Should buttonhole cannulation be discontinued?

The arteriovenous fistula remains the preferred type of vascular access because of its higher patency and lower infection rates (1). Cannulation of the fistula remains a challenge for patients, caregivers, and nurses, often resulting in increased pain, stress, and vessel injury. The use of rotating puncture sites along the entire length of the fistula, or rope-ladder cannulation (RLC), was intended to heal the site and prevent hematoma, stenosis, infection, and pseudoaneurysm. Constant site needling or buttonhole cannulation (BHC) was introduced in the 1970s to address cannulation of short fistula segments (2). The technique is based on repeated sharp needle insertion into the same site, preferably by a single cannulator, at the same angle over the course of six to nine hemodialysis sessions. This produces a scar tissue tract across the veinwall, which can subsequently be cannulated with blunt needles. The initial observational studies of BHC, many of which were retrospective, reported lower rates of hematomaand infiltration injury, less needling time (2,3), and, importantly, less pain (3–7). BHCwas adopted bymany home dialysis programs for its ease of cannulation and decreased training time, as well as in-center units, with support from the National Kidney Foundation’s guidelines on vascular access (1) and promotion as the preferred type of cannulation by theUnitedKingdomRenal Association (8). The recent reporting of increased infection rates (4,6,9–11) with the use of BHC and debate as to its benefit in reducing pain (9,11–14) and improving patency (15) has tempered its use in many units and calls into question whether BHC should be used at all. Before addressing this question, it is important to review the complete body of evidence, the study population (particularly differentiating home or in-center patients), outcomes, follow-up, and the specifics of the buttonhole technique. Recent systematic reviews of both in home and in-center dialysis (16) and intensive dialysis (17), as well as a narrative review (18), highlight the concerns regarding observational and crossover study designs, short follow-up, operator dependency, and differences in the BHC technique. These reviews highlight the ongoing uncertainty surrounding the benefits and risks of BHC (14,19). In this issue of CJASN, Muir at al. (20) report on a retrospective, single-center experience comparing BHC to RLC in a home hemodialysis population, accompanied by a systematic review of the related literature. The study reports a significant 3-fold increase in total fistula-related infections/1000 fistula-days in the BHC cohort (0.39 events) compared with the RLC cohort (0.10 events), with little change in the effect size after adjustment for differences in populations. The rate of systemic fistula infections is more difficult to interpret because events rates were low and results were not reported as per 1000 fistula-days, with different reporting periods: 2003 to mid-2004 for RLC and 2004–2009 for BHC. In addition, interpretation of the rate of fistula loss and need for surgical intervention is hampered by the exclusion of radiologic interventions that represent the majority of interventions in patients with fistula and have been significantly associated with fistula patency outcomes (21). The utilizationrelated outcomes were as follows: Training time and ongoing support encounters were significantly higher with BHC (median, 46 days [interquartile range, 37–60 days] versus 37 days [interquartile range, 25–58 days] in the RLC and 1 encounter per 13.2 fistula-days versus 1 encounter per 19 fistula-days for RLC). These results, however, cannot be generalized because each facility will develop its own proficiency with training, patient selection, and technique, which may lead to different results than those reported by Muir et al. (20) in their single-center experience. As the authors point out, the accompanying systematic review is limited by a small number of studieswith relatively short follow-up. However, they also report an approximate 3-fold increase in systemic infection risk with BHC compared with RLC in pooled analyses across various studydesigns, themost reliable ofwhich is based on randomized controlled trials, with a pooled relative risk of 3.34 (95% confidence interval [95% CI], 0.91 to 12.20; I2515%). The concordance between this pooled estimate and the authors’ local experience seems to reassure that the measured effect is accurate, and the I2 statistic suggests little heterogeneity in this body of evidence. It is, however, likely that this analysis (which included only four small studies) was not adequately powered to detect heterogeneity. The relative risk reported in these trials ranged between 1.00 (95% CI, 0.14 to 6.90) (i.e., no increased risk of infection with BHC) to 25.00 (95%CI, 1.51 to 414.12) (i.e., significant risk of infection with BHC), indicating that an unexplained clinical heterogeneity probably exists. This marked variability in infection risk across studies is likely to be attributed tounmeasuredvariables in theuse ofBHCandRLC. The study by Muir et al. (20) is representative of the literature on cannulation techniques. It describes a *Schulich School of Medicine and Dentistry and Department of and Epidemiology and Biostatistics, Western University, London, Ontario, Canada; Kidney Clinical Research Unit, London Health Sciences Centre, London, Ontario, Canada; The Li Ka Shing Knowledge Institute, Keenan Research Centre, St. Michael’s Hospital, Toronto, Ontario, Canada; and Department of Nephrology, Humber River Hospital, Toronto, Ontario, Canada