One-Stage vs. Two-Stage Brachio-Basilic Arteriovenous Fistula for Dialysis Access: A Systematic Review and a Meta-Analysis

Introduction A brachiobasilic arteriovenous fistula (BB-AVF) can provide access for haemodialysis in patients who are not eligible for a more superficial fistula. However, it is unclear whether one- or two-stage BB-AVF is the best option for patients. Aim To systematically assess the difference between both procedures in terms of access maturation, patency and postoperative complications. Methods Online search for randomised controlled trials (RCTs) and observational studies that compared the one-stage versus the two-stage technique for creating a BB-AVF. Results Eight studies were included (849 patients with 859 fistulas), 366 created using a one-stage technique, while 493 in a two-stage approach. There was no statistically significant difference between the two groups in the rate of successful maturation (Pooled risk ratio = 0.95 [0.82, 1.11], P = 0.53). Similarly, the incidence of postoperative haematoma (Pooled risk ratio = 0.73 [0.34, 1.58], P = 0.43), wound infection (Pooled risk ratio = 0.77 [0.35, 1.68], P = 0.51) and steal syndrome (Pooled risk ratio = 0.65 [0.27, 1.53], P = 0.32) were statistically comparable. Conclusion Although more studies seem to favour the two-stage BVT approach, evidence in the literature is not sufficient to draw a final conclusion as the difference between the one-stage and the two-stage approaches for creation of a BB-AVF is not statistically significant in terms of the overall maturation rate and postoperative complications. Patency rates (primary, assisted primary and secondary) were comparable in the majority of studies. Large randomised properly conducted trials with superior methodology and adequate sub-group analysis are needed before making a final recommendation.


Introduction
The superiority of haemodialysis (HD) access created by means of an Arteriovenous Fistula (AVF) in patients with end stage renal disease (ESRD) has been shown before. Stenosis and thrombosis is less likely to occur in a well-functioning mature AVF when compared to arteriovenous grafts (AVG) and central venous catheters (CVC), resulting in prolonged patency rates for AVFs as has been described previously [1]. Also, AVFs carry a lower risk for infection [2,3]. However, around 20%-50% of all fistulas fail to mature into a useful HD access [4][5][6][7].
The preferred location for placing an AVF for the first time is distally at the radius, thus making it possible to place a second fistula proximally if the first one failed to mature. The order of preference for creating an AVF [8][9][10] This order is in agreement with the National Kidney Foundation Kidney Disease Outcomes Quality Initiative (NKF KDOQI) guidelines [11]. However, fistulas created distally at the wrist are less likely to mature compared to proximal AVF, at the same time proximal AVF require less intervention and are likely to last longer [12]. The decision of where to create the AVF can be helped by preoperative vascular mapping using ultrasound imaging which is expected to improve chances of creating an AVF that will likely mature into a useful dialysis access [13,14]. Placement of a primary forearm fistula is feasible in 40% to 50%, with an upper arm fistula possible in an additional 25% to 35% of patients [15]. An AVF prevalence of 65% has been recommended in the KDOQI guidelines for patients undergoing HD [11], this prevalence is currently higher in Europe (67%-91%) compared to the US (24%-47%) [15][16][17][18]; however, the prevalence of AVFs in the US varies significantly among different dialysis units [15,19].
Dagher was the first to describe the use of basilic vein to create an AVF in the upper arm between the end of basilic vein and the side of the brachial artery to act as access for long term haemodialysis [20]. Since then, the procedure has seen several changes and modifications. Superficialisation of a brachiobasilic fistula to make it more susceptive to cannulation can be achieved either by an elevation technique without mobilisation to bring the vein superficial to the surgically reconstructed deep fascia and subcutaneous tissue in the anatomic location of the basilic vein [21], or by a transposition technique by mobilising the entire length of the basilic vein to position the vein anterolaterally through a subcutaneous flap [22].
Some of the debate surrounding brachiobasilic arteriovenous fistulas (BB-AVF) has been focused on the decision to choose between one-stage vs the two-stage techniques. The one-stage procedure aims to create a fistula between the basilic vein and the brachial artery in the upper arm in one procedure. This would require a long incision to gain access and mobilise the basilic vein making sure the anastomosis is not placed under tension and no obvious stenosis is present proximally. The main advantage of this technique is the shorter waiting time required to cannulate the fistula. Also the one-stage will prevent the patient from having to undergo another procedure and is more cost effective as hospital resources will be used only once. One of the main disadvantages of this technique is the long incision which will require a longer time to heal and also carries a higher risk for wound-related complications. Also the procedure takes longer and is more demanding [23][24][25]. Moreover, in a study by Anaya-Ayala et al assessing the anatomy of basilic vein found that only 66% of patients are expected to have a "normal" basilic vein entering one of two paired brachial veins close to the axilla, while up to 34% will have an "abnormal" variant that would negatively influence the newly created fistula maturation [26].
The two-stage procedure allows the basilic vein to become arterialised and as such, more resistant to torque and will become easier to mobilise in the second procedure as it gets transformed into a bigger and stronger structure. The hope is that operative difficulty and complications would be reduced with improved patency rates [27].
This review was designed to systematically assess the difference between both procedures in terms of access maturation and survival, as well as complications and interventions required to maintain patency for haemodialysis.

