ABSTRACT
Aim
The Australian and New Zealand Society of Nephrology (ANZSN) established a quality indicator (QI) program to measure performance and reduce variation of nephrology care. This program relies on registry data from the Australia and New Zealand Dialysis and Transplant Registry (ANZDATA). ANZDATA also conducts surveys to ascertain more granular characteristics of participating nephrology units to better understand variation in practice. This study assessed ANZDATA and nephrology unit surveys to determine patient‐ and centre‐level factors that may account for variation in dialysis access QIs.
Methods
Dialysis access QIs were evaluated over a 5‐year period using ANZDATA (2016–2020). Patient characteristics were derived from the same period and dataset. Centre characteristics were determined from the 2020‐unit binational survey. Statistical analysis involved multivariable regression with fixed effects for patient‐ and centre‐characteristics and a random effect by centre.
Results
Wide variation was seen across dialysis access QIs. Using multivariable regression, patient‐level characteristics associated with definitive dialysis access at commencement of dialysis included male gender and body mass index decreasing from 30 kg/m2. Centres with a home haemodialysis nurse and a ‘peritoneal dialysis (PD) first policy’ were associated with higher rates of definitive dialysis access. There was also a progressive relationship with the number of full‐time equivalent (FTE) medical staff if the unit already had an FTE greater than 10. Inclusion of centre‐level variables reduced the variance in the model by 68%.
Conclusion
This study demonstrates certain patient‐ and centre‐level characteristics are associated with definitive dialysis access at the commencement of dialysis in Australia and New Zealand. Interventional trials for PD first programmes and different models of care for vascular access are needed.
Keywords: dialysis access, kidney failure, kidney replacement therapy, quality indicators, registry
1. Introduction
Definitive dialysis access is the cornerstone of care for patients with kidney failure treated with dialysis. All forms of dialysis require access either to the peritoneal cavity for peritoneal dialysis (PD) or the bloodstream for haemodialysis (HD). Ideally, dialysis access should be established in a safe and timely manner to facilitate initiation of PD or HD. PD cannot be commenced without definitive access in the form of a PD catheter. For HD, options for definitive vascular access are in the form of an arteriovenous fistula (AVF) or graft (AVG). However, many patients commencing HD do so with temporary access delivered by central venous catheters (CVC). Definitive vascular access is the preferred choice of dialysis access for patients commencing HD, as early definitive access is associated with significantly reduced risk of sepsis and mortality [1] and lower cost and length of hospital stay [2].
Despite this, there is wide variation in patients commencing dialysis with definitive compared to temporary dialysis access across Australia and New Zealand. In Australia in 2022, only 39% of patients commenced HD with definitive access, while in New Zealand this proportion was only 20% [3]. Patient‐level factors associated with the use of definitive dialysis access previously described include age, comorbidities, likelihood of long‐term survival, treatment goals, and timing of dialysis initiation.
Quality indicators (QIs) can be utilised to improve healthcare service delivery and outcomes and help identify and understand variation in practice. The Australian and New Zealand Society of Nephrology (ANZSN), in collaboration with the Australia and New Zealand Dialysis and Transplant Registry (ANZDATA), a binational registry that collects data on all patients receiving kidney replacement therapy (KRT) in Australia and New Zealand, developed and implemented a QI programme in 2021 after a review of international literature for key performance indicators and quality measures and after extensive consultation with numerous key stakeholders across Australia and New Zealand. Consideration was given to the relevance and ease of collection of QIs in the Australian and New Zealand environment, with the ANZDATA Registry determined to be most appropriate to facilitate the collection and reporting of QIs. Dialysis access planning at commencement of dialysis was considered an important QI.
Periodically, ANZDATA conducts surveys to ascertain more detailed characteristics of participating nephrology units in Australia and New Zealand to correlate these characteristics to ANZDATA variables and outcomes. Introduction of the ANZSN/ANZDATA QI programme, coupled with information from unit characteristic surveys, provides an opportunity for better understanding variations in practice across Australia and New Zealand at both a patient‐ and centre‐level. The ANZSN/ANZDATA dialysis access QIs are shown in Table 1. We aimed to describe patterns of dialysis access in patients with kidney failure in Australia and New Zealand using the dialysis access QIs and evaluate patient‐ and centre‐level factors that may be associated with variability in practice.
TABLE 1.
ANZSN/ANZDATA dialysis access quality indicators (QIs).
| Dialysis access QI | Numerator | Denominator |
|---|---|---|
| [A] All dialysis access planning | New KRT patients on HD with AVF/AVG or PD with PD catheter | All patients commencing KRT with dialysis |
| [B] Vascular access at first HD | New KRT patients on HD with AVF/AVG | All new KRT patients on HD |
| [C] Vascular access planning | New KRT patients on HD with AVF/AVG | Patients on HD at 3 months post commencing |
| [D] PD planning | New KRT patients on PD with PD catheter | Patients on PD at 3 months post commencing |
Abbreviations: AVF, arteriovenous fistula; AVG, arteriovenous graft; HD, haemodialysis; KRT, kidney replacement therapy; PD, peritoneal dialysis; QI, quality indicator.
