‘Optimal’ vs. ‘Suboptimal’ Haemodialysis Start with Central Venous Catheter—A Better Way to Assess a Vascular Access Service?

Abstract

Background: Whether patients commence haemodialysis with a central venous catheter (CVC), or an arteriovenous fistula (AVF) is used to audit the quality of a vascular access service. However, this crude metric of measurement can miss the increasing nuance and complexity of vascular access planning. We aimed to understand whether commencing haemodialysis with a CVC represented an ‘optimal’ or ‘suboptimal’ outcome and how this could influence the assessment of a vascular access service. Methods: From a prospective clinical database, patients known to nephrology >90 days prior to initiating haemodialysis as first-ever renal replacement therapy (2011–2020) from a single centre were included. Results: A total of 158/254 patients started haemodialysis with a CVC, and 96 with arteriovenous fistula. For 91 patients, the CVC was deemed ‘optimal’ care due to factors such as unpredictable deterioration in renal function (n = 41) and inadequate veins for AVF creation (n = 24). For 67 patients, the CVC was ‘suboptimal’ due to factors such as no/late referral to access assessment (n = 25) and delays in the AVF creation pathway (n = 13). There was no difference in mean survival between the AVF and ‘suboptimal’ groups (2.53 vs. 2.21 years, p = 0.31). There was a survival difference between AVF versus CVC (2.53 vs. 1.97 years, p = 0.002) and ‘suboptimal’ versus ‘optimal’ CVC cohorts (2.21 vs. 1.40 years, p = 0.16). Conclusions: Understanding whether a CVC is ‘optimal’ or ‘suboptimal’ allows a more nuanced analysis of service provision. High mortality in the ‘optimal’ group suggests a frailer cohort where CVC is potentially the best care. Studying ‘suboptimal’ CVC starts helps identify practice and system issues preventing ‘optimal’ care.

1. Introduction

Traditionally there has been an associated survival benefit for patients on haemodialysis (HD) who have an arteriovenous fistula (AVF) compared to a central venous catheter (CVC) [1]. CVCs in previous studies have been associated with infections, sepsis episodes, and higher rates of mortality in the HD population [2,3]. In the past, this led to a ‘fistula first’ approach where clinical guidelines encouraged aggressive pursuit of arteriovenous access creation for patients requiring HD. [4]. National bodies such as the United Kingdom (UK) Renal Registry routinely collect data on a patient’s incident and prevalent HD access i.e., CVC vs. AVF, allowing comparison between different treatment centres [5,6]. Best practice tariffs in the UK’s National Health Service (NHS) had set targets of 75–85% of prevalent HD patients having established access to either an AVF or AVG, with targets to minimize those starting HD with a CVC [5]. These metrics were often used as a barometer to assess the quality of a vascular access service. However, several factors have emerged that challenge the validity and comprehensiveness of this metric [7].

Rates of both chronic kidney disease (CKD) and end-stage kidney disease (ESKD) are increasing globally [8]. This means there is an ever-rising number of patients being assessed for, and commencing on, HD. Further to the increasing numbers of patients commencing HD, these patients are often older, frailer, and have more comorbidities. In countries such as the United States of America and the UK, the largest increase in patients requiring kidney replacement therapy is in those aged ≥75 years old [6,9]. The vast majority (>90%) of this cohort of patients will be commenced on HD with very few transferring either to peritoneal dialysis or kidney transplantation [6,9]. The evidence base for the survival benefit due to an AVF in these older comorbid HD patients is less clear, where selection bias in studies may overestimate any observed effect [10,11]. Furthermore, older patients with comorbidities such as vascular disease often present unique challenges that can complicate the creation and maturation of an AVF. For instance, maturation rates may be lower and the competing risk of death may lead to mature AVFs that are never used for dialysis, necessitating the use of CVCs as a practical and sometimes preferable alternative [12]. Thus, a rigid focus on incident AVF use may inadvertently penalise centres for making clinically appropriate decisions tailored to individual patient circumstances.

Clinical practice and priorities in both nephrology and vascular access have also shifted significantly. Guidelines now promote a more patient-centred and less paternalistic model for vascular access delivery where patient choice is paramount [13]. There is an awareness of the burden of multiple AVF creation attempts or post-creation interventions to maintain patency on patients’ quality of life and well-being [14,15]. This, coupled with better infection-control procedures and improved catheter designs which have reduced CVC-associated infections, requires vascular access treatment decisions to be more nuanced and holistic [16]. The move away from ‘fistula first’ to ‘right access for the right patient at the right time’ has created challenges in defining what truly represents ‘optimal’ care and how services can audit their practice [17].