Methods
This systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines [28]. No published protocol exists for this review.

Eligibility criteria
We searched for randomised controlled trials (RCTs) and observational studies that compared the one-stage technique with the two-stage technique for creating a brachiobasilic arteriovenous fistula (BBAVF) for haemodialysis access. Case series and review articles were excluded from this review.

Search strategy
A search of the literature for relevant studies was conducted in August 2014 using the following terms: (["Basilic Vein" OR "Basilic"] AND ["Fistula" OR "Arteriovenous" OR "Access"] AND "dialysis"). We searched the online databases of: Medline, CINAHL, EMBASE, the Cochrane library and Google Scholar. We did not restrict our search by publication date or status, however, we only included studies published in English language and those conducted on humans. We also searched the bibliographies of included trials for additional studies. A summary of the study selection process can be found in the PRISMA flow diagram below [ Fig. 1]. Studies were not restricted based on the duration of follow-up.
The main outcome measures for this review were successful maturation and development of postoperative complications, namely wound haematoma, wound infection and steal syndrome. Secondary outcomes were primary and secondary patency rates. Definitions for "maturation", "primary patency" and "secondary patency" were those specified in individual studies.

Data collection
KB and DH independently extracted the data from included studies on a Microsoft Excel spreadsheet. Any differences in recording the outcomes of interest were discussed between two Eligibility for inclusion was determined by two researchers separately (KB, DH) by going through the abstracts of the relevant citations. Differences were settled by examining the full article by both authors, and then any remaining uncertainties regarding eligibility of studies were settled following a discussion with a third author (SRW).
authors (KB, DH), and if remained unsettled, a third author was consulted to resolve the issue (SRW). The following characteristics regarding participants were recorded: age, sex, presence of co-morbidities, primary patency rate, secondary patency rate, maturation rate and postoperative complications (Haematoma, wound infection and steal syndrome). Usability of fistula for Haemodialysis, time to first use for HD and interventions needed to maintain patency were recorded when possible. We also extracted data on compliance with the Society for Vascular Surgery (SVS) recommended standards for reports dealing with arteriovenous haemodialysis accesses [29]. To this end, we assessed whether studies provided SVS standard-based grading of factors that affect outcomes and whether studies provided SVS standard-based grading of severity of arteriovenous access complications. The studies' inclusion and exclusion criteria were also recorded [ Table 1].

Quality assessment for risk of bias
The Downs and Black Tool was used for quality assessment [36]. This tool consists of a total of 27 questions assessing the quality of reporting, external validity and internal validity generating scores between 0 to 32 which includes a score of 0-5 for sample size justification, however, this has been modified by awarding one point for studies that reported on sample size calculations, and zero for those that did not report a methods of sample size calculation. Hence, the modified score ranged from 0 to 27, with higher scores reflecting higher quality. Details of the quality assessment can be found in a separate supplemental table [S1 Table].

Data analysis
Statistical analyses were performed using Review Manager version 5.3 [37]. We used the random effects model of DerSimonian and Laird [38] to calculate pooled risk ratios for categorical outcomes measures. The Cochran's Q test was used to determine statistical heterogeneity among studies. 95% confidence interval and P-values < 5% were used to determine statistical significance. We compared between the fixed and random effects modelling to produce a sensitivity analysis aimed at detection of the influence of publication bias of small-study effects [39]. Regarding the meta-analysis we additionally, performed a sensitivity analysis limited to published articles only.

