- Email: [email protected]
Surgical Management of Infected PTFE Hemodialysis Grafts: Analysis of a 15-Year Experience
Surgical Management of Infected PTFE Hemodialysis Grafts: Analysis of a 15.Year Experience Marwan R. Tabbara, MD, Patrick J. O'Hara, MD, Norman R. Hertzer, MD, Leonard P. Krajewski, MD, and Edwin G. Beven, MD, Cleveland, Ohio
The records of 52 consecutive patients who underwent surgical treatment for 57 episodes of hemodialysis graft infection (HGI) from 1977 to 1993 were reviewed to determine the mortality and morbidity associated with this complication and to clarify guidelines for its management. The study group consisted of 35 women and 17 men whose mean age was 57 years at initial graft placement. Thirty-three (58%) HGIs involved straight grafts in the upper arm, 12 (21%) straight forearm grafts, 11 (19%) loop forearm grafts, and 1 (2%) a loop groin fistula. All of these grafts were constructed with polytetrafluoroethylene (PTFE). All 57 cases of HGI showed at least local evidence and 41 (72%) caused systemic symptoms. Thirty-seven (65%) HGIs were associated with positive blood cultures. The predominant infecting organism was Staphylococcus, which was isolated alone or in combination with other organisms from 40 (70%) graft or wound sites. Seventy-eight percent (31/40) of the staphylococcal infections involved Staphylococcus aureus. The median time from graft implantation to diagnosis of HGI was 7 months (mean 16 months, range 0 to 77 months) and from diagnosis to surgical treatment, 4 days (mean 6 days, range 0 to 26 days). Initial surgical management consisted of complete excision of all prosthetic material in 43 (75%) cases and partial excision in 14. The 30-day mortality rate following the last operation for the treatment of HGI was 12% (6/52) and was not significantly increased by incomplete excision. Six (86%) of the early deaths were related to sepsis and each of these patients had positive blood cultures. None of the infected grafts could be salvaged without removal of at least part of the original graft. None of the 43 complete excisions was complicated by recurrent infection at the same time, whereas this complication did occur following six (43%) of the 14 procedures during which residual prosthetic material was left in place (p = 0.00008, Fisher's exact test). Prosthetic HGI is a serious complication that is optimally treated by excision of all infected PTFE. Complete removal of synthetic material offers a significantly reduced risk of recurrent graft sepsis at the same site. (Ann Vasc Surg 1995;9:378-384.)
Hemodialysis is still considered the optimal treatment for mhny patients with end-stage renal disease) Unfortunately, maintenance of adequate angioaccess is often problematic, especially for patients requiring chronic hemodialysis. The un-
From the Department of Vascular Surgery, The Cleveland Clinic Foundation, Cleveland, Ohio. Reprint requests: Patrick J. O'Hara, MD, Department of Vascular Surgery, The Cleveland Clinic Foundation, 9500 Euclid Ave., Cleveland, OH 44195.
378
availability of suitable autogenous vein for construction of an arteriovenous fistula in some patients, or physician and technician preferences in other instances, has led to more frequent use of synthetic grafts for hemodialysis access. Polytetrafluoroethylene (PTFE) grafts are now most commonly used, accounting for up to 80% of the total hemodialysis angioaccess procedures in selected reports, z3 However, compared to autogenous access vein grafts, PTFF. grafts are less durable with higher rates of thrombosis and infection. 4 The reported incidence of infection in-
Vol. 9, No. 4 1995
volving PTFE angioaccess is 11% to 35%, substantially higher than the 2% to 3% reported for autogenous angioaccess. 4 Although thrombosis is usually amenable to thrombectomy and revision or placement of a new angioaccess site at another location, the optimal m a n a g e m e n t of infected synthetic angioaccess grafts remains controversial. Reported treatment recommendations have ranged from simple incision and drainage with or without flap coverage2"5-9 to partial or complete graft excision. 4'5'1~ To determine the mortality and morbidity associated with this complication and to clarify our own guidelines for its optimal management, we have reviewed our experience with infected PTFE angioaccess grafts at The Cleveland Clinic.