Furthermore, arbitrary targets for HD populations dialysing with an AVF ignore the health economic realities of most healthcare systems. There is an ever-increasing demand for vascular surgical and interventional radiology pathways and prioritisation for patients who would definitely benefit from arteriovenous access is crucial [18,19]. Finally, simply comparing the proportion of patients using a CVC vs. an AVF gives little to no information about where possible problems in service delivery lie. It is often unclear whether delayed CKD referral to nephrology, delay in fistula creation pathways or access to other treatment modalities (peritoneal dialysis or transplantation) is affecting the care delivered by a vascular access service [20,21]. Without this information, it can be difficult for services to identify priorities for quality improvement and service reconfiguration.

The aim of this study was to demonstrate a more nuanced technique of auditing the vascular access outcomes for patients commencing HD in a single centre. For patients commencing HD with a CVC, there was a holistic assessment of whether this represented ‘optimal’ or ‘suboptimal’ care. We then describe how this process helped identify priorities for quality improvement within the local vascular access service.

2. Materials and Methods

2.1. Study Design

This was a single-centre retrospective cohort study using a prospectively designed clinical database—the Northern Ireland Vascular Access Database for Chronic Kidney Disease (ethics approval reference number 20/NI/0069). The database prospectively collects clinical and vascular access data of patients with advanced chronic kidney disease. Recorded patient characteristics include gender, age, race, ethnicity, primary renal disease, type of vascular access, pre-operative ultrasound measurements, co-morbidities, RRT status, patient survival, and access survival data.

2.2. Study Setting and Population

The study was set in the Regional Nephrology and Transplant Unit within the Belfast Health and Social Care Trust in Northern Ireland, a region of the United Kingdom. The centre provides dialysis access procedures to the whole of the Northern Irish population. Within the team, there are two consultant nephrologists with a specialist interest in vascular access. They provide the pre-operative duplex ultrasound examination of vascular mapping using a Sonosite M-Turbo ultrasound machine (Sonosite, Bothell, WA, USA) with a high frequency (13–6 MHz) linear probe. All patients for consideration of HD receive vascular mapping with the only absolute contraindication to referral being patient choice. Following mapping, clinical assessment, and discussion with the patient, if a suitable AVF creation site is identified, the patient is then referred to a team of six senior vascular access and kidney transplant surgeons. The service is also supported by a specialist nursing team and an interventional radiology team with experience in vascular access. Patients also have access to a dedicated clinical psychologist. Most CVC insertions at the centre, including tunnelled CVCs, are performed by nephrologists, with a small minority requiring placement by interventional radiology.

The study population included all patients known to nephrology for >90 days prior to initiating haemodialysis as the first-ever renal replacement therapy between 2011 and 2020. The 90-day limit was set to avoid including those initiated on HD due to first presentation with acute kidney injury as part of the centre’s inpatient consult service.

2.3. Data Collection

Key demographic data collected from the clinical database included age, gender, race, and primary renal disease. Renal-specific data was also collected including renal referral date, HD start date, previous access attempts, and AVF outcomes. Each individual case of a patient initiating HD with a CVC was then assessed holistically by the vascular access team to explore whether this represented ‘optimal care’ for this individual or if there were barriers to a more appropriate access option, i.e., ‘suboptimal care’. The results were presented annually to the whole team. Listed below are the questions the team posed to themselves to identify whether the initial access of an HD patient was ‘optimal’ vs. ‘suboptimal’.

(1)

Was the choice of vascular access type (AVF, AVG, or CVC) tailored to the patient’s individual risk factors, including age, comorbidities, and life expectancy?

(2)

Were patient preferences and lifestyle factors considered in deciding the type and timing of vascular access?

(3)

Was the trajectory of kidney function predictable that allowed adequate planning of access type?

(4)

For those starting HD with a CVC, could another more appropriate permanent access type have been created?

(5)

Did disruption to the access creation pathway prevent the individual from commencing HD with their preferred access?

(6)

Was vascular access planned and placed with coordination between nephrologists, surgeons, and the dialysis team to ensure timely access creation and minimize CVC dependence if AVF/AVG was the most appropriate option?

