- Email: [email protected]
Infectious Complications Following Conversion to Buttonhole Cannulation of Native Arteriovenous Fistulas: A Quality Improvement Report
Original Investigation Infectious Complications Following Conversion to Buttonhole Cannulation of Native Arteriovenous Fistulas: A Quality Improvement Report Laura Labriola, MD,1 Ralph Crott, MD, PhD,2 Christine Desmet, RN,1 Geneviève André, RN,1 and Michel Jadoul, MD1 Background: Constant-site or buttonhole cannulation of native arteriovenous fistulas (AVFs) has gained in popularity compared with rope-ladder cannulation. However, cannulating nonhealed skin might increase the risk of (AVF-related) infectious events, as suggested by small reports. Study Design: Quality improvement report. Setting & Participants: All patients on in-center hemodialysis therapy using a native AVF from January 1, 2001, to June 30, 2010. Quality Improvement Plan: Shift to buttonhole cannulation between August 2004 and January 2005. Because the infectious event rate increased after the shift, educational workshops were held in May 2008 for all nurses, with review of every step of buttonhole protocol. Outcomes: Infectious events (unexplained bacteremia caused by skin bacteria and/or local AVF infection) and complicated infectious events (resulting in metastatic infection, death, or AVF surgery) were ascertained during 4 periods: (1) rope-ladder technique in all, (2) switch to buttonhole, (3) buttonhole in all before workshops, and (4) buttonhole in all after workshops. Results: 177 patients (aged 70.4 ⫾ 11.5 years) with 193 AVFs were analyzed, including 186,481 AVF-days. 57 infectious events occurred (0.31 events/1,000 AVF-days). The incidence of infectious events increased after the switch to the buttonhole method (0.17 [95% CI, 0.086-0.31], 0.11 [95% CI, 0.0014-0.63], and 0.43 [95% CI, 0.29-0.61] events/1,000 AVF-days in periods 1, 2, and 3, respectively; P ⫽ 0.003). This reached significance during only the second full year of buttonhole cannulation. During period 4, the incidence tended to decrease (0.34 events/1,000 AVF-days). Complicated infectious events (n ⫽ 12) were virtually restricted to period 3 (n ⫽ 11; 0.153 [95% CI, 0.076-0.273] events/1,000 AVF-days), with a significant decrease in period 4 (n ⫽ 1; 0.024 [95% CI, 0.001-0.118] events/1,000 AVF-days; RR for period 3 vs period 4, 6.37 [95% CI, 1.09-138.4]; P ⫽ 0.04). Limitations: Observational partly retrospective design. Conclusion: Intensive staff education regarding strict protocol for the buttonhole procedure was associated with a decrease in infectious events. Am J Kidney Dis. 57(3):442-448. © 2011 by the National Kidney Foundation, Inc. INDEX WORDS: Arteriovenous fistula; buttonhole; cannulation; hemodialysis; infection.
C
onstant-site or buttonhole cannulation of native arteriovenous fistulas (AVFs) with blunt needles was reported first in 1977 by Twardowski1 and recently has gained in popularity, especially in home hemodialysis (HD). In buttonhole cannulation, the needles are inserted at the same AVF sites at each session, instead of moving the cannulation sites along the whole access length (rope-ladder method). Using a constant site during 6 HD sessions creates a From the Departments of 1Nephrology and 2Public Health, Cliniques Universitaires Saint-Luc, Université Catholique de Louvain, Brussels, Belgium. Received June 3, 2010. Accepted in revised form October 1, 2010. Originally published online January 10, 2011. Address correspondence to Laura Labriola, MD, Department of Nephrology, Cliniques Universitaires Saint-Luc, Université Catholique de Louvain, Ave Hippocrate 10, B 1200 Brussels, Belgium. E-mail: [email protected] © 2011 by the National Kidney Foundation, Inc. 0272-6386/$36.00 doi:10.1053/j.ajkd.2010.10.045 442
tunnel from the skin to the vessel. This tunnel is used during subsequent cannulations to guide the blunt needle to access the vessel, after careful disinfection and withdrawal of the scab formed at the cannulation site. Potential advantages of buttonhole cannulation over the standard rope-ladder technique include easier cannulation, fewer missed sticks, less pain, faster hemostasis after needle removal, and fewer hematomas and aneurysms.2-5 However, studies comparing cannulation techniques are very scarce.4,5 Infection rates with native AVFs are low.6,7 However, the possibility of an increased incidence of infectious events with buttonhole cannulation recently has been suggested, although it is poorly documented.4,5,8,9 Because cannulating nonhealed skin might increase the risk of infectious events, we conducted an observational study in our HD patients with AVFs to examine the impact of switching to buttonhole cannulation on the infectious event rate. Am J Kidney Dis. 2011;57(3):442-448
AVF-Related Infections and Buttonhole Cannulation
METHODS Participants and Quality Improvement Plan We included all patients on maintenance HD therapy using a native AVF from January 1, 2001, to June 30, 2010, in our hospital HD unit. They were dialyzed 3 times a week (except for 2 patients dialyzed 6 times weekly and 2 others dialyzed only twice weekly) for 3.5-5 (median, 4) h/session using hollow-fiber high-flux or superflux polysulfone dialyzers (Fresenius Medical Care, www. fresenius.com). Standard blood flow was 360 mL/min. Selfcannulation was not performed. From August 4, 2004, to January 31, 2005, the rope-ladder technique using sharp needles was progressively replaced by the buttonhole cannulation method with blunt needles. After January 31, 2005, all AVFs were cannulated using blunt needles (15 Gauge, except in 3 patients, 14 Gauge) using the buttonhole method. Because the rate of infectious events (defined later) increased progressively in 2006-2007, educational workshops were organized for all nurses in late May 2008. Every step of the buttonhole cannulation method was reviewed carefully (discussed later), including scab removal and disinfection before cannulation and after needle removal. Four periods thus were considered in our analysis: (1) period 1 (January 1, 2001, to August 3, 2004): all patients on rope-ladder technique with sharp needles; (2) period 2 (August 4, 2004, to January 31, 2005): progressive switch to buttonhole cannulation method with blunt needles; (3) period 3 (February 1, 2005, to May 19, 2008): all patients on buttonhole method, before educational workshops; and (4) period 4 (May 20, 2008, to June 30, 2010): all patients on buttonhole cannulation method, after workshops. For annual comparisons, 2008 was divided into A (before educational workshops: months 1-5) and B (after workshops: months 6-12). Age, sex, diabetes, number of AVF-days, location of AVF (forearm or upper arm), and infectious events were recorded. The primary end point was the incidence of infectious events. The secondary end point was the incidence of complicated infectious events. Infectious events included both unexplained bacteremia caused by skin bacteria and/or local AVF infection. Local AVF infection was defined as nonallergic erythema, pain or tenderness close to cannulation sites, necrotic scabs, or drainage from cannulation site(s). In every case, a local swab specimen was obtained. Bacteremia caused by a skin micro-organism (ie, Staphylococcus aureus or Staphylococcus epidermidis) was considered an infectious event originating from the AVF, even without local signs, in the absence of an alternative source (such as diabetic foot, etc). Complicated infectious events included metastatic complications (ie, endocarditis, septic arthritis, etc), death related to an infectious event, and AVFs requiring surgery (overt or imminent rupture). Indications for measuring C-reactive protein and obtaining blood cultures were very broad and unchanged in our unit during the study period (not only fever or chills, but also unusual symptoms, such as diarrhea, vomiting, or altered mental status). Because the incidence of methicillin-resistant S aureus (MRSA) in our HD unit is very low, cefazolin alone was used as first-line empirical therapy when an infectious event was suspected.10 The antibiotic then was adapted according to susceptibility tests. All culture results (blood and skin swabs) were extracted electronically from the hospital database. No patient received nasal or skin antistaphylococcal prophylaxis, such as mupirocin.
