Impact of arteriovenous fistula aneurysms on a UK dialysis populations’ perception of vascular access

Abstract

Background

Clinically-oriented outcome measures are increasingly being recognized as lacking in consideration of factors important to patients. There is an emerging move of guideline bodies advocating a more patient-centred approach. Aneurysms in autogenous arteriovenous fistula (AVF) can be considered unsightly and a constant reminder for patients of their dependence on dialysis. However, their impact on patient’s perception has not previously been reported.

Methods

Between April 2017–18, the Vascular Access Questionnaire (VAQ) was administered to prevalent haemodialysis patients across ten dialysis units via structured interviews, as part of a quality improvement project. Data for the subgroup of patients with aneurysmal AVF (categorised as per classification by Valenti et al.), were retrospectively evaluated and compared to the wider cohort.

Results

Data were collected for 539 patients (median age: 66 years; 59% male), of whom 195 (36%) had aneurysmal AVF, with Type 2 morphology (cannulation site) being the most common (75%). Duration of AVF was found to be significantly associated with aneurysmal development, with estimated likelihoods of 11%, 43% and 61% after one, five and ten years, respectively. Interestingly, patients with diabetes had a significantly lower prevalence of aneurysmal development than those that were non-diabetic (25% vs. 43%, p < 0.001). Overall VAQ scores were not found to differ significantly by aneurysm status (p = 0.816) or across morphology types (p = 0.277). However, patients with aneurysmal AVF were significantly more concerned with the appearance of their AVF (p < 0.001) than the wider cohort. Despite this, patients with aneurysmal AVF gave significantly higher scores for satisfaction and ease of use and lower scores for bruising and clotting (p < 0.05).

Conclusions

Aneurysmal AVF are often cited as an important factor by patients for not proceeding with fistula formation. In this evaluation of patient reported experiences, those with aneurysmal AVF reported high satisfaction levels. This may help clinicians highlight positive patient reported outcomes of aneurysmal AVF during preprocedural consent processes.

Background

Guidelines and improvement initiatives regarding a permanent form of vascular access (VA) for dialysis suggest an optimum access pathway of arteriovenous fistulas (AVF) over arteriovenous grafts (AVGs), with tunnelled central venous catheters (TCVCs) as a last choice [1,2,3,4,5]. However, emerging updates have suggested a trend towards a more patient tailored theme of “the right access for the right patient at the right time” [6]. Clinically orientated “hard” outcome measures are increasingly being recognized as lacking in consideration of factors important to patients, and may not be the only way of comparing options for the patient [7,8,9].

Aneurysms within AVF are reported as being relatively common, although the reported incidence differs considerably between studies, ranging from 5–60% [10,11,12,13,14]. These may necessitate surveillance, and potentially require intervention [15]. Whilst there are debates on the most appropriate definitions and classifications of AVF aneurysms, consensus opinion seems to suggest an expansion of the vessel wall to over 18 mm is a reasonable definition [11, 16]. AVF aneurysms are a very visible and often considered unsightly reminder of a patient’s dependence on dialysis. The cosmetic aspects are cited as a potential reason by patients for refusing AVF [17, 18].

The reporting of patient perspectives is slowly increasing in the published literature [19,20,21,22,23]. Whilst descriptions and classifications have been developed for aneurysms in AVF, to the best of our knowledge, the impact of these from the patient’s perspective has not been extensively reported [11, 15]. As part of a quality improvement audit, the Vascular Access Questionnaire (VAQ), described by Quinn et al. and consisting of a patient-reported questionnaire composed of 17 VA related questions, which are scored on a 0–4 Likert scale, was administered to a large regional dialysis population at a tertiary referral centre, to identify patient characteristics that influence their perceptions and outcomes of VA for dialysis [21].

In this paper, we analysed the subgroup of patients with clinically-identified aneurysmal AVF, and compared their results to the remainder of the AVF cohort, as well as comparing across the different morphological types. This analysis included demographic factors, presence of other medical conditions, duration of dialysis and previous access, satisfaction with access and ease of use.

Methods

Between April 2017 and April 2018, the VAQ was administered to established haemodialysis patients at ten dialysis units within a UK region (West Midlands) served by a single tertiary referral centre. The aim of this analysis was to further characterise the sub-group of patients with clinically identified aneurysmal AVF and compare their results to the remainder of the cohort with arteriovenous fistulas without aneurysms, and to compare between the different aneurysmal types.