(7)

Were local guidelines followed regarding infection control, access maturation monitoring, and complication management post-placement?

(8)

Did the care team conduct follow-up assessments to evaluate access functionality and adjust care as needed over time?

(9)

Was temporary vascular access appropriately used for a short period of time to bridge a patient to their preferred RRT modality, i.e., transplantation or peritoneal dialysis?

(10)

Were individuals supported adequately to engage with access creation and initial needling of AVF?

Premise: in most centres, including our own, the default modality for ‘crash landers’, including crash landers known to nephrology, is haemodialysis with a central venous catheter. Successfully commencing HD with an AVF, PD, or receiving a pre-emptive transplant requires a coordinated successfully managed pre-treatment pathway. With this in mind, we audit HD catheter starts to ensure they are never accepted as the default for ‘suboptimal’ medical practice/system issues.

Factors taken into account when deciding on ‘optimal’ versus ‘suboptimal’ are as follows:

• Decline in kidney function: was this predictable? If the eGFR was 20 for those aged under 70 years or 15 for those aged over 70 at 3 months pre-dialysis, these patients were considered to have an acute decline and a CVC start ‘optimal’. If eGFR was below these targets and the patient was not referred for vascular access assessment, this was ‘suboptimal’.
• Modality choices: did the patient start on their modality of choice? We acknowledge that on occasion a brief time on HD may be required before PD or pre-emptive transplant. If this was less than 6 weeks, the patient was still considered to have ‘started’ on their chosen modality. If the patient did not get their modality of choice, they were considered to be a ‘suboptimal’ CVC start prompting a review of whether the modality choice was appropriate or whether there was a service delay in providing the modality.
• Surgical pathways: were there any delays in the AVF creation pathway? If there were delays and the patient required HD to start with a CVC, this was considered suboptimal.
• Non-adherence: non-adherence contributing to HD start with a CVC was considered a ‘suboptimal’ start. The rationale for this was to better understand this cohort of patients and ideally improve services to help those who find it more difficult to engage with services.

2.4. Statistical Analysis

Statistical analysis was performed using SPSS Statistics version 24 (IBM Corp, Armonk, NY, USA). Regression models were used to identify variables of statistical significance that predicted a suboptimal line outcome. The independent t-test was used to compare baseline clinical outcomes between the AVF, optimal CVC, and suboptimal CVC groups.

3. Results

Between 2011–2020, 254 patients known to nephrology services >90 days were commenced on HD. This represented 60% of the initiated renal replacement therapy modality with a peritoneal dialysis rate of 17.8% and a pre-emptive transplantation rate of 22.4%. The clinical characteristics of those included in the study can be seen in

Table 1. Clinical characteristics of 254 individuals who commenced dialysis between January 2011 and December 2020 who were known to nephrology services for over 90 days.

Clinical Characteristics	Number (% Total Population)
Age (years), median; range	65; 17–94
Gender
Male	152 (60)
Female	102 (40)
Caucasian	249 (98)
Primary renal disease
Diabetic nephropathy	70 (27)
Glomerulonephritis	47 (18)
Interstitial disease	44 (17)
Other	36 (14)
Unknown	33 (13)
Polycystic kidney disease	24 (9)

. Within this cohort, 96 (38%) of the patients initiated HD with an AVF or an AVG, whilst 158 (62%) patients started HD with a CVC. Following individual case reviews by the vascular access team, the CVC group patients were divided into ‘optimal’ and ‘suboptimal’ groups. A total of 91 patients (58% of the CVC group) were classified as receiving the optimal vascular access management despite commencing HD with a CVC. A total of 67 patients (42% of the CVC group) upon review of their individual cases had potential barriers to them receiving the most appropriate vascular access option. The ‘optimal start’ group (combined AVF/AVG and optimal CVC) represented 187 (74%) of the entire cohort.

For those patients who commenced HD with an AVF/AVG at three months, 13 patients no longer used the permanent access created for the following reasons: six died, four transplanted, two recovered self-supporting kidney function, and one withdrew from dialysis treatment for conservative care management. At 1 year, a further 38 individuals no longer utilised their permanent access (23 transplants, 14 died, one withdrew from haemodialysis).

