Hospital management and outcomes of stroke in Indigenous Australians: evidence from the 2009 Acute Care National Stroke Audit


  • Conflict of interest: None declared.
  • Funding: The National Stroke Foundation is a not-for-profit non-government organization that funds the Audit program.

Correspondence: Monique F. Kilkenny, Translational Public Health Unit, Stroke and Ageing Research Centre (STARC), Department of Medicine, Monash Medical Centre, Southern Clinical School, Monash University, Level 1/43-51 Kanooka Grove, Clayton Vic 3168, Australia.




Indigenous Australians have greater stroke mortality rates than non-Indigenous people (97% Indigenous Australians are aged <65 years). Quality of care in hospital is an important factor for avoiding death and disability. No national review of acute stroke care for Indigenous Australians is available.


We aimed to compare adherence with clinical processes and outcomes among Indigenous and non-Indigenous patients with acute stroke admitted to hospital.


Hospitals participating in the National Stroke Audit of acute services in 2009 that provided data for at least one Indigenous patient were included (n = 33, 37%). Differences in death/dependency (modified Rankin Score 3–6) at discharge were determined using two-level multivariate analysis adjusting for hospital site and patient variables. A matched subgroup analysis in those aged 18 to 64 years was also undertaken.


Among 1162 eligible patients (60% male; 18–64 years n = 305), 7% were Indigenous (18–64 years: 18%). Indigenous patients had a greater prevalence of stroke risk factors, e.g. diabetes, more had intracerebral hemorrhages (25% vs. 16% non-Indigenous, P = 0·05), and were less likely be treated in a stroke unit and receive timely allied health assessments than non-Indigenous patients. Indigenous patients aged 18–64 years had a threefold odds of dying or being dependent at discharge (Adjusted odds ratio = 3·09, 95% confidence interval = 1·07–8·95).


Australian Indigenous patients with stroke received a reduced quality of care in hospitals and experienced worse outcomes than non-Indigenous patients. Indigenous patients require the provision of evidence-based care to increase their opportunities for optimal health outcomes following stroke. Further research to explain the differences is needed.


Stroke is Australia's second largest cause of death [1] and a leading cause of adult disability [2]. Indigenous Australians are distinguished as either Aboriginal and/or Torres Strait Islanders (ATSI). In 2006, the numbers of Indigenous people living in Australia was estimated to be 517 200, or 2·5% of the total Australian population [3]. The burden of stroke from fatal and nonfatal events is much greater in the Indigenous Australian population [4]. The hospital admission rate for stroke among Indigenous Australians is about 1·5 times greater than in non-Indigenous Australians [5]. Deaths attributed to stroke among Indigenous Australians are about 1·9 times greater than for non-Indigenous Australians, and the largest disparity in mortality occurs in Indigenous people aged less than 65 years [4, 5]. For example, among Indigenous people aged 35–54 years, the mortality rate associated with cerebrovascular disease is about five times greater than in non-Indigenous people [5].

The quality of hospital care has been shown to be an important factor in reducing death and disability after stroke [6-8]. Importantly, treatment in stroke units, where specialized care is provided in the one hospital location by expert health professionals, ensures that a greater number of clinical processes are adhered to [9] and is associated with a 22% reduction in death and disability [8]. Despite the disparities in survival and differences in hospitalizations, there have been few studies designed to review variations in hospital management and the associated affects on health outcomes for Indigenous people. Previously, there has been a review of Indigenous patients admitted with stroke (1988–1992) in Perth metropolitan teaching hospitals [10] and an audit of stroke among Indigenous and non-Indigenous patients (2001–2002) at the Royal Darwin Hospital [11]. The authors of these studies reported an increased prevalence of stroke risk factors such as diabetes, smoking, and alcohol consumption in Indigenous patients compared with non-Indigenous patients [11]. Pepper et al. reported a reduced use of evidenced-based medications for ischemic stroke in Indigenous patients [11]. These previous, single-site studies had small sample sizes of less than 90 patients, and the findings may not be generalizable to other settings. Since the quality of hospital care may play an important role in the outcome of stroke, studies designed to quantify differences in clinical management in hospitals provided to Indigenous and non-Indigenous Australians are needed [12].

A unique opportunity to review hospital management and outcomes of Indigenous patients with stroke at a national level was available using data from the 2009 National Stroke Audit of acute hospital services [13].


We aimed to compare the adherence to recommended clinical processes of care and the health outcomes of Indigenous and non-Indigenous patients admitted to hospital with acute stroke for all ages and among those aged 18–64 years.