Buttonhole Cannulation Method Track Creation Before starting the buttonhole method, the AVF was evaluated by a nurse specialized in vascular access (reference nurse). For at least 2 weeks (6 sessions), cannulation was performed using sharp Am J Kidney Dis. 2011;57(3):442-448
needles by one reference nurse to create the buttonhole track. When the track had been created, a blunt needle was inserted at least twice consecutively by the reference nurse, and if no problem occurred, subsequent cannulations could be performed by any member of the nursing staff.
General Procedure First, autonomous patients wash the access area with water and soap. Second, the access area is disinfected carefully with alcoholic povidone iodine before scab removal. Third, the scab is removed very carefully with the nonsharp edge of a sterile 19 Gauge sharp needle (BD Microlance 3; Nipro Medical Corporation, www.nipro.com). This step should not be performed hastily. Fourth, the access area is disinfected again using 2 gauzes soaked with alcoholic povidone iodine. As for the second step, it is crucial to respect the contact time of the disinfectant (at least 30 seconds) before cannulation. Fifth, the nurse cannulates using blunt needles, maintaining aseptic technique and following the angle of the track. If the blunt needle does not slip into the track, it must be pulled back a little and the angle must be changed slightly to access the vessel.
Educational Workshops Every step of the buttonhole technique (including scab removal and disinfection before cannulation and after needle removal) was reviewed by nephrologists and reference nurses. The workshops included a video showing the procedure, as well as pictures of infected AVFs. A list of frequent mistakes was compiled and discussed, with particular attention to skin disinfection (including contact time with disinfectant) and scab removal. Another important point was the (over)use of sharp needles in case of cannulation failure with blunt needles. A standardized protocol was created: nurses were no longer allowed to use sharp needles; after 2 unsuccessful attempts with a blunt needle by any nurse, only a reference nurse could use a sharp needle, and then only after 2 additional failed attempts with a blunt needle. The decision also was made to put into the chart of each patient a picture (A4-sized print of jpeg file) of his or her AVF with the blunt needles in place to show the exact angle of the created track. In case of a suspicious scab (hypertrophic and/or irregular or nodular, even without local erythema, pain, or tenderness), a new cannulation site had to be chosen by the reference nurse.
Statistical Analysis For each patient, total time on HD in the unit was retrieved from the charts, together with any infectious event. Exposure to the risk of infectious events was defined as total number of AVF-days. These were determined per period. The incidence of infectious events per period was defined as a person-time rate with the numerator equal to the number of observed episodes of infection during the period of interest, standardized to 1,000 AVF patientdays. For complicated infectious events, similar calculations were performed. Repeated events in the same patient were counted as new events when they occurred more than 3 weeks after completion of antibiotic therapy or when the causal micro-organism was different. Incidence rates between periods were compared using Fischer mid-P exact test (2 tailed), and relative risk (RR) was calculated as the ratio of the incidence rates.11 Parameter estimates were reported with 95% confidence intervals (CIs). Statistical significance was set at P ⬍ 0.05. When necessary, HolmBonferroni correction for multiple testing was performed. In case of a zero event rate, 0.5 was added in each count of the table to allow calculations (Haldane continuity correction). Changes in 443
Labriola et al Table 1. Characteristics of Included Patients
No. of AVFs No. of patients No. of AVF-days Age (y) Men (%) Diabetes (%)
Period 1 (rope-ladder technique)
Period 2 (progressive shift to BH)
Period 3 (BH before workshops)
Period 4 (BH after workshops)
111 105 64,132 71 ⫾ 9 64.9 34.0
56 54 8,887 69.8 ⫾ 9 59.3 31.5
119 112 71,849 70.5 ⫾ 9 65.9 29.7
87 85 41,613 69.8 ⫾ 10.1 60 37.6
Note: Values expressed as mean ⫾ standard deviation or number. Abbreviations: AVF, arteriovenous fistula; BH, buttonhole.
demographic and clinical characteristics of patients between periods were analyzed using RxC tables using Pearson 2 test. Multivariate Poisson regression analysis also was performed on number of events, controlling for age (as a continuous variable), sex, diabetes, AVF location (forearm or upper arm), and periods 2-3-4 (coded as dummy variables), with the reference period by default in regressions therefore period 1.
RESULTS Study Participants A total of 186,481 AVF-days, 193 AVFs, and 177 patients were available for analysis (ie, 100% of candidate patients and AVF-days). Overall, patients were aged 70.4 ⫾ 11.5 years, 65.8% were men, and 32.9% had diabetes. Table 1 lists numbers of AVFdays, AVFs, and patients in each period, as well as main demographic characteristics. Primary End Point (incidence of infectious events) Fifty-seven infectious events occurred during follow up (0.31 events/1,000 AVF-days): 24 local infections without bacteremia, 18 cases of AVF-related bacteremia without local AVF infection, and 15 cases of combined local infection and bacteremia. Per-Period Analysis
Distributions of infectious events and calculated absolute incidences per 1,000 AVF-days are listed in Table 2. The RR of infectious events was significantly lower (0.39; 95% CI, 0.19-0.78; P ⫽ 0.006) during period 1 than period 3, as well as for periods 1 and 2 combined compared with period 3 (RR, 0.38; 95% CI, 0.19-0.73; P ⫽ 0.003). Compared with period 4, the RR of infectious events in period 3 also was higher; however, not significantly (RR, 1.29; 95% CI, 0.692.49; P ⫽ 0.4). During the transition to the buttonhole method (period 2), there was no difference in infectious event rates compared with period 1 (P ⫽ 0.8). The per-period incidence of infectious events is listed in Table 2. Corrected for multiple comparisons, only period 1 and period 1 plus 2 versus period 3 were significant at the level of ␣ ⫽ 0.05. 444
In univariate separate RxC analyses, none of the patients’ characteristics (age, sex, diabetes, or AVF type) was associated with the incidence of infectious events (data not shown). In a multivariate Poisson regression analysis with robust variance estimation, none of the controlling variables was associated significantly with the infection count, except period 3 (RR, 2.28; P ⫽ 0.03). The next highest association was AVF location (upper arm vs forearm), with an RR of 1.71 (P ⫽ 0.09). Annual Analysis
As shown in Fig 1, the absolute annual incidence of infectious events increased progressively after implementation of the buttonhole method: 0.11 (95% CI, 0.0120.4), 0.28 (95% CI, 0.091-0.66), 0.21 (95% CI, 0.0570.54), 0.11 (95% CI, 0-0.38), 0.11 (95% CI, 0.012-0.39), 0.33 (95% CI, 0.13-0.67), 0.69 (95% CI, 0.39-1.12), and 0.68 (95% CI, 0.25-1.78) in 2001, 2002, 2003, 2004, 2005, 2006, 2007, and 2008A, respectively. Nevertheless, the increase only reached statistical significance during the third year after implementation of the buttonhole method, in 2007 (P ⫽ 0.004, 0.07, 0.03, 0.04, and 0.03 for 2007 compared with 2001, 2002, 2003, 2004, and 2005, respectively). Compared with 2006, the increase observed in 2007 was borderline significant (2-tailed P ⫽ 0.09; 1-tailed P ⫽ 0.05). Concerning 2008A, there was no significant difference compared with 2007 (P ⫽ 0.9). Corrected for multiple comparisons, only 2001 was statistically significant compared with 2007 (P ⫽ 0.004 compared with a corrected threshold P ⫽ 0.007). During 2008B (immediately after the educational workshops), the incidence of infectious events per 1,000 AVF days significantly decreased compared with 2008A, with an RR of 0.16 (95% CI, 0.017-0.56; P ⫽ 0.03). In 2009 and 2010, a nonsignificant increase was observed compared with 2008B, with RRs of 2.95 (95% CI, 0.70-20.06; P ⫽ 0.2) and 2.11 (95% CI, 0.31-17.73; P ⫽ 0.4), respectively. Table 3 lists micro-organisms causing AVF-related bacteremia. S aureus was found in 66.7% of cases. In Am J Kidney Dis. 2011;57(3):442-448
AVF-Related Infections and Buttonhole Cannulation Table 2. Incidence of Infectious Events Per Period Period 1 (rope-ladder technique)
Period 2 (progressive shift to BH)
Period 3 (BH before workshops)
Period 4 (BH after workshops)
Local infection alone No. No. of AVFs requiring surgery
5 0
0 0
11 3
8 0
Bacteremia alone No. No. of AVFs requiring surgery
3 0
1 0
12 0
2 0
Combined local infection and bacteremia No. No. of AVFs requiring surgery
3 0
0 0
8 2
4 0
Metastatic infections No. No. of AVFs requiring surgery
0 0
0 0
4 0
1 0
Deaths No. with endocarditis No. with septicemia
0 0
0 0
2 2
1 0
11 0.17 (0.086-0.31)
1 0.11 (0.0014-0.63)
31 0.43 (0.29-0.61)
14 0.34 (0.19-0.55)
0
0
11
1
0
0
0.153 (0.076-0.273)
0.024 (0.001-0.118)
Infectious Events
Total no. of infectious events Absolute incidence of all infectious eventsa Complicated infectious events (n) Absolute incidence of complicated infectious eventsa
Note: When provided, confidence intervals are in parentheses. Abbreviations: AVF, arteriovenous fistula; BH, buttonhole. a Per 1,000 AVF-days.