Infected PTFE hemodialysis grafts
379
center, whereas the remaining 16 (28%) had been implanted at other facilities. The anatomic locations of the HGIs are summarized in Table I. All patients except one with a loop groin fistula had procedures performed in an upper extremity, which is the preferred location at our center. Statistical comparisons w h e n appropriate were carried out using Fisher's exact test (two-tailed). RESULTS
The presenting symptoms and signs in this series are summarized in Table II. According to the inclusion criteria, all HGIs produced at least local symptoms and 41 (72%) caused systemic manifestations.
PATIENTS AND METHODS
The hospital records of 52 patients who underwent surgical intervention for infected PTFE angioaccess grafts from 1977 to 1993 were reviewed retrospectively. Patients with infection involving autogenous angioaccess constructions or synthetic grafts other than PTFE were excluded from this analysis. Five patients developed a second graft infection in either a contralateral angioaccess site (n = 4) or a new PTFE graft placed in the same extremity 19 months after complete removal and subsequent resolution of the initial graft infection (n = 1). These remote graft infections were considered to represent separate events, yielding a total of 57 episodes of hemodialysis graft infection (HGI) among the 52 patients. Nevertheless, the reappearance of infection at the site of previous graft involvement was judged to represent a continuation of the original septic process and was classified as recurrent graft sepsis. The diagnosis of HGI was based on the clinical appearance of the exposed synthetic material, the presence of purulence or a draining sinus contiguous with a PTFE graft, or isolation of microorganisms from the excised graft material in the presence of pseudoaneurysm formation, anastomotic disruption, or adjacent soft tissue sepsis. The study group consisted of 17 men and 35 w o m e n whose mean age was 57 years (median 59 years, range 23 to 80 years) at the time of HGI treatment. Thirty-six (69%) were hypertensive, 28 (54%) were diabetic, 17 (33%) were cigarette smokers, 20 (38%) had clinical evidence of coronary artery disease, and 13 (25%) were obese. Forty-one (72%) of the infected angioaccess grafts had originally been constructed at our
Table I. Anatomic location of PTFE angioaccess grafts in 57 episodes of HGI
Graft locationand type
No.
%
25 7 1
44 12 2
11 12
19 21
57
2 100
Proximal arm Brachioaxillary Brachiocephalic Brachiointernal jugular Forearm Loop Straight Groin Loop
Table II. Presenting symptoms in 57 episodes of PTFE angioaccess graft infection
Presenting symptoms
No.
%
38 20 28 3 16
67 35 50 5 28
28 27 18 14 14 6 6
49 47 32 25 25 11 11
Abscess
4
7
Sinus tract
3
5
Systemic Fever Chills Malaise Generalized sepsis None Local Induration Cellulitis Drainage Hemorrhage Tenderness/pain Pseudoaneurysm Exposed graft
Annals of Vascular Surgery
380 Tabbara et al.
Table III. Culture results for all 57 episodes of HGI (52 patients) Graft/wound culture
Blood culture
No.
%
No.
%
Not available No growth
4 2
7 4
7 13
12 23
Candida
2
4
1
2
6 3 40 (5)
11 5 70 (9)
4 3 29 (7)
7 5 51 (12)
(31) (4) 57
(54) (7) 100
(22) (0) 57
(39) (0) 100
Gram-negative rods Streptococcal species Staphylococcal species Methicillin-re sistant S. aureus S. aureus S. epidermidis
Total
Thirty-seven (65%) HGIs were associated with bacteremia and the predominant organism was Staphylococcus isolated from 78% (29/37) of positive blood cultures and 51% (29/57) of HGIs (Table III). Four HGIs involved multiple organisms, but again the predominant pathogen isolated from excised graft material or from the infected tissue contiguous with the graft was Staphylococcus in 40 (70%) HGIs, either alone or in combination with other microorganisms. Seventy-eight percent (31/40) of the staphylococcal infections involved Staphylococcus aureus (Table III). Considering the 57 episodes of HGI in which the interval could be determined, the median time from graft implantation to the clinical diagnosis of HGI was 7 months (mean 16 months, range 0 to 77 months) and from clinical diagnosis to surgical treatment, 4 days (mean 6 days, range 0 to 26 days). Two patients (two episodes of HGI) underwent partial graft excision at another hospital prior to subsequent complete excision at our center. The appropriate intervals could not be determined for these two patients. Complete excision was defined as removal of all of the original synthetic graft material, whereas partial excision permitted a portion of the original graft to remain in place. For patients treated with partial excision, the exact amount of original graft material left in situ could not be precisely quantified. Consequently, if any original graft material was left in place, irrespective of the amount, the excision was classified as partial. Initial surgical management of the 57 cases of HGI consisted of partial excision of the infected synthetic material and placement of a PTFE interposition graft in two (4%),
