | — |
| Normal oxygen saturation (≥ 92%) a | 28,274 (93.6) | 28,032 (93.6) | 242 (90.6) | 0.051 |
| Missing | 1940 (6.0) | 1920 (6.0) | 20 (7.0) | — |
| Normal blood glucose a | 7986 (28.0) | 7908 (28.0) | 78 (31.0) | 0.296 |
| Missing | 3652 (11.4) | 3617 (11.3) | 35 (12.2) | — |
| Normal systolic blood pressure (≥ 100 mmHg) a | 30,436 (97.2) | 30,170 (97.2) | 266 (95.3) | 0.062 |
| Missing | 845 (2.6) | 837 (2.6) | 8 (2.8) | — |
| Respiratory rate (breaths/min) | ||||
| 12–16 a | 17,449 (55.7%) | 17,308 (55.7%) | 141 (50.5%) | 0.084 |
| 17–20 a | 10,411 (33.2%) | 10,308 (33.2%) | 103 (36.9%) | 0.186 |
| Missing | 813 (2.5%) | 805 (2.5%) | 8 (2.8%) | — |
| Heart rate (beats/min) | ||||
| Normal (60–100) a | 25,472 (81.3%) | 25,264 (81.3%) | 208 (74.6%) | 0.004 |
| Moderately rapid (100–120) a | 3195 (10.2%) | 3151 (10.1%) | 44 (15.8%) | 0.002 |
| Irregular (> 120 or < 60) a | 2681 (8.6%) | 2654 (8.5%) | 27 (9.7%) | 0.500 |
| Missing | 815 (2.5%) | 807 (2.5%) | 8 (2.8%) | — |
Abbreviations: COPD, chronic obstructive pulmonary disease; EMS, emergency medical service; GCS, Glasgow Coma Scale; TIA, transient ischaemic attack.
Presented as a proportion of patients for whom data were not missing.
3.3. Prehospital Care Quality
Prehospital care is presented in Table 2. Identification of stroke within the emergency call was similar between groups. However, Indigenous patients were significantly less likely to receive a MASS assessment from paramedics (166/285 [58.2%] vs. 21,271/31,605 [67.3%]; p = 0.001), even after adjusting for age (adjusted odds ratio, 0.721 [95% CI, 0.569–0.915]). This included less frequent assessment for hand grip weakness, facial droop and speech deficits. However, among assessed patients, there were no differences in the rate of deficits identified. Indigenous patients were also significantly less likely to receive an electrocardiogram. Across the state, Indigenous patients were more frequently taken to stroke‐capable hospitals; however, fewer were taken to EVT‐capable hospitals. Among patients attended in metropolitan Melbourne, there was no difference in the rate of transport to stroke‐capable hospitals (138/150 [92.0%] vs. 19,938/22,586 [88.3%]; p = 0.157) or EVT‐capable hospitals (64/150 [42.7%] vs. 9559/22,586 [42.3%]; p = 0.932). No differences were identified in ambulance response times or time spent on scene, although transport time to hospital was shorter for Indigenous patients.
TABLE 2.
Prehospital care of Indigenous and non‐Indigenous patients.