All patients undergoing haemodialysis in the centres who were able to consent and did not present a linguistic barrier were approached. Patients undergoing acute in-patient dialysis or home haemodialysis were excluded. Data collected also included clinical diagnosis of aneurysmal AVF presence, based on physical examination by members of the team administering the questionnaire. These were classified according to the types as proposed by Valenti et al. [11]. This classification was chosen as it is undertaken on the basis of external appearance and without the requirement for ultrasound assessment.

The VAQ was verbally administered in the patient’s preferred language by clinicians who specialise in working with the target population and disease process to patients whilst they were undergoing dialysis. Questions were asked using the terminology and phrases in common use by clinicians at a patient’s dialysis unit; this was considered sufficient for the purposes of linguistic validation in the context of this study. Data relating to patient health and co-morbidity, including the presence of diabetes and its management, was gained on direct questioning. For patients who elected not to participate, the reasons for non-participation were recorded, and have been reported previously, along with further details of the survey process [24]. Data was collated and managed using REDCap electronic data capture instrument hosted at the University of Birmingham, and exported for analysis to IBM SPSS version 22 (IBM Corp. Armonk, NY), and GraphPad Prism version 7.0 (GraphPad Software, San Diego, Calif. USA) [25]. Institutional audit committee approval was obtained (CARMS-12695).

Statistical methods

Comparisons between patients with and without aneurysmal AVF were performed using Mann-Whitney U tests for continuous or ordinal variables. Nominal factors were analysed using Fisher’s exact test, where possible, with Chi-square tests used where there were too many categories for an exact test to be calculable. These analyses were then repeated for the subgroup of patients that developed aneurysmal AVF, to allow comparisons across the aneurysm morphological types. Further assessment of significant variables was performed using univariable binary logistic regression models. Prior to the analysis, the goodness of fit of factors were assessed using Hosmer-Lemeshow tests, with transformations (e.g. logarithmic) applied where significant poor fit was detected. Throughout the analysis, p < 0.05 was deemed to be indicative of statistical significance.

Results

Patient demographics

Of the dialysis cohort of 920 patients, 749 completed the questionnaire. Of those who did not complete the questionnaire 64.3% (n = 110) were not present at the time of the visit to their dialysis unit, 16.4% (n = 28) did not want to participate and 19.3% (n = 33) could not consent. From this cohort of 749 there were 539 patients who were achieving dialysis through a native arteriovenous fistula were included in the analysis. These patients had a median age of 66 years (Interquartile range (IQR): 56–76), and the majority were male (59%) and of White (50%) or Asian (36%) racial background. The median duration of haemodialysis was 4 years (IQR 2–7), of which a median of 3 years (IQR 1–6) was on the current AVF. Most patients had brachiocephalic (51%) or radiocephalic (39%) AVF, with brachiobasilic AVF in the remainder (10%). All patients underwent standard sharp needles for cannulation with the aim of developing ropeladder cannulation tracts. A total of 195 (36%) AVF were clinically classed as aneurysmal, of which 40 (21%) were of type 1 morphology, 146 (75%) were type 2, and 9 (5%) were type 3, which was in keeping with reported literature. Due to the small number of type 3 aneurysms, these were combined with type 2 aneurysms for subsequent analysis; this was deemed clinically appropriate since type 3 aneurysms are a natural progression from type 2b.

Factors associated with aneurysmal AVF

Of the demographic factors, aneurysmal AVF rates were not found to differ significantly with age (p = 0.255), gender (p = 0.927) or racial background (p = 0.198) (Table 1).

Table 1 Comparisons of patient demographics between those with non-aneurysmal and aneurysmal AVFs, and across aneurysm types

Patients with diabetes were found to be significantly less likely to have aneurysmal AVF, with rates being lowest in tablet-controlled diabetics at 15%, compared to 43% in non-diabetics (p < 0.001). There was also considerable variability in the aneurysmal AVF rates across the ten centres included in the study, ranging from 19 to 48%, although this did not reach statistical significance (p = 0.053). Within those that had developed aneurysms, none of the demographic factors considered were found to be significantly associated with the type of morphology (Table 1).

