Preferred haemodialysis vascular access for diabetic chronic kidney disease patients: a systematic literature review

Introduction

The incidence and prevalence of end-stage kidney disease (ESKD) has been growing over the last decade by 4%-8% per annum worldwide, with diabetes mellitus (DM) as one of the leading causes (1). In parallel, the number of surgical and interventional procedures required to establish and maintain the arteriovenous vascular access for haemodialysis (HD) keeps rising every year (2). Despite many efforts, many patients are still dialysed on a permanent tunnelled catheter, although there is considerable geographic variation. DOPPS I data indicate that in Europe, HD patients were threefold more likely to have an autogenous arteriovenous fistula (AVF) as compared to North America. However, between DOPPS I and III, AVF use increased to 47% in the United States and decreased slightly from 80% to 74% in Europe (3). Actually, the proportion of prevalent HD patients with permanent catheters in Europe has been estimated to be as high as 25% (4). The increase of co-morbidities such as DM (from 18% to 33%) and vascular disease (from 22% to 34%) in HD patients between DOPPS I and III probably led to higher proportions of patients at risk for unsuccessful AVF creation.

In an effort to improve vascular access outcomes, the National Kidney Foundation and the European Renal Association/European Dialysis and Transplant Association published guidelines for vascular access (5, 6). Based on these recommendations, a special project was launched in the United States, known as the Fistula First Breakthrough Initiative (http://www.FistulaFirst.org/). The purpose of this initiative was to increase the likelihood that every patient would receive an autogenous vascular access. The Work Group however recognized after a while that in some cases, the “fistula first at all costs” approach leads to non-maturation and access failure despite repetitive interventions in certain subgroups, including diabetics, elderly and those with peripheral vascular disease (3, 4). Therefore, it is uncertain whether attempting to create a fistula first in these high-risk patients is the most cost-effective or optimal solution for each individual.

With this background in mind, we performed a systematic review of the available evidence to clarify what is the most advisable strategy of vascular access planning in diabetic ESKD patients (with respect to catheter type, autogenous fistula or graft and position) in terms of impact on patient- and technique-centred outcomes.

Methods

Data source and search strategy

MEDLINE, EMBASE and CENTRAL databases were searched for English-language articles without time restriction through focused, high-sensitive search strategies (Supplementary Table I, available online at vascular-access.info). References from relevant studies and reviews published on the same topic were screened for supplementary articles.

Study selection

We included any study providing outcome data on diabetics on chronic HD treatment on the basis of the type of vascular access primarily attempted. Studies were considered without restrictions of duration of follow-up. Diabetes (type I or II) was considered when it was either a cause of end-stage renal disease or a superimposed condition. We considered any possible type of vascular access, including tunnelled catheters placed in any position (jugular vein, femoral vein, subclavian vein), grafts placed in any position (radial artery, brachial artery) or autogenous fistulas placed in any position (radial artery, brachial artery). Outcomes of interest were vascular access patency, vascular access infections, all cause and cardiovascular mortality. Studies were excluded if 1) not dealing with diabetics; 2) not providing the above-mentioned outcome data in relationship to the type of first placed vascular access; 3) dealing with vascular accesses not related to HD. Case reports, reviews, editorials, letters and studies performed on children (age <18 years) or animals were excluded as well, although they were screened as potential sources of additional references. Selection of relevant studies was independently performed by two authors (DB and LC). Discrepancies were solved collegially by discussion amongst DB and LC.

Quality assessment

We used the Newcastle-Ottawa Scale to assess the study quality for observational studies. This scale considers a quality score calculated on the basis of three major items: study participants (0 to 4 points), adjustment for confounding (0 to 2 points) or ascertainment of the exposure or outcome of interest (0 to 3 points) with a maximum score of 9 points which represents the highest methodological quality.

Data extraction and analysis

Data extraction and analysis were performed by two reviewers independently (DB and LC). In studies considering mixed populations, the subgroup of patients with documented diabetes was selectively described only if corresponding data were available.