agreement with the very low prevalence of MRSA in our in-center HD unit,10 no bacteremia caused by MRSA was observed during the studied 9-year period. Secondary End Point (incidence of complicated infectious events) As listed in Table 2, some infectious events were particularly severe and all except one were restricted to period 3, when 3 patients developed endocarditis,
of which 2 were caused by S aureus (both fatal) and 1 was caused by S epidermidis (the patient survived). One more endocarditis caused by S epidermidis occurred during early period 4 (fatal). These 4 patients presented with fever and bacteremia without local AVF infection. Two other deaths related to an infectious event were observed during period 3, with 2 cases of local AVF infection with fatal S aureus bacteremia (2006). No death was related to an infec-
Figure 1. Annual incidence of infectious events. Lines around the squares indicate 95% confidence intervals; *P ⬍ 0.05 (compared with 2001, 2002, 2003, 2004, 2005, 2006, and 2008B); **P ⬍ 0.05 (compared with 2001, 2003, 2004, 2005, and 2008B). Period 1 (January 1, 2001, to August 3, 2004): all patients using rope-ladder technique with sharp needles; period 2 (August 4, 2004, to January 31, 2005): progressive switch to buttonhole (BH) method using blunt needles; period 3 (February 1, 2005, to May 19, 2008): all patients using BH method, before educational workshops; and period 4 (May 20, 2008, to June 30, 2010): all patients using BH method, after educational workshops. Abbreviation: AVF, arteriovenous fistula. Am J Kidney Dis. 2011;57(3):442-448
445
Labriola et al Table 3. Microorganisms Causing AVF-Related Bacteremia
Methicillin-sensitive Staphylococcus aureus Staphylococcus epidermidis Total (n)
Period 1 (rope-ladder technique)
Period 2 (progressive shift to BH)
Period 3 (BH before workshops)
Period 4 (BH after workshops)
2
0
15
5
4 6
1 1
5 20
1 6
Abbreviations: AVF, arteriovenous fistula; BH, buttonhole.
tious event originating from the AVF during the other periods. In addition to the 4 cases of endocarditis, 1 more metastatic infection was observed: 1 patient with bacteremia caused by S epidermidis developed metastatic costovertebral arthritis (cured after 6 weeks of intravenous antibiotic therapy). Concerning local AVF infections, 34 episodes were cured using intravenous antibiotic therapy alone. However, 5 required surgery: 4 because of spontaneous rupture and 1 because of imminent risk of rupture (very thin skin surrounding a large necrotic cannulation site area), all during period 3. Two patients developed subsequent bacteremia and died. Thus, 11 complicated infectious events occurred during period 3 and 1 occurred during period 4, whereas none was observed during periods 1 and 2. As listed in Table 2, absolute incidences of complicated infectious events were 0.153 (95% CI, 0.0760.273) and 0.024 (95% CI, 0.001-0.118) per 1,000 AVF-days in periods 3 and 4, respectively. The RR of complicated infectious events was significantly lower during periods 1 and 2 combined compared with period 3 (0.051; 95% CI, 0.003-0.87; P ⫽ 0.01). Only one complicated infectious event was observed after intervention (period 4). The decrease in incidence of complicated infectious events in period 4 in comparison to period 3 was statistically significant (RR for period 3 vs period 4, 6.37; 95% CI, 1.09-138.4; P ⫽ 0.04). In a univariate RxC table analysis, only period 3 compared with all other periods grouped and female sex were associated significantly with the occurrence of complicated infectious events, even when corrected for multiple comparisons (Bonferroni threshold, P ⫽ 0.010). This was confirmed using logistic regression analysis of the probability of a complicated infectious event, controlling for the same potential confounding variables and exposure as previously performed for the primary end point. Only female sex and period 3 were associated significantly with complicated infectious events (odds ratios, 8.79 [P ⫽ 0.008] and 27.17 [P ⫽ 0.002], respectively). 446
DISCUSSION Although infection rates with native AVFs are low,6,7 there recently has been concern that infection rates may increase using the buttonhole cannulation method.4,5,8,9,12,13 However, reports concerning infections with buttonhole cannulation are mostly anecdotal or describe small cohorts of self-cannulating home dialysis patients. To our knowledge, this is the first report documenting an increase in infections associated with buttonhole cannulation in an in-center HD unit. A recent observational study reported infection as a negative outcome of buttonhole cannulation in self-cannulating home HD patients in Canada.13 However, this finding might tentatively be ascribed to poorer hygienic precautions by patients than health care professionals. In our hospital HD unit, the staff includes 40 RNs (many part-time), 5 of whom are considered reference nurses. Patients are at least 15 years older than in previous reports4,13 and have multiple comorbid conditions. Thus, our report provides important new information about the risk of infectious events and especially complicated infectious events associated with implementation of the buttonhole method in an in-center HD unit. It describes our educational campaign to decrease the rate of infectious events with the buttonhole method and the improvement after rigorous re-education and constant training of staff. The role of the vascular access coordinator(s) or reference nurse(s) is essential not only for teaching buttonhole cannulation to other staff members, but also for track creation and difficult cannulations. We documented a significant increase in incidence of infectious events after implementation of the buttonhole method. Interestingly, this increase did not reach statistical significance until the third year after its implementation. We strongly suspect that the nurses, after initial meticulous learning and training, progressively paid less attention to the hygiene protocol of the buttonhole technique. The decrease in rate of infectious events after staff re-education supports this hypothesis. Other factors that theoretically could have had a role in the rate of AVF infections, such as type of disinfectant (alcoholic povidone iodine) or patient to Am J Kidney Dis. 2011;57(3):442-448
AVF-Related Infections and Buttonhole Cannulation