2. Methods
This observational study involved Australian and New Zealand patients with kidney failure on KRT, with 5 years of data extracted from the ANZDATA Registry including all patients aged 18 years and older who initiated KRT in the form of PD or HD from 1 January 2016 to 31 December 2020. Paediatric units were excluded. Full details of the structure and methods of ANZDATA have been published elsewhere [4]. In brief, ANZDATA collects demographic and clinical information about all patients who are residents in Australia or New Zealand and are receiving long‐term KRT for kidney failure with either dialysis or kidney transplantation. Patient data is provided by the patient's treating unit through an annual survey. Additional data is also provided in “real‐time” for pre‐specified events, such as change in dialysis modality or patient death. ANZDATA measures performance across the dialysis access QIs. The composite dialysis access QI (QI [A] from Table 1), including definitive access for patients starting HD together with those starting PD, was the focus of this analysis given it most likely demonstrates a unit's overall preparedness for commencement of dialysis and prioritises avoiding the risks of CVCs.
Patient‐level characteristics were extracted from ANZDATA. Characteristics (at dialysis initiation) included age, sex, ethnicity (non‐indigenous or First Nations, with separate variables for First Nations ethnicity in Australia and New Zealand), body mass index (BMI), primary kidney disease (diabetic kidney disease or not), diabetes (as a comorbidity, defined as type 1 or type 2), comorbid vascular disease (defined as one or more of coronary artery disease, cerebrovascular disease or peripheral vascular disease), smoking history (current/previous or not), late referral (known to the nephrology community < 3 months prior to dialysis initiation), quintile of socioeconomic disadvantage (derived from the Australian Bureau of Statistics' Index of Relative Socio‐economic Advantage and Disadvantage for socio‐economic indexes for areas [SEIFA] [5]) and remoteness based on postcode of residence collapsed into three categories (major cities, regional or remote) [6].
Centre‐level characteristics were obtained from the 2020 Australian and New Zealand nephrology unit characteristic survey [7] and included type of nephrology unit (managed in a government public healthcare service or in the private sector), specialist nephrologist staff (measured by full‐time equivalent [FTE]), nursing staff composition (including chronic kidney disease [CKD] educator, PD nurse, vascular access nurse, nurse practitioner), vascular access surgeon, vascular access multidisciplinary meeting (MDM) and whether the unit had a “PD first” policy. No definitions were provided for these parameters, and the responses were at the discretion of the responding unit. Centre size, based on incident dialysis patient numbers, was extracted from ANZDATA. Only centres that responded to the survey were included in the analysis.
Statistical analyses were performed using Stata (version 18.0, College station, TX, StataCorp LLC). Univariate modelling was performed on all patient‐ and centre‐level factors. For the continuous variables of patient age at commencement of dialysis, BMI, unit size (incident dialysis number) and FTE of consultant medical staff, linear splines were utilised given their relationship with establishing definitive dialysis access. The knots in the linear splines were chosen based on the changing relationship of these continuous variables with the outcome of definitive access. For age, this was at 80 years; for BMI at 30 kg/m2; unit size was incident dialysis number of 50 and 100; and for FTE this was at 5 and 10. They were then included in the multi‐level mixed‐effects logistic regression if their univariate p‐value was < 0.05. Centre was included as a random effect and year as a fixed effect. Variables were removed from the multivariate model one at a time using backwards elimination, excluding the covariate with the highest p‐value, until all covariates had a p < 0.05. The study was approved by The Royal Melbourne Hospital Human Research Ethics Committee (RMH HREC number 2020.090).
3. Results
Between 1 January 2016 and 31 December 2020, 17 951 patients commenced dialysis at 105 nephrology units across Australia and New Zealand. Of these, 10 248 (57%) patients commenced with definitive dialysis access. The proportion of patients commencing dialysis with definitive access varied widely among centres. There was substantial variability between each unit's performance for this measure over the 5‐year study period (Figure 1).
FIGURE 1.
Variation in quality indicator of definitive dialysis access at initiation of dialysis (QI [A] from Table 1), 2016–2020. QI, quality indicator.
Patient characteristics are shown in Table 2. There were significant differences in patient characteristics between those who commenced dialysis with versus without any definitive dialysis access. Patients who started dialysis without definitive dialysis access were more likely to be female, to be referred late to nephrology, to be of First Nations ethnicity, to have rural residence, and to have primary kidney disease listed as diabetic nephropathy with comorbid diabetes and vascular disease.