Of the dialysis-related factors considered, aneurysmal AVF rates were not found to differ significantly between the different AVF types (p = 0.659) (Table 2). This is perhaps interesting given the perception that radiocephalic fistulas are traditionally felt to be more at risk of aneurysmal change, however, it may be that this reflects the longevity of the radiocephalic fistulas in comparison to the brachiocephalic however this was not adjusted for in this cohort.

Table 2 Comparisons of dialysis-related factors between those with and without aneurysmal AVF, and across aneurysm types

However, patients with aneurysmal AVF were found to have significantly longer durations of haemodialysis (median 7 vs. 3 years, p < 0.001) and of the current access (6 vs. 2 years, p < 0.001). Further analysis of the former returned an odds ratio of 2.13 (95% CI: 1.80–2.53, p < 0.001) per doubling of haemodialysis duration, with an estimated likelihood of developing aneurysms of 11%, 43% and 61% after one, five and ten years of dialysis, respectively (Fig. 1).

Fig. 1

Association between the duration of haemodialysis and aneurysmal AVF rates. Points and whiskers represent the observed rates within percentiles of the distribution, with 95% confidence intervals, and are plotted at the midpoint of the interval. The trendline is from a binary logistic regression model, with Log₂ [duration of haemodialysis] as a covariate. The log-transformation was applied in order to improve model fit, as the Hosmer-Lemeshow test indicated significant poor fit when using the untransformed variable (p = 0.009). The resulting model returned an odds ratio for aneurysmal AVFof 2.13 (95% CI: 1.80–2.53, p < 0.001) per doubling of haemodialysis duration

Patients with type 2–3 aneurysmal AVF had significantly longer durations of haemodialysis (median: 7 vs. 5 years, p = 0.046) and of the current access (6 vs. 4 years, p = 0.008) than those with type 1 aneurysms (Table 2). This subgroup analysis also found that, where aneurysms developed, these were significantly more likely to be type 2–3 in radiocephalic AVF (90%), than in brachiocephalic (72%) or brachiobasilic (79%) AVF (p = 0.010). Patients who had undergone radiological intervention in the previous year were significantly less likely to have had type 2–3 aneurysms than those that did not (68% vs. 83%, p = 0.047).

Regarding patient views, those who reported greater degrees of satisfaction (p = 0.022) and ease of use (p = 0.010) were significantly more likely to have aneurysmal AVF (Table 3).

Table 3 Comparisons of patient views between those with and without aneurysmal AVF, and across aneurysm types

Within those developing aneurysms, none of the patient views considered were found to be significantly associated with the classified type of aneurysmal morphology.

VAQ scores

For the whole cohort, the VAQ score followed a positively skewed distribution, with a median of 3 (IQR 1–7) and mean of 5.07. The total VAQ score was not found to differ significantly between those with and without aneurysmal AVF (mean: 5.22 vs. 4.99, p = 0.816, Fig. 2a) or between type 2–3 and 1 aneurysms (mean 5.06 vs. 5.80, p = 0.277, Table 4).

Table 4 Comparisons of the VAQ score between those with and without aneurysmal AVFs, and across aneurysm types

Fig. 2

Comparisons of the Vascular Access Questionnaire (VAQ) score between those with and without aneurysmal AVF. Data for the overall VAQ score is plotted, along with components of the score that were found to differ significantly in those with aneurysmal AVF, as per the analysis in Table 4. Unlabelled bars represent < 10% of the cases

Analysis of the individual components of the VAQ score found patients with aneurysmal AVF to have significantly lower (better) scores for bruising (mean: 0.24 vs. 0.33, p = 0.030) and clotting (0.07 vs. 0.18, p = 0.004). However, patients with aneurysmal AVF had significantly higher (worse) scores for the appearance component of the score (mean 0.65 vs. 0.25, p < 0.001, Fig. 2b). A subgroup analysis on the appearance component of the VAQ found those with type 2–3 aneurysms to have significantly lower (better) scores than those with type 1 aneurysms (mean: 0.55 vs. 1.03, p = 0.016).

Discussion

Within this large cohort of patients, the overall aneurysmal AVF rate was 36%, as evaluated clinically on physical examination. This is broadly in line with the quoted incidence reported by Valenti et al., and is the classification we at our centre use in our daily practice [11].