Data from the Australian National Stroke Audit: Acute Services Clinical Audit 2009 were used [13]. A nested sample of audit data from hospitals that had treated at least one Indigenous patient was selected to reduce heterogeneity by hospitals. Therefore, these data provide a good representation from sites that treat both Indigenous and non-Indigenous patients. In addition, since over 97% of the Indigenous population are aged less than 65 years [3], a component of the analyses was restricted to patients audited who were aged 18–64 years (referred to here as the younger subgroup).

Methods for the National Stroke Audit Program have been previously published [14]. Trained data abstractors at each participating hospital, using a web-based data entry tool, conducted at least 40 consecutive medical records audits. Data included demographic characteristics (including age, gender, and Indigenous status), history of risk factors (including diabetes, hypertension, etc.), process of care indicators (such as stroke unit (SU) care, aspirin within 48 h, and assessment by allied health), and health outcomes at discharge. Process of care indicators were designed to measure adherence to clinical practice recommendations in the 2007 Acute Clinical Guidelines for Stroke [15]. In this study, we included the subset of clinical indicators identified as reliable for representing the full suite of indicators in the measurement of the quality of stroke care and predicting outcome in Australia (n = 14) [16]. Stroke outcomes were measured as health status at time of discharge using the modified Rankin Score (mRS), where 0 is no symptoms and 6 means death. Patients were classified as being ‘Indigenous’ or ‘non-Indigenous’ based on the ATSI status question routinely asked on admission in all hospitals and recorded on the admission sheet. In the 2009 Acute Stroke Audit, interrater reliability was used to measure agreement among auditors for 222 cases. The level of agreement for recording of Indigenous status was substantial (kappa 0·79, 95% confidence interval 0·43–0·94).

Statistical analysis

Data were analyzed using STATA software (version 10.1, Stata Corporation PL). Chi-square tests were used for categorical variables, Fisher's exact test for dichotomous variables, and the Wilcoxon Mann–Whitney Rank Sum test for continuous variables. Bivariate and multivariate logistic regression models were used to predict the probability of stroke outcomes. Demographic variables found to differ between the Indigenous and non-Indigenous patients and type of stroke were included in multivariable models. In addition, adjustment for patient case mix to account for stroke severity was based on a validated prognostic model for comparing patient outcomes [17]. These variables included presence of an arm deficit, speech impairment, incontinence within 72 h, and ability to walk unaided on admission. Adjustments were also made for patient clustering within hospitals. Since certain types of patients may cluster within particular settings or geographical locations and are often more likely to respond in a similar manner, these samples cannot be assumed to act independently [18]. Therefore, statistical adjustment for patient clustering within hospitals, in addition to individual patient factors, provides a further measure against overstating differences in outcomes between Indigenous and non-Indigenous groups. We present adjusted odds ratios and 95% CIs.


Audit sample

Figure 1 shows there were 96 Australian hospitals that participated in the Clinical Audit component of the 2009 National Stroke Audit and 34% of hospitals had audited at least one Indigenous patient with stroke. Of the total 2009 audit sample of 3307 patients, 80 (2·41%) were identified as Indigenous. Selecting only the hospitals that had audited at least one Indigenous patient the sample decreased to 1162. The median number of Indigenous patient audits per hospital was two (interquartile range: 1–3; minimum 1, maximum 14). Figure 2 shows that these patients were from hospitals located in all states and territories of Australia. Two-thirds of Indigenous patients admitted with stroke were aged less than 65 years (Fig. 3). Restricting the patients to just those aged 18 to 64 years, the 33 hospitals provided data for 305 patients, whereby 18% were Indigenous.

Figure 1.

Number of patients and hospitals included for Indigenous patient analyses.

Figure 2.

Distribution of hospitals treating at least one Indigenous patient in the 2009 National Stroke Audit. New South Wales includes Australian Capital Territory; SU, stroke unit; ATSI: Aboriginal & Torres Strait Islander.

Figure 3.

Age and gender distributions by Indigenous status; ATSI: Aboriginal & Torres Strait Islander.

Baseline characteristics

Table 1 shows baseline demographic characteristics, risk factors, type of stroke, and level of stroke severity between the Indigenous and non-Indigenous patients treated at the eligible hospitals for all age groups and also for the younger age group. The samples were similar for gender, history of high cholesterol, history of hypertension, and level of independence prior to stroke. However, the Indigenous patients were significantly younger (mean age 59 (standard deviation, SD, 15) versus 74 (SD 13) years non-Indigenous), including when the sample was restricted to those aged 18–64 years. The Indigenous patients were also more likely to have diabetes, report being a smoker (current or past), and have a history of high-risk alcohol consumption as compared with the non-Indigenous patients. The Indigenous patients were also more likely to be admitted with a hemorrhagic stroke. In the younger sub-subgroup, there were more Indigenous patients that were incontinent in the first 72 h of stroke onset than non-Indigenous patients (P = 0·04).

