Fistulas are considered the preferred form of vascular access and are widely promoted by vascular access experts and guideline committees. These recommendations are based on observational data showing an association between fistula use and improved survival, lower procedure rates, and lower costs. There has never been a randomized controlled trial comparing the outcomes of patients treated with fistulas to those treated with catheters, and it was not clear if nephrologists would be willing to participate in a trial.

This survey suggests that nephrologists in Canada, Europe, and Australia/New Zealand would be willing to participate in a future randomized controlled trial comparing outcomes of patients treated with fistulas to those treated with catheters.

The majority of people with kidney failure are treated with hemodialysis and require access to the bloodstream. While vascular access (“access”) is a lifeline, it is also a key driver of morbidity, mortality, and cost [1]. Arteriovenous fistulas, arteriovenous grafts, and central venous catheters are the three main options for vascular access, but the majority of hemodialysis patients use either a fistula or catheter [2]. Guidelines recommend a fistula for of access based on observational data indicating that use of fistulas is associated with a lower risk of complications and access-related costs and longer patient survival [1, 3–9]. This had led to a number of local, regional, and national initiatives designed to increase the proportion of individuals treated with them [10].

The observational literature comparing fistulas to catheters has important limitations that may have biased previous comparisons and may be difficult to overcome in the absence of randomization [11, 12]. Patients who start dialysis urgently, those who are acutely ill, and those with the highest burden of comorbidity are treated almost exclusively with catheters. Studies including these patients in the catheter group are at high risk of indication bias. Further, most studies did not ascertain access outcomes rigorously or did not report them comprehensively. These studies are at high risk of ascertainment and information bias [13].

The adoption of health-care interventions in the absence of sufficient evidence of safety and efficacy can lead to inefficient use of health-care resources and, potentially, unintended harm to patients. For example, increasing the dose of dialysis delivered to patients was associated with better outcomes in observational studies, but subsequent randomized trials have failed to show benefit [14, 15]. Normalizing hemoglobin levels in patients with kidney disease using erythropoietin-stimulating agents was also felt to be desirable until clinical trials showed that it increased the risk of cardiovascular events [16–21]. However, it is not clear if physicians would be willing to participate in a clinical trial comparing fistulas to catheters given that the observational data have shown a strong and consistent benefit of fistulas, there are incentives for dialysis programs to increase the numbers of patients treated with them in many jurisdictions, and the proportion of patients treated with a fistula is considered a quality measure.

We surveyed nephrologists from Canada, Europe, Australia, and New Zealand to assess their willingness to participate in a randomized trial comparing fistulas to catheters in incident hemodialysis patients. We also sought to elicit physician views about vascular access, their familiarity with the literature, and their opinions regarding current guideline recommendations.

A total of 267 nephrologists and nephrology trainees participated in the survey. Of those, 248 respondents fully completed the survey (including the question on their willingness to participate in a trial). The majority of respondents were from Canada (68 %), followed by Australia/New Zealand (21 %) and then Europe (12 %). As the total number of survey recipients from Canada, Europe, and Australia/New Zealand was not disclosed, we were unable to calculate the final response rates.

Willingness to participate in a future RCT

Figure 1 illustrates the respondent’s willingness to participate in a randomized controlled trial comparing fistulas to catheters. In total, 79 % of the total respondents agreed that a future trial was needed. While the majority of respondents from Europe (66 %) and Australia/New Zealand (64 %) were supportive of a trial, the proportion in favor was lower than in Canada (86 %).

Thirty-eight percent of nephrologists who were in favor of a trial were supportive of a study in all hemodialysis patients who were candidates for a fistula (37 % in Canada; 45 % in Europe; 40 % in Australia/New Zealand), while 62 % of nephrologists indicated that it should be restricted to individuals at high risk of primary fistula failure (63 % in Canada; 55 % in Europe; 60 % in Australia/New Zealand).

Our findings confirm that the majority of nephrologists surveyed from Canada, Europe, and Australia/New Zealand would be willing to participate in a randomized trial comparing fistulas to catheters in incident hemodialysis patients. However, the majority of respondents felt that it should be limited to patients who are at high risk of primary failure of their fistula. Nephrologists from Canada, those who had been in practice for a short duration of time, practiced in an urban setting, and self-identified as vascular access experts were more likely to indicate a willingness to be involved in a trial comparing fistulas to catheters. While fistulas are aggressively promoted, our results would suggest that many practicing nephrologists feel that there is sufficient equipoise to warrant a randomized comparison of catheters and fistulas.

The majority of nephrologists who were in support of a randomized trial comparing catheters to fistulas felt that it should be restricted to individuals with a higher risk of primary failure. Risk factors for failure to mature include age over 65, female gender, the presence of peripheral vascular disease and coronary artery disease, obesity, diabetes, location of the fistula, small-caliber veins, and ethnicity other than Caucasian [22]. A high risk of primary failure may alter the risk-benefit ratio in those undergoing a fistula attempt, as 25–69 % of fistulas created will fail to mature, depending on the patient’s risk profile [22]. If a fistula never works, the patient does not have the opportunity to experience the potential benefits of avoiding a hemodialysis catheter. In addition, failure of fistula maturation has previously been shown to lead to repeated surgical interventions, disruption of the hemodialysis schedule, and patient distress [22]. It is also associated with inefficient use of health-care resources; a trial that tested the benefits of attempting a fistula in this high-risk patient population would inform clinical practice and help providers select the patients most likely to benefit.