TABLE 2.
Patient‐level characteristics for definitive dialysis access at dialysis initiation quality indicator (QI [A] from Table 1), 2016–2020.
| Patient characteristics | Definitive dialysis access | No definitive dialysis access | p | |
|---|---|---|---|---|
| Total patients | 10 248 | 7702 | ||
| Country/State | Northern Territory | 255 (2.5%) | 307 (4%) | < 0.01 |
| New South Wales | 2721 (26.6%) | 1677 (21.8%) | ||
| Victoria | 2144 (20.9%) | 1491 (19.4%) | ||
| Queensland | 1737 (17%) | 1300 (16.9%) | ||
| South Australia | 636 (6.2%) | 399 (5.2%) | ||
| Western Australia | 864 (8.4%) | 844 (11%) | ||
| Tasmania | 134 (1.3%) | 126 (1.6%) | ||
| Australian Capital Territory | 157 (1.5%) | 144 (1.9%) | ||
| New Zealand | 1587 (15.5%) | 1411 (18.3%) | ||
| Gender | Female | 3697 (36.1%) | 2978 (38.7%) | < 0.01 |
| Male | 6551 (63.9%) | 4724 (61.3%) | ||
| Age at KRT entry, median (IQR) | 63 (52, 72) | 62 (50, 72) | < 0.01 | |
| BMI, median (IQR) | 28.4 (24.6, 33.2) | 27.9 (23.9, 33.4) | < 0.01 | |
| Late referral | 680 (6.7%) | 2346 (31%) | < 0.01 | |
| First Nations | 1380 (13.5%) | 1430 (18.6) | < 0.01 | |
| First Nations‐ Australia | 804 (7.8%) | 887 (11.5%) | < 0.01 | |
| Māori New Zealand | 576 (5.6%) | 543 (7.1%) | < 0.01 | |
| Primary kidney disease diabetic nephropathy | 4140 (40.4%) | 3297 (42.8%) | 0.001 | |
| Smoker | 5010 (48.9%) | 3881 (50.4%) | 0.046 | |
| Comorbid diabetes | 5427 (53%) | 4416 (57.3%) | < 0.01 | |
| Comorbid vascular disease | 4234 (41.3%) | 3807 (49.4%) | < 0.01 | |
| Remoteness | Major city | 6528 (69.3%) | 4644 (67.7%) | 0.029 |
| Outside major city | 2895 (30.7%) | 2219 (32.3%) | ||
Abbreviations: BMI, body mass index; IQR, interquartile range; KRT, kidney replacement therapy.
Centre characteristics are shown in Table 3. Although 105 nephrology centres across Australia and New Zealand provide data to the ANZDATA Registry, only 69 centres (66%) completed the centre characteristics survey representing 76% of dialysis patients (Table S1). There was no significant difference in the rates of definitive access between units that responded to the survey and those that did not (57.2% v 56.9%). There were some differences in patient characteristics between centres that responded to the survey and those that did not, namely that centres that did not respond had patients who were older, had lower BMI, and were less likely to be of First Nations ethnicity, have comorbid diabetes, or smoke.
TABLE 3.
Centre‐level characteristics for definitive dialysis access at dialysis initiation quality indicator (QI [A] from Table 1) a .
| Centre characteristics | ||
|---|---|---|
| Total number | 105 | |
| Parent centre | New South Wales | 28 (26.9%) |
| Victoria | 21 (20.2%) | |
| Queensland | 29 (27.9%) | |
| South Australia | 3 (2.9%) | |
| Western Australia | 5 (4.8%) | |
| Tasmania | 2 (1.9%) | |
| Australian Capital Territory | 2 (1.9%) | |
| Northern Territory | 2 (1.9%) | |
| New Zealand | 12 (11.5%) | |
| Incident dialysis size (number of patients in 2020) | < 50 | 77 (74%) |
| 50–100 | 17 (16.3%) | |
| > 100 | 10 (9.6%) | |
| Remoteness | Major city | 67 (72%) |
| Regional | 24 (26%) | |
| Remote | 2 (2%) | |
| Full time equivalent consultant nephrology staff | < 5 | 42 (65%) |
| 5–10 | 19 (29%) | |
| > 10 | 4 (6%) | |
| Private nephrology unit | 11 (16%) | |
| Chronic kidney disease nurse educator | 53 (77%) | |
| Peritoneal dialysis nurse | 52 (75%) | |
| Home haemodialysis nurse | 40 (58%) | |
| Vascular access nurse | 37 (54%) | |
| Nurse practitioner | 28 (41%) | |
| Vascular surgeons | 64 (93%) | |
| Multidisciplinary access service | 39 (57%) | |
| Peritoneal dialysis first policy | 32 (46%) | |
From 2020 ANZDATA centre survey.
