Geographic remoteness and chronic kidney disease outcomes: a population-based cohort study using linked pathology data

Abstract

Background

Evidence on inequities in long-term chronic kidney disease (CKD) outcomes from population-based cohorts is limited, and geographic inequities across the full spectrum of CKD in Australia remain under-examined. We investigated how clinical outcomes and healthcare utilisation vary by geographic remoteness in Western Australia (WA), leveraging a state-wide linked pathology dataset spanning over 15 years.

Methods

A cohort study in WA of adults (≥ 18 years) with incident CKD (2010–2017) identified from pathology data linked to hospital, emergency department (ED), and death records. CKD was defined by two estimated glomerular filtration rate (eGFR) measurements, 3–12 months apart, within the same CKD stage (3a: 45–59, 3b: 30–44, or 4: 15–29 mL/min/1.73 m²). Clinical outcomes included early CKD presentation (stage 3a), CKD progression (to a higher CKD stage or kidney failure based on eGFR), kidney replacement therapy (KRT), and all-cause mortality. Healthcare utilisation outcomes included 3-year frequency of hospital separations (completed admitted hospital stays), ED presentations, and serum creatinine testing. Associations with geographic remoteness were examined using multivariable logistic, Cox proportional hazards, and negative binomial regression, adjusting for age, sex, comorbidities, and CKD stage at cohort entry.

Results

Among 78,244 individuals (58,845 major city; 15,990 regional; 3,409 remote), remote residents had lower odds of early presentation (adjusted odds ratio [aOR], 0.87; 95% confidence interval [CI], 0.79–0.95) and higher hazards of progression (adjusted hazard ratio [aHR], 1.15; 95% CI, 1.09–1.22), KRT (aHR, 1.94; 95% CI, 1.71–2.20), and mortality (aHR, 1.12; 95% CI, 1.06–1.19) compared with major city residents. Differences were most pronounced among those aged 18–60 years (early presentation: aOR, 0.76; 95% CI, 0.64–0.89; mortality: aHR, 1.31; 95% CI, 1.15–1.49), and 18–60 years with diabetes (progression: aHR, 1.88; 95% CI, 1.70–2.09; KRT: aHR, 2.67; 95% CI, 2.27–3.14). Healthcare utilisation was particularly elevated in the remote 18–60-year group with diabetes (ED presentations: adjusted incidence rate ratio [aIRR], 3.01; 95% CI, 2.73–3.31; hospitalisations: aIRR, 2.29; 95% CI, 2.10–2.50; serum creatinine testing: aIRR, 1.16; 95% CI, 1.07–1.26). Regional residents had comparable clinical outcomes to major city residents.

Conclusions

Population-level data from WA shows significant geographic disparities in CKD outcomes. Determining and addressing drivers of these disparities, especially among remote adults aged 18–60 years with diabetes, should be a priority for equitable kidney health policy and service planning.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12882-026-04998-8.

Background

Chronic kidney disease (CKD) is a major and growing global public health concern. Analysis of the Global Burden of Disease Study 2021 estimated 637.7 million people had CKD in 2021, a 92% increase since 1990 [1, 2]. Consistent with this, the most recent International Society of Nephrology Global Kidney Health Atlas reported CKD affects approximately 10% of the global population [3, 4].

CKD often remains undiagnosed at early stages due to its asymptomatic nature and diagnostic requirement for repeated abnormal test results more than 3 months apart [5, 6]. Its frequent comorbidity with other chronic conditions such as cardiovascular disease and diabetes [7] may also shift clinical priorities, contributing to underdiagnosis and affecting ongoing management.

Equity in CKD healthcare is an important consideration in geographically dispersed countries like Australia. Approximately 28% of Australians reside in regional and remote areas [8, 9]. These populations have a higher prevalence of health risk factors and often face barriers to accessing healthcare, including reduced availability of specialist services, primary care, and coordinated care [9]. Consequently, they often experience poorer health outcomes than those in metropolitan areas, including higher rates of hospitalisation and death [9].

International research demonstrates regional inequities in the burden of CKD influenced by socio-economic factors and healthcare access [2, 4]. However, few population studies have examined inequities longitudinally across the full clinical course of CKD from early stage through to kidney failure [10]. In Australia, reporting on CKD has relied on hospital data, dialysis and transplant registries, screening programs, or population health surveys [11, 12]. While valuable, these data sources are biased towards people with later-stage disease or based on single measurements that are insufficient to diagnose CKD.