| Overall | Non‐Indigenous | Indigenous | p | |
|---|---|---|---|---|
| Total | 32,163 | 31,876 | 287 | |
| Call taker identified stroke, n (%) | 16,107 (50.1) | 15,965 (50.1) | 142 (49.5) | 0.838 |
| Paramedic intervention, n (%) | ||||
| MASS assessment completed | 21,437 (67.2) | 21,271 (67.3) | 166 (58.2) | 0.001 |
| Speech assessed a | 26,040 (81.7) | 25,834 (81.7) | 206 (72.3) | < 0.001 |
| Speech deficit b | 11,961 (45.9) | 11,858 (45.9) | 103 (50.0) | 0.240 |
| Hand grips assessed a | 26,839 (84.2) | 26,615 (84.2) | 224 (78.6) | 0.010 |
| Hand grips deficit b | 12,163 (45.3) | 12,053 (45.3) | 110 (49.1) | 0.253 |
| Facial droop assessed a | 24,150 (75.7) | 23,954 (75.8) | 196 (68.8) | 0.006 |
| Facial droop deficit b | 9450 (39.1) | 9375 (39.1) | 75 (38.3) | 0.803 |
| Blood glucose taken a | 28,511 (89.4) | 28,259 (89.4) | 252 (88.4) | 0.588 |
| Missing | Not provided | 271 (0.9) | < 5 | — |
| Electrocardiogram taken a | 29,237 (91.7) | 28,987 (91.7) | 250 (87.7) | 0.015 |
| Missing | Not provided | 271 (0.9%) | < 5 | — |
| Any paramedic intervention a | 16,746 (53.3) | 16,590 (53.3) | 156 (55.9) | 0.388 |
| Intubation a | Not provided | 473 (1.5) | < 5 | 0.906 |
| Intravenous access a | 11,032 (35.1) | 10,939 (35.2) | 93 (33.3) | 0.524 |
| Analgesia | ||||
| Analgesia given a | 4773 (15.2) | 4724 (15.2) | 49 (17.6) | 0.271 |
| Analgesia refused a | 418 (1.3) | 409 (1.3) | 9 (3.2) | 0.006 |
| Missing interventions, n (%) | 774 (2.4) | 766 (2.4) | 8 (2.8) | — |
| Stroke or TIA identified by paramedics, n (%) a | 17,334 (55.6) | 17,193 (55.6) | 141 (50.9) | 0.114 |
| Missing, n (%) | 990 (3.1) | 980 (3.1) | 10 (3.5) | — |
| Time metrics | ||||
| 000 cal to ambulance arrival at patient (min), median (IQR) | 11.5 (8.5–17.3) | 11.5 (8.5–17.3) | 10.8 (8.3–18.2) | 0.485 |
| Missing, n (%) | 872 (2.7%) | 863 (2.7%) | 9 (3.1%) | — |
| Time on scene (min), median (IQR) | 18 (13–25) | 18 (13–25) | 18 (13–25) | 0.527 |
| Missing, n (%) | 683 (2.1) | 676 (2.1) | 7 (2.4) | — |
| Transport duration (min), median (IQR) c | 19 (12–28) | 19 (12–28) | 16 (9–25) | < 0.001 |
| Transported to hospital, n (%) a | 31,313 (99.0) | 31,034 (99.0) | 279 (99.3) | 0.617 |
| Missing, n (%) | 531 (1.7) | 525 (1.6) | 6 (2.1) | — |
| Lights and sirens transport to hospital, n (%) c | 14,394 (46.0) | 14,267 (46.0) | 127 (45.5) | 0.880 |
| Type of hospital, n (%) c | ||||
| Stroke‐capable hospital | 27,628 (89.4) | 27,371 (89.3) | 257 (93.5) | 0.027 |
| EVT‐capable hospital | 10,082 (32.6) | 10,015 (32.7) | 67 (24.4) | 0.003 |
| Missing | Not provided | 387 (1.2) | < 5 | — |
Abbreviations: EVT, endovascular thrombectomy; IQR, interquartile range; MASS, Melbourne Ambulance Stroke Screen; TIA, transient ischaemic attack.
Presented as a proportion of patients for whom data were not missing.
Presented as a proportion of assessed patients.
Presented for transported patients for whom data were not missing.
3.4. In‐Hospital Care Quality
Hospital care is presented in Table 3. Indigenous patients were more likely to leave against medical advice after both ED presentation and hospital admission, and less likely to be admitted to hospital, but more likely to be admitted to an intensive care unit (ICU) during admission. Unadjusted in‐hospital mortality was higher in non‐Indigenous patients (12/248 [4.8%] vs. 2965/28,830 [10.3%]; p = 0.005).
TABLE 3.