Univariate analyses for all dialysis access planning (QI [A] as per Table 1) are presented in Table 4. An indicator variable for units that were non‐responders to the survey was not a significant predictor of definitive access (odds ratio [OR] 0.99, p = 0.77, 95% confidence interval [CI] 0.92–1.06). Multivariable analysis including both patient‐ and centre‐level characteristics is presented in Figure 2. There was a positive association with the patient factors of male gender, BMI increasing between 15 and 30 kg/m2 and age at start of dialysis increasing between 18 and 80 years, and a negative association with late referral, BMI increasing from 30 kg/m2, age at start of dialysis increasing from 80 years, First Nations Australians, and comorbid diabetes and vascular disease. Centre‐level characteristics in the multivariable analysis associated with an increased likelihood of definitive dialysis access were a progressively greater consultant nephrology FTE beyond 10 FTE, having home haemodialysis nurses, having a PD first policy, and the Australian state of treatment. Compared to the state of New South Wales (NSW) in Australia, jurisdictions with less likelihood of definitive dialysis access were Victoria, Queensland, Western Australia, Tasmania, Australian Capital Territory (ACT) and New Zealand. There were no interactions between the location of centres and any of the centre‐level characteristics. By introducing centre‐level variables into the model, the variance of the random effect was reduced by 68% with jurisdiction contributing most to this effect.
TABLE 4.
Univariate analysis for patient‐ and centre‐level characteristics for definitive dialysis access at dialysis initiation quality indicator (QI [A] from Table 1).
| Characteristics | Odds ratio (confidence interval) | ||
|---|---|---|---|
| Patient‐level | Age at start of dialysis (years) a | 18–80 | 1.006 (1.004–1.008) |
| > 80 | 0.93 (0.91–0.96) | ||
| Gender (Male) | 1.11 (1.05–1.19) | ||
| BMI (kg/m2) a | 15–30 | 1.04 (1.03–1.05) | |
| > 30 | 0.98 (0.97–0.99) | ||
| State/Country | NSW | Reference | |
| VIC | 0.89 (0.81–0.97) | ||
| QLD | 0.82 (0.75–0.91) | ||
| SA | 0.98 (0.85–1.13) | ||
| WA | 0.63 (0.56–0.71) | ||
| TAS | 0.66 (0.51–0.84) | ||
| ACT | 0.67 (0.53–0.85) | ||
| NT | 0.51 (0.43–0.61) | ||
| NZ | 0.69 (0.63–0.76) | ||
| Remoteness | Major city | Reference | |
| Regional | 1.02 (0.95–1.1) | ||
| Remote | 0.64 (0.35–1.46) | ||
| SEIFA quintiles | 1st | 1.01 (0.92–1.11) | |
| 2nd | 1.08 (0.98–1.2) | ||
| 3rd | Reference | ||
| 4th | 0.98 (0.88–1.08) | ||
| 5th | 1.001 (0.91–1.1) | ||
| Primary kidney disease diabetic nephropathy | 0.91 (0.85–0.96) | ||
| Comorbid diabetes mellitus | 0.84 (0.79–0.89) | ||
| Comorbid vascular disease | 0.72 (0.68–0.76) | ||
| Smoking | 0.94 (0.89–0.99) | ||
| Late referral | 0.16 (0.15–0.18) | ||
| First Nations | 0.68 (0.63–0.74) | ||
| First Nations‐ Australia | 0.65 (0.59–0.72) | ||
| Māori New Zealand | 0.79 (0.7–0.88) | ||
| Centre‐level | Incident dialysis patients a | 1–50 | 0.99 (0.99–1) |
| 50–100 | 0.99 (0.99–1) | ||
| > 100 | 1.003 (1.001–1.004) | ||
| Full time equivalent consultant nephrology staff a | 1–5 | 0.99 (0.96–1.04) | |
| 5–10 | 0.99 (0.97–1.02) | ||
| > 10 | 1.16 (1.1–1.39) | ||
| Private nephrology unit | 0.084 (0.69–1.03) | ||
| CKD nurse educator | 0.98 (0.84–1.14) | ||
| PD nurse | 1.18 (1.02–1.34) | ||
| Home HD nurse | 1.24 (1.14–1.35) | ||
| Vascular access nurse | 0.93 (0.86–1.02) | ||
| Nurse Practitioner | 1.19 (1.11–1.27) | ||
| Vascular surgeon | 0.84 (0.76–0.93) | ||
| MDM vascular access service | 1.07 (0.99–1.15) | ||
| PD first policy | 1.27 (1.19–1.36) | ||
Abbreviations: ACT, Australian Capital Territory; BMI, body mass index; CKD, chronic kidney disease; HD, haemodialysis; MDM, multi‐disciplinary meeting; NSW, New South Wales; NZ, New Zealand; PD, peritoneal dialysis; QLD, Queensland; SA, South Australia; SEIFA, socio‐economic indexes for area; TAS, Tasmania; VIC, Victoria; WA, Western Australia.