Results

Search results

The flow diagram of the selection process is depicted in Figure 1. Two hundred and sixty-two potentially relevant references were initially found. A total of 213 citations were excluded after title/abstract skimming because they were clearly not pertinent for the topic of our review or because of search overlap. Amongst the 49 studies selected for full text examination, 36 studies were excluded because of the following: dealing with an inappropriate population/problem (n = 16), dealing with an inappropriate intervention (n = 4) or not including a proper comparator (n = 13), no outcome data available (n = 3). A total of 13 studies were therefore reviewed in detail and included in the review. Main characteristics of these studies are summarized in Table I.

Fig. 1 - TABLE I -

Flow diagram of the study selection process.

Summary of studies included in this review

Author (Ref.)	Year	Study type	Population	Follow-up	Vascular access	Outcome(s)	Results
AVF = arteriovenous fistula; AVG = arteriovenous graft; CVC = central venous catheter; DM = diabetes mellitus.
*No effect measure reported.
Chan et al (9)	2007	Retrospective multicentre cohort study	Prevalent haemodialysis patients	25 months	AVF AVG	Patient survival  AVF vs. AVG Survival of the technique  AVF vs. AVG	60% vs. 50% (~) OR = 1.34 (p = 0.123) 60% vs. 50% (~) OR = 1.49 (p = 0.244)
David et al (10)	2010	Retrospective single-centre cohort study	Incident haemodialysis patients	80 months	Distal AVF Middle-arm AVF Proximal AVF	Survival of the technique  Distal AVF  Middle-arm AVF  Proximal AVF	57%* 55% 30%
Dhingra et al (11)	2001	Retrospective multicentre cohort study	Haemodialysis patients	2 years	AVF AVG CVC	Patient survival  CVC vs. AVF  AVG vs. AVF Vascular access-related infection associated mortality:  CVC vs. AVF  AVG vs. AVF	60% vs. 70% (~) RR = 1.54 (p<0.002) 65% vs. 70% (~) RR = 1.41 (p<0.003) RR = 2.30 (p = 0.06) RR = 2.47 (p = 0.02)
Diehm et al (12)	2010	Retrospective single-centre cohort study	Chronic kidney disease patients	2 years	AVF AVG CVC	Survival of the technique  DM vs. non-DM	Primary patency rate OR 0.6 (95% CI 0.3-1.0) Secondary patency rate OR 0.4 (95% CI 0.2-0.7)
Field et al (13)	2008	Retrospective single-centre cohort study	Incident haemodialysis patients		Distal AVF Proximal AVF	Survival of the technique  DM vs. non-DM  DM	34% vs. 26% (p = 0.11) Better survival of proximal vs. distal AVFs*
Hammes et al (14)	2008	Retrospective single-centre cohort study	Incident haemodialysis patients who underwent fistulae angiography	78 months	AVF	Survival of the technique	Similar rate of subsequent stenosis between patients with/without cephalic arch stenosis*
Konner et al (15)	2000	Retrospective single-centre cohort study	Incident haemodialysis patients	72 months	Distal AVF Proximal AVF	Patient survival Survival of the technique	Lower survival rates in diabetic patients* Similar primary patency rates between groups*
Leapman et al (18)	1996	Retrospective single-centre cohort study	Incident haemodialysis patients	5 years	AVF	Survival of the technique DM vs. non-DM	1 year* 42% vs. 63% 5 years* 18% vs. 36%
Murphy et al (17)	2002	Retrospective single-centre cohort study	Incident haemodialysis patients	1 year	Proximal AVF	Survival of the technique  DM vs. non-DM  DM <65 vs. >65 yo  DM male vs. female	39% vs. 40% (p = N.S.) 59% vs. 59%*69% vs. 47%
Ravani (7)	2002	Prospective single-centre cohort study	Incident haemodialysis patients	3 years	Distal AVF Proximal AVF	Survival of the technique  DM vs. non-DM  DM	Primary patency HR = 1.85, p = 0.01 Cumulative patency HR = 2.38, p = 0.04 Similar results between distal and proximal AVF*
Saxena (8)	2002	Prospective single-centre cohort study	Haemodialysis patients	4 years	AVF AVG Permanent CVC Subclavian CVC Femoral CVC	Vascular access-related infection-associated mortality  AVF (Ref.)  AVG  permanent CVC  subclavian CVC  femoral CVC Vascular access-related infections  AVF (Ref.)  AVG  permanent CVC  subclavian CVC  femoral CVC	15% 42% (p<0.0006) 33% (p<0.03) 37.5% (p<0.001) 100% (p<0.0005) 0.04/patient-year 1.07/patient-year 1.15/1,000 catheter-days 1.3/1,000 catheter-days 1.5/1,000 catheter-days
Yeager (19)	2002	Retrospective single-centre case-control study	Haemodialysis patients with finger gangrene	3 years	AVF	Patient survival (finger gangrene vs. no-finger gangrene)	52% vs. 49%*
Konner (16)	2002	Retrospective single-centre cohort study	Chronic kidney disease patients		Distal AVF Proximal perforating AVFProximal non-perforating AVF	Primary patency rate  Distal AVF  Proximal perforating AVF  Proximal non-perforating AVF Cumulative patency rate:  Distal AVF  Proximal perforating AVF  Proximal non-perforating AVF Thrombosis rate:  Distal AVF  Proximal perforating AVF  Proximal non-perforating AVF	80%* >80% 50% 90% 80% 50% 6.3 0.8 3.0