nurse ratio (3:1),14 were unchanged during the study period. The turnover rate of our nursing staff did not change during the study, with a median working time in the unit exceeding 10 years, a testimony of the experience with AVFs of our staff (working time was even longer for reference nurses). During period 3, the incidence of catheter-related bacteremia remained low in our unit,15 thus suggesting the absence of general problems with hygienic precautions in our unit, but on the contrary, a specific buttonhole-related problem. Our study has some limitations. This long-term study had a partly retrospective pre-post design. However, our HD charts are completed meticulously and always document the indication for and duration of every antibiotic treatment. Furthermore, our policy of systematic blood cultures according to very broad indications (as discussed) did not change for decades. All culture results were extracted electronically from the hospital database. These 3 characteristics make underestimation of infectious events in periods 1 and 2 very unlikely. Moreover, the dramatic events defined as “complicated infectious events” were unlikely to be missed, even with a partly retrospective design. All hospitalizations took place in our hospital, thus providing easy access to a single database of electronic records of every hospitalization. In our study, age, sex, AVF location, and diabetes were not predictive of infectious events in general. However, complicated infectious events were significantly more frequent in women (8 complicated infectious events in a total of 12). Nonetheless, because sex ratio did not change between periods, this result could be a chance finding. Alternatively, this finding may reflect the difficulties related to AVF characteristics in women, making cannulation more difficult and thus potentially increasing the risk of severe AVF infection and rupture. AVF failure rates are higher in women than men in some, but not all, studies.16-18 In view of the dramatic infectious events observed after the switch to buttonhole cannulation, one might question the ongoing use of this technique in our unit. However, the buttonhole method has been used for more than 10 years and is still used in our home HD and low-care patients, with excellent outcomes and no increase in infectious complications (T. Goovaerts, data not shown). Second, potential advantages of the buttonhole method (as described) are expected to improve AVF longevity and thus avoid catheter insertion (and the associated infectious risk). Nonetheless, we acknowledge that in units lacking adequate RN staffing or training, the risk may outweigh the potential benefits. Ultimately, only a randomized controlled trial could help assess the pros and contras of buttonhole cannulation. Am J Kidney Dis. 2011;57(3):442-448
A key factor explaining the increase in infectious event rate in buttonhole cannulation is nonadherence to hygienic rules during every step of the procedure. Buttonhole cannulation is a demanding technique and requires strict adherence to the protocol. Rigorous disinfection not only before, but also after scab removal is crucial. Scab removal often is difficult and must be complete to prevent scab particles, which contain bacteria, from entering the blood. We recommend that these steps should not be performed hastily, even in a busy in-center HD facility. However, to avoid tissue injury at the mouth of the track, scab removal should not be performed using sharp instruments. In some units, scabs are soaked with saline for 10 minutes before removal. Hadley and Moran19 very recently reported the use of gauze soaked with tap water and liquid antibacterial soap over each scab for a few minutes to make scab removal easier. Another crucial point is careful attention to the angle of cannulation. Use of videos or pictures can help show the exact cannulation angle, especially in large units with many staff members. Poor attention to the angle of cannulation may trigger intermittent use of sharp instead of blunt needles. As noted by van Loon et al,5 some nurses prefer to recannulate using a sharp needle when cannulation with a blunt needle seems difficult. This makes cannulation much easier. However, use of sharp needles may damage the created track, resulting in the formation of one or more “false tracks,” which can predispose to larger scabs difficult to remove, more track injury, and finally, infection. In our unit, the rate of sharp needle use decreased dramatically from 48% in 2007 to 1.5% at the end of 2008 (data not shown). Nesrallah et al13 recently observed an increase in bacteriemia caused by S aureus (with metastatic complications in 4 cases) after implementation of the buttonhole method in 56 patients receiving home nocturnal HD. All cases of bacteremia coincided with local AVF infection, in contrast to our cases without obvious AVF infection despite bacteremia. Interestingly, Nesrallah et al13 reported a significant decrease after using topical mupirocin 2% cream on each buttonhole site with a sterile cotton swab, without dressings, after hemostasis achievement. No infection occurred during treatment with mupirocin. Although numbers of patients and events were relatively small, as the investigators note, these results are promising and should trigger randomized controlled trials. In in-center HD units, the risk of resistance to mupirocin might represent a substantial barrier to the long-term success of this strategy.20 In conclusion, units considering a shift to buttonhole cannulation should be warned that the method requires strict respect of every step of the procedure to 447
Labriola et al
avoid severe infectious complications. Rigorous education and constant training of the nurses are mandatory to reach this goal.
ACKNOWLEDGEMENTS Support: None. Financial Disclosure: The authors declare that they have no relevant financial interests.
REFERENCES 1. Twardowski Z. The “buttonhole” method of needle insertion takes center stage in the attempt to revive daily home hemodialysis. Contemp Dial Nephrol. 1977;18(2):18-19. 2. Twardowski Z, Kubara H. Different sites versus constant sites of needle insertion into arteriovenous fistulas for treatment of repeated dialysis. Dial Transplant. 1979;8:978-980. 3. Krönung G. Plastic deformation of Cimino fistula by repeated puncture. Dial Transplant. 1984;13:635-638. 4. Verhallen AM, Kooistra MP, van Jaarsveld BC. Cannulating in haemodialysis: rope-ladder or buttonhole technique? Nephrol Dial Transplant. 2007;22(9):2601-2604. 5. Van Loon MM, Goovaerts T, Kessels AG, van der Sande FK, Tordoir JH. Buttonhole needling of haemodialysis arteriovenous fistulae results in less complications and interventions compared to the rope-ladder technique. Nephrol Dial Transplant. 2010;25(1): 225-230. 6. Rayner HC, Pisoni RL. The increasing use of hemodialysis catheters: evidence from the DOPPS on its significance and ways to reverse it. Semin Dial. 2010;23(1):6-10. 7. Hoen B, Paul-Dauphin A, Hestin D, Kessler M. EPIBACDIAL: a multicenter prospective study of risk factors for bacteremia in chronic hemodialysis patients. J Am Soc Nephrol. 1998;5:869-876. 8. Doss S, Schiller B, Moran J. Buttonhole cannulation—an unexpected outcome. Nephrol Nurs J. 2008;35:417-419.