Linear splines.
FIGURE 2.
Multivariate logistic regression of definitive dialysis access at initiation of dialysis, adjusted for patient‐ and centre‐level characteristics. New South Wales is the reference state. *Linear splines. BMI, body mass index; FTE, full time equivalent consultant nephrology staff.
Univariate and multivariate analyses for different variations of dialysis access QIs at initiation of HD and PD (QIs [B], [C] and [D] as shown in Table 1) are presented in the Supporting Information (Table S2 and Figures S1–S3). There were similar observations in these analyses where patient factors associated with an increased likelihood of definitive dialysis access included male gender, BMI decreasing from 30 kg/m2 and age at start of dialysis decreasing from 80 years. For the two HD‐specific QIs ([B] and [C]), comorbid vascular disease was associated with a lower likelihood of definitive vascular access. Additionally, at 3 months, New Zealand Māori and patients with comorbid diabetes were variables associated with a lower rate for definitive vascular access for HD. Centre‐level variables of centres that had multidisciplinary access services were more likely to have definitive vascular access for HD at dialysis initiation although centres with CKD nurse educators were less likely to have definitive HD access at 3 months. Centres with a vascular access nurse and a progressive increase in FTE beyond 10 FTE consultant nephrology staff were more likely to have patients with a PD catheter inserted at initiation of dialysis for patients on PD by 3 months (QI [D]). By including centre‐level variables into the respective models for QIs [B], [C] and [D], variance of the centre level random effect in each was reduced by 80%, 62% and 81% respectively.
4. Discussion
QIs are important to understand trends in healthcare delivery, benchmark performance and to identify opportunities for improvement. There is wide variation in the performance of dialysis access QIs across nephrology units in Australia and New Zealand, indicating significant differences in the way nephrology care is delivered. Analysis of this variation with inclusion of centre‐level variables showed a substantial reduction in variation and identified unit characteristics associated with better performance of dialysis access QIs. Whether these associations are causal is not known, but they do indicate opportunities for testing interventions at a process level to improve outcomes.
As expected, patient‐level factors are also important determinants of dialysis access at initiation of dialysis. Late referral to a nephrology service was the patient‐level factor with the largest effect. While dialysis access QIs measured by ANZSN/ANZDATA exclude those who were referred late, we included them in our analysis as late referral is an important factor—both common and powerful. In many cases, particularly for those referred to a nephrology unit 1–3 months prior to dialysis start, it may be possible to achieve permanent access prior to dialysis commencement. Late referrals are associated with higher morbidity, poorer health outcomes, and reduced access to kidney transplantation [8]. As such, the creation of expedited pathways of clinical care for patients who are referred late is likely to provide multiple benefits.
Aboriginal and Torres Strait Islander people of Australia [9] and New Zealand Māori [10] people experience significant disparity when it comes to the burden of CKD and all aspects of nephrology care. The ANZSN/ANZDATA QIs demonstrate the association of First Nations ethnicity and dialysis access with First Nations Australians less likely to have definitive dialysis access at dialysis initiation and Māori New Zealanders less likely to have definitive vascular access by 3 months after HD commencement. There may be no physiological rationale for this, but there are examples in other areas of systematic differences in access to care such as coronary angiography in the setting of acute cardiac ischaemia [11]. There are models of care [12] that have been co‐designed with First Nations communities that include principles of community‐based and culturally safe care and developing a First Nations health workforce. These principles have been applied to kidney transplantation [13] although the benefit has not yet been established. These principles have not been applied to dialysis modality and access choice, creation, and management in Indigenous communities [14]. A co‐designed model of care for dialysis access in Indigenous communities should be a priority for an implementation study in this area. Uptake and utilisation of these recommendations could be included as a novel centre‐level variable in future studies to better understand variation in QI performance.
Female gender was associated with lower likelihood of definitive dialysis access in the QIs that include dialysis access for HD, and this has also been reported in many international settings [15, 16]. There is an increased rate of AVF‐related complications seen in women, including thrombosis, primary failure, and overall failure [17]. It has been hypothesised that smaller vessel size in women may contribute; however, evidence is inconclusive [18, 19]. This may potentially dissuade nephrologists and dialysis access surgeons from AVF creation in women. This difference between genders was not seen in the PD exclusive QI [D] and therefore this difference overall may predominantly be attributed to HD access.