Study characteristics

Amongst the 13 studies reviewed, 2 were prospective cohort studies (7, 8), 10 were retrospective cohort studies (9-10-11-12-13-14-15-16-17-18) and 1 was a case-control study (19). The number of patients ranged from 127 (14) to 5,198 (11). Diabetes was present in 22% (7) to 55% (19) of the study populations. Follow-up duration ranged from 24 (10) to 80 (11) months. The overall study quality was low to moderate.

Ravani et al (7) analysed a cohort of 197 incident HD patients (22% DM) who underwent distal and proximal AVF creation by nephrologists in a single centre. At the start of HD therapy, 117 patients (59.7%) had a dialysis catheter and the remaining patients had an AVF. Saxena et al (8) analysed the vascular access-related sepsis and mortality among 218 HD patients (29% DM) with different types of vascular access (AVF, arteriovenous graft (AVG), temporary and permanent dialysis catheters). In the study of Chan et al (9), a cohort of 764 incident HD patients >65 years old (43% DM) who underwent AVF and AVG creation were studied. Patients with dialysis catheters were excluded. David et al (10), analysed the vascular access patency in a cohort of 274 chronic kidney disease patients (26% DM) referring to AVF creation at several locations (distal, middle-arm and proximal AVF). Dhingra et al (11) analysed the all-cause, cardiovascular and infection-related mortality among a cohort of 5,189 HD patients (31% DM) with AVF, AVG and dialysis catheters. Diehm et al (12) analysed the vascular access patency on a cohort of 244 HD patients (25% DM) with different types of vascular access (AVF, AVG and dialysis catheters). In the study of Field et al (13), a cohort of 289 incident HD patients (36% DM) who underwent distal and proximal AVF creation was studied. Hammes et al (14) analysed a cohort of 127 incident HD patients (41% DM) who underwent AVF angiography aiming to determine the time to development of clinically significant stenosis among patients with and without cephalic arch lesions. Konner (15) analysed the vascular access patency and patient survival in a cohort of 247 chronic kidney disease patients (23% DM) who underwent distal or proximal AVF creation in a single centre. In a later study of Konner et al (16), the authors analysed the primary and cumulative patency rates in a cohort of 748 chronic kidney disease patients (24% DM) who underwent distal, proximal perforating or non-perforating vein AVF creation in a single centre. Murphy and Nicholson (17), analysed a cohort of 293 chronic kidney disease patients (23% DM) who underwent proximal AVF creation in a single centre, comparing <65- and >65-year-olds, and male versus female patients. Leapman et al (18) analysed a cohort of 150 chronic kidney disease patients (34% DM) who underwent wrist AVF creation, aiming to determine the cumulative patency of the vascular access. In the study of Yeager et al (19), 222 HD patients (54% DM) were analysed. Patient survival was determined among those with finger gangrene and those without it.