Study selection
The results of the study selection process are summarised in the PRISMA flow diagram [ Fig. 1]. We started with a total of 969 citations. Following the removal of duplicates and limiting the search criteria to studies conducted on humans and in English language, we were left with 295 citations. We then screened the titles of those papers, and found 80 potentially relevant citations. The abstracts of those titles were examined for relevant outcomes, and 22 papers were evaluated for eligibility criteria, of those 8 citations met our criteria and were included in the systematic review [21,22,[30][31][32][33][34][35]. Of those 8 studies, 1 was a randomised controlled studies (RCT). Five were retrospective cohort studies [22,30,31,33,34] and 2 studies were cohort studies but it was unclear whether they were retrospective or prospective [21,35]. This last citation was a conference presentation which we included in the review, however we also ran a group of sensitivity tests excluding the data from this citation and including data extracted from published papers only [35].
Six of the included studies compared outcomes between 1-stage versus 2-stage BB-AVF formation techniques, while Hossny et al compared 3 different groups, first group of patients had traditional 1-stage basilic vein transposition (BVT), while the second group had 1-stage basilic vein elevation, and the third group underwent a 2-stage BB-AVF. For the sake of this metaanalysis, we pooled the first 2 groups from this particular study together [21]. Similarly, Effat had 3 groups of patients in his conference paper, the first group had standard 1-stage BVT, whereas the second group had a 2-stage BVT and the last group consisted of patients who had 2-stage superficialization of the basilic vein to create BB-AVF. We pooled the data from the 1stage procedures in this last study together in the meta-analysis [35].

Participants
The studies included a total of 849 patients who had 859 fistulas, of those 366 fistulas were formed using a

Successful maturation rate
Successful maturation rates were reported in 6 of included studies [21,22,[31][32][33][34]; the criteria used for reporting maturation are found in [ Table 3]. Those studies had a combined total of 683 fistulas, 301 of those were created in the one stage group, whereas 382 were created in the two stage group. The difference between the two groups was not significant in pooled analysis (Pooled risk ratio = 0.95 [0.82, 1.11], 95% CI, P = 0.53) [ Fig. 2]. Heterogeneity was detected statistically (Cochran's Q = 14.48; degree of freedom (DF) = 5; P = 0.001; I 2 = 65%). The significance of the results was not altered when using the fixed effects analysis model as a sensitivity test to detect publication bias (Pooled risk ratio = 0.92 [0.84, 1.01], 95% CI, P = 0.07).

Postoperative complications
Haematoma. The incidence of postoperative wound haematoma was reported in 6 of the included studies [21,22,30,31,33,35] with a total of 711 fistulas, of those, 295 fistulas were created in the 1-stage group and 416 fistulas in the 2-stage group. Analysis of pooled data showed the difference was not significant (Pooled risk ratio = 0.73 [0.34, 1.58], 95% CI, P = 0.43) [Fig. 3]. Heterogeneity was not detected statistically (Cochran's Q = 9.76; degree of freedom (DF) = 5; P = 0.08; I 2 = 49%). The results were not changed significantly when using the fixed effects analysis model as a sensitivity test to detect publication bias (Pooled risk ratio = 0.67 [0.41, 1.11], 95% CI, P = 0.12). A sensitivity test by excluding the data from the conference paper by Effat [35] was carried out, and no significant difference was found in the incidence of postoperative haematoma between the two groups (Pooled risk ratio = 0.67 [0.27, 1.64], 95% CI, P = 0.38). Wound infection. Six of the included studies [30][31][32][33]35] reported on the incidence of postoperative wound infection with a total number of 681 fistulas, 265 of those belonged to the 1-stage group, while 416 consisted of 2-stage fistulas. Meta-analysis of the pooled data showed the difference between groups not to be significant (Pooled risk ratio = 0.77 [0.35, 1.68], 95% CI, P = 0.51) [Fig. 4]. There was no evidence of statistical heterogeneity (Cochran's Q = 5.76; degree of freedom (DF) = 5; P = 0.51; I 2 = 13%). The results were not changed significantly when using the fixed effects analysis model (Pooled risk ratio = 0.73 [0.39, 1.37], 95% CI, P = 0.32). A sensitivity test by excluding the data from the conference paper by Effat [35] was carried out, and no significant difference was found in the incidence of postoperative wound infection between the two groups (Pooled risk ratio = 0.57 [0.25, 1.27], 95% CI, P = 0.17).
Steal syndrome. Six of the studies [22,[30][31][32][33]35] reported on the risk of developing significant postoperative ischaemia (steal syndrome). Those studies had a combined total of 681 Successful maturation rates were derived from these data.
The number of AVFs that required intervention to assist maturation is unclear.
Kakkos [31] 67 / 76 69 / 98 Maturation rates were reported. Maturation was based upon clinical judgement (development of basilic vein dilatation and thrill for a sufficient length).
Includes fistulas that required intervention to assist maturation for dialysis. 7 one stage fistulas required such intervention and 3 two stage fistulas required such intervention.
El-Mallah [32] 12 / 20 18 / 20 Patency at 4 weeks was reported and we used this figure to determine successful maturation. The authors did not provide a definition for patency.
The number of AVFs that required intervention to assist maturation is unclear.
Syed [33] 6 / 29 14 / 77 Maturation rates were reported. Fistula maturation was defined as dilation of the vein to allow cannulation and support dialysis at a minimum flow rate of 350ml/min for at least 3 sessions.
The number of AVFs that required intervention to assist maturation is unclear.
Agarwal [34] 55 / 61 62 / 83 Maturation rates were reported. Maturation was defined as the use of the fistula for haemodialysis for any amount of time or, if it was not used, documentation in surgical or renal records that the fistula was mature and ready for use based upon successful cannulation and/or physical examination by vascular surgery.
Includes an unspecified number of fistulae that needed percutaneous intervention to assist maturation.
No patients needed reintervention to assist achievement of successful dialysis at 6 weeks. A sensitivity test by excluding the data from the conference paper by Effat [35] was carried out, and no significant difference was found in the incidence of postoperative haematoma between the two groups (Pooled risk ratio = 0.79 [0.32, 1.94], 95% CI, P = 0.60).