448
9. Ludlow V. Buttonhole cannulation in hemodialysis: improved outcomes and increased expense—is it worth it? CANNT J. 2010;20(1):29-37. 10. Labriola L, Jadoul M. New recommendations in the treatment of gram positive bacteraemia in dialysis patients. Nephrol Dial Transplant. 2008;23(7):2431-2432. 11. Lancaster HO. Significance tests in discrete distributions. JASA. 1961;56:223-234 12. Murcutt G. Buttonhole cannulation: should this become the default technique for dialysis patients with native fistulas? J Renal Care. 2008;34(2):101-108. 13. Nesrallah GE, Cuerden M, Wong JH, Pierratos A. Staphylococcus aureus bacteraemia and buttonhole cannulation: long-term safety and efficacy of mupirocin prophylaxis. Clin J Am Soc Nephrol. 2010;5(6):1047-1053. 14. Elseviers MM, Van Waeleghem JP; European Dialysis and Transplant Nurses Association/European Renal Care Association. Complications of vascular access: results of a European multi centre study of the EDTNA/ERCA Research Board. EDTNA ERCA J. 2003;29(3):163-167. 15. Labriola L, Crott R, Jadoul M. Catheter-related bacteraemia is not a fatality: 8 year-follow-up in a Belgian hemodialysis unit. Poster presented at the World Congress of Nephrology 2009; May 22-26, 2009; Milan, Italy. 16. Allon M, Lockhart ME, Lilly RZ, et al. Effect of preoperative sonographic mapping on vascular access outcomes in hemodialysis patients. Kidney Int. 2001;60(5):2013-2020. 17. Peterson WJ, Barker J, Allon M. Disparities in fistula maturation persist despite preoperative vascular mapping. Clin J Am Soc Nephrol. 2008;3(2):437-441. 18. Caplin N, Sedlacek M, Teodorescu V, Falk A, Uribarri J. Venous access: women are equal. Am J Kidney Dis. 2003;41:429432. 19. Hadley K, Moran J. Buttonhole access: noninvasive scab removal. [Annual Dialysis Conference; March 7-9, 2010; Seattle, Washington; abstract]. Hemodial Int. 2010;14(1):98. 20. Patel JB, Gorwitz RG, Jernigan JA. Mupirocin resistance. Clin Infect Dis. 2009;49(6):935-941.
Am J Kidney Dis. 2011;57(3):442-448
C
onstant-site or buttonhole cannulation of native arteriovenous fistulas (AVFs) with blunt needles was reported first in 1977 by Twardowski1 and recently has gained in popularity, especially in home hemodialysis (HD). In buttonhole cannulation, the needles are inserted at the same AVF sites at each session, instead of moving the cannulation sites along the whole access length (rope-ladder method). Using a constant site during 6 HD sessions creates a From the Departments of 1Nephrology and 2Public Health, Cliniques Universitaires Saint-Luc, Université Catholique de Louvain, Brussels, Belgium. Received June 3, 2010. Accepted in revised form October 1, 2010. Originally published online January 10, 2011. Address correspondence to Laura Labriola, MD, Department of Nephrology, Cliniques Universitaires Saint-Luc, Université Catholique de Louvain, Ave Hippocrate 10, B 1200 Brussels, Belgium. E-mail: [email protected] © 2011 by the National Kidney Foundation, Inc. 0272-6386/$36.00 doi:10.1053/j.ajkd.2010.10.045 442
tunnel from the skin to the vessel. This tunnel is used during subsequent cannulations to guide the blunt needle to access the vessel, after careful disinfection and withdrawal of the scab formed at the cannulation site. Potential advantages of buttonhole cannulation over the standard rope-ladder technique include easier cannulation, fewer missed sticks, less pain, faster hemostasis after needle removal, and fewer hematomas and aneurysms.2-5 However, studies comparing cannulation techniques are very scarce.4,5 Infection rates with native AVFs are low.6,7 However, the possibility of an increased incidence of infectious events with buttonhole cannulation recently has been suggested, although it is poorly documented.4,5,8,9 Because cannulating nonhealed skin might increase the risk of infectious events, we conducted an observational study in our HD patients with AVFs to examine the impact of switching to buttonhole cannulation on the infectious event rate. Am J Kidney Dis. 2011;57(3):442-448
AVF-Related Infections and Buttonhole Cannulation
METHODS Participants and Quality Improvement Plan We included all patients on maintenance HD therapy using a native AVF from January 1, 2001, to June 30, 2010, in our hospital HD unit. They were dialyzed 3 times a week (except for 2 patients dialyzed 6 times weekly and 2 others dialyzed only twice weekly) for 3.5-5 (median, 4) h/session using hollow-fiber high-flux or superflux polysulfone dialyzers (Fresenius Medical Care, www. fresenius.com). Standard blood flow was 360 mL/min. Selfcannulation was not performed. From August 4, 2004, to January 31, 2005, the rope-ladder technique using sharp needles was progressively replaced by the buttonhole cannulation method with blunt needles. After January 31, 2005, all AVFs were cannulated using blunt needles (15 Gauge, except in 3 patients, 14 Gauge) using the buttonhole method. Because the rate of infectious events (defined later) increased progressively in 2006-2007, educational workshops were organized for all nurses in late May 2008. Every step of the buttonhole cannulation method was reviewed carefully (discussed later), including scab removal and disinfection before cannulation and after needle removal. Four periods thus were considered in our analysis: (1) period 1 (January 1, 2001, to August 3, 2004): all patients on rope-ladder technique with sharp needles; (2) period 2 (August 4, 2004, to January 31, 2005): progressive switch to buttonhole cannulation method with blunt needles; (3) period 3 (February 1, 2005, to May 19, 2008): all patients on buttonhole method, before educational workshops; and (4) period 4 (May 20, 2008, to June 30, 2010): all patients on buttonhole cannulation method, after workshops. For annual comparisons, 2008 was divided into A (before educational workshops: months 1-5) and B (after workshops: months 6-12). Age, sex, diabetes, number of AVF-days, location of AVF (forearm or upper arm), and infectious events were recorded. The primary end point was the incidence of infectious events. The secondary end point was the incidence of complicated infectious events. Infectious events included both unexplained bacteremia caused by skin bacteria and/or local AVF infection. Local AVF infection was defined as nonallergic erythema, pain or tenderness close to cannulation sites, necrotic scabs, or drainage from cannulation site(s). In every case, a local swab specimen was obtained. Bacteremia caused by a skin micro-organism (ie, Staphylococcus aureus or Staphylococcus epidermidis) was considered an infectious event originating from the AVF, even without local signs, in the absence of an alternative source (such as diabetic foot, etc). Complicated infectious events included metastatic complications (ie, endocarditis, septic arthritis, etc), death related to an infectious event, and AVFs requiring surgery (overt or imminent rupture). Indications for measuring C-reactive protein and obtaining blood cultures were very broad and unchanged in our unit during the study period (not only fever or chills, but also unusual symptoms, such as diarrhea, vomiting, or altered mental status). Because the incidence of methicillin-resistant S aureus (MRSA) in our HD unit is very low, cefazolin alone was used as first-line empirical therapy when an infectious event was suspected.10 The antibiotic then was adapted according to susceptibility tests. All culture results (blood and skin swabs) were extracted electronically from the hospital database. No patient received nasal or skin antistaphylococcal prophylaxis, such as mupirocin.
Buttonhole Cannulation Method Track Creation Before starting the buttonhole method, the AVF was evaluated by a nurse specialized in vascular access (reference nurse). For at least 2 weeks (6 sessions), cannulation was performed using sharp Am J Kidney Dis. 2011;57(3):442-448
needles by one reference nurse to create the buttonhole track. When the track had been created, a blunt needle was inserted at least twice consecutively by the reference nurse, and if no problem occurred, subsequent cannulations could be performed by any member of the nursing staff.
General Procedure First, autonomous patients wash the access area with water and soap. Second, the access area is disinfected carefully with alcoholic povidone iodine before scab removal. Third, the scab is removed very carefully with the nonsharp edge of a sterile 19 Gauge sharp needle (BD Microlance 3; Nipro Medical Corporation, www.nipro.com). This step should not be performed hastily. Fourth, the access area is disinfected again using 2 gauzes soaked with alcoholic povidone iodine. As for the second step, it is crucial to respect the contact time of the disinfectant (at least 30 seconds) before cannulation. Fifth, the nurse cannulates using blunt needles, maintaining aseptic technique and following the angle of the track. If the blunt needle does not slip into the track, it must be pulled back a little and the angle must be changed slightly to access the vessel.