Hospital care for Indigenous and non‐Indigenous patients.
| Overall | Non‐Indigenous | Indigenous | p | |
|---|---|---|---|---|
| Total | 32,163 | 31,876 | 287 | |
| Admitted to ED, n (%) | 30,538 (94.9) | 30,254 (94.9) | 284 (99.0) | 0.002 |
| ED triage category, n (%) a | ||||
| Resuscitation | 1833 (6.0) | 1811 (6.0) | 22 (7.7) | 0.214 |
| Emergency | 16,102 (52.7) | 15,954 (52.7) | 148 (52.1) | 0.835 |
| Urgent | 9790 (32.1) | 9700 (32.1) | 90 (31.7) | 0.894 |
| Ambulance offload < 40 min, n (%) a | 25,003 (85.3) | 24,770 (85.3) | 233 (86.3) | 0.739 |
| Missing, n (%) | 1214 (4.0) | 1201 (4.0) | 14 (4.9) | — |
| ED length of stay (min), median (IQR) b | 302 (199–516) | 302 (199–516) | 311 (202.5–488.5) | 0.875 |
| Discharged within 4 h, n (%) a | 12,023 (39.4) | 11,912 (39.4) | 111 (39.1) | 0.921 |
| Discharge direction after ED, n (%) a | ||||
| Ward | 28,086 (92.0) | 27,829 (92.0) | 257 (90.5) | 0.357 |
| Home | 1413 (4.6) | 1399 (4.6) | 14 (4.9) | 0.807 |
| Left against advice | 114 (0.4) | 106 (0.4) | 8 (2.8) | < 0.001 |
| Aged care facility | Not provided | 92 (0.3) | < 5 | 0.884 |
| Died in ED | Not provided | 220 (0.7) | < 5 | 0.458 |
| Other | Not provided | 608 (2.0) | < 5 | 0.253 |
| Admitted to hospital, n (%) | 29,078 (90.4) | 28,830 (90.4) | 248 (86.4) | 0.021 |
| Admitted to ICU, n (%) c | 2627 (9.0) | 2590 (9.0) | 37 (14.9) | 0.001 |
| ICU length of stay (h), median (IQR) | 44 (22–92.5) | 44 (22–92) | 47 (28.5–132) | 0.467 |
| Admission discharge direction, n (%) c | ||||
| Transfer to other facility | 10,009 (34.4) | 9930 (34.4) | 79 (31.9) | 0.393 |
| Discharge to private accommodation/home | 14,422 (49.6) | 14,289 (49.6) | 133 (53.6) | 0.202 |
| Discharge to aged care facility | 1494 (5.1) | 1482 (5.1) | 12 (4.8) | 0.830 |
| Left against medical advice | 176 (0.6) | 164 (0.6) | 12 (4.8) | < 0.001 |
| Death | 2977 (10.2) | 2965 (10.3) | 12 (4.8) | 0.005 |
| Hospital admission length of stay (days), median (IQR) | 3 (1–7) | 3 (1–7) | 3 (1–6) | 0.617 |
Abbreviations: ED, emergency department; ICU, intensive care unit; IQR, interquartile range.
Presented as a proportion of patients admitted to the ED for whom data were not missing.
For patients admitted to the ED.
Presented as a proportion of patients admitted to hospital.
3.5. Patient Outcomes
Age‐standardised incidence of 90‐day mortality was higher within the Indigenous than the non‐Indigenous population (39 vs. 24 per 100,000 person‐years; IRR, 1.621 [95% CI, 1.031–2.398], Figure S1). Unadjusted and adjusted patient outcomes are presented in Table 4. Indigenous patients had lower unadjusted mortality at 30 and 90 days. However, after adjustment, there was no difference in mortality (30 days: HR, 0.814 [95% CI, 0.518–1.279]; 90 days: HR, 0.805 [95% CI, 0.534–1.214]). Instead, age, pre‐existing conditions and socioeconomic status were associated with mortality outcomes (Table S3). Indigenous patients more often recontacted the EMS for any reason within 90 days (sHR, 1.548 [95% CI, 1.194–2.007]).