Table 1. Selected characteristics of Indigenous patients and non-Indigenous patients with stroke admitted to Australian hospitals participating in the 2009 audit
CharacteristicsAll ages P-valueAge 18–64 years P-value


n = 80

n (%)


n = 1,082

n (%)


n = 53

n (%)


n = 243

n (%)

  1. aModified Rankin score (mRS) 0–1; ATSI: Aboriginal or Torres Strait Islander.
  2. SD, standard deviation; IQR, interquartile range.
Age mean (SD)59 (15)74 (13)<0·00150 (10)54 (9)0·004
Female33 (41)511 (47)0·3020 (38)88 (36)0·84
Independent prior to strokea 49 (69)643 (66)0·5934 (72)182 (80)0·23
Hypercholesterolemia35 (54)413 (48)0·3624 (55)92 (45)0·24
Hypertension53 (72)737 (75)0·5729 (60)135 (61)0·90
Diabetes mellitus37 (49)240 (27)<0·00126 (51)50 (23)<0·001
Unknown type of stroke7 (9)70 (7)0·466 (11)10 (4)0·04
Hemorrhagic stroke20 (25)173 (16)0·0517 (32)43 (18)0·02
Ischemic stroke53 (66)819 (77)0·0330 (57)188 (78)0·001
Partial anterior circulation infarct15 (19)228 (22) 9 (17)51 (21) 
Lacune infarct7 (9)185 (17) 2 (4)41 (18) 
Posterior circulation infarct2 (3)101 (10) 0 (0)22 (9) 
Total anterior circulation infarct4 (5)100 (10) 3 (6)24 (10) 
Unknown25 (31)205 (19) 16 (30)10 (4) 
Current or past smoker48 (74)394 (46)<0·00134 (79)139 (64)0·05
High alcohol consumption27 (34)91 (8)<0·00123 (51)42 (23)<0·001
Impaired speech on admission41 (58)693 (70)0·0425 (54)142 (64)0·22
Weak arm on admission51 (64)748 (69)0·3130 (64)163 (70)0·43
Able to walk on admission28 (37)289 (28)0·0722 (46)114 (49)0·69
Incontinent <72 h of stroke onset35 (46)482 (47)0·8620 (40)59 (26)0·04
Length of stay: median (IQR)7 (4–12)7 (3–14)0·255 (2–10)5 (2–9)0·74

Hospital management

Overall, there were several differences in clinical care provided in-hospital between the Indigenous and non-Indigenous patients admitted with acute stroke (Fig. 4). Over half of the patients (59%) were admitted to a hospital with a SU. However, the non-Indigenous patients were more likely to receive care in an SU than the Indigenous patients. The non-Indigenous patients with an ischemic stroke were also more likely to receive aspirin within 48 h and more likely to be discharged on antithrombotic medication than the Indigenous patients. In addition, the non-Indigenous patients were more likely to receive assessments by allied health within 48 h of admission compared with Indigenous patients. No Indigenous patient received intravenous thrombolysis. However, the Indigenous patients were more likely to receive a mood assessment during admission compared with the non-Indigenous patients. Similar patterns of hospital management (received stroke unit care, aspirin within 48 h, mood assessment and assessed by physiotherapy within 48 h) were found for the 18- to 64-year age group comparisons (Fig. 5).

Figure 4.

Clinical management for Indigenous patients and non-Indigenous patients with stroke (all ages). *P < 0·05; **P = 0·06.

Figure 5.

Clinical management for Indigenous patients and non-Indigenous patients with stroke if aged 18–64 years. *P < 0·05.

Health outcomes

Overall, in-hospital mortality was 14% (n = 163; Indigenous 13% and non-Indigenous 14%) and patients dependent at time of discharge (mRS 3–5) was 72% for the Indigenous and 67% for the non-Indigenous patients. These differences for the whole cohort were not statistically significant. However, for the younger subgroup, in-hospital mortality was 7% (n = 22; Indigenous 13% and non-Indigenous 6%, P = 0·077) and patients dependent at time of discharge (mRS 3–5) was 67% for the Indigenous and 45% for the non-Indigenous patients (P = 0·016). These findings for worse health outcomes (mRS 3–6) in the younger subgroup for the Indigenous patients remained significant after adjusting for hospital site, age, risk factors (e.g. diabetes, smoking, alcohol), stroke severity variables (speech impairment, incontinent within 72 h), stroke subtype, and access to a SU (Table 2).