partial excision of the PTFE alone in 11 (19%), and complete excision of all synthetic material in the remaining 43 (75%). One patient (2%) who had an exposed and presumably infected graft was treated with brachioradialis muscle flap coverage. The arterial and venous anastomotic sites were managed with ligation, primary repair, or patch closure and in one instance with construction of an arteriovenous anastomosis. The wounds were treated with appropriate drainage and localized care. Specific antibiotic therapy was employed in 93% (53/57) of the HGIs. Four patients, all of w h o m were eventually treated with complete excision of the graft material, received no supplemental antibiotics. Three patients were initially treated with total excision. A fourth patient, w h o had recurrent infection after a partial excision at another institution, underwent total excision at our hospital but was subsequently lost to followup. There were no documented recurrent infections in this small subgroup, presumably because the septic focus was completely removed. None of the 43 complete excisions was associated with early recurrence of infection at the same site (Fig. 1), whereas this complication did occur following six (43%) of the 14 procedures during which residual prosthetic material was left in place (p = 0.00008). None of the infected grafts could be salvaged without removal of at least part of the original graft. The single patient whose exposed graft was treated with brachioradialis flap coverage alone had a recurrence of sinus drainage 6 weeks later but refused another operation. He subsequently underwent reoperation at another hospital l0 weeks after the recurrence. In one of the two cases of HGI treated with partial excision and interposition grafting, Pseudomonas infection recurred 6 months later and resulted in death from sepsis. The other patient, infected with S. aureus, has remained free of infection for 18 months. There have been f o u r , recurrent infections among the 11 patients w h o underwent partial graft excision without direct reconstruction. Two of these were caused by S. aureus and the third by Pseudornonas. The fourth patient, w h o was lost to late follow-up, did not have culture data available for review. The patient with Pseudomonas died of persistent sepsis 3 weeks later despite complete excision of all residual graft material. One of the two remaining patients with staphylococcal sepsis required complete graft excision 2 months later for refractory sepsis, and the other was treated with local w o u n d dressings and died 2 years later of a stroke but had no documented recurrent sepsis.
Vol. 9, No. 4 1995
Infected PTFE hemodialysis grafts 381
43 45 40 35 30 INFECTION EPISODES
25 20 15 10 5 0
COMPLETE (0)* EXCISION
METHOD
PARTIAL (43%)* (% RECURRENCE)
Fig. 1. Incidence of recurrent graft infection for all 57 episodes of graft sepsis (52 patients). There were significantly fewer recurrent graft infections among patients who underwent complete excision of infected synthetic material. (*p = 0.00008, Fisher's exact test.)
T a b l e IV. Follow-up s u m m a r y for all 52 patients (57 HGIs) Excision method Complete (38 patients/43 HGIs) Deaths < = 30 days > 30 days Late survivors Lost to late follow-up Total
Partial (14 patients/14 HGIs)
Entire series (52 patients/57 HGIs)
No.
Rec.
No.
Rec.
No.
Rec.
25 4 21 12 1 38
0 0 0 0 0 0
8 2 6 5 1 14
5 2 3 0 1 6
33 6 27 17 2 52
5 2 3 0 1 6
Rec. = recurrent infection.
N o n e of the 38 patients (43 episodes of HGI) w h o were initially treated w i t h complete excision of the infected graft material developed r e c u r r e n t sepsis at the graft site, a l t h o u g h four of t h e m died w i t h i n 30 days of graft removal, three f r o m systemic sepsis (one of w h o m also h a d a m y o c a r d i a l infarction) and one f r o m myocardial i n f a r c t i o n alone. T w e n t y - o n e others w h o u n d e r w e n t complete excision are k n o w n to have died during the late follow-up period; causes of d e a t h included u r e m i a in 11, stroke in two, a n d cardiac disease in four. Four others died of sepsis. In three of t h e m the cause of d e a t h was clearly p n e u m o n i a , colon perforation, or u r i n a r y sepsis, respectively, w h e r e a s the f o u r t h d e a t h was related to either u n d e t e r m i n e d sepsis or p u l m o n a r y embolism. The last of these patients, w h o initially p r e s e n t e d w i t h Staphylococcus bacteremia, died of Candida sepsis.