Spanning over 2.5 million square kilometres, Western Australia (WA) is the largest geographic area in the world covered by a single health authority [13]. Its vast size and low population density presents unique challenges to the delivery of equitable and effective kidney healthcare. Between 2010 and 2020, the age-sex standardised prevalence of CKD in WA increased from 4.7% to 6.0%, with rates higher in regional and remote areas compared with major cities [14]. These findings were derived from a novel, state-wide linked pathology dataset that enables population-level monitoring of CKD across the full spectrum of the disease. This study aimed to use this dataset to examine how clinical outcomes and healthcare utilisation varies among people with CKD in WA according to geographic remoteness, to better understand disparities in kidney health outcomes and care. We acknowledge that Aboriginal people are over-represented in remote and chronic disease cohorts; however, ethnicity variables were unavailable in the linked dataset and therefore could not be examined in this study.

Methods

A cohort study using linked pathology data was conducted, reported in accordance with RECORD guidelines [15, 16].

Data

We used the de-identified CKD.WA dataset which links pathology, hospital, emergency department (ED), and death records via Privacy Preserving Record Linkage (PPRL) methodologies [17–19]. It includes all adults tested for serum creatinine at four pathology providers, PathWest, Western Diagnostic, Clinipath, and Australian Clinical Labs (ACL), from 2006 to 2022 (2.7 million individuals), with data from one provider only available from 2017 onwards. Together, these providers cover nearly the entire WA pathology market [20]. Three providers supplied additional test results including blood glucose and haemoglobin A1C (HbA1c), if ordered with serum creatinine. State hospital, ED, and mortality data were sourced from the WA Department of Health Hospital Morbidity Data Collection, Emergency Department Data Collection, and Cause of Death Unit Record File, respectively. Demographic information was available from all data sources. Ethnicity variables were unavailable, precluding analysis of Aboriginal subgroups.

Data cleaning was performed in Python (version 3.11.4) [21] to prepare the dataset for cohort selection and analysis, including deduplication, implausible or invalid value exclusion, and categorical variable recoding for consistency across datasets.

Study cohort

The study cohort included individuals with two estimated glomerular filtration rate (eGFR) measurements, taken 3–12 months apart, within the same CKD stage: 3a (45–59 mL/min/1.73 m²), 3b (30–44 mL/min/1.73 m²), or 4 (15–29 mL/min/1.73 m²). eGFR was calculated from serum creatinine using the CKD-EPI formula [22]. Individuals were classified as having CKD if they met this diagnostic criterion during the study period. An individual’s baseline was defined as the date of the second eGFR measurement in the first qualifying test pair confirming CKD. Individuals entered the cohort at their respective baseline date. To ensure inclusion of incident cases from 2010 onwards, a look-back period (2006–2009) was applied excluding individuals with prior qualifying tests. To allow at least 4 years of follow-up, 2017 was the final cohort entry year. Individuals were included who at baseline were aged 18–90 years and had a WA residential postcode that mapped to an Accessibility/Remoteness Index of Australia (ARIA+) remoteness category [23]. Individuals were excluded if prior to baseline they received dialysis or a kidney transplant based on clinical codes in hospital and ED data available from 2002 onwards (Supplementary Table 1), or had evidence of kidney failure from pathology data, defined as two eGFR measurements 3–12 months apart < 15 mL/min/1.73 m².

Remoteness

Baseline residential postcode was mapped to ARIA+ remoteness area which classifies Australia into five remoteness classes [23]. For statistical power and interpretability these were aggregated into: (1) major cities, (2) regional (inner and outer regional), and (3) remote (remote and very remote).

Covariates

Covariates included age at baseline, sex, CKD stage at baseline (3a, 3b or 4), and baseline comorbidities of diabetes, cardiovascular disease, and cancer. Age in years was grouped into 18–60, 61–75, and 76–90 based on sample size distribution and key health transitions. Sex was assigned using the most frequently recorded value. Comorbidities were identified using diagnostic and procedure codes from hospital and ED data (Supplementary Table 1). Diabetes status was additionally determined using pathology data, defined as two abnormal test results above diagnostic thresholds (HbA1c ≥ 6.5%, fasting glucose ≥ 7.0 mmol/L, or random glucose ≥ 11.1 mmol/L) [24], with the second test date taken as the qualification date.