Study outcomes

Mortality

Dhingra et al (11) reported a higher all-cause, infection-related and cardiovascular-related mortality among patients with a dialysis catheter, in comparison to those with an AVF (RR = 1.54, p<0.002). Also, all-cause and infection-related mortality was significantly higher among those with an AVG versus AVF (RR = 1.41, p<0.003). On the other hand, in the study of Chan et al (9), mortality was not significantly higher in patients with an AVG compared to those with an AVF (RR = 1.34, p = 0.123). Finally, Konner (15) described a higher mortality rate among DM patients with an AVF versus non-DM patients (70% versus 40% at 60 months follow-up, respectively).

Vascular access patency

Ravani et al (7), Diehm et al (12) and Konner (15) reported lower patency rates of AVF among DM versus non-DM patients (HR 2.38, p = 0.04; OR 0.4, 95% CI 0.2-0.7, respectively). On the other hand, Murphy and Nicholson (17) and Field et al (13) reported similar AVF patency rates between DM and non-DM patients (approximately 40% and 30%, respectively; no effect measure reported). Within the DM group, both Ravani et al (7) and Konner (15) reported similar secondary patency rates among those with distal versus proximal AVF, and Murphy and Nicholson (17) reported similar cumulative patency rates between young and older patients. On the other hand, Field et al (13) reported a higher patency rate for DM patients with a proximal versus distal AVF, and Murphy and Nicholson (17) reported a higher patency rate in male versus female DM patients. In these studies, comparisons within the DM population were entirely descriptive. In another study, Konner et al (16), reported a lower primary patency rate in patients with non-perforating proximal AVF versus perforating proximal AVF and distal AVF (approximately 50%, 80% and >80%, respectively); the cumulative patency rates among the three study groups were similar (approximately 90%, 80% and 80%, respectively) and the thrombosis rate was lower among those with a proximal perforating AVF (6.3, 3.0 and 0.8 per 100 patients at risk; no effect measure reported). In the study of Chan et al (9), the authors reported similar vascular access patency rates between patients with an AVF and an AVG (60% versus 50%, OR = 1.49, p = 0.244). David et al (10) described similar patency rates between distal, middle-arm and proximal AVF (57%, 55% and 30%, respectively). Hammes et al (14) reported that the presence of cephalic arch stenosis in DM patients with an AVF was not a risk factor for the development of a subsequent stenosis. Finally, Yeager et al (19), reported that DM and premature atherosclerotic disease were independent risk factors for finger gangrene.

Vascular access-related infections

The study of Saxena et al (8), showed that vascular access-related sepsis was significantly lower among patients with an AVF (8.3%) in comparison with those with an AVG (33.3%) or a permanent dialysis catheter (27.3%) (AVG versus AVF, RR = 4.02, p<0.0006; permanent catheter versus AVF, RR = 3.29, p<0.03). Patients with temporary femoral catheters presented the highest sepsis-related mortality (100%, RR = 5.78, 95% CI 1.55-21.54). Dhingra et al (11), reported a higher vascular access-related infection-associated mortality among DM patients with permanent catheters and AVGs, in comparison with those with AVFs (RR = 2.30, p = 0.06; RR = 2.47, p = 0.02, respectively).

Discussion

In our systematic review, including 13 studies comprising over 2,800 participants with DM, we found that DM patients using a dialysis catheter apparently experience a higher risk of death and infection compared with patients who successfully achieved and maintained an AVF as HD access. Primary patency rates appeared to be lower in diabetics versus non-diabetics. The comparison between the use of an AVG or an AVF as HD access generated conflicting results.