Twelve surviving patients, followed Ior a m e a n of 31 m o n t h s ( m e d i a n 11 m o n t h s , range 0.4 to 90 m o n t h s ) after graft excision, have not h a d recurrent sepsis. The r e m a i n i n g patient has b e e n lost to follow-up. Late follow-up i n f o r m a t i o n for all 52 patients is s u m m a r i z e d in Table IV. The 30-day m o r t a l i t y rate following the last operation for t r e a t m e n t of HGI was 12% (6/52). The early m o r t a l i t y rate for complete excision (11%, 4/38) was not significantly different f r o m that for partial excision (14%, 2/14). The c u m u lative 5-year survival rate for the entire series was 29% (Fig. 2). Five of the early deaths were related to sepsis a n d e a c h of these patients h a d positive blood cultures. The remaining early d e a t h was caused by a ~m y o c a r d i a l infarction. The causes of early a n d late d e a t h a m o n g all 52 patients are s u m m a r i z e d in Table V.
Annals of Vascular Surgery
382 Tabbara et al.
100 90
~
80 70
ACTUARIAL 6o SURVIVAL so ( ~176 )
~
40
" ~ 1 3 ( 1 3 )
30
~
- ~
-~;,~(6)
20
~-'o
v,W
(:)~;c (2)
10 0 0
I
I
I
I
I
I
I
I
I
~0
30 days
1
2
3
4
5
6
7
8
9
POSTOPERATIVE YEAR Fig. 2. Actuarial survival of 52 patients who were treated for 57 episodes of PTFE angioaccess graft infection. Ctanulative 5-year survival rate was 29% for the entire series of patients. The number of patients still eligible for consideration at each interval is in parentheses.
Table V. Causes of all 33 deaths among the 52 patients (57 HGIs) in the entire series Cause of death
Complete excision (n = 38)
Sepsis Myocardial infarction
Early deaths 50% (3/6) 17% (1/6)
Early deaths 33% (2/6) 0
Early deaths 83% (5/6) 17% (1/6)
Uremia Sepsis Cardiac Stroke
Late deaths 41% (11/27) 15% (4/27) 15% (4/27) 7% (2/27)
Late deaths 4% (1/27) 11% (2/27) 0 7% (2/27)
Late deaths 44% (12/27) 26% (7/27) 15% (4/27) 15% (4/27)
DISCUSSION Synthetic HGI is a serious problem for several reasons. The infection itself m a y be fatal if septicemia occurs and at best it usually results in termination of the use of a scarce angioaccess site. Furthermore, this infection can be costly to treat. In a recent report HGI accounted for 27% of all patients admitted for construction or revision of a hemodialysis vascular access site who required a hospital stay of more than 14 days. ~6This consideration is becoming increasingly important as the population ages, since it is estimated that more than 300,000 Medicare patients will require treatment for end-stage renal disease (ESRD) by the year 2000, an increase of nearly 50%) y AI-
excision (n = 14)
Partial
Entire series (n = 52)
though it is clear that HGI is best avoided, rapid eradication and prevention of recurrences are imperative once it develops. According to our departmental registry, 700 angioaccess grafts were placed in 335 (48%) men and 365 (52%) w o m e n at The Cleveland Clinic from January 1989 through May 1993. Forty-four percent (146/335) of the m e n and 63% (231/365) of the w o m e n required synthetic grafts because of inadequate or unavailable autogenous vein. Women m a y require synthetic grafts for angioaccess more frequently than m e n because they possess smaller veins. This observation may explain the larger proportion of women (66%) with HGI compared to men (34%) in the current series.