From a centre‐level perspective, units that have a PD first policy were shown to have a greater proportion of patients more likely to start dialysis with definitive access. Having a dedicated PD first policy may suggest that units have an engaged model of pre‐dialysis care, including processes that identify suitable patients and timely CKD education, home assessment, and surgical review [20]. These units may also have an ability to more rapidly insert PD catheters through better access to surgeons and theatres or have interventional nephrologists/radiologists who perform PD insertions; and these catheters have the advantage of being able to be used shortly after insertion. Furthermore, the time between PD‐catheter insertion and use for commencement of PD can be minimal, especially when compared to AVF creation, and therefore may be an option in urgent situations, provided centres have capability. There can be unintended consequences of PD first policies, however, as increased utilisation of PD has been associated with reduced technique survival in some circumstances [21], highlighting that patient selection may ultimately be more important than the policy itself. There is also conflicting data regarding the survival benefit of PD [22, 23]. A PD‐first policy should be used in tandem with a shared decision‐making framework for modality selection that also respects patient autonomy.
Units that have home HD nurses were also shown to have a greater proportion of patients more likely to start dialysis with definitive access. There is no current literature that supports this finding, and this variable is likely also a surrogate for units with an engaged model of pre‐dialysis care and relative supporting clinical pathways.
The relationship between unit consultant FTE and definitive dialysis access was a complex one. Increasing FTE beyond 10 was associated with definitive dialysis access; however, trends observed among units with lesser FTE are not consistent with this. Of nephrology units in our study, only four had FTE greater than 10. These units with ≥ 10 FTE have very large inpatient numbers and are all metropolitan public “teaching” hospitals with CKD nurse educators, home HD nurses, and vascular access nurses and hence may qualitatively differ from smaller units. Among these larger units, those with more FTE may have more resources for staff training and ongoing education and better processes around multidisciplinary care [24]. With more FTE, the unit is more likely to have nephrologists with sub‐specialty interests such as advanced CKD care, dialysis access, and PD, which could support better definitive dialysis access planning. Staff may also have non‐clinical portfolios including quality improvement, and these units may have well‐defined models of care that support timely dialysis access. Our study suggests that there is a role for units already established with at least 10 FTE consultant nephrology staff to increase their senior medical staff. However, units with FTE between 5 to 10, increasing FTE was associated with reduced rates of definitive dialysis access. It therefore could be hypothesised that some units with FTE between 5 and 10 may benefit from merging to consolidate or centralise services, resources, and infrastructure, if possible, to improve the likelihood of establishing definitive dialysis access.
There remains significant variation in how jurisdictions within Australia and New Zealand perform across dialysis access QIs. There was no interaction between any of the jurisdictions and any of the centre‐level characteristics within this analysis that account for the variation seen. In Australia, while there is some Federal funding for healthcare, most allocation of healthcare resources is determined at a state/territory level, and each jurisdiction coordinates nephrology services differently. This may affect various dialysis access processes including resources, clinical care pathways, and management, dialysis modality selection, access to theatre, and surgical waitlist management. The nature of these differences at a jurisdictional level is not known. A qualitative approach to understanding high performing units across Australia and New Zealand may identify new centre‐level variables that could be included and tested in future centre surveys and analyses of variation. With this approach, it would be helpful to also explore clinicians' attitude and units' culture of practice towards dialysis education, modality selection, and dialysis access creation as well as broader health service concepts like patient reported experience measures, quality improvement, and consumer engagement as they may represent other centre‐level characteristics that contribute to timely dialysis access creation.
A significant strength of this study is the use of ANZDATA, a binational registry that captures clinical data for all patients starting and continuing KRT in Australia and New Zealand. Also, covering a 5‐year study period, this study included nearly 18 000 patients for analysis. Unique to this study is the inclusion of a broader set of centre‐specific factors to further explore centre‐level variation. This included information detailing medical, surgical, and nursing staff composition of nephrology units as well as whether units utilised a multidisciplinary approach to dialysis access or a PD first policy and helped explain variation previously seen.
A limitation of the study was that only two‐thirds of nephrology units in Australia and New Zealand completed the centre‐specific survey used for analysis. However, there were only minimal significant differences between patient characteristics; hence, findings are likely to be representative of all units. Another limitation of our study was the lack of more granular detail on blood vessel characteristics in patients which may influence the success of AVF creation, such as arterial and venous diameter and brachial artery flow volume, and therefore may have explained why some patients required a CVC if there were no suitable vessels for an AVF. Additionally, no data was collected as to whether dialysis access was guided by the End‐Stage Kidney Disease Life‐Plan strategy outlined in the 2019 Kidney Disease Outcomes Quality Initiative guidelines [25] as this may impact decision making with respect to definitive dialysis access formation.