Educational Workshops Every step of the buttonhole technique (including scab removal and disinfection before cannulation and after needle removal) was reviewed by nephrologists and reference nurses. The workshops included a video showing the procedure, as well as pictures of infected AVFs. A list of frequent mistakes was compiled and discussed, with particular attention to skin disinfection (including contact time with disinfectant) and scab removal. Another important point was the (over)use of sharp needles in case of cannulation failure with blunt needles. A standardized protocol was created: nurses were no longer allowed to use sharp needles; after 2 unsuccessful attempts with a blunt needle by any nurse, only a reference nurse could use a sharp needle, and then only after 2 additional failed attempts with a blunt needle. The decision also was made to put into the chart of each patient a picture (A4-sized print of jpeg file) of his or her AVF with the blunt needles in place to show the exact angle of the created track. In case of a suspicious scab (hypertrophic and/or irregular or nodular, even without local erythema, pain, or tenderness), a new cannulation site had to be chosen by the reference nurse.
Statistical Analysis For each patient, total time on HD in the unit was retrieved from the charts, together with any infectious event. Exposure to the risk of infectious events was defined as total number of AVF-days. These were determined per period. The incidence of infectious events per period was defined as a person-time rate with the numerator equal to the number of observed episodes of infection during the period of interest, standardized to 1,000 AVF patientdays. For complicated infectious events, similar calculations were performed. Repeated events in the same patient were counted as new events when they occurred more than 3 weeks after completion of antibiotic therapy or when the causal micro-organism was different. Incidence rates between periods were compared using Fischer mid-P exact test (2 tailed), and relative risk (RR) was calculated as the ratio of the incidence rates.11 Parameter estimates were reported with 95% confidence intervals (CIs). Statistical significance was set at P ⬍ 0.05. When necessary, HolmBonferroni correction for multiple testing was performed. In case of a zero event rate, 0.5 was added in each count of the table to allow calculations (Haldane continuity correction). Changes in 443
Labriola et al Table 1. Characteristics of Included Patients
No. of AVFs No. of patients No. of AVF-days Age (y) Men (%) Diabetes (%)
Period 1 (rope-ladder technique)
Period 2 (progressive shift to BH)
Period 3 (BH before workshops)
Period 4 (BH after workshops)
111 105 64,132 71 ⫾ 9 64.9 34.0
56 54 8,887 69.8 ⫾ 9 59.3 31.5
119 112 71,849 70.5 ⫾ 9 65.9 29.7
87 85 41,613 69.8 ⫾ 10.1 60 37.6
Note: Values expressed as mean ⫾ standard deviation or number. Abbreviations: AVF, arteriovenous fistula; BH, buttonhole.
demographic and clinical characteristics of patients between periods were analyzed using RxC tables using Pearson 2 test. Multivariate Poisson regression analysis also was performed on number of events, controlling for age (as a continuous variable), sex, diabetes, AVF location (forearm or upper arm), and periods 2-3-4 (coded as dummy variables), with the reference period by default in regressions therefore period 1.
RESULTS Study Participants A total of 186,481 AVF-days, 193 AVFs, and 177 patients were available for analysis (ie, 100% of candidate patients and AVF-days). Overall, patients were aged 70.4 ⫾ 11.5 years, 65.8% were men, and 32.9% had diabetes. Table 1 lists numbers of AVFdays, AVFs, and patients in each period, as well as main demographic characteristics. Primary End Point (incidence of infectious events) Fifty-seven infectious events occurred during follow up (0.31 events/1,000 AVF-days): 24 local infections without bacteremia, 18 cases of AVF-related bacteremia without local AVF infection, and 15 cases of combined local infection and bacteremia. Per-Period Analysis
Distributions of infectious events and calculated absolute incidences per 1,000 AVF-days are listed in Table 2. The RR of infectious events was significantly lower (0.39; 95% CI, 0.19-0.78; P ⫽ 0.006) during period 1 than period 3, as well as for periods 1 and 2 combined compared with period 3 (RR, 0.38; 95% CI, 0.19-0.73; P ⫽ 0.003). Compared with period 4, the RR of infectious events in period 3 also was higher; however, not significantly (RR, 1.29; 95% CI, 0.692.49; P ⫽ 0.4). During the transition to the buttonhole method (period 2), there was no difference in infectious event rates compared with period 1 (P ⫽ 0.8). The per-period incidence of infectious events is listed in Table 2. Corrected for multiple comparisons, only period 1 and period 1 plus 2 versus period 3 were significant at the level of ␣ ⫽ 0.05. 444
In univariate separate RxC analyses, none of the patients’ characteristics (age, sex, diabetes, or AVF type) was associated with the incidence of infectious events (data not shown). In a multivariate Poisson regression analysis with robust variance estimation, none of the controlling variables was associated significantly with the infection count, except period 3 (RR, 2.28; P ⫽ 0.03). The next highest association was AVF location (upper arm vs forearm), with an RR of 1.71 (P ⫽ 0.09). Annual Analysis
As shown in Fig 1, the absolute annual incidence of infectious events increased progressively after implementation of the buttonhole method: 0.11 (95% CI, 0.0120.4), 0.28 (95% CI, 0.091-0.66), 0.21 (95% CI, 0.0570.54), 0.11 (95% CI, 0-0.38), 0.11 (95% CI, 0.012-0.39), 0.33 (95% CI, 0.13-0.67), 0.69 (95% CI, 0.39-1.12), and 0.68 (95% CI, 0.25-1.78) in 2001, 2002, 2003, 2004, 2005, 2006, 2007, and 2008A, respectively. Nevertheless, the increase only reached statistical significance during the third year after implementation of the buttonhole method, in 2007 (P ⫽ 0.004, 0.07, 0.03, 0.04, and 0.03 for 2007 compared with 2001, 2002, 2003, 2004, and 2005, respectively). Compared with 2006, the increase observed in 2007 was borderline significant (2-tailed P ⫽ 0.09; 1-tailed P ⫽ 0.05). Concerning 2008A, there was no significant difference compared with 2007 (P ⫽ 0.9). Corrected for multiple comparisons, only 2001 was statistically significant compared with 2007 (P ⫽ 0.004 compared with a corrected threshold P ⫽ 0.007). During 2008B (immediately after the educational workshops), the incidence of infectious events per 1,000 AVF days significantly decreased compared with 2008A, with an RR of 0.16 (95% CI, 0.017-0.56; P ⫽ 0.03). In 2009 and 2010, a nonsignificant increase was observed compared with 2008B, with RRs of 2.95 (95% CI, 0.70-20.06; P ⫽ 0.2) and 2.11 (95% CI, 0.31-17.73; P ⫽ 0.4), respectively. Table 3 lists micro-organisms causing AVF-related bacteremia. S aureus was found in 66.7% of cases. In Am J Kidney Dis. 2011;57(3):442-448
AVF-Related Infections and Buttonhole Cannulation Table 2. Incidence of Infectious Events Per Period Period 1 (rope-ladder technique)
Period 2 (progressive shift to BH)
Period 3 (BH before workshops)
Period 4 (BH after workshops)
Local infection alone No. No. of AVFs requiring surgery
5 0
0 0
11 3
8 0
Bacteremia alone No. No. of AVFs requiring surgery
3 0
1 0
12 0
2 0
Combined local infection and bacteremia No. No. of AVFs requiring surgery
3 0
0 0
8 2
4 0
Metastatic infections No. No. of AVFs requiring surgery
0 0
0 0
4 0
1 0
Deaths No. with endocarditis No. with septicemia
0 0
0 0
2 2
1 0
11 0.17 (0.086-0.31)
1 0.11 (0.0014-0.63)
31 0.43 (0.29-0.61)
14 0.34 (0.19-0.55)
0
0
11
1
0
0
0.153 (0.076-0.273)
0.024 (0.001-0.118)
Infectious Events
Total no. of infectious events Absolute incidence of all infectious eventsa Complicated infectious events (n) Absolute incidence of complicated infectious eventsa
Note: When provided, confidence intervals are in parentheses. Abbreviations: AVF, arteriovenous fistula; BH, buttonhole. a Per 1,000 AVF-days.