Vol. 9, No. 4 1995
There is uncertainty regarding the optimal management of synthetic graft infections even with respect to traditional vascular reconstructions for aneurysmal or occlusive disease. ~3'~8-2~ Some of this controversy may reflect the lack of a suitable severity index that could be universally applied to all such infections. Traditionally, complete resection of the involved graft material with either direct autogenous or extra-anatomic synthetic bypass has been advocated to minimize the risk of recurrent local infection. ~3 Recently, innovative reports have suggested that it may be possible to entirely debride infected synthetic grafts in conjunction with in situ replacement under selected circumstances. ~8'22 This technique may be more likely to be successful in the treatment of localized graft infections caused by Staphylococcus epiderrnidis confined to a glycoprotein biofilm that is completely removed with the graft itself. 22 Consequently, it is not a reasonable consideration for most of the patients in this series. In the presence of bacteremia or a large inoculum of bacteria capable of producing proteolytic enzymes (e.g., Pseudornonas or S. aureus), however, contamination of the newly placed synthetic graft still seems likely. Furthermore, unlike the synthetic graft material used for conventional arterial or venous reconstructions, dialysis grafts are routinely placed through skin puncture, presumably increasing the opportunities for primary and recurrent graft inoculation with microorganisms. Because late follow-up information for all patients who have had synthetic angioaccess grafts placed during the span of this series is unavailable, the precise incidence of synthetic graft sepsis at our center remains unknown. Patients with ESRD are often debilitated and immunosuppressed and their projected life expectancy is shorter than that of the general population. The actuarial 2-year survival rate for all Medicare patients with ESRD who are 65 to 69 years of age is only 52%. The adjusted 1-year actuarial survival rate from 3 months after the onset of ESRD for Medicare patients requiring hemodialysis is 70% among those with diabetes and 79% in those without diabetes. Sepsis is the second most common cause of death, exceeded only by fatal complications of cardiac disease) In the current series the 1- and 2-year actuarial survival rates are only 54% and 43%, respectively, despite a mean age of only 57 years. This finding, which is undoubtedly influenced by the 30-day mortality rate (12%) associated with HGI in our experience, probably reflects the underlying debilitation of our patient population as well as the
Infected PTFE hemodialysis grafts
383
advanced stage of infection present at the time of its detection. All but one of the early deaths in our series were caused by uncontrolled sepsis. Nearly two thirds of the patients in this study presented with documented bacteremia, indicating serious systemic involvement at the outset. All of the deaths related to sepsis were among patients who had positive blood cultures. The most common organism isolated was Staphylococcus, either alone or in combination, a finding consistent with other reports 23-25 The advanced stage of the infection by the time it was detected clinically reflects the difficulty in early diagnosis of graft infection among patients who often experience septic events unrelated to HGI. Early signs and symptoms are often nonspecific and therefore HGI is frequently a diagnosis of exclusion unless there is obvious clinical evidence localizing the infection to the synthetic graft. Some patients may have multiple occluded synthetic grafts, all of which may be potential sources of sepsis. Moreover, some cases of HGI in this series were likely caused by earlier episodes of bacteremia from another source. Clinically it may be impossible to distinguish primary from secondary HGI in its late stages. As a practical matter, early death is commonly related to failure to control sepsis and treatment should be directed toward achieving this objective. In addition to specific antibiotic therapy and nutritional support, early and complete excision of the infected synthetic material with appropriate debridement and drainage is usually required. In this series none of the infected synthetic grafts could be salvaged without removal of at least part of the original graft, a finding consistent with other reports. 12-~4 In fact the risk of recurrent sepsis was significantly increased w h e n the PTFE graft was not completely excised. The native artery and vein may be managed by ligation, primary repair, or autogenous patching w h e n necessary. Immediate construction of an autogenous arteriovenous fistula with the native artery and vein was feasible in one of our patients and has been reported previously, x~ An aggressive search for other treatable sources of sepsis should also be carried out if prompt improvement does not occur.
It was not possible from this study to determine the minimum safe interval before reimplantation of another synthetic angioaccess graft at a previously infected site. To ensure that the original source ~of sepsis has been eliminated, the treatment interval should be lengthened as much as the clinical situation will permit. Alternative
384
Tabbara et al.
dialysis options such as the use of peritoneal lavage, repeated vascular cannulation, or indwelling catheters may be required while control of sepsis is achieved, but each of these methods has limitations. If another permanent vascular access site is needed, construction of an autogenous arteriovenous fistula is the most desirable alternative provided it is feasible. In this respect intraoperative venography is a useful adjunct to determine the availability of autogenous vein.
CONCLUSION Prosthetic HGI is a serious complication that is optimally treated by early diagnosis and excision of all infected PTFE. Complete removal of synthetic material offers a significantly reduced risk of recurrent graft sepsis at the same site. Prosthetic HGI is best avoided by the preferential use of autogenous primary procedures for hemodialysis access. The statistical analysis was performed by Mariane Piedmont, Department of Biostatistics and Epidemiology.