Systematic Review
El-Mallah in his paper titled "Staged basilic vein transposition for dialysis angioaccess" published in 1998 [32] compared outcomes in two groups randomly allocated to receive either 1stage BVT or 2-stage BVT. The difference in early patency rates was significant and favoured the 2-stage approach (60% of 1-stage vs 90% of 2-stage, P < 0.05), as well as overall patency rates at the end of follow-up (50% of 1-stage vs 80% of 2-stage, P < 0.05) [32]. Postoperative wound infection rate also favoured the 2-stage approach with one case compared to 3 in the 1stage group. There was no difference in postoperative aneurysmal dilatation, and there was no significant ischaemia (steal syndrome) reported in either of the two groups [32]. Hossny looked at the different surgical techniques used in creation of a BB-AVF in 2003 [21]. He compared patency rates and dialysis related complications in 70 patients divided in 3 groups, 30 of those patients had traditional BVT whereas 20 patients had 1-stage BB-AVF with elevation and the remaining 20 patients had 2-stage fistula with elevation of the vein [21]. Cumulative secondary patency rates were comparable among the 3 groups at 1 year and 2 years; at 1 year (86.7% for the BVT group, 90% for the 1-stage elevation group and 84.2% for the last group), while at 2 years (82.8%, 70%, and 68.4%, respectively) [21]. Similarly, no significant difference was found in his study in postoperative early thrombosis across the groups. Postoperative arm oedema occurred in a total of 14 patients, all of whom had temporary subclavian access catheters sited on the same arm on which the new fistula was created. All 14 cases were managed conservatively with success. The difference in developing postoperative haematoma significantly favoured the traditional BVT approach compared to the two elevation methods with fewer haematomas reported in the first group. Interestingly, the dialysis staff were more satisfied with the 1-stage BVT technique, whereas only 53.3% reported satisfaction with the elevation technique (1-stage and 2-stage techniques) (P < .001) [21].
Kakkos et al also tried to answer the question of "What is the Optimal Technique" for performing a BVT in a retrospective study of 173 patients published in 2010 [31]. They found that the incidence of venous hypertension (17% vs 4%, P = 0.004), wound infection (13% vs 3%, P = 0.012) and all complications (43% vs 11%, P < 0.001) were significantly higher in patients who had 1-stage BVT when compared to those who had a 2-stage BVT [31]. Time to fistula use in HD was-expectedly-shorter in the 1-stage group (Median = 68 (49-103) days) compared to the 2-stage group (Median = 132 (102-166) days). This difference was significant (P < 0.001) [31]. However maturation rates were similar (85% for the 1-stage groups versus 82% for the 2stage group, P = 0.49). Median time to use the fistulas for HD in the 2-stage group was 66 in patients who developed postoperative complications, compared to 50 days in those who did not (P = 0.019). They also found that wound infection occurred more in patients who were operated under general anaesthetic compared to those who had their procedures done under local anaesthetic (OR 38, P < 0.001). Also, venous hypertension was found to occur more frequently in patients who developed postoperative wound haematoma, but the difference was not statistically significant (18% vs 6%, P 0.12) [31]. A trend was noted towards steal being more common in patients with previous vascular access than in those who did not have such access (4.9% vs 1.1%, P = 0.19) [31].