TABLE 4.
Outcomes of Indigenous and non‐Indigenous patients.
| Overall | Unadjusted comparisons | Adjusted HR/sHR (95% CI) | |||
|---|---|---|---|---|---|
| Non‐Indigenous | Indigenous | p | |||
| Total | 32,163 | 31,876 | 287 | ||
| Mortality, n (%) | |||||
| Within 48 h of 000 call | 1536 (4.8) | 1528 (4.8) | 8 (2.8) | 0.113 | — |
| 30 days | 4644 (14.4) | 4620 (14.5) | 24 (8.4) | 0.003 | 0.814 (0.518–1.279) |
| 90 days | 5671 (17.6) | 5643 (17.7) | 28 (9.8) | < 0.001 | 0.805 (0.534–1.214) |
| Repeat stroke, n (%) | |||||
| 30 days | 522 (1.6) | 517 (1.6) | 5 (1.7) | 0.872 | 0.900 (0.333–2.428) |
| 90 days | 927 (2.9) | 918 (2.9) | 9 (3.1) | 0.796 | 0.909 (0.427–1.937) |
| Recontact for any reason, n (%) | |||||
| 30 days | 2612 (8.1) | 2575 (8.1) | 37 (12.9) | 0.003 | 1.214 (0.816–1.807) |
| 90 days | 5532 (17.2) | 5455 (17.1) | 77 (26.8) | < 0.001 | 1.548 (1.194–2.007) |
Abbreviations: CI, confidence interval; HR, hazard ratio; sHR, subdistribution hazard ratio.
3.6. Sensitivity Analyses
Sensitivity analyses in which Indigenous status was classified according to individual hospital admission records corroborate the findings of our primary analysis, except for the rate of hospital admission and adjusted 90‐day recontact, which no longer differed significantly between Indigenous and non‐Indigenous patients (Tables S4–S6). Sensitivity analyses using multiple imputation to account for missing data supported the primary mortality findings (Table S7).
4. Discussion
Our investigation into the epidemiology of stroke and TIA, and quality of care provided to Indigenous and non‐Indigenous patients with stroke or TIA attended by EMSs in Victoria has several key findings. Age‐standardised incidence within the Indigenous population was more than twice that in the non‐Indigenous population, and Indigenous patients were a median of 15 years younger than non‐Indigenous patients. Rates of stroke identification in the emergency call and stroke or TIA hospital diagnoses were comparable, demonstrating strengths within several components of the care pathway. However, Indigenous patients less frequently received a stroke assessment from paramedics, were more likely to leave hospital against medical advice and were more likely to recontact the EMS for any reason within 90 days. Although there was no difference in adjusted mortality, the age‐standardised incidence of 90‐day mortality was higher for Indigenous Australians.
Our findings suggest that improving paramedics' completion of stroke assessments (MASS) should be a priority in prehospital care. It is not clear from our findings why Indigenous patients were less likely to receive a stroke assessment or electrocardiogram. One explanation may be that Indigenous patients were younger (median age, 62 years), highlighting the need for stroke assessment pathways to be responsive to populations where stroke may occur earlier than expected. Data from the Stroke Foundation suggest that only 25% of stroke patients in Australia are under the age of 65 years [1]. Importantly, our age‐adjusted analysis also demonstrated reduced odds of receiving a MASS assessment among Indigenous patients, suggesting that factors beyond age, including differences in presentation and health system processes, are likely to account for this disparity.
Another explanation may be that Indigenous populations have previously reported higher rates of haemorrhagic strokes [2, 18]. Symptom presentation of haemorrhagic stroke, including headaches, vomiting and facial weakness, overlaps with the presentation of other conditions, including migraines, seizures, hypoglycaemia and Bell palsy [19]. As such, stroke symptoms may not always be recognised through standard assessment approaches. These findings suggest that there remains a potential need to increase EMS education surrounding the recognition of non‐hallmark stroke symptoms, and to ensure that stroke assessment tools are used equitably across different populations.