Outcomes

Outcomes included early CKD presentation, time in days to CKD progression, time in days to kidney replacement therapy (KRT), time in months to death, and healthcare utilisation. Early presentation was defined as stage 3a CKD at baseline (yes/no). CKD progression was defined using eGFR as transition to a more advanced stage than at baseline or to kidney failure. This was confirmed by two eGFR measurements taken 3–12 months apart: within the same CKD stage for stage progression, or both < 15 mL/min/1.73 m² for kidney failure. KRT (dialysis or transplantation) was identified from hospital and ED clinical codes (Supplementary Table 1). All-cause mortality was obtained from state death records (month of death only). Healthcare utilisation was assessed over 3 years following baseline, including the total number of hospital separations, ED presentations, and serum creatinine tests. Hospital separations were defined as completed admitted hospital stays with separations ≤ 1 day apart combined as a single episode of care.

Statistical analysis

Individuals’ characteristics, all categorical, were summarised using descriptive statistics and are presented as counts and percentages (n, %) stratified by remoteness. Chi-squared (χ²) tests were used to identify associations between remoteness and other covariates, and outcome variables.

Logistic regression was used to examine the outcome of early CKD presentation. Cox proportional hazards regression was used to examine progression, KRT, and all-cause mortality outcomes. For progression and KRT, censoring occurred at death or study end; for all-cause mortality, censoring occurred at study end. Time to event for mortality models was measured in months rather than days as only month of death was available; a small offset was applied for deaths occurring in the same month as baseline to allow model estimation. Proportional hazards assumptions were assessed for all covariates using Schoenfeld residuals, and corresponding plots for the main exposure in primary models are presented in Supplementary Fig. 2. 3-year healthcare utilisation outcomes were analysed using negative binomial regression to account for count data overdispersion. Follow-up time (in years) was incorporated as a log-offset to account for varying observation periods, with observation from baseline until death or 3 years. In the progression, KRT and healthcare utilisation models, deaths were assigned to the last day of the recorded month.

Models were adjusted for covariates to control for potential confounding. Adjusted estimates (odds ratios, hazard ratios or incidence rate ratios) for remoteness are presented with 95% confidence intervals. Interaction effects between remoteness and selected demographic and clinical variables were included a priori based on clinical relevance. Only meaningful and statistically significant interactions from this set are reported. Effect estimates comparing remoteness categories within relevant subgroups were derived by combining main and interaction effects, with 95% confidence intervals calculated for each comparison. Unadjusted models and full results are in the Supplementary (Tables 3–11 and 15–21). Following cohort selection and covariate derivation, no missing data remained. Analyses were conducted in R (version 4.3.0) [25], with modelling performed using the stats, survival, and MASS packages. Statistical significance was defined at p < 0.05.

Sensitivity analysis

To address potential bias from data availability, methodological decision-making, and competing risks, three sensitivity analyses were conducted: (1) exclusion of pathology records from the provider with data only available from 2017, (2) application of an alternative CKD definition based on the sustained eGFR method described by Liu et al. [26] (Supplementary Methods), and (3) application of Fine–Gray subdistribution hazard models for progression and KRT outcomes, treating death as a competing event.

Ethics approval

Ethics approval was received from the Curtin University Human Research Ethics Committee (HREC; HRE2019–0303) and WA Health HREC (RGS0000001183).

Results

Study cohort

The cohort included 78,244 distinct individuals who first met pathological CKD criteria between 2010 and 2017 (Supplementary Fig. 1), contributing 627,648 person-years of follow up (mean 8 years). Baseline characteristics are shown in Table 1. There were 58,845 (51.2% female) living in major cities, 15,990 (49.6% female) in regional areas, and 3,409 (47.7% female) in remote areas. The remote subgroup had a younger demographic, with 30.9% aged 18–60 years versus 11.5% in major city and regional groups. CKD stage distribution was similar across groups: stage 3a (81.2–83.2%), stage 3b (14.1–15.0%), and stage 4 (2.7–3.8%), though remote residents had the highest proportions in stages 3b and 4. A higher proportion in remote areas had diabetes (39.9%) than in regional (26.1%) and major city (25.8%) areas. Conversely, the proportion with cancer was lower in remote areas (15.0%) than in regional (18.6%) and major city (18.9%) areas. Cardiovascular disease affected between 61.1% and 62.7% across all remoteness groups. Ethnicity data was not available in the linked dataset and therefore not reported among baseline characteristics.