Although associations found in this study are only hypothesis‐generating, findings may suggest several possible implementation projects to improve clinical care and reduce variation in practice. For example, to establish true benefits of a PD‐first policy, and whether this is a causal factor for definitive dialysis access at commencement of dialysis, an interventional study in the form of a cluster‐randomised trial of a PD‐first programme may be required with the primary outcome of definitive dialysis access at dialysis initiation. Other outcomes of interest might be PD utilisation, definitive vascular access at HD initiation, technique survival, peritonitis rates, and patient‐reported outcome measures. Whilst this would be challenging, there is a precedent for such trials in PD [26]. Unforeseen impacts would also need to be monitored. Implementation of a rapid access model of establishing vascular access for late referrals, regardless of the reason for late referral, could also be considered. A proposed model would include streamlined CKD education, prioritised surgical review and planning, dedicated theatre access, and nurse‐led case management with the aim of offering an efficient and effective way to ensure definitive dialysis access at dialysis initiation. This multidisciplinary model could be trialled in a major nephrology unit for 12 months and allow comparison within the unit before and after and in parallel with other comparable units.
Centre‐level characteristics are an important addition to the QI programme as they may help individual units benchmark their performance relative to other units. Further support and education for units in interpreting their performance, including variability in performance year‐to‐year, would assist units in understanding their performance and guide future quality improvement programmes.
5. Conclusion
Dialysis access remains a critical element of dialysis care. This study demonstrates variation in practice and challenges in establishing dialysis access in patients with kidney failure in Australia and New Zealand across a suite of dialysis access QIs. It highlights several patient‐level factors that may affect the likelihood of starting with definitive dialysis access, and the inclusion of centre‐level variables helped reduce a significant amount of the variation. Qualitative studies are needed to ascertain new centre‐level variables that could be utilised to help understand centre variation further. Based on the impact of late referrals and having a PD‐first policy, intervention trials are required to definitively determine their benefit on timely dialysis access.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Data S1. Supporting Information.
Acknowledgements
The authors thank the entire Australian and New Zealand nephrology community that provides information to and maintains the Australian and New Zealand Dialysis and Transplant Registry (ANZDATA) databases. The data reported here have been supplied by the Australian and New Zealand Dialysis and Transplant Registry. The interpretation and reporting of these data are the responsibility of the authors and should not be seen as an official policy or interpretation of ANZDATA. The authors received no financial support for the research, authorship, and/or publication of this article. The results presented in this paper have not been published previously in whole or part, except in abstract format. Open access publishing facilitated by The University of Melbourne, as part of the Wiley ‐ The University of Melbourne agreement via the Council of Australian University Librarians.
Steinberg A. G., Toussaint N. D., Davies C. E., Gray N. A., and McDonald S. P., “Variability in Dialysis Access Over Time and Across Centres in Australia and New Zealand: Effect of Centre‐ and Patient‐Related Factors on Dialysis Access Quality Indicators,” Nephrology 30, no. 7 (2025): e70076, 10.1111/nep.70076.
Subject Editor: Szu‐Chun Hung
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- 1. Ravani P., Palmer S. C., Oliver M. J., et al., “Associations Between Hemodialysis Access Type and Clinical Outcomes: A Systematic Review,” J Am Soc Nephrol 24 (2013): 465–473. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Wu L. C., Lin M. Y., Hsieh C. C., et al., “Planned Creation of Vascular Access Saves Medical Expenses for Incident Dialysis Patients,” Kaohsiung Journal of Medical Sciences 25, no. 10 (2009): 521–529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. https://www.anzdata.org.au/report/anzdata‐46th‐annual‐report‐2023‐data‐to‐2022/.
- 4. https://www.anzdata.org.au/anzdata/services/data‐management/data‐collection/.
- 5. Australian Bureau of Statistics , “Socio‐Economic Indexes for Areas (SEIFA): Technical Paper,” ABS, 2021, https://www.abs.gov.au/statistics/detailed‐methodology‐information/concepts‐sources‐methods/socio‐economic‐indexes‐areas‐seifa‐technical‐paper/2021.
- 6. Australian Bureau of Statistics , “Australian Statistical Geography Standard (ASGS) Edition 3,” ABS, Jul 2021–Jun 2026, https://www.abs.gov.au/statistics/standards/australian‐statistical‐geography‐standard‐asgs‐edition‐3/jul2021‐jun2026.