agreement with the very low prevalence of MRSA in our in-center HD unit,10 no bacteremia caused by MRSA was observed during the studied 9-year period. Secondary End Point (incidence of complicated infectious events) As listed in Table 2, some infectious events were particularly severe and all except one were restricted to period 3, when 3 patients developed endocarditis,
of which 2 were caused by S aureus (both fatal) and 1 was caused by S epidermidis (the patient survived). One more endocarditis caused by S epidermidis occurred during early period 4 (fatal). These 4 patients presented with fever and bacteremia without local AVF infection. Two other deaths related to an infectious event were observed during period 3, with 2 cases of local AVF infection with fatal S aureus bacteremia (2006). No death was related to an infec-
Figure 1. Annual incidence of infectious events. Lines around the squares indicate 95% confidence intervals; *P ⬍ 0.05 (compared with 2001, 2002, 2003, 2004, 2005, 2006, and 2008B); **P ⬍ 0.05 (compared with 2001, 2003, 2004, 2005, and 2008B). Period 1 (January 1, 2001, to August 3, 2004): all patients using rope-ladder technique with sharp needles; period 2 (August 4, 2004, to January 31, 2005): progressive switch to buttonhole (BH) method using blunt needles; period 3 (February 1, 2005, to May 19, 2008): all patients using BH method, before educational workshops; and period 4 (May 20, 2008, to June 30, 2010): all patients using BH method, after educational workshops. Abbreviation: AVF, arteriovenous fistula. Am J Kidney Dis. 2011;57(3):442-448
445
Labriola et al Table 3. Microorganisms Causing AVF-Related Bacteremia
Methicillin-sensitive Staphylococcus aureus Staphylococcus epidermidis Total (n)
Period 1 (rope-ladder technique)
Period 2 (progressive shift to BH)
Period 3 (BH before workshops)
Period 4 (BH after workshops)
2
0
15
5
4 6
1 1
5 20
1 6
Abbreviations: AVF, arteriovenous fistula; BH, buttonhole.
tious event originating from the AVF during the other periods. In addition to the 4 cases of endocarditis, 1 more metastatic infection was observed: 1 patient with bacteremia caused by S epidermidis developed metastatic costovertebral arthritis (cured after 6 weeks of intravenous antibiotic therapy). Concerning local AVF infections, 34 episodes were cured using intravenous antibiotic therapy alone. However, 5 required surgery: 4 because of spontaneous rupture and 1 because of imminent risk of rupture (very thin skin surrounding a large necrotic cannulation site area), all during period 3. Two patients developed subsequent bacteremia and died. Thus, 11 complicated infectious events occurred during period 3 and 1 occurred during period 4, whereas none was observed during periods 1 and 2. As listed in Table 2, absolute incidences of complicated infectious events were 0.153 (95% CI, 0.0760.273) and 0.024 (95% CI, 0.001-0.118) per 1,000 AVF-days in periods 3 and 4, respectively. The RR of complicated infectious events was significantly lower during periods 1 and 2 combined compared with period 3 (0.051; 95% CI, 0.003-0.87; P ⫽ 0.01). Only one complicated infectious event was observed after intervention (period 4). The decrease in incidence of complicated infectious events in period 4 in comparison to period 3 was statistically significant (RR for period 3 vs period 4, 6.37; 95% CI, 1.09-138.4; P ⫽ 0.04). In a univariate RxC table analysis, only period 3 compared with all other periods grouped and female sex were associated significantly with the occurrence of complicated infectious events, even when corrected for multiple comparisons (Bonferroni threshold, P ⫽ 0.010). This was confirmed using logistic regression analysis of the probability of a complicated infectious event, controlling for the same potential confounding variables and exposure as previously performed for the primary end point. Only female sex and period 3 were associated significantly with complicated infectious events (odds ratios, 8.79 [P ⫽ 0.008] and 27.17 [P ⫽ 0.002], respectively). 446
DISCUSSION Although infection rates with native AVFs are low,6,7 there recently has been concern that infection rates may increase using the buttonhole cannulation method.4,5,8,9,12,13 However, reports concerning infections with buttonhole cannulation are mostly anecdotal or describe small cohorts of self-cannulating home dialysis patients. To our knowledge, this is the first report documenting an increase in infections associated with buttonhole cannulation in an in-center HD unit. A recent observational study reported infection as a negative outcome of buttonhole cannulation in self-cannulating home HD patients in Canada.13 However, this finding might tentatively be ascribed to poorer hygienic precautions by patients than health care professionals. In our hospital HD unit, the staff includes 40 RNs (many part-time), 5 of whom are considered reference nurses. Patients are at least 15 years older than in previous reports4,13 and have multiple comorbid conditions. Thus, our report provides important new information about the risk of infectious events and especially complicated infectious events associated with implementation of the buttonhole method in an in-center HD unit. It describes our educational campaign to decrease the rate of infectious events with the buttonhole method and the improvement after rigorous re-education and constant training of staff. The role of the vascular access coordinator(s) or reference nurse(s) is essential not only for teaching buttonhole cannulation to other staff members, but also for track creation and difficult cannulations. We documented a significant increase in incidence of infectious events after implementation of the buttonhole method. Interestingly, this increase did not reach statistical significance until the third year after its implementation. We strongly suspect that the nurses, after initial meticulous learning and training, progressively paid less attention to the hygiene protocol of the buttonhole technique. The decrease in rate of infectious events after staff re-education supports this hypothesis. Other factors that theoretically could have had a role in the rate of AVF infections, such as type of disinfectant (alcoholic povidone iodine) or patient to Am J Kidney Dis. 2011;57(3):442-448
AVF-Related Infections and Buttonhole Cannulation
nurse ratio (3:1),14 were unchanged during the study period. The turnover rate of our nursing staff did not change during the study, with a median working time in the unit exceeding 10 years, a testimony of the experience with AVFs of our staff (working time was even longer for reference nurses). During period 3, the incidence of catheter-related bacteremia remained low in our unit,15 thus suggesting the absence of general problems with hygienic precautions in our unit, but on the contrary, a specific buttonhole-related problem. Our study has some limitations. This long-term study had a partly retrospective pre-post design. However, our HD charts are completed meticulously and always document the indication for and duration of every antibiotic treatment. Furthermore, our policy of systematic blood cultures according to very broad indications (as discussed) did not change for decades. All culture results were extracted electronically from the hospital database. These 3 characteristics make underestimation of infectious events in periods 1 and 2 very unlikely. Moreover, the dramatic events defined as “complicated infectious events” were unlikely to be missed, even with a partly retrospective design. All hospitalizations took place in our hospital, thus providing easy access to a single database of electronic records of every hospitalization. In our study, age, sex, AVF location, and diabetes were not predictive of infectious events in