Table 1.

Baseline characteristics of the study cohort by remoteness

Characteristic	Major cities	Regional	Remote
Number of people	58,845	15,990	3409
Age group (years)
18–60	6785 (11.5%)	1844 (11.5%)	1055 (30.9%)
61–75	22,805 (38.8%)	6754 (42.2%)	1445 (42.4%)
76–90	29,255 (49.7%)	7392 (46.2%)	909 (26.7%)
Sex
Female	30,119 (51.2%)	7934 (49.6%)	1625 (47.7%)
Male	28,726 (48.8%)	8056 (50.4%)	1784 (52.3%)
CKD stage
3a	48,430 (82.3%)	13,303 (83.2%)	2769 (81.2%)
3b	8594 (14.6%)	2249 (14.1%)	511 (15.0%)
4	1821 (3.1%)	438 (2.7%)	129 (3.8%)
Comorbidities
Diabetes	15,180 (25.8%)	4167 (26.1%)	1361 (39.9%)
Cardiovascular disease	36,870 (62.7%)	9768 (61.1%)	2101 (61.6%)
Cancer	11,100 (18.9%)	2969 (18.6%)	512 (15.0%)

Values are presented as number of individuals (percentage within each remoteness category). Percentages may not sum to 100% due to rounding

Clinical outcomes associated with remoteness in CKD: early presentation, progression, kidney replacement therapy, and all-cause mortality

Compared with major cities, individuals in remote areas had poorer CKD outcomes, with lower odds of early presentation (adjusted odds ratio [aOR], 0.87; 95% confidence interval [CI], 0.79–0.95) and higher hazards of progression (adjusted hazard ratio [aHR], 1.15; 95% CI, 1.09–1.22), KRT (aHR, 1.94; 95% CI, 1.71–2.20), and all-cause mortality (aHR, 1.12; 95% CI, 1.06–1.19). In contrast, differences between regional and major city residents were weakly or not statistically significant for early presentation, progression, and mortality. However, regional residents had a lower hazard of KRT (aHR, 0.82; 95% CI, 0.72–0.92) compared with major cities (Table 2, clinical outcome counts and full models in Supplementary Tables 2–4, 6, 8, and 10). No major violations of the proportional hazards assumption were observed for the progression, KRT, and all-cause mortality Cox models (Supplementary Fig. 2).

Table 2.

Adjusted associations between remoteness and clinical outcomes in CKD

	Early presentation		Progression		Kidney replacement therapy		All-cause mortality
	aOR (95% CI)	p-value	aHR (95% CI)	p-value	aHR (95% CI)	p-value	aHR (95% CI)	p-value
Major cities	1.0		1.0		1.0		1.0
Regional	1.04 (0.99–1.09)	0.11	1.02 (1.00–1.05)	0.10	0.82 (0.72–0.92)	0.001	1.04 (1.01–1.07)	0.008
Remote	0.87 (0.79–0.95)	0.002	1.15 (1.09–1.22)	< 0.001	1.94 (1.71–2.20)	< 0.001	1.12 (1.06–1.19)	< 0.001

aOR = adjusted odds ratio; aHR = adjusted hazard ratio; CI = confidence interval. Early CKD presentation was modelled using logistic regression. CKD progression, initiation of kidney replacement therapy, and all-cause mortality were modelled using Cox proportional hazards regression. All models were adjusted for baseline age group, sex, and comorbidities (diabetes, cardiovascular disease, and cancer). The Cox models were additionally adjusted for baseline CKD stage