Physicians from Canada were more likely to indicate a willingness to participate in a randomized trial comparing fistulas to catheters. While guideline recommendations from Canada, Europe, and Australia/New Zealand consistently recommend fistulas as the preferred form of vascular access for hemodialysis, the prevalent use of fistulas varies. The DOPPS study showed that the prevalence of fistulas in Canada was 50 %, whereas in Europe the range was 57–83 % [23]. In New Zealand, the rates were closer to 60 % [24]. This may be due to differences in the availability of infrastructure or the access to, or organization of services, although our results would suggest that the infrastructure to support fistula creation and maintenance are more developed in Canada than in European centers. It more likely reflects differences in the attitudes and beliefs of health-care providers.

Interestingly, nephrologists who have practiced for 5–10 years were more likely to participate in a trial. Those who were new to practice (<5 years) and those who had been in practice longer (>10 years) were less likely to be willing to participate in a clinical trial. The reasons for this are not clear. New physicians may be more likely to adhere to the principles taught to them during training, whereas clinicians who had been in practice for a long time were more likely ingrained in their practice patterns and less open to the idea of a clinical trial. Individuals who spent more of their time in direct patient care were less likely to want to participate in a trial. The reasons for this are not clear. It may be that physicians with a more clinical focus are more likely to adhere to the teaching they received during their fellowship training and less likely to read the primary literature for themselves or to have had additional training in critical appraisal. This is a potential threat to the feasibility of an RCT, given that the majority of patients are cared for by this group of physicians. Finally, self-identified experts likely have a greater familiarity with the vascular access literature. As a consequence, they may be more aware of the limitations of existing studies—that they are of low quality and prone to bias—and may be more willing to participate in a clinical trial [4]. However, experts are more likely to be involved in the generation of guideline recommendations, and there appears to be discordance between the views expressed in our survey and current guideline recommendations.

Our study has several strengths. First, we recruited respondents from a number of countries in order to explore geographical differences in opinions. Second, this survey is the first, to our knowledge, to ask physicians directly about their willingness to participate in a randomized controlled trial comparing fistulas and catheters. Demonstrating the presence of equipoise is a necessary step in justifying an RCT examining this issue. Our results highlight the discordance between guideline recommendations that consistently recommend fistulas and the views of nephrologists who responded to our survey. Third, our results provide insight into the issues felt to be important considerations in the design of a RCT from the perspective of the practicing nephrologists. This facilitates development of a trial protocol that is feasible and maximizes the likelihood of participation.

Our study also has limitations. The intent of this survey was to capture the perception of nephrologists internationally to inform the development of a future trial. Outside of Canada, we distributed the survey via the European Renal Association (ERA) and the Australian and New Zealand Society for Nephrology. European participants were identified by the ERA and were typically vascular access experts from different countries. This ultimately led to a discrepancy in numbers of European to non-European respondents overall, even though Europe has a much larger population. It was however, felt that European respondents’ views were likely reflective of physician views in their countries of origin, but we are unable to confirm that. As a consequence, Canadian views are likely over-represented in the overall outcomes, but similar results were found for each of the geographic regions examined. The response rates are not calculated because we were unable to obtain an accurate count of the physicians that the survey was distributed to in any region. In addition, we attempted to engage US participants in the survey but were unable to do so. As a consequence, we were unable to gauge physician views and opinions in the USA, which would have been of interest. Our results reflect the views of those physicians who responded to our survey and may not reflect the views of non-respondents or an individual physician’s actual practice. Finally, we have only captured the views of nephrologists. The willingness of patients to be assigned randomly to different vascular access strategies has not been explored and is an important consideration. In a recent study by Manns et al., patients were engaged to identify their priorities for research in kidney disease and identified vascular access as a key area of investigation [25]. Although this does not imply that these patients would participate in an RCT, it does show the awareness regarding the importance of further research in this area from a patient perspective. We are currently testing the feasibility of enrolling patients in a randomized trial that will randomly assign people starting hemodialysis with a catheter to either attempt fistula creation or continued use of a tunneled catheter (NCT02675569).

In summary, the majority of nephrologists we surveyed from Canada, Europe, Australia, and New Zealand would be willing to participate in a randomized controlled trial comparing fistulas to catheters in incident hemodialysis patients. Our survey results suggest that nephrologists believe that the existing evidence forming the base of guideline recommendations about vascular access is of low quality, that there is equipoise surrounding the optimal vascular access strategy for hemodialysis.

The optimal hemodialysis vascular access is not yet known. The current trend towards fistulas over catheters is based on observational evidence that is limited in scope and prone to bias. A randomized control trial comparing fistulas and catheters would be the most objective and comprehensive study to prove the superiority of one form of access over the other. We surveyed nephrologists from Canada, Europe, Australia, and New Zealand to assess their willingness to participate in the aforementioned trial. Although the number of Canadian respondents was greater than those from other regions, the distribution of responses was similar in all regions. That is, a majority of our respondents were willing to participate in a randomized trial but more so if limited to patients who are at high risk of primary fistula failure.

ANZSN, Australia and New Zealand society of nephrology; CSN, Canadian Society of Nephrology; DOPPS, Dialysis Outcomes and Practice Patterns Study; ERA/EDTA, European Renal Association/European Dialysis and Transplantation Association