REFERENCE S 1. U.S. Renal Data System, 1991 Annual Data Report, The National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, Md., August 1991. V. Survival probabilities and causes of death. Am J Kidney Dis 1991;18:49-60. 2. Connolly JE, Brownell DA, Levine EE et al. Complications of renal dialysis access procedures. Arch Surg 1984; 119:13251328. 3. Hinsdale J, Lipkowitz G, Hoover E. Vascular access for hemodialysis in the elderly: Results a n d perspectives in a geriatric population. Dialysis Transplant 1985;14:560-565. 4. Zibari GB, Rohr MS, Landreneau MD, et al. Complications from permanent hemodialysis vascular access. Surgery 1988; 104:681-688.
5. Tellis VA, Weiss E Matas AJ, et al. Skin-flap coverage of polytetrafluqroethylene vascular access graft exposed by previous infection. Surgery 1988; 103:118-121. 6. Bhat D, Tellis V, Kohlberg W, et al. Management of sepsis involving expanded polytetrafluoroethylene grafts for hemodialysis access. Surgery 1980;87:445-450.
Annals of Vascular Surgery
7. Okadome K, watanabe T, I~na M, et al. Skin sliding closure technique is effective for management of infected prosthesis in cases of arterial reconstruction with synthetic grafts. Jpn J Surg 1982;12:349-355. 8. Elliot M, Bazzaniga A, Thomas J, et al. Use of expanded polytetrafluoroethylene grafts for vascular access in hemodialysis: Laboratory and clinical evaluation. Am Surg 1977;43: 455-459. 9. Raju S. PTFE grafts for hemodialysis access. Ann Surg 1987;206:666-673. 10, Gifford RR. Management of tunnel infections of dialysis polytetrafluoroethylene grafts. J Vasc Surg 1985;2:854-858. 11. Padberg FJ, Lee BC, Curl GR. Hemoaccess site infection. Surg Gynecol Obstet 1992;174:103-108. 12. Corry RJ, Patel NP, West JC. Surgical management of complications of vascular access for hemodialysis. Surg Gynecol Obstet 1980;151:49-54. 13. O'Hara P, Hertzer N, Beven E, et al. Surgical management of infected abdominal aortic grafts: Review of a 25-year experience. J Vasc Surg 1986;3:725-731. 14. Baker L, Johnson J, Goldfarb D. Expanded polytetrafluoroethylene (PTFE) subcutaneous arteriovenous conduit: An improved vascular access for chronic hemodialysis. Trans Am Soc Artif Intern Organs 1976;22:382-387. 15. McMullen K, Hayes D, Hussey JL, et al. Salvage of hemodialysis access in infected arteriovenous fistulas. Arch Surg 1991;126:1303-1305. 16. Mayers JD, Markell MS, Cohen LS, et al. Vascular access surgery for maintenance hemodialysis. Variables in hospital stay. ASAIO Trans 1992;38:113-115. 17. Iglehart J. Health Policy Report: The American Health Care System, The End Stage Renal Disease Program. N Engl J Med 1993;328:366-371. 18. Kieffer E, Bahnini A, Koskas F, et al. In situ allograft replacement of infected infrarenal aortic prosthetic grafts: Results in forty-three patients. J Vase Surg 1993;17:349-356. 19. Yeager RA, Porter JM. Basic data underlying clinical decision making in vascular surgery: Arterial and prosthetic graft infection. Ann Vase Surg 1992;6:485-491. 20. Francois F, Thdvenet A. Conservative treatment of prosthetic aortic graft infection with irrigation. Ann Vasc Surg 199l;5: 199-201. 21. Calligaro KD, Veith FJ. Clinical review: Diagnosis and management of infected prosthetic aortic grafts. Surgery 1991; 110:805-813. 22. Bandyk DF, Bergamini TM, Kinney EV, et al. In situ replacement of vascular prostheses infected by bacterial biofihns. J Vasc Snrg 1991;13:575-583. 23. Kirami N, Tuazon C, Murray H, et al. Staphylococcus aureus carriage rate of patients receiving long-term hemodialysis. Arch Intern Med 1978;138:1657-1659. 24. Bunt T. Synthetic vascular graft infections. I. Graft infections. Surgery 1983;93:733-746. 25. Fivush BA, Bock GH, Guzzetta PC, et al. Vancomycin prevents polytetrafluoroethylene graft infections in pediatric patients receiving chronic hemodialysis. Am J Kidney Dis 1985;5:120-123.