Significant interactions were observed between remoteness and age group, and remoteness and diabetes. Lower odds of early presentation in the remote group were seen in age groups 18–60 years (aOR, 0.76; 95% CI, 0.64–0.89) and 61–75 years (aOR, 0.80; 95% CI, 0.69–0.93), with no significant difference in those 76–90 years. Among adults 18–60 years, particularly those with diabetes, remote residence was associated with an increased hazard of progression (aHR, 1.88; 95% CI, 1.70–2.09). This association was absent, weaker, or reversed in older age groups such as remote residents 76–90 years without diabetes (aHR, 0.85; 95% CI, 0.75–0.96). The elevated hazard of KRT in remote areas was seen in the 18–60 (aHR, 2.67; 95% CI, 2.27–3.14) and 61–75-year age groups (aHR, 1.98; 95% CI, 1.48–2.64) with diabetes, while other subgroups showed non-significant or weak associations. The higher hazard of mortality in the remote group was observed only in younger groups (18–60 years: aHR, 1.31; 95% CI, 1.15–1.49; 61–75 years: aHR, 1.19; 95% CI, 1.08–1.31). These findings highlight among remote residents, those aged 18–60 years, especially with diabetes, consistently experienced poorer clinical outcomes. These key subgroups are outlined in red boxes in Fig. 1 (full models in Supplementary Tables 5, 7, 9, and 11).

Fig. 1.

Forest plots of clinical outcomes in CKD by remoteness showing subgroup-specific contrasts. (a) Early presentation, remoteness x age group interaction. (b) Progression, remoteness x age group and remoteness x diabetes interactions. (c) Kidney replacement therapy, remoteness x age group and remoteness x diabetes interactions. (d) All-cause mortality, remoteness x age group interaction. Forest plots present adjusted odds ratios (aORs) and hazard ratios (aHRs) with 95% confidence intervals from models including interaction terms. Age group in years. Early CKD presentation was modelled using logistic regression. CKD progression, initiation of kidney replacement therapy, and all-cause mortality were modelled using Cox proportional hazards regression. All models were adjusted for baseline age group, sex, and comorbidities (diabetes, cardiovascular disease, and cancer). The Cox models were additionally adjusted for baseline CKD stage. Subgroup contrasts were derived by combining main effects and relevant interaction terms from the fully adjusted models. Red boxes highlight the remote 18–60-year subgroup or remote 18–60-year subgroup with diabetes, which showed consistently poorer clinical outcomes across all models in comparison with their major city reference group

3-year healthcare utilisation by remoteness: emergency department presentations, hospital separations, and serum creatinine pathology tests

Descriptive statistics for healthcare utilisation outcomes are presented in Supplementary Table 14. Compared with major cities, ED presentation rates were higher in regional (adjusted incidence rate ratio [aIRR], 1.43; 95% CI, 1.40–1.47) and remote areas (aIRR, 2.39; 95% CI, 2.28–2.51). Hospitalisation rates were higher in remote areas (aIRR, 1.21; 95% CI, 1.16–1.27) and slightly lower in regional areas (aIRR, 0.94; 95% CI, 0.92–0.96). Serum creatinine testing rates were slightly lower in regional areas (aIRR, 0.93; 95% CI, 0.91–0.95) and comparable in remote areas (Table 3; full models in Supplementary Tables 15, 16, 18, and 20).

Table 3.

Adjusted associations between remoteness and 3-year healthcare utilisation

	Emergency department presentations		Hospital separations		Serum creatinine pathology tests
	aIRR (95% CI)	p-value	aIRR (95% CI)	p-value	aIRR (95% CI)	p-value
Major cities	1.0		1.0		1.0
Regional	1.43 (1.40–1.47)	< 0.001	0.94 (0.92–0.96)	< 0.001	0.93 (0.91–0.95)	< 0.001
Remote	2.39 (2.28–2.51)	< 0.001	1.21 (1.16–1.27)	< 0.001	1.00 (0.96–1.04)	0.90

aIRR = adjusted incidence rate ratio; CI = confidence interval. The rate of emergency department presentations, hospital separations, and serum creatinine testing in the 3 years following baseline were modelled using negative binomial regression. All models were adjusted for baseline age group, sex, CKD stage, and comorbidities (diabetes, cardiovascular disease, and cancer)

Significant interactions were observed between remoteness and both age group and diabetes across all three outcomes. Among remote residents aged 18–60 years with diabetes, rates of ED presentations (aIRR, 3.01; 95% CI, 2.73–3.31), hospitalisation (aIRR, 2.29; 95% CI 2.10–2.50), and serum creatinine testing (aIRR, 1.16; 95% CI, 1.07–1.26) were particularly elevated compared with the same major city subgroup. This key subgroup is highlighted with red boxes in Fig. 2 (full models in Supplementary Tables 17, 19, and 21).