Table 2. Factors associated with death or dependency at discharge
Outcome at discharge (died or dependent – mRS 3–6)All agesAge 18–64 years
Adjusted odds ratioa (95% CI)Adjusted odds ratioa (95% CI)
  1. aAdjusted for all factors listed in table including clustering of patients within hospitals.
  2. bDiabetes mellitus, current or past smoker, or high alcohol consumption.
Indigenous2·32 (0·85–6·36)3·09 (1·07–8·95)
Age1·03 (1·01–1·05)0·97 (0·94–1·01)
Hemorrhagic stroke1·29 (0·68–2·49)1·80 (0·57–5·78)
One or more vascular risk factorsb1·09 (0·69–1·71)0·96 (0·57–1·61)
Unable to walk on admission4·56 (3·17–6·54)7·74 (3·56–16·78)
Impaired speech on admission2·46 (1·49–4·04)4·04 (1·81–9·00)
Incontinent <72 h of stroke5·08 (2·74–9·42)2·61 (0·81–8·45)
Weak arm on admission1·07 (0·67–1·72)1·16 (0·48–2·77)
Treatment on a stroke unit0·53 (0·32–0·92)0·51 (0·26–1·01)

Predictors for death or dependency at discharge

The predictors of patients dying or being dependent for all ages included advancing age and stroke severity factors, such as being unable to walk on admission, having impaired speech, or being incontinent within 72 h of stroke (Table 2). Access to an SU was associated with reduced death and disability. There were insufficient data to determine the predictors for death or dependency at discharge by Indigenous status.


For the first time, we present data from hospitals across Australia that were used to compare the care provided to Indigenous and non-Indigenous patients managed in the same hospitals. One of the study's strengths is that it provides a unique sample of Indigenous patients from 33 hospitals located in all states and territories of Australia. Although a limitation of this study may be considered the sample size (Indigenous n = 80 and non-Indigenous n = 1182), this is a large sample of Indigenous patients relative to other similar studies in this area. Previous published audits have focused only on one hospital or there was no comparison made with non-Indigenous patients [10, 11]. We present new and important data to show that the quality of clinical care within hospitals may be different based on Indigenous status, and this influences health outcomes. Clinicians and health administrators are able to consider these findings and review what practices or polices might be modified within health services to ensure the same quality of care for all patients admitted with stroke.

We acknowledge that there may be several additional potential weaknesses of this study. For example, there were small numbers of Indigenous patients per hospital (median of two patients), but this actually reflects the known distribution of the Indigenous population in Australia. In addition, level of dependency at discharge, when assessed as an outcome measure, may be confounded by the length of stay. Since there were no statistically significant differences in lengths of stay between the Indigenous and non-Indigenous patients, we did not include this variable in the logistic regression models. Socioeconomic status may also be a confounding factor in relation to stroke outcome [19]; however, we are unable to report the influence of this factor because this is not recorded in the audit.

We found that Indigenous patients admitted with stroke were younger and had a greater prevalence of stroke risk factors than non-Indigenous patients. These findings are consistent with other Australian and overseas studies [10, 11, 20, 21]. For example, a greater proportion of Indigenous patients are admitted with diabetes (present study 49%, Crowley et al. reported 41%, and Pepper et al. 44%) and history of cigarette smoking is also consistently reported in the majority of Indigenous patients with stroke (present study 74%, Crowley et al. 67%, and Pepper et al. 64%) [10, 11]. However, alcohol consumption tends to differ (present study 34%, Crowley et al. 49%, and Pepper et al. 22%) [10, 11] and may reflect temporal changes from when the data were collected or the fact that with small samples from specific geographic locations, the data from single-center studies may not be representative for Australia. Nevertheless, management of risk factors and prevention of diabetes continue to be an area that needs targeted health programs specifically for Indigenous people in Australia. Greater comorbidity among Indigenous patients demands the provision of evidence-based care to increase the opportunities for optimal health outcomes following stroke (and to prevent avoidable strokes occurring in the first place).

An interesting finding was that Indigenous patients were more likely to have a hemorrhagic stroke compared with non-Indigenous patients. The known risk factors for hemorrhagic stroke include hypertension, smoking, and excessive alcohol consumption [22]. In our study, the Indigenous patients had a greater prevalence of these risk factors compared with the non-Indigenous patients. The numbers are small for hemorrhagic stroke, but further investigation is needed in a larger population cohort.