The preference of AVF over all other forms of access arises from its functional advantages because of a lower rate of complications. Autogenous fistulae have lower rates of infections than catheters and AVGs, and the lowest rate of thrombosis, providing longer survival of the access (5, 6). Perl et al (20) reported that patients starting HD using a central venous catheter had a higher risk of death in the first year compared to those who started HD with an AVF or AVG. Ravani et al (21) performed a systematic review aiming to quantify the associations between vascular access type and mortality, infection and cardiovascular events. The authors showed that persons using central venous catheters for HD experience a much higher risk of death, infection, cardiovascular events and hospitalization compared with persons who achieve an AVF or an AVG as HD access. However, AVG use was also associated with increased risk of death, infection and hospitalization, compared to the use of an AVF. Nevertheless, since most of the data on this field were obtained from observational studies, there is always the reservation that adjustment for baseline co-morbidity cannot be complete. As a consequence, the presence of a functioning AVF is probably a marker of a patient’s health and adherence, and so all or even most of the superior outcome may not be related to the AVF itself but rather to selection bias (21). On the contrary, catheter use is associated with acute illness and late presentation for dialysis, factors that are associated with high mortality and that may be difficult to adjust for. Dhingra et al (11) reported a lower survival among those patients using an AVG and Chan et al (9) reported similar outcomes between end-stage renal disease patients achieving an AVF or an AVG. However, the study populations on these two studies were quite different – Dhingra et al (11) included HD patients aged >15 years and Chan et al (9) included only HD patients aged >65 years. Although overall ESKD diabetic patients probably do better with an AVF, in comparison with a dialysis catheter or an AVG, diabetic patients aged 65 years and older probably may experience similar outcomes either with an AVF or graft (22).

Our review suggests that diabetic patients have decreased odds for vascular access long-term survival, often resulting in repetitive interventions. There are no sufficient data to allow meaningful comparison of different techniques and locations on the arm (wrist/forearm/elbow), and existing data are conflictive. It is likely that this is just a reflection of different case-mix, bias by indication and experience of involved surgeons. It seems obvious in the light of good surgical practice that when planning permanent access placement, one should always consider the most distal site possible because it preserves more proximal vessels and has fewer complications (5). However, the major disadvantage of distal AVF is the relatively high primary failure rate. In view of the more limited life expectancy, a primary choice for more proximal places can be discussed, especially in the elderly and in those with additional co-morbidities. In this regard, vascular mapping in preparation for the creation of a vascular access should be performed in all patients in order to maximize the chance of AVF placement success (5, 6).

Our review has some strengths and limitations that deserve mentioning. Strengths include that we performed a systematic search of medical databases, and that data extraction and analysis were made by two independent reviewers according to current methodological standards. However, although comprehensive search strategies were implemented, publication bias cannot be excluded. In order to maximize the number of included studies we decided to adopt broad criteria, considering any paper including at least a subpopulation of HD patients with acknowledged DM and outcome data available according to the first type of vascular access placed. Yet, in most studies diabetics often represented only a minor subpopulation of the whole study cohort. This may therefore hamper the generalizability of findings to the whole diabetic HD population. There was a high heterogeneity among studies with respect to the study design, number of subjects enrolled, severity and vintage of diabetes, presence of co-morbidities and, above all, age, which prevented us to perform data pooling. Furthermore, all the studies had an observational design (mostly retrospective) and we were unable to find even a single randomized trial providing useful data for our review purpose. Also of note, data on the rates of vascular access patency were often only descriptive. This, again, makes it highly challenging to draw even a preliminary conclusion on what is the optimal vascular access for HD to be universally recommended in diabetics.

In conclusion, although it is widely recognized that an AVF appears to be the access of choice for younger and healthier HD diabetic patients, the everlasting question concerning older, sicker patients with risk factors for AVF failure and associated complications still remains unresolved. Patients should be well informed on the available evidence on vascular access. A strategy whereby reasonable effort is done to create an autogenous vascular access in those with good prognosis, both with regard to primary patency as to life expectancy, seems to be a defendable approach based on the available evidence. Much more clinical investigations in this important field are urgently needed, in view of the importance for this increasing patient group.