Syed et al carried out a similar study comparing the outcomes of 1-stage (29 patients) and 2-stage (77 patients) BVT and published their findings in 2012 [33]. 79% of patients in the 1stage group had a history of a previously failed access for HD compared to 51% of the 2-stage group. They found that the rate of primary failure was comparable between both groups (21% vs 18%) [33]. In their study, patients who had 1-stage BVT had better patency rates when compared to those who had 2-stage procedures at 1 year (82% vs 67%), 2 years (81% vs 27%) and 3 years (51% vs 18%) respectively. The same finding was reported for secondary patency (91%, 80% and 58% for 1-stage BVT compared to 81%, 61% and 45% for the 2-stage group at 1, 2 and 3 years respectively) [33]. Reintervention rate in this study was 62% for the 1-stage vs 66% for the 2-stage group. It is worth noting that 87% of the patients in the 2-stage group were using catheters for dialysis, whereas 55% of the 1-stage group were dialysing through a catheter at the time of access formation [33].
The number of AVFs that failed to progress from the first stage to the second stage in the two-staged BVT approach were unclear in four studies [22,30,34,35]. In the remaining four, El-Mallah [32] reported 1/20 patient which had an occluded shunt, while Hossny [21] also had 1/ 20 patient failing to progress due to spontaneous thrombosis within the first 4 weeks postoperatively. Syed [33] had 2/77 patients that never progressed to the second stage of the procedure. Kakkos [31] reported that 26/98 of his patients never had a second stage procedure (thrombosed (n = 4), failed to mature and was abandoned during the re-exploration (n = 12), patient refused the procedure (n = 3), lost to follow-up (n = 1), died (n = 2), venous hypertension (n = 2), venous monomelic neuropathy (n = 1) requiring ligation, moved out of state (n = 1)).
Number of interventions required to maintain patency or to improve the fistula maturation rates were not reported clearly in all studies. Hossny reported that in the one-stage group one patient underwent ligation and another had a surgical revision, same numbers occurred in the two-stage group. All ligations were done to treat venous hypertension, whereas all revisions were performed to improve poor flow [21]. Kakkos reported that in the two-stage group 6 patients had endovenous angioplasty interventions, 3 had surgical revisions compared to 3 patients and 1 patient in the one-stage group respectively [31]. Syed et al performed 37 angioplasty interventions, 9 surgical revisions and 5 thrombectomy procedures in their twostage group, compared to 14, 2 and 2 patients respectively [33]. The remaining studies either did not report data related to fistula salvage procedures or it was reported in poor details making it difficult to quantify those interventions.
With the exception of the studies by El-Mallah et al [32] and the one by Vrakas et al [22] which both reported significantly superior patency rates in the two-stage groups, and the paper by Syed [33] which conversely reported a significantly better patency rates in the one-stage BVT group, the remaining studies all reported comparable patency rates [21,30,31,34,35] [ Table 2]. However it is important to point out that patency rate data were reported as percentages with the lack of clearly identifiable denominators in the majority of those studies, thus making pooling those data in a meta-analysis not feasible. Also, definitions used in individual studies included in this review for patency rates (primary, assisted primary and secondary) differed significantly.