general. However, complicated infectious events were significantly more frequent in women (8 complicated infectious events in a total of 12). Nonetheless, because sex ratio did not change between periods, this result could be a chance finding. Alternatively, this finding may reflect the difficulties related to AVF characteristics in women, making cannulation more difficult and thus potentially increasing the risk of severe AVF infection and rupture. AVF failure rates are higher in women than men in some, but not all, studies.16-18 In view of the dramatic infectious events observed after the switch to buttonhole cannulation, one might question the ongoing use of this technique in our unit. However, the buttonhole method has been used for more than 10 years and is still used in our home HD and low-care patients, with excellent outcomes and no increase in infectious complications (T. Goovaerts, data not shown). Second, potential advantages of the buttonhole method (as described) are expected to improve AVF longevity and thus avoid catheter insertion (and the associated infectious risk). Nonetheless, we acknowledge that in units lacking adequate RN staffing or training, the risk may outweigh the potential benefits. Ultimately, only a randomized controlled trial could help assess the pros and contras of buttonhole cannulation. Am J Kidney Dis. 2011;57(3):442-448
A key factor explaining the increase in infectious event rate in buttonhole cannulation is nonadherence to hygienic rules during every step of the procedure. Buttonhole cannulation is a demanding technique and requires strict adherence to the protocol. Rigorous disinfection not only before, but also after scab removal is crucial. Scab removal often is difficult and must be complete to prevent scab particles, which contain bacteria, from entering the blood. We recommend that these steps should not be performed hastily, even in a busy in-center HD facility. However, to avoid tissue injury at the mouth of the track, scab removal should not be performed using sharp instruments. In some units, scabs are soaked with saline for 10 minutes before removal. Hadley and Moran19 very recently reported the use of gauze soaked with tap water and liquid antibacterial soap over each scab for a few minutes to make scab removal easier. Another crucial point is careful attention to the angle of cannulation. Use of videos or pictures can help show the exact cannulation angle, especially in large units with many staff members. Poor attention to the angle of cannulation may trigger intermittent use of sharp instead of blunt needles. As noted by van Loon et al,5 some nurses prefer to recannulate using a sharp needle when cannulation with a blunt needle seems difficult. This makes cannulation much easier. However, use of sharp needles may damage the created track, resulting in the formation of one or more “false tracks,” which can predispose to larger scabs difficult to remove, more track injury, and finally, infection. In our unit, the rate of sharp needle use decreased dramatically from 48% in 2007 to 1.5% at the end of 2008 (data not shown). Nesrallah et al13 recently observed an increase in bacteriemia caused by S aureus (with metastatic complications in 4 cases) after implementation of the buttonhole method in 56 patients receiving home nocturnal HD. All cases of bacteremia coincided with local AVF infection, in contrast to our cases without obvious AVF infection despite bacteremia. Interestingly, Nesrallah et al13 reported a significant decrease after using topical mupirocin 2% cream on each buttonhole site with a sterile cotton swab, without dressings, after hemostasis achievement. No infection occurred during treatment with mupirocin. Although numbers of patients and events were relatively small, as the investigators note, these results are promising and should trigger randomized controlled trials. In in-center HD units, the risk of resistance to mupirocin might represent a substantial barrier to the long-term success of this strategy.20 In conclusion, units considering a shift to buttonhole cannulation should be warned that the method requires strict respect of every step of the procedure to 447
Labriola et al
avoid severe infectious complications. Rigorous education and constant training of the nurses are mandatory to reach this goal.
ACKNOWLEDGEMENTS Support: None. Financial Disclosure: The authors declare that they have no relevant financial interests.
REFERENCES 1. Twardowski Z. The “buttonhole” method of needle insertion takes center stage in the attempt to revive daily home hemodialysis. Contemp Dial Nephrol. 1977;18(2):18-19. 2. Twardowski Z, Kubara H. Different sites versus constant sites of needle insertion into arteriovenous fistulas for treatment of repeated dialysis. Dial Transplant. 1979;8:978-980. 3. Krönung G. Plastic deformation of Cimino fistula by repeated puncture. Dial Transplant. 1984;13:635-638. 4. Verhallen AM, Kooistra MP, van Jaarsveld BC. Cannulating in haemodialysis: rope-ladder or buttonhole technique? Nephrol Dial Transplant. 2007;22(9):2601-2604. 5. Van Loon MM, Goovaerts T, Kessels AG, van der Sande FK, Tordoir JH. Buttonhole needling of haemodialysis arteriovenous fistulae results in less complications and interventions compared to the rope-ladder technique. Nephrol Dial Transplant. 2010;25(1): 225-230. 6. Rayner HC, Pisoni RL. The increasing use of hemodialysis catheters: evidence from the DOPPS on its significance and ways to reverse it. Semin Dial. 2010;23(1):6-10. 7. Hoen B, Paul-Dauphin A, Hestin D, Kessler M. EPIBACDIAL: a multicenter prospective study of risk factors for bacteremia in chronic hemodialysis patients. J Am Soc Nephrol. 1998;5:869-876. 8. Doss S, Schiller B, Moran J. Buttonhole cannulation—an unexpected outcome. Nephrol Nurs J. 2008;35:417-419.
448
9. Ludlow V. Buttonhole cannulation in hemodialysis: improved outcomes and increased expense—is it worth it? CANNT J. 2010;20(1):29-37. 10. Labriola L, Jadoul M. New recommendations in the treatment of gram positive bacteraemia in dialysis patients. Nephrol Dial Transplant. 2008;23(7):2431-2432. 11. Lancaster HO. Significance tests in discrete distributions. JASA. 1961;56:223-234 12. Murcutt G. Buttonhole cannulation: should this become the default technique for dialysis patients with native fistulas? J Renal Care. 2008;34(2):101-108. 13. Nesrallah GE, Cuerden M, Wong JH, Pierratos A. Staphylococcus aureus bacteraemia and buttonhole cannulation: long-term safety and efficacy of mupirocin prophylaxis. Clin J Am Soc Nephrol. 2010;5(6):1047-1053. 14. Elseviers MM, Van Waeleghem JP; European Dialysis and Transplant Nurses Association/European Renal Care Association. Complications of vascular access: results of a European multi centre study of the EDTNA/ERCA Research Board. EDTNA ERCA J. 2003;29(3):163-167. 15. Labriola L, Crott R, Jadoul M. Catheter-related bacteraemia is not a fatality: 8 year-follow-up in a Belgian hemodialysis unit. Poster presented at the World Congress of Nephrology 2009; May 22-26, 2009; Milan, Italy. 16. Allon M, Lockhart ME, Lilly RZ, et al. Effect of preoperative sonographic mapping on vascular access outcomes in hemodialysis patients. Kidney Int. 2001;60(5):2013-2020. 17. Peterson WJ, Barker J, Allon M. Disparities in fistula maturation persist despite preoperative vascular mapping. Clin J Am Soc Nephrol. 2008;3(2):437-441. 18. Caplin N, Sedlacek M, Teodorescu V, Falk A, Uribarri J. Venous access: women are equal. Am J Kidney Dis. 2003;41:429432. 19. Hadley K, Moran J. Buttonhole access: noninvasive scab removal. [Annual Dialysis Conference; March 7-9, 2010; Seattle, Washington; abstract]. Hemodial Int. 2010;14(1):98. 20. Patel JB, Gorwitz RG, Jernigan JA. Mupirocin resistance. Clin Infect Dis. 2009;49(6):935-941.
Am J Kidney Dis. 2011;57(3):442-448