Fig. 2.

Forest plots of 3-year healthcare utilisation by remoteness showing subgroup-specific contrasts. (a) Emergency department presentations, remoteness x age group and remoteness x diabetes interactions. (b) Hospital separations, remoteness x age group and remoteness x diabetes interactions. (c) Serum creatinine pathology tests, remoteness x age group and remoteness x diabetes interactions. Forest plots present adjusted incidence rate ratios (aIRRs) with 95% confidence intervals from models including interaction terms. Age group in years. The rate of emergency department presentations, hospital separations, and serum creatinine testing in the 3 years following baseline were modelled using negative binomial regression. All models were adjusted for baseline age group, sex, CKD stage, and comorbidities (diabetes, cardiovascular disease, and cancer). Subgroup contrasts were derived by combining main effects and relevant interaction terms from the fully adjusted models. Red boxes highlight the remote 18–60-year subgroup with diabetes which showed elevated healthcare utilisation across all measures compared with the same subgroup from major cities

Sensitivity analyses

In the first sensitivity analysis, excluding pathology data from the provider with incomplete data reduced the cohort by 0.4%. In the second, applying a sustained eGFR-based CKD definition yielded a cohort of 72,600 individuals, 81.4% of whom overlapped with the main cohort. Results from both analyses were consistent with the main analysis. In competing risk analyses for CKD progression and KRT, effect estimates were consistent with those from the primary Cox models, suggesting minimal impact of competing mortality (Supplementary Tables 12 and 13).

Discussion

This study provides novel insights into CKD disparities in WA, leveraging a unique linked pathology dataset to track individuals longitudinally across the disease course. Our findings indicate individuals living in remote areas experienced significantly poorer clinical outcomes compared with those in major cities, including 13% lower odds of presenting with CKD early at stage 3a, a 15% higher hazard of disease progression, a 94% higher hazard of initiating KRT, and a 12% higher hazard of all-cause mortality. In contrast, clinical outcomes among regional residents were largely comparable with those from major cities. Among remote populations, younger adults aged 18–60 years with diabetes emerged as a particularly high-risk subgroup. Disparities were also evident among those aged 61–75 years and 61–75 years with diabetes in remote areas.

These findings align with and extend prior Australian and international research demonstrating elevated CKD burden and adverse outcomes in remote populations. Nationally, reporting by the Australian Institute of Health and Welfare highlights higher CKD prevalence, mortality rates, hospitalisations and incidence of treated kidney failure in remote areas [12, 27]. Similarly, Canadian studies report greater CKD prevalence, increased risks of mortality and hospitalisation, and poorer quality of care among remote populations [28–30]. However, as noted in a systematic review of rural chronic disease research across the United Kingdom, United States, Canada, Australia, and New Zealand, much of the literature is focused on describing patterns and prevalence with limited data on comparative health outcomes between rural/remote and urban populations [31]. Additionally, much CKD research focuses on advanced stages with few population-level studies examining outcomes across the full disease spectrum. One recent example is a study of Scotland’s Grampian region, which, while investigating socioeconomic factors, found individuals in more disadvantaged areas experienced worse CKD outcomes [10]. Our study is among the first to examine this with a focus on geographic remoteness. It adds important nuance by demonstrating disparities are not uniform but vary by age and comorbidities, with people with diabetes in younger subgroups consistently experiencing poorer outcomes.

To explore potential contributors to poorer clinical outcomes, we examined 3-year healthcare utilisation. Remote residents used hospital and ED services more frequently than major city residents, suggesting heavier reliance on acute care, consistent with national trends [9]. However, serum creatinine testing rates were comparable between remote and major city areas, indicating poorer outcomes may not be solely attributable to reduced access to care. Healthcare utilisation patterns were most pronounced in the remote 18–60-year age group with diabetes who had significantly higher rates of ED presentations, hospitalisations, and serum creatinine testing compared with the equivalent major city subgroup.