Our data were used to show that there were fewer adherences to most recommended processes of care for Indigenous patients within the same hospitals that treated both Indigenous and non-Indigenous patients. Indigenous patients were less likely to receive SU care than non-Indigenous patients, even though there is level 1 evidence [8, 9, 23] that conclusively shows that receiving these processes of care improves health outcomes [15, 23]. Indigenous Australians were also less likely to be receiving timely assessments by allied health professionals than non-Indigenous Australians [24]. Importantly, no Indigenous patients received intravenous thrombolysis, which is an effective treatment for reducing death and disability from ischemic stroke [25]. Further evaluation of this disparity based on Indigenous status in hospitals is urgently needed.

It has been shown that Indigenous Australians are estimated to suffer a mortality disadvantage compared with non-Indigenous Australians [26], and the greatest disparity in mortality occurs in the younger age groups [5]. Using multivariable logistic regression modeling with adjustment for patient clustering within hospitals and stroke severity factors, we have shown that Indigenous patients in the younger subgroup (18–64 years) tended to have three times the chance of dying or being dependent at discharge compared with non-Indigenous Australians. This is consistent with recent data from a large sample using linked health data from Western Australia [4]. It is relevant that our findings demonstrate poorer compliance with evidence-based recommendations, known to improve health outcomes, for Indigenous patients when compared with non-Indigenous patients. Future research may be required to confirm this finding using a larger Indigenous sample.

The interrater reliability results show that we have substantial agreement between auditors for recording of Indigenous status, but we have no data to show how accurate these data represent the number of Indigenous patients admitted with stroke since Indigenous status is often underreported [27]. Nevertheless, our Indigenous sample represented 2·4% of the total audit sample, which is consistent with representation reported for the Australian population (approximately 2·5%) [12]. It is important to acknowledge in light of the aforementioned findings that it has been shown that the quality of routinely collected data on Indigenous status is poor [5]. There are different methods of data collection in hospitals and by region for identifying a person's Indigenous status at the time of hospital admission or for completion of death certificates. Moreover, there are uncertainties about the number of people who identify themselves as Indigenous Australians. In this study, 3% of the patients (n = 9) had their Indigenous status missing. We accept that underreporting of Indigenous status may be a source of underestimation in the prevalence of risk factors or health outcomes at discharge. Conversely, there may also have been an overestimation of the gaps in clinical care. However, reporting bias for these variables should have been similar for those recognized as Indigenous or non-Indigenous within each of the hospitals. The Australian Institute of Health and Welfare have developed national best practice guidelines for collecting Indigenous status in health data sets, and by December 2012, all states and territories will adopt these guidelines [28]. These guidelines will ensure a greater systematic approach to recording Indigenous status to reduce the likelihood of this form of bias.

In 2004, the National Strategy for Heart, Stroke and Vascular Health Strategies Group identified some actions to reduce the impact of stroke among Indigenous Australians [29]. These actions included providing a primary health care program, improving management of risk factors, and training and support of health-care providers. It is unclear whether these actions have been successfully undertaken since this strategy was published. Our clinical audit data provide evidence that further work is needed to improve hospital management and outcomes of Indigenous Australians with stroke. There continues to be disparities in health-care delivery and health outcomes following stroke between Indigenous and non-Indigenous Australians, which requires urgent action. Our data provide complimentary evidence to the recently published study on the burden of stroke in Western Australia whereby the burden profile between Indigenous and non-Indigenous stroke was substantial (approximately three times greater for age-standardized rates in both sexes than corresponding non-Indigenous rates) with rate ratios largest at younger ages [4]. Further research to explain these differences are needed including prospective cohort studies with longer term follow-up. This may also include trials to test interventions aimed at reducing health care and outcome inequalities. Future audits undertaken by the National Stroke Foundation may also be designed to oversample in hospitals that treat Indigenous patients in order to increase the sample size.

In conclusion, the health disparities of ATSI demand special attention and resources [5]. As highlighted by Hill et al., the health gap that is associated with the increased risk of mortality if an Indigenous Australian becomes ill may be influenced by a combination of factors such as late presentations, shortcomings in acute management, and poor follow-up during the course of the disease [26]. Our data provide evidence to support these assertions in the area of acute stroke management. Initiatives to increase the equitable access to evidence-based stroke care services in Australia are important if we are to reduce the burden of this disease in Indigenous people.