The lower rate of aneurysm formation in AVF amongst diabetics is consistent with larger studies investigating the association with diabetes and lower rates of abdominal aortic aneurysm development, or slower growth, although the reasons for this are not entirely clear [26]. The potential for diabetic medication to slow aneurysm growth has been suggested based on experimental and clinical data, and this would appear to be supported in this data on AVF [27]. However, it must be acknowledged that development of aneurysms in the setting of AVF is associated with cannulation in contrast to their development in aortic aneurysm disease [16].

This dataset confirms the logical conclusion that the likelihood of developing an aneurysm is increased with increasing longevity of the AVF. For the purposes of pre-operative consent, it may be useful to give an indication that for an AVF that is still being used for ten years, there is a 60% chance of it becoming aneurysmal. Similarly, within those patients that developed AVF aneurysms, those with type 2–3 morphology tended to have longer AVF duration.

Patients with aneurysmal AVF reported significantly higher satisfaction with their access while at the same time reporting worse scores for appearance, especially in type 1 morphology. Hypothetically, this might relate to aneurysmal AVF being inherently easier to cannulate. The combination of these two findings suggests that, although the appearance of aneurysmal AVF is a deterrent to creation of AVF for patients, this is offset by an improvement in ease of use and may not be sufficient to render patients dissatisfied with their access in the longer term. It should be recognised however, that patients may be satisfied with their fistula but may not offset this with the potential need for further intervention due to the aneurysm. As such, the ability to create an AVF with a lower risk of aneurysmal formation would be of benefit in this situation, if the appearance concern is a significant enough issue to deter AVF formation [28]. In addition, the high level of satisfaction with aneurysmal AVF would be a useful component of discussion with new patients at the outpatient clinic.

Wide variation existed between our units in terms of rate of aneurysmal AVF (19–48%), which narrowly missed statistical significance (p = 0.053). It is conceivable that this difference between units could relate to different practices and levels of experience of dialysis nurses with respect to cannulation and represents an area of practice that could be targeted to improve outcomes. As particular cannulation practice (buttonhole, rope-ladder, area-cannulation or combinations) was not captured at initial data collection, this would require further exploration, and would be targeted in the subsequent audit and quality improvement cycles.

In keeping with traditional teaching on arterial aneurysms, size based quantitative analysis may be a useful prognostic factor, and merit evaluation quantitatively. However, this was not within the scope of this evaluation of patient perceptions of AVF aneurysms. It could additionally be argued that these vary widely, and a size cut off of growth in millimetres or increase per year may not be an appropriate measure to use in patient satisfaction evaluation. Capturing differences in perceptions may be more appropriate when using a clinical physical evaluation-based classification, such as that proposed by Valenti el al., which was the approach used in the present study. Such an approach may also be more in line with what the patient is “seeing” on their arm.

The authors acknowledge the importance of aneurysmal morphology quantification and risk factors such as cannulation practice, high flow, pre-existing comorbidities and temporal trends in combination with size-based classification. The authors also acknowledge that the VAQ was administered only once, and further longitudinal analysis to capture changing patient views alongside the natural history of AVF that evolve into specific morphology is required. This is an often neglected area of study, which in our experience is labour intensive and has multiple influencing factors. Nevertheless, we believe this would be useful in evaluating management methods or possibly cannulation techniques for reducing risks associated with AVF aneurysms and would form part of future work.

The authors are not aware of any previous published studies reporting in such detail of a large dialysis population of over 500 patients and aneurysmal AVF, catered to by a single tertiary centre. We equally do also acknowledge this forms an early building block for this evidence base and further work is required to gather pertinent data on patient perspectives.

Conclusions

This is the first report the authors are aware of that examines the impact of aneurysmal AVF from the patient perspective. In this cohort, whilst the presence of an aneurysm was associated with concerns regarding appearance, interestingly, it did not reduce patients’ overall satisfaction with their AVF. This might be expected, given the perception that these AVF are easier to cannulate, and this may be a positive aspect to perhaps highlight during the consent process. Our findings highlight the importance of understanding the patient’s perspective on cosmesis as well as the day-to-day functionality perceived with aneurysmal AVF.

Rates of aneurysmal formation varied between units, which certainly merits further research. Further quantitative analysis such as sizes, cannulation practice, or survival analysis specific to this cohort would be useful. If the reasons for such differences could be quantified, (e.g. cannulation practices), then this could form the basis for an educational or procedural intervention to help enact change in practice, in line with patient satisfaction.