The poorer CKD outcomes observed among younger adults with diabetes in remote areas likely reflect a combination of factors. While serum creatinine testing rates were higher in this subgroup compared with major cities, this does not necessarily reflect formal diagnosis or appropriate follow-up, suggesting potential gaps in early intervention and sustained care. Increased dependence on acute services may indicate deficiencies in accessible and ongoing primary care, a known challenge in remote Australia [9]. Cultural factors may reduce engagement with health services, increasing the risk of later-stage presentation and impacting adherence to treatment plans. Two qualitative studies of CKD patient experiences in remote Australia, one with a mean age of 55.5 years and another with 75% of participants aged 18–60 years, highlight key themes of disease knowledge gaps, fragmented care, travel and relocation burden, and significant out-of-pocket costs [32, 33]. The presence of diabetes as a comorbidity complicates disease management, potentially compounding diagnosis and treatment challenges especially where coordinated care may be lacking. Importantly, disparities appear more pronounced in younger than older remote residents, where survivor effects, competing mortality risks, and limited time to manifest outcomes may mask the true extent of remote disadvantage.

These findings have important implications for health service planning and policy in WA. The similar outcomes observed between regional and major city populations likely reflect many regional centres in WA, such as Bunbury and Geraldton which have populations over 30,000 people, have relatively good access to healthcare infrastructure. Extending elements of coordinated care available in larger centres to smaller, more remote communities may help support earlier detection and improved management for high-risk groups. Further research is needed to understand the drivers of poorer CKD outcomes in younger adults with diabetes in remote areas. These findings support exploring coordinated models of care that can effectively manage comorbid conditions like CKD and diabetes [34, 35], particularly in resource-limited remote settings.

Limitations

Remoteness was assessed using residential postcode at baseline, without accounting for changes in residence over time. According to the 2021 Australian Census, approximately 52% of people in major cities and regional areas, and 38% of people in remote areas reported the same address as 5 years prior in WA [36]. However, this does include moves between addresses within the same area, so the proportion remaining within the same remoteness category is likely higher. Our dataset did not include primary care or pharmacy records limiting insight into overall healthcare usage and treatment. One pathology provider contributed data only from 2017 to 2022, but they represent a small share of the WA market, and sensitivity analyses excluding their records had minimal impact on the results. CKD definition is challenging, and staging based on arbitrary eGFR cutoffs may not accurately reflect the underlying variation in kidney function. In addition, our requirement for persistent eGFR reduction to confirm chronicity, while reducing misclassification of transient kidney dysfunction, may have excluded some individuals with true CKD who had only a single low eGFR measurement. We acknowledge Aboriginal people are over-represented in chronic disease and remote cohorts, but Aboriginal status was not available in the dataset and is a priority for future research.

Conclusion

This study highlights substantial disparities in CKD outcomes among adults aged 18–60 years with diabetes living in remote areas of WA, despite apparently equal access to health services. Understanding the social, cultural, and health system factors underlying these disparities will be essential to developing targeted strategies to reduce inequities in kidney health outcomes in remote Australia.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Abbreviations

aHR

Adjusted hazard ratio

aIRR

Adjusted incidence rate ratio

aOR

Adjusted odds ratio

ARIA+

Accessibility/Remoteness Index of Australia

CI

Confidence interval

CKD

Chronic kidney disease

ED

Emergency department

eGFR

Estimated glomerular filtration rate

HbA1c

Haemoglobin A1C

KRT

Kidney replacement therapy

PPRL

Privacy Preserving Record Linkage

WA

Western Australia

Author contributions

DH, AC, ET, and SB supervised the study, with ET and AC additionally responsible for funding acquisition and project administration. DH, AC, ET, SB, and JS conceived and designed the study, with KC, CL, and SRa also contributing to the methodology. JS, KC, and SRa curated the data, with JS and SB responsible for formal analysis. JS drafted the first version of the manuscript, and all authors contributed to critical review and editing of the final manuscript.

Funding

This research was supported by the Digital Health CRC Limited (DHCRC) and is part of a larger 4-year collaborative partnership between Curtin University, La Trobe University, Deakin University, WA Department of Health, WA Country Health Services (in particular Justin Manuel), WA Primary Health Alliance, and the DHCRC. DHCRC is funded under the Australian Commonwealth’s Cooperative Research Centres (CRC) Program.

Data availability

The datasets used in this study are not publicly available due to privacy considerations. Researchers may apply to the relevant data custodians for access.

Declarations

Ethics approval and consent to participate

Ethics approval was received from the Curtin University Human Research Ethics Committee (HREC; HRE2019–0303) and the WA Health HREC (RGS0000001183), including a waiver of consent. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.