3.1. Optimal CVC Group

A total of 91 patients who commenced HD with a CVC were deemed to have received optimal access care. A total of 41 patients (45%), who were known to nephrology services and had early-stage CKD for which dialysis access planning would not yet be appropriate, subsequently had an unpredictable acute deterioration in kidney function, e.g., from an episode of sepsis not related to their kidney disease. A total of 25 patients (27%) had been mapped by the access team and found to have no suitable AVF site options and hence proceeded with CVC insertion for HD. Twelve individuals (13%) had an AVF creation attempt which failed to mature and had no further options for AVF creation. Seven patients (8%) were commenced on HD with a CVC for a short period of time (<90 days) to optimise them prior to switching to their preferred kidney replacement therapy option of either peritoneal catheter insertion (n = 2) or kidney transplantation (n = 5). Finally, six patients (7%) had a needling complication at the time of commencing HD that required the use of a CVC for a period of <4 weeks before CVC removal and successful use of the incident AVF.

3.2. Suboptimal CVC Group

Of the 67 patients who initiated HD with a CVC, 25 individuals (37%) either had no referral to the access team or were referred late, i.e., close to requiring HD. This is despite being known to nephrology services. It was unclear in the cases of non-referral to the access service if this represented patient choice or a clinician’s judgement that AVF was not in the patient’s best interests. A total of 19 patients (28%) appeared to have switched their dialysis preference from peritoneal dialysis/conservative care to HD shortly before requiring HD. A total of 13 patients (19%) did not receive an AVF in time for commencement on HD due to delays in the AVF creation pathways. A total of 10 individuals (15%) commenced HD with a CVC due to a lack of patient engagement with the vascular access service. This may have reflected a patient’s personal preference, but there were individuals for whom the reviewing team felt better psychosocial support may have assisted these patients receive a more definitive vascular access plan.

3.3. Predicting Patients at Risk of Commencing HD with a Suboptimal CVC

The results of the logistic regression analysis of multiple clinical variables exploring predictive variables for suboptimal CVC are shown in

Table 2. Logistic regression analysis of predictors of receiving a suboptimal CVC as initial HD access.

Clinical Variable	Odds Ratio	95% Cl	p Value
Age	1.53	0.47–2.31	0.57
Female Gender	1.87	1.03–2.05	0.36
Non-white race	0.03	−0.22–0.39	0.86
Diabetic Nephropathy	2.34	1.65–3.11	0.20

. No clinical variables examined were significantly associated with being in the suboptimal CVC group.

The cohort was divided into quartiles of time known to nephrology services. Those known the shortest amount of time to a nephrologist (Q1) and those known longest to the service (Q4) appeared to be at increased risk of suboptimal access. Q1 OR = 8.39 (95% Cl 2.02–12.54 p = 0.02) Q4 OR = 2.67 (95% Cl 1.39–5.12 p = 0.04).

3.4. Patient Survival

Two-year mortality was 53% for optimal CVC HD start, 37% for suboptimal CVC, and 30% for AVF start. There was no difference in mean survival between AVF and suboptimal group (2.53 vs. 2.21 years, p = 0.31). There was a survival difference between AVF versus total CVC (2.53 vs. 1.97 years, p = 0.002) and suboptimal versus ‘optimal’ CVC cohorts (2.21 vs. 1.40 years, p = 0.16). This potentially demonstrates that the cohort in the optimal CVC group was frailer than those in the AVF and suboptimal line groups.

4. Discussion

The results of this study demonstrate the utility of a more holistic approach in assessing whether vascular outcomes are optimal or suboptimal for individual patients. This is particularly pertinent given the increasingly complex and comorbid patient population initiating HD. Notably, the lack of significant clinical predictors for suboptimal CVC use underscores the complexity of these cases and suggests that factors beyond the traditional clinical variables may influence access outcomes. Whilst risk factors for unplanned dialysis initiation have been well-described in the literature, patients who are at risk of poor vascular access planning despite being known to nephrology are less well understood [21,22].

On initial review of access outcomes from our centre, only 38% of patients commencing HD with an AVF would fall below previously reported targets. These rates would be consistent with other modern centres that reported AVF at initiation HD rates [6,20]. However, our analysis demonstrates that at least 74% of patients commencing HD likely received the optimal access for their individual requirements. Further to this, our data needs to be understood in the context of coming from a centre with high pre-emptive kidney transplant (22.4%) and peritoneal dialysis rates (17.8%) [23,24], suggesting that our HD population is particularly frail. The finding that patients with both the shortest and longest durations of nephrology follow-up were at increased risk for suboptimal dialysis access has been described in other studies and indicates that timing and coordination of care are critical factors impacting vascular access outcomes [20,25]. Within our current database, we do not capture AVF/AVG that are made but never utilised. This would be an additional important metric to measure to further understand the appropriate use of the access services. It is important to note that, following 1-year of initiation with HD with an AVF/AVG, 25% of this cohort were transplanted and 15% had died.