The records of 52 consecutive patients who underwent surgical treatment for 57 episodes of hemodialysis graft infection (HGI) from 1977 to 1993 were reviewed to determine the mortality and morbidity associated with this complication and to clarify guidelines for its management. The study group consisted of 35 women and 17 men whose mean age was 57 years at initial graft placement. Thirty-three (58%) HGIs involved straight grafts in the upper arm, 12 (21%) straight forearm grafts, 11 (19%) loop forearm grafts, and 1 (2%) a loop groin fistula. All of these grafts were constructed with polytetrafluoroethylene (PTFE). All 57 cases of HGI showed at least local evidence and 41 (72%) caused systemic symptoms. Thirty-seven (65%) HGIs were associated with positive blood cultures. The predominant infecting organism was Staphylococcus, which was isolated alone or in combination with other organisms from 40 (70%) graft or wound sites. Seventy-eight percent (31/40) of the staphylococcal infections involved Staphylococcus aureus. The median time from graft implantation to diagnosis of HGI was 7 months (mean 16 months, range 0 to 77 months) and from diagnosis to surgical treatment, 4 days (mean 6 days, range 0 to 26 days). Initial surgical management consisted of complete excision of all prosthetic material in 43 (75%) cases and partial excision in 14. The 30-day mortality rate following the last operation for the treatment of HGI was 12% (6/52) and was not significantly increased by incomplete excision. Six (86%) of the early deaths were related to sepsis and each of these patients had positive blood cultures. None of the infected grafts could be salvaged without removal of at least part of the original graft. None of the 43 complete excisions was complicated by recurrent infection at the same time, whereas this complication did occur following six (43%) of the 14 procedures during which residual prosthetic material was left in place (p = 0.00008, Fisher's exact test). Prosthetic HGI is a serious complication that is optimally treated by excision of all infected PTFE. Complete removal of synthetic material offers a significantly reduced risk of recurrent graft sepsis at the same site. (Ann Vasc Surg 1995;9:378-384.)
Hemodialysis is still considered the optimal treatment for mhny patients with end-stage renal disease) Unfortunately, maintenance of adequate angioaccess is often problematic, especially for patients requiring chronic hemodialysis. The un-
From the Department of Vascular Surgery, The Cleveland Clinic Foundation, Cleveland, Ohio. Reprint requests: Patrick J. O'Hara, MD, Department of Vascular Surgery, The Cleveland Clinic Foundation, 9500 Euclid Ave., Cleveland, OH 44195.
378
availability of suitable autogenous vein for construction of an arteriovenous fistula in some patients, or physician and technician preferences in other instances, has led to more frequent use of synthetic grafts for hemodialysis access. Polytetrafluoroethylene (PTFE) grafts are now most commonly used, accounting for up to 80% of the total hemodialysis angioaccess procedures in selected reports, z3 However, compared to autogenous access vein grafts, PTFF. grafts are less durable with higher rates of thrombosis and infection. 4 The reported incidence of infection in-
Vol. 9, No. 4 1995
volving PTFE angioaccess is 11% to 35%, substantially higher than the 2% to 3% reported for autogenous angioaccess. 4 Although thrombosis is usually amenable to thrombectomy and revision or placement of a new angioaccess site at another location, the optimal m a n a g e m e n t of infected synthetic angioaccess grafts remains controversial. Reported treatment recommendations have ranged from simple incision and drainage with or without flap coverage2"5-9 to partial or complete graft excision. 4'5'1~ To determine the mortality and morbidity associated with this complication and to clarify our own guidelines for its optimal management, we have reviewed our experience with infected PTFE angioaccess grafts at The Cleveland Clinic.
Infected PTFE hemodialysis grafts
379
center, whereas the remaining 16 (28%) had been implanted at other facilities. The anatomic locations of the HGIs are summarized in Table I. All patients except one with a loop groin fistula had procedures performed in an upper extremity, which is the preferred location at our center. Statistical comparisons w h e n appropriate were carried out using Fisher's exact test (two-tailed). RESULTS
The presenting symptoms and signs in this series are summarized in Table II. According to the inclusion criteria, all HGIs produced at least local symptoms and 41 (72%) caused systemic manifestations.