However, clinician and/or system‐level bias may have also affected Indigenous patients, as has previously been reported in emergency settings [20] and stroke care [21]. Systemic inequities in prehospital stroke care have been demonstrated in the United States, where white patients were more likely to be evaluated by a stroke team than African American patients [22]. Similarly, Hispanic and Asian patients were less likely than non‐Hispanic white patients to be correctly identified as having a stroke by EMSs [23]. Together, these findings highlight the need for stroke services to better recognise how individual and process biases may affect stroke outcomes, especially for Indigenous Australians who may present younger and atypically [1].
A higher proportion of Indigenous patients were taken to stroke‐capable hospitals. This finding likely reflects the higher proportion of Indigenous patients living in regional and remote areas, where most major public hospitals have access to the VST service. Importantly, transport to a stroke‐capable facility does not imply that thrombolysis was administered, and additional investigation is warranted. The increased proportion of Indigenous patients residing in regional and remote areas also explains the reduced likelihood of their transport to EVT‐capable hospitals. Notably, no such disparity was observed among patients attended within metropolitan Melbourne. Currently, only five EVT‐capable hospitals exist in Victoria, and all are in metropolitan Melbourne. For metropolitan patients, paramedics may have favoured transport to EVT‐capable hospitals, even if this involved bypassing smaller facilities and an extended transport duration.
In hospital, a higher proportion of Indigenous patients were admitted to ICUs. Regionality may also explain this finding, with ICU admission being the closest alternative to EVT‐capable centres in regional areas. Our study also suggests that Indigenous patients were more likely to leave against medical advice. This finding has been reported consistently across Australian health services and is increasingly understood as reflecting health system factors rather than individual patient characteristics alone. Previous studies suggest that experiences of cultural unsafety, racism, communication barriers, competing family and community responsibilities, and limited responsiveness of services to Indigenous needs may contribute to discharge decisions [24]. Viewed through this lens, the increased hazard of EMS recontact within 90 days may reflect ongoing unmet healthcare needs and opportunities to strengthen continuity of culturally safe stroke care following hospital discharge.
There was no significant difference in adjusted mortality at 30 or 90 days; however, age‐standardised incidence of 90‐day mortality per 100,000 population was higher in the Indigenous population than in the non‐Indigenous population. Until now, age‐standardised mortality has not been studied in an Indigenous cohort within Victoria, although rates in a Western Australian, South Australian and Northern Territory cohort were reported to be around twofold higher in Indigenous patients [2]. Previous studies investigating the impact of age on patient recovery and mortality have suggested that younger patients have a higher chance of optimal recovery, lower stroke recurrence rates and lower mortality rates than older patients [25]. Despite being substantially younger at stroke onset, Indigenous patients did not experience the survival advantage typically observed in younger stroke populations. This finding highlights the enduring influence of broader social, structural and healthcare determinants on outcomes.
Understanding how stroke services respond to Indigenous peoples' health needs is an important step towards strengthening culturally safe, responsive and accountable systems of care. Our findings identify opportunities to improve care across the stroke continuum, while also highlighting strengths within existing pathways upon which future improvements can be built. Consistent with contemporary Indigenous health frameworks, these improvements should be developed through genuine partnerships with Indigenous communities, organisations and leaders, recognising Indigenous peoples as experts in the design, delivery and evaluation of services that affect their health [5, 26]. Strengthening Indigenous leadership, cultural safety and accountability within mainstream stroke services has the potential to support Indigenous concepts of health and wellbeing, while improving experiences and outcomes of care.
4.1. Limitations
Our study has some limitations. Data on previously reported confounders such as stroke severity, disability scores [25, 26] and new placement in a nursing home upon hospital discharge were not available, and neither were data on stroke subtypes and ED management. This study was based on hospital‐confirmed stroke cases attended by a single EMS, and the linkage rate with hospital data for patients known to be transported by EMSs to public hospitals has been reported to be 83.3% [12]. Individuals with stroke who did not engage EMSs, whether due to cultural, geographical or other barriers to access, are not accounted for in incidence calculations. As such, the true incidence of stroke in our population is likely to be under‐reported.