• 1. Ethier J.,Mendelssohn DC.,Elder SJ. Vascular access use and outcomes: an international perspective from the Dialysis Outcomes and Practice Patterns Study. Nephrol Dial Transplant 2008; 23: 3219-3226
• 2. USRDS. Annual Data Report: Atlas of Chronic Kidney Disease and End-Stage Renal Disease in the United States, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda 2010
• 3. Young EW.,Dykstra DM.,Goodkin DA.,Mapes DL.,Wolfe RA.,Held PJ. Hemodialysis vascular access preferences and outcomes in the Dialysis Outcomes and Practice Patterns Study (DOPPS). Kidney Int 2002; 61: 2266-2271
• 4. Ravani P.,Spergel LM.,Asif A.,Roy-Chaudhury P.,Besarab A. Clinical epidemiology of arteriovenous fistula in 2007. J Nephrol 2007; 20: 141-149
• 5. Clinical practice guidelines for vascular access. Am J Kidney Dis 2006; 48: S248-S273
• 6. Tordoir J.,Canaud B.,Haage P. EBPG on vascular access. Nephrol Dial Transplant 2007; 22: ii88-ii117
• 7. Ravani P.,Marcelli D.,Malberti F. Vascular access surgery managed by renal physicians: the choice of native arteriovenous fistulas for hemodialysis. Am J Kidney Dis 2002; 40: 1264-1276
• 8. Saxena AK.,Panhotra BR.,Naguib M. Outcome of dialysis access-related septicemia among diabetics following optimized AV-fistula placement. Kidney Blood Press Res 2002; 25: 109-114
• 9. Chan MR.,Sanchez RJ.,Young HN.,Yevzlin AS. Vascular access outcomes in the elderly hemodialysis population: a USRDS study. Semin Dial 2007; 20: 606-610
• 10. David P.,Navino C.,Capurro F. L’accesso vascolare per dialisi su vasi nativi nel paziente diabetico: esperienza di un singolo centro. G Ital Nefrol 2010; 27: 522-526
• 11. Dhingra RK.,Young EW.,Hulbert-Shearon TE.,Leavey SF.,Port FK. Type of vascular access and mortality in U.S. hemodialysis patients. Kidney Int 2001; 60: 1443-1451
• 12. Diehm N.,van den Berg JC.,Schnyder V. Determinants of haemodialysis access survival. Vasa 2010; 39: 133-139
• 13. Field M.,Bailey G.,Jaipersad A.,Morgan RH.,Pherwani AD. Primary patency rates in AV fistulae and the effect of patient variables. J Vasc Access 2008; 9: 45-50
• 14. Hammes M.,Funaki B.,Coe FL. Cephalic arch stenosis in patients with fistula access for hemodialysis: relationship to diabetes and thrombosis. Hemodial Int 2008; 12: 85-89
• 15. Konner K. Primary vascular access in diabetic patients: an audit. Nephrol Dial Transplant 2000; 15: 1317-1325
• 16. Konner K.,Hulbert-Shearon TE.,Roys EC.,Port FK.,Porto F. Tailoring the initial vascular access for dialysis patients. Kidney Int 2002; 62: 329-338
• 17. Murphy GJ.,Nicholson ML. Autogeneous elbow fistulas: the effect of diabetes mellitus on maturation, patency, and complication rates. Eur J Vasc Endovasc Surg 2002; 23: 452-457
• 18. Leapman SB.,Boyle M.,Pescovitz MD. The arteriovenous fistula for hemodialysis access. Gold standard or archaic relic? Am Surg 1996; 62: 652-656
• 19. Yeager RA.,Moneta GL.,Edwards JM.,Landry GJ.,Taylor LM.,McConnell DB. Relationship of hemodialysis access to finger gangrene in patients with end-stage renal disease. J Vasc Surg 2002; 36: 245-249
• 20. Perl J.,Wald R.,McFarlane P.,Bargman JM.,Vonesh E.,Na Y.,Jassal SV. Hemodialysis Vascular Access Modifies the Association between Dialysis Modality and Survival. J Am Soc Nephrol 2011; 22: 1113-1121
• 21. Ravani P.,Palmer SC.,Oliver MJ.,Quinn RR.,MacRae JM.,Tai DJ.,Pannu NI. Associations between Hemodialysis Access Type and Clinical Outcomes: A Systematic Review. J Am Soc Nephrol 2013; 24: 465-473
• 22. Lee HW.,Allon M. When Should a Patient Receive an Arteriovenous Graft Rather Than a Fistula? Semin Dial 2013; 26: 6-10