- 7. McDonald S., Hewawasam E., Hurst K., Irish G., and Clayton P., ANZDATA Special Reports 1: Unit Survey 2020 (Australia and New Zealand Dialysis and Transplant Registry, 2021). [Google Scholar]
- 8. Cass A., Cunningham J., Snelling P., and Ayanian J. Z., “Late Referral to a Nephrologist Reduces Access to Renal Transplantation,” American Journal of Kidney Diseases 42, no. 5 (2003): 1043–1049. [DOI] [PubMed] [Google Scholar]
- 9. Hoy W. E., Mott S. A., and Mc Donald S. P., “An Expanded Nationwide View of Chronic Kidney Disease in Aboriginal Australians,” Nephrology 21 (2016): 916–922. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Huria T., Palmer S., Beckert L., Williman J., and Pitama S., “Inequity in Dialysis Related Practices and Outcomes in Aotearoa/New Zealand: A Kaupapa Māori Analysis,” International Journal for Equity in Health 17 (2018): 27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Randall D. A., Jorm L. R., Lujic S., O'Loughlin A. J., Eades S. J., and Leyland A. H., “Disparities in Revascularization Rates After Acute Myocardial Infarction Between Aboriginal and Non‐Aboriginal People in Australia,” Circulation 127 (2013): 811–819. [DOI] [PubMed] [Google Scholar]
- 12. Bateman S., Riceman R., Owen K., et al., “Models of Care to Address Disparities in Kidney Health Outcomes for First Nations People,” Kidney International 104 (2023): 681–689. [DOI] [PubMed] [Google Scholar]
- 13. Hughes J. T., Cundale K., Webster A., Owen K. J., and McDonald S. P., “Towards Equity in Kidney Transplantation: The Next Steps,” Medical Journal of Australia 219, no. 8 (2023): S19–S22. [DOI] [PubMed] [Google Scholar]
- 14. Tunnicliffe D. J., Bateman S., Arnold‐Chamney M., et al., “Recommendations for Culturally Safe and Clinical Kidney Care for First Nations Australians,” Medical Journal of Australia 219, no. 8 (2023): 374–385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Markell M., Brar A., Stefanov D. G., and Salifu M. O., “Gender Disparity in Fistula Use at Initiation of Hemodialysis Varies Markedly Across ESRD Networks‐Analysis of USRDS Data,” Hemodialysis International 22, no. 2 (2018): 168–175. [DOI] [PubMed] [Google Scholar]
- 16. Angelici L., Marino C., Umbro I., et al., “Gender Disparities in Vascular Access and One‐Year Mortality Among Incident Hemodialysis Patients: An Epidemiological Study in Lazio Region, Italy,” Journal of Clinical Medicine 10, no. 21 (2021): 5116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Caplin N., Sedlacek M., Teodorescu V., Falk A., and Uribarri J., “Venous Access: Women Are Equal,” American Journal of Kidney Diseases 41, no. 2 (2003): 429–432. [DOI] [PubMed] [Google Scholar]
- 18. Miller C. D., Robbin M. L., and Allon M., “Gender Differences in Outcomes of Arteriovenous Fistulas in Hemodialysis Patients,” Kidney International 63 (2003): 346–352. [DOI] [PubMed] [Google Scholar]
- 19. Pounds L. L. and Teodorescu V. J., “Chronic Kidney Disease and Dialysis Access in Women,” Journal of Vascular Surgery 57, no. 4S (2013): 49–53. [DOI] [PubMed] [Google Scholar]
- 20. Li P. K. and Chow L. M., “Peritoneal Dialysis‐First Policy Made Successful: Perspective and Actions,” American Journal of Kidney Diseases 62, no. 5 (2013): 993–1005. [DOI] [PubMed] [Google Scholar]
- 21. Toussaint N. D., McMahon L. P., Dowling G., et al., “Introduction of Renal Key Performance Indicators Associated With Increased Uptake of Peritoneal Dialysis in a Publicly Funded Health Service,” Peritoneal Dialysis International 37, no. 2 (2017): 198–204. [DOI] [PubMed] [Google Scholar]
- 22. Lee M. B. and Bargman J. M., “Survival by Dialysis Modality‐Who Cares?,” Clinical Journal of the American Society of Nephrology 11, no. 6 (2016): 1083–1087. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Marshall M. R., “The Benefit of Early Survival on PD Versus HD‐ Why This Is (Stiil) Very Important,” Peritoneal Dialysis International 40, no. 4 (2020): 405–418. [DOI] [PubMed] [Google Scholar]
- 24. Kuo G., Lee T. H., Chen J. J., et al., “The Dialysis Facility Levels and Sizes Are Associated With Outcomes of Incident Hemodialysis Patients,” Scientific Reports 11, no. 1 (2021): 20560. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Lok C. E., Huber T. S., Lee T., et al., “KDOQI Clinical Practice Guideline for Vascular Access: 2019 Update,” American Journal of Kidney Diseases 75 (2020): S1–S164. [DOI] [PubMed] [Google Scholar]
- 26. Chow J. S. F., Boudville N., Cho Y., et al., “Multi‐Center, Pragmatic, Cluster‐Randomized, Controlled Trial of Standardized Peritoneal Dialysis (PD) Training Versus Usual Care on PD‐Related Infections (The TEACH‐PD Trial): Trial Protocol,” Trials 24 (2023): 730, 10.1186/s13063-023-07715-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1. Supporting Information.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.