4.2. Conclusion
The age‐standardised incidence of EMS attendance to patients with stroke in the Indigenous population was twice that for the non‐Indigenous population, and Indigenous patients were a median of 15 years younger at stroke onset. Although rates of stroke identification by call‐takers were comparable, Indigenous patients were less likely to receive a stroke assessment by paramedics. Systemic barriers to care quality may exist for Indigenous patients and opportunities exist to improve care across the stroke continuum.
Author Contributions
Candice Menezes: formal analysis and writing – original draft. Ziad Nehme: supervision, methodology, and writing – review and editing. Luke J. Burchill: conceptualization, and writing – review and editing. Tegwyn McManamny: writing – review and editing. Angela Dos Santos: writing – review and editing. Michelle Crilly: writing – review and editing. Luke P. Dawson: writing – review and editing. Benjamin Clissold: writing – review and editing. David Anderson: writing – review and editing. Emily Nehme: data curation, project administration, supervision, methodology, and writing – review and editing.
Funding
Ziad Nehme is supported by a Future Leader Fellowship from the National Heart Foundation of Australia (no. 105690) and an Emerging Leadership Investigator Grant from the National Health and Medical Research Council (no. 2034615).
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Table S1: STROBE statement—checklist of items that should be included in reports of observational studies.
Table S2: Crude, age‐standardised and age‐specific incidence of EMS attendance for stroke among Indigenous and non‐Indigenous populations, per 100,000 person‐years.
Figure S1: Crude and age‐standardised incidence of 90‐day mortality following EMS attendance for stroke, per 100,000 person‐years, within Indigenous and non‐Indigenous populations.
Table S3: Multivariable analyses of 30‐ and 90‐day mortality.
Table S4: Baseline characteristics of Indigenous and non‐Indigenous patients, with Indigenous status classified according to individual hospital admission records.
Table S5: Pre‐hospital and hospital care of Indigenous and non‐Indigenous patients, with Indigenous status classified according to individual hospital admission records.
Table S6: Patient outcomes, with Indigenous status classified according to individual hospital admission records.
Table S7: Adjusted mortality outcomes at 30 and 90 days, using multiple imputation to account for missingness in comorbidities and socio‐economic status.
Acknowledgements
We recognise Aboriginal and Torres Strait Islander peoples as the traditional custodians of this Country. Reflecting more than 500 Aboriginal and Torres Strait Islander Nations, we use the term ‘Indigenous’ respectfully and with the knowledge that Indigenous language is as diverse as the communities this term refers to. We acknowledge the Victorian Department of Health as the source of Victorian Admitted Episodes Dataset and Victorian Emergency Minimum Dataset data for this study, the Victorian Department of Government Services as the source of the Victorian Death Index data, and the Centre for Victorian Data Linkage for the provision of data linkage. Aboriginal and Torres Strait Islander leadership, governance and engagement: This project followed the Maiam nayri Wingara Indigenous Data Sovereignty principles. The research team worked in partnership with Indigenous authors who guided contextualisation of the study findings. Open access publishing facilitated by Monash University, as part of the Wiley ‐ Monash University agreement via the Council of Australasian University Librarians.
Data Availability Statement
Data are available upon reasonable request from the corresponding author and will be subject to ethics committee and data custodian approval.
References
- 1. Stroke Foundation , “Top 10 Facts About Stroke,” [Internet], accessed 10 September 2026, https://strokefoundation.org.au/About‐Stroke/Learn/facts‐and‐figures.