The observed difference in survival further complicates the traditional narrative that AVF use is always superior. While patients with AVFs had better survival outcomes compared to those with CVCs, the differences in survival between the optimal and suboptimal CVC groups suggest that factors such as frailty and comorbidities play a significant role in patient outcomes. The higher mortality in the optimal CVC group, despite the CVC being deemed appropriate, likely reflects the underlying health status of these patients rather than the type of access itself. These findings would be consistent with the modern interpretation of the effect of vascular access on survivability in HD populations [26,27]. In elderly haemodialysis patients (≥80 years), transitioning from an initial CVC to an AVF within the first year of dialysis shows a similar mortality risk compared to those who start with an AVF, suggesting this transition can be a viable approach [28]. However, prolonged use of CVCs is associated with significantly increased mortality. This again likely reflects the reduced frailty of those who are suitable/selected for AVF creation compared to those who are not.

For those with a limited prognosis on dialysis, the ethos of our unit is to avoid multiple attempts at AVF creation. Our goal is a maximum of two options to achieve functioning access (e.g., one attempt at AVF, then a CVC; PD catheter, then a CVC). This ethos is reflected in those who started HD over the age of 80 years. There were 57 patients aged over 80 years. In this group, 28 (49%) had pre-dialysis AVF creation; 26 had only one AVF creation; and 2 had 2 AVF creations and used their 2nd AVF as incident access. In total, 20 patients started HD with their AVF, 4 further patients eventually successfully used the AVF created pre-dialysis, and 4 patients’ AVFs failed.

As a team, the process of identifying both optimal and suboptimal CVC outcomes for our patients was instructive where potential service improvements could be made. Within the optimal cohort, 25 patients did not have suitable veins for AVF creation. Whilst this reality could not be changed for those individuals, it has prompted a reinvigorated vein preservation strategy for the whole of Northern Ireland by the access team. As a service, we are also attempting to expand our use of AVG in patients who do not have suitable veins for AVF creation. In the suboptimal group, the high number of individuals changing modality preference at a late stage in treatment planning has prompted us to review our dialysis education information, particularly assessing its accessibility to individuals with lower health literacy. We have started assessing within our CKD population how potential socioeconomic factors may act as potential barriers to optimal vascular access care and we are utilising clinical psychology expertise to assist patients to actively engage with the service. We are also exploring how to best record patient preferences with regard to vascular access and measuring vascular access-specific Patient Reported Outcome Measures (PROMS) have been recommended by the Kidney Health Initiative to enable the patient voice to be heard [29]. Vascular access guidelines have increasingly emphasised the importance of incorporating frailty, predicted dialysis duration, and tools like the Kidney Failure Risk Equation (KFRE) to optimize individualized care. Recent studies suggest that the KFRE may play a critical role in future vascular access planning by improving risk stratification and guiding clinical decision-making [30,31], again highlighting the ever-increasing number of factors that require assessment prior to access creation.

Our work has several limitations beyond those inherent to being a single-centre retrospective study. There is obvious potential for selection bias, where our own team decided whether CVC use was optimal versus suboptimal. The small sample size may mean the study lacks sufficient statistical power to detect clinically significant differences between groups and prevents subgroup analysis. Finally, the study covers nearly a decade of evolving clinical practice where priorities and guidelines in vascular access creation have changed significantly.

Future research should focus on refining the criteria for optimal versus suboptimal CVC use and exploring interventions that can reduce the barriers to AVF creation where appropriate. Additionally, studies that further investigate the role of patient preference and quality of life in vascular access decisions will be crucial in developing guidelines that truly reflect the needs of the modern HD population.

For any service to maintain standards and ultimately improve care, it is crucial that there are indices and measurements to audit. Unfortunately, the traditional tools used by vascular access services to monitor quality of care no longer fit with modern clinical practice. We would argue the approach we have outlined above could be a useful initial step but would be further improved with increased focus on factors such as health literacy, socioeconomic barriers, patient preference data, and quality of life measurements. We intend to complete work assessing these aspects in our own centre and work collaboratively with colleagues in other units to further understand how best to measure the performance of a vascular access service.