This was a prospective cohort study of PAD participants recruited from an outpatient clinic in Townsville Hospital (Ethics [HREC/13/QTHS/125; SSA/13/QTHS/203; HREC/14/QTHS/203; H6028 & H5537] [13, 14]. A 5-member Indigenous reference committee was consulted for approval of the research. The recruitment period for this cohort was between May 2003 and February 2016. Patients who provided written informed consent and met the following eligibility criteria were included: diagnosis of PAD by a vascular specialist as previously described, [13] including any of the following presentations: intermittent claudication, critical limb ischemia, asymptomatic or symptomatic carotid stenosis, and aortic or peripheral aneurysms. Patients who identified themselves as Aboriginal and/or Torres Strait Islanders were considered Indigenous. Patients who did not identify as Indigenous were considered non-Indigenous.

Clinical characteristics collected for each participant included: presenting complaint, age, sex, diabetes, hypertension, smoking history, dyslipidaemia, renal function (estimated glomerular filtration rate [eGFR]), ischaemic heart disease (IHD), history of stroke, and current medications. Hypertension and diabetes were defined by history or treatment for these conditions. Cigarette smoking classification was based on smoking history, and defined as current smoker (smoked within the last month), previous smoker or never smoked. IHD was defined by a history of myocardial infarction, angina or coronary revascularization. Serum creatinine was measured using a spectrophotometry method in a pathology laboratory as previously described [1]. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration formula, as this has previously been shown to best predict complications [1]. Abdominal aortic aneurysm (AAA) was defined as an orthogonal infra-renal aortic diameter of at least 30 mm on ultrasound or computed tomography angiography [13, 14]. A significant carotid artery stenosis was defined as ≥50% using Australian Society for Ultrasound in Medicine criteria [13, 14]. Current prescriptions for aspirin, other antiplatelet agents, warfarin, beta-blockers, calcium channel blockers, angiotensin pathway inhibitors, metformin, insulin, frusemide and statins were recorded. Follow-up data was retrieved from out-patient visits, hospital chart reviews, linked data and /or inpatient admissions as previously described [13, 14]. The primary outcome was to evaluate the differences in the combined incidence of major cardiovascular events (MI, stroke or death) between Indigenous and non-Indigenous Australians with PAD. The diagnoses of MI and stroke were made by Royal Australasian College of Physicians accredited physician in accordance with international guidelines [15, 16]. Death was defined to include deaths from all causes. Linked data were based on International Classification of Diseases (ICD)-10 coding for MI (I21.0-I22.9) and stroke (I60.0-I64.0). Patients were censored at the first major event, or at the date of last review or linked data request, if they did not experience an event.

The required sample size was estimated according to a planned Cox regression analysis, examining the association of Indigenous status with major cardiovascular events. On the basis of previous PAD studies conducted at The Townsville Hospital, the incidence of major cardiovascular events was estimated to be 30% over 3 years, [17] and the model was designed to include the following variables: age, sex, IHD, diabetes, hypertension, smoking and Indigenous status. It was therefore estimated that approximately 230 patients in total would be needed to be included in order to attain 10 outcome events per degree of freedom for each predictor variable to be included in the Cox regression analysis [18].

The presenting symptoms and risk factors were compared between Indigenous and non-Indigenous patients using Pearson’s χ2 test and Mann Whitney U test. The combined incidence of MI, stroke or death in Indigenous and non-Indigenous groups was assessed with Kaplan-Meier analysis with differences compared using the log rank test. Multivariate Cox proportional hazards analyses were undertaken to assess the association between Indigenous status and cardiovascular events, adjusting for relevant confounding risk factors and medications (hypertension, current smoking, diabetes, IHD, age, critical limb ischemia, insulin, frusemide, angiotensin converting enzyme [ACE] inhibitor, anti-platelet or anti-coagulant prescription) across 3 different models. The selection of variables to adjust for was based on those which have been established as risk factors for PAD outcome, or were found to be significantly different between Indigenous and non-Indigenous patients. No violations of the proportional hazards assumption were observed. All analyses were performed using STATA version 14.1 (StataCorp, College Station, Texas, USA), and SPSS version 20.0 (IBM SPSS Inc., Chicago, Illinois, USA).

Four hundred and one patients with PAD were included, of which 16 (4.0%) were Indigenous and 385 (96.0%) were non-Indigenous Australians. Indigenous participants were younger at entry (median age 63.3 [54.7–67.8] vs 69.6 [63.3–75.4], p = 0.005), were more likely to be current smokers (56.3% vs 31.4%, p = 0.041) and have insulin-treated diabetes (18.8% vs 5.2%, p = 0.022). Indigenous participants were also more likely to be prescribed frusemide for heart failure (Table 1). A lower proportion of Indigenous Australians were prescribed a statin, although this difference was not significant. There were no significant differences in presenting complaint at entry between Indigenous and non-Indigenous Australians (Table 2).

Median (inter-quartile range) follow-up in Indigenous patients and non-Indigenous patients was 2.5 (0.3–5.3) and 2.5 (0.4–5.4) years, respectively (p = 0.973). In total, five major cardiovascular events (MI, stroke, or death) were recorded in Indigenous Australians, and 45 major cardiovascular events were recorded in non-Indigenous Australians. Kaplan Meier curves illustrating the cumulative proportion of Indigenous and non-Indigenous patients who had major cardiovascular events are presented in Fig. 1. Differences between both groups were considered to be significant (log-rank test p = 0.004). By unadjusted Cox proportional hazard analysis, Indigenous Australians had a 3.59-fold higher risk of major cardiovascular events during follow-up compared to non-Indigenous Australians (p = 0.007; Table 3). By multivariate Cox proportional hazard analysis, Indigenous Australians were at ~ 5-fold greater risk of major cardiovascular events (HR 4.72 [1.41–15.78], p = 0.012), after adjusting for hypertension, current smoking, diabetes, IHD, age, critical limb ischemia, insulin, frusemide, ACE inhibitor, anti-platelet or anti-coagulant prescription (Table 3).