Discussion
The number of patients with end stage renal disease (ESRD) requiring haemodialysis (HD) is steadily rising, a trend that is expected to continue [2]. A well-functioning AVF is superior to grafts and central catheters in providing access for haemodialysis efficiently and at the same time with the least rate of access related complications. This has lead vascular surgeons to resort to the basilic vein which by virtue of its anatomical position is less likely to be damaged by repeated cannulation as with the more superficial veins of the arm and forearm. However, a consensus on how to form a brachiobasilic AVF does not exist as some surgeons choose to do this in a one-stage operation, while others prefer a two-stage procedure with the first procedure usually involving making the anastomosis between the basilic vein and the brachial artery, while in the second stage the arterialised vein is mobilised and brought closer to the skin surface to facilitate cannulation for HD sessions.
This review identified eight studies [21,22,[30][31][32][33][34][35], including data from a conference paper [35] in order to increase the rigour of the review. The pooled data referred to 849 patients with a total of 859 fistulas, 366 of those fistulas belonged to patients who underwent a 1-stage BB-AVF, while 493 fistulas were performed using a 2-stage technique to create the access. The data from 6 of the included studies [21,22,30,31,33,35] were used to compare the difference between the two groups in developing postoperative haematoma which was not significant (Pooled risk ratio = 0.69 [0.30, 1.56], 95% CI, P = 0.37). Excluding the data from the conference paper by Effat [35] in a sensitivity test did not alter the result (Pooled risk ratio = 0.61 [0.23, 1.60], 95% CI, P = 0.31).
Similarly, Vrakas et al reported a smaller mean vein diameter of (3.6 ± 1.3 mm) for the 2stage, versus (4.0 ± 1.1 mm) fistulas created in the 1-stage group, yet their results favoured the 2-stage approach with primary functional patency at 1 and 2 years for the 1-stage group of 71% and 53% versus 87% and 75% in the 2-stage group. Assisted Primary functional patency at 1 and 2 years for the 1-stage group was 77% and 57% versus 95% and 77% in the 2-stage group, while secondary functional patency at 1 and 2 years for the 1-stage group was 79% and 57% versus 95% and 77% in the 2-stage group.
Conversely, in study by Agarwal et al [34], their patients in the 1-stage group achieved better maturation rate than those who had a 2-stage BVT fistulas (90% vs 75%, P = 0.02). They did not include any analysis between the two groups based on vein diameter. Vein diameter has been shown to negatively influence maturation and patency rates in AVFs, and is one of the main predictors of those outcomes in fistulas [40,41], and indeed has been shown to be the only independent predictor of maturation in some studies [42,43]. Syed et al reported similar findings with better primary and cumulative patencies in the 1-stage group with primary patency at 1, 2 and 3 years (82% vs 67%), (81% vs 27%) and (51% vs 18%), while (secondary patency at 1, 2 and 3 years (91%, 80% and 58% for 1-stage BVT and 81%, 61% and 45% for the 2stage group. Variations in vein diameter between the two groups were not reported in this study. Kakkos et al [31] did not find a significant difference in maturation between the two groups, as 15% of fistulas in the 1-stage group did not mature, compared to 18% in the second group (P = 0.49). They did however find significant difference in developing postoperative haematoma (13% vs 3%, P = 0.012), venous hypertension (17% vs 4%, P = 0.004) and overall complications (43% vs 11%, P < 0.001), all in favour of the 2-stage BVT technique.
Kim et al compared the 2-stage approach to all other AVF procedures including 1-stage BVT, radiocephalic and brachiocephalic fistulas. All of the 2-stage BB-AVFs in their study successfully matured compared to a pool consisting of all different types that showed a combined maturation rate of 52% (P = 0.001). Fistula failure occurred in 7% of the 2-stage BVT compared to 59% of other fistulas (P = 0.001), and more 2-stage BVT fistulas were used successfully for HD compared to all other fistula types (87% vs 48%, P = 0.024). Also, the patency rate at 1 year was superior in the 2-stage group compared to other AVFs (91% vs 47%, P = 0.003).
One of the limitations of this review is the low number of randomised trials included-1 study was randomised-while the remaining 7 were cohort studies. Most of these studies were retrospective. Another limitation is the variation in surgical approaches, those variations include technical differences in performing the procedure, as well as differences in equipment used and expertise among participating surgeons. Those limitations can be addressed by conducting a large randomised multi-centre trial that would adhere to a rigid protocol in patients' selection process and performing the procedures. Another limiting factor is the lack of sufficient sub-group analysis among included studies, particularly analysis taking into account factors that are known to be associated with fistula maturation such as vein diameter. Finally, we highlight that included studies were not compliant with SVS reporting recommendations regarding baseline factors that affect outcomes or severity of complications.

Conclusion
Although more studies seem to favour the 2-stage BVT approach, evidence in the literature is not sufficient to draw a final conclusion as the difference between the 1-stage and the 2-stage approaches for creation of a BB-AVF is not statistically significant in terms of the overall maturation rate and postoperative complications. Patency rates (primary, assisted primary and secondary) were comparable in the majority of studies. Large randomised properly conducted trials with adequate sub-group analysis are needed before making a final recommendation. Future studies should aim for compliance with established reporting standards.