- 2. Balabanski A. H., Nedkoff L., Brown A., et al., “Incidence of Stroke in the Aboriginal and Non‐Aboriginal Populations of Australia: A Data Linkage Study,” Stroke 54, no. 8 (2023): 2050–2058. [DOI] [PubMed] [Google Scholar]
- 3. Kilkenny M. F., Harris D. M., Ritchie E. A., Price C., and Cadilhac D. A., “Hospital Management and Outcomes of Stroke in Indigenous Australians: Evidence From the 2009 Acute Care National Stroke Audit,” International Journal of Stroke 8, no. 3 (2013): 164–171. [DOI] [PubMed] [Google Scholar]
- 4. Dos Santos K., Mohr M., Jude M., et al., “Prospective Analysis of Stroke Recognition, Stroke Risk Factors, Thrombolysis Rates and Outcomes in Indigenous Australians From a Large Rural Referral Hospital,” Internal Medicine Journal 52, no. 3 (2022): 468–473. [DOI] [PubMed] [Google Scholar]
- 5. Gardiner F. W., Rallah‐Baker K., Dos Santos A., et al., “Indigenous Australians Have a Greater Prevalence of Heart, Stroke, and Vascular Disease, Are Younger at Death, With Higher Hospitalisation and More Aeromedical Retrievals From Remote Regions,” EClinicalMedicine 42 (2021): 101181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Katzenellenbogen J. M., Vos T., Somerford P., Begg S., Semmens J. B., and Codde J. P., “Burden of Stroke in Indigenous Western Australians: A Study Using Data Linkage,” Stroke 42, no. 6 (2011): 1515–1521. [DOI] [PubMed] [Google Scholar]
- 7. Thompson S. C., Woods J. A., and Katzenellenbogen J. M., “The Quality of Indigenous Identification in Administrative Health Data in Australia: Insights From Studies Using Data Linkage,” BMC Medical Informatics and Decision Making 12 (2012): 133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Kennedy B., Howell S., and Breckell C., Indigenous Identification in Administrative Data Collections and the Implications for Reporting Indigenous Health Status. Technical Report No. 3. (Queensland Health, Health Statistics Centre, 2009). [Google Scholar]
- 9. Australian Bureau of Statistics , “Estimates of Aboriginal and Torres Strait Islander Australians,” 2021, [Internet] ABS, accessed 2024, March 10, https://www.abs.gov.au/statistics/people/aboriginal‐and‐torres‐strait‐islander‐peoples/estimates‐aboriginal‐and‐torres‐strait‐islander‐australians/latest‐release.
- 10. Zhao H., Coote S., Easton D., et al., “Melbourne Mobile Stroke Unit and Reperfusion Therapy: Greater Clinical Impact of Thrombectomy Than Thrombolysis,” Stroke 51, no. 3 (2020): 922–930. [DOI] [PubMed] [Google Scholar]
- 11. Ambulance Victoria , “Victorian Stroke Telemedicine,” [Internet], 2020, accessed March 2024, https://www.ambulance.vic.gov.au/victorian‐stroke‐telemedicine.
- 12. Andrew E., Cox S., and Smith K., “Linking Ambulance Records With Hospital and Death Index Data to Evaluate Patient Outcomes,” International Journal of General Medicine 15 (2022): 567–572. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Australian Bureau of Statistics , “Remoteness Areas,” [Internet], 2023, accessed February 28, 2024, https://www.abs.gov.au/statistics/standards/australian‐statistical‐geography‐standard‐asgs‐edition‐3/jul2021‐jun2026/remoteness‐structure/remoteness‐areas.
- 14. Australian Bureau of Statistics , “Census of Population and Housing: Socio‐Economic Indexes for Areas (SEIFA), Australia, 2016 [Internet],” 2018, accessed October 2023, https://www.abs.gov.au/ausstats/abs@.nsf/Lookup/by%20Subject/2033.0.55.001~2016~Main%20Features~IRSD~19.
- 15. Ambulance Victoria , “Clinical Practice Guidelines, 2022,” accessed July 2023, https://www.ambulance.vic.gov.au/clinical‐practice‐guidelines.
- 16. Safer Care Victoria , “Endovascular Clot Retrieval for Acute Stroke: Statewide Service Protocol,” [Internet], 2026, accessed September 2026, https://www.safercare.vic.gov.au/best‐practice‐improvement/clinical‐guidance/stroke‐clinical‐network/endovascular‐clot‐retrieval‐protocol.
- 17. Australian Bureau of Statistics , TableBuilder [Internet] (ABS, 2021), accessed November 2023, https://www.abs.gov.au/statistics/microdata‐tablebuilder/tablebuilder. [Google Scholar]
- 18. Balabanski A. H., Newbury J., Leyden J. M., et al., “Excess Stroke Incidence in Young Aboriginal People in South Australia: Pooled Results From Two Population‐Based Studies,” International Journal of Stroke 13, no. 8 (2018): 811–814. [DOI] [PubMed] [Google Scholar]
- 19. Buck B. H., Akhtar N., Alrohimi A., Khan K., and Shuaib A., “Stroke Mimics: Incidence, Aetiology, Clinical Features and Treatment,” Annals of Medicine 53, no. 1 (2021): 420–436. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Quigley A., Hutton J., Phillips G., et al., “Implicit Bias Towards Aboriginal and Torres Strait Islander Patients Within Australian Emergency Departments,” Emergency Medicine Australasia 33, no. 1 (2021): 9–18. [DOI] [PubMed] [Google Scholar]
- 21. Ikeme S., Kottenmeier E., Uzochukwu G., and Brinjikji W., “Evidence‐Based Disparities in Stroke Care Metrics and Outcomes in the United States: A Systematic Review,” Stroke 53, no. 3 (2022): 670–679. [DOI] [PubMed] [Google Scholar]
- 22. Bhattacharya P., Mada F., Salowich‐Palm L., et al., “Are Racial Disparities in Stroke Care Still Prevalent in Certified Stroke Centers?,” Journal of Stroke and Cerebrovascular Diseases: The Official Journal of National Stroke Association 22, no. 4 (2013): 383–388. [DOI] [PubMed] [Google Scholar]
- 23. Govindarajan P., Friedman B. T., Delgadillo J. Q., et al., “Race and Sex Disparities in Prehospital Recognition of Acute Stroke,” Academic Emergency Medicine 22, no. 3 (2015): 264–272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Coombes J., Hunter K., Bennett‐Brook K., et al., “Leave Events Among Aboriginal and Torres Strait Islander People: A Systematic Review,” BMC Public Health 22, no. 1 (2022): 1488. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Bevan H., Sharma K., and Bradley W., “Stroke in Young Adults,” Stroke 21, no. 3 (1990): 382–386. [DOI] [PubMed] [Google Scholar]
- 26. Australian Institute of Health and Welfare , “Aboriginal and Torres Strait Islander Health Performance Framework ‐ Summary Report,” [Internet], 2025, accessed June 2026, https://www.indigenoushpf.gov.au/report‐overview/overview/summary‐report.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: STROBE statement—checklist of items that should be included in reports of observational studies.
Table S2: Crude, age‐standardised and age‐specific incidence of EMS attendance for stroke among Indigenous and non‐Indigenous populations, per 100,000 person‐years.
Figure S1: Crude and age‐standardised incidence of 90‐day mortality following EMS attendance for stroke, per 100,000 person‐years, within Indigenous and non‐Indigenous populations.
Table S3: Multivariable analyses of 30‐ and 90‐day mortality.
Table S4: Baseline characteristics of Indigenous and non‐Indigenous patients, with Indigenous status classified according to individual hospital admission records.
Table S5: Pre‐hospital and hospital care of Indigenous and non‐Indigenous patients, with Indigenous status classified according to individual hospital admission records.
Table S6: Patient outcomes, with Indigenous status classified according to individual hospital admission records.
Table S7: Adjusted mortality outcomes at 30 and 90 days, using multiple imputation to account for missingness in comorbidities and socio‐economic status.
Data Availability Statement
Data are available upon reasonable request from the corresponding author and will be subject to ethics committee and data custodian approval.