PATIENTS AND METHODS
The hospital records of 52 patients who underwent surgical intervention for infected PTFE angioaccess grafts from 1977 to 1993 were reviewed retrospectively. Patients with infection involving autogenous angioaccess constructions or synthetic grafts other than PTFE were excluded from this analysis. Five patients developed a second graft infection in either a contralateral angioaccess site (n = 4) or a new PTFE graft placed in the same extremity 19 months after complete removal and subsequent resolution of the initial graft infection (n = 1). These remote graft infections were considered to represent separate events, yielding a total of 57 episodes of hemodialysis graft infection (HGI) among the 52 patients. Nevertheless, the reappearance of infection at the site of previous graft involvement was judged to represent a continuation of the original septic process and was classified as recurrent graft sepsis. The diagnosis of HGI was based on the clinical appearance of the exposed synthetic material, the presence of purulence or a draining sinus contiguous with a PTFE graft, or isolation of microorganisms from the excised graft material in the presence of pseudoaneurysm formation, anastomotic disruption, or adjacent soft tissue sepsis. The study group consisted of 17 men and 35 w o m e n whose mean age was 57 years (median 59 years, range 23 to 80 years) at the time of HGI treatment. Thirty-six (69%) were hypertensive, 28 (54%) were diabetic, 17 (33%) were cigarette smokers, 20 (38%) had clinical evidence of coronary artery disease, and 13 (25%) were obese. Forty-one (72%) of the infected angioaccess grafts had originally been constructed at our
Table I. Anatomic location of PTFE angioaccess grafts in 57 episodes of HGI
Graft locationand type
No.
%
25 7 1
44 12 2
11 12
19 21
57
2 100
Proximal arm Brachioaxillary Brachiocephalic Brachiointernal jugular Forearm Loop Straight Groin Loop
Table II. Presenting symptoms in 57 episodes of PTFE angioaccess graft infection
Presenting symptoms
No.
%
38 20 28 3 16
67 35 50 5 28
28 27 18 14 14 6 6
49 47 32 25 25 11 11
Abscess
4
7
Sinus tract
3
5
Systemic Fever Chills Malaise Generalized sepsis None Local Induration Cellulitis Drainage Hemorrhage Tenderness/pain Pseudoaneurysm Exposed graft
Annals of Vascular Surgery
380 Tabbara et al.
Table III. Culture results for all 57 episodes of HGI (52 patients) Graft/wound culture
Blood culture
No.
%
No.
%
Not available No growth
4 2
7 4
7 13
12 23
Candida
2
4
1
2
6 3 40 (5)
11 5 70 (9)
4 3 29 (7)
7 5 51 (12)
(31) (4) 57
(54) (7) 100
(22) (0) 57
(39) (0) 100
Gram-negative rods Streptococcal species Staphylococcal species Methicillin-re sistant S. aureus S. aureus S. epidermidis
Total
Thirty-seven (65%) HGIs were associated with bacteremia and the predominant organism was Staphylococcus isolated from 78% (29/37) of positive blood cultures and 51% (29/57) of HGIs (Table III). Four HGIs involved multiple organisms, but again the predominant pathogen isolated from excised graft material or from the infected tissue contiguous with the graft was Staphylococcus in 40 (70%) HGIs, either alone or in combination with other microorganisms. Seventy-eight percent (31/40) of the staphylococcal infections involved Staphylococcus aureus (Table III). Considering the 57 episodes of HGI in which the interval could be determined, the median time from graft implantation to the clinical diagnosis of HGI was 7 months (mean 16 months, range 0 to 77 months) and from clinical diagnosis to surgical treatment, 4 days (mean 6 days, range 0 to 26 days). Two patients (two episodes of HGI) underwent partial graft excision at another hospital prior to subsequent complete excision at our center. The appropriate intervals could not be determined for these two patients. Complete excision was defined as removal of all of the original synthetic graft material, whereas partial excision permitted a portion of the original graft to remain in place. For patients treated with partial excision, the exact amount of original graft material left in situ could not be precisely quantified. Consequently, if any original graft material was left in place, irrespective of the amount, the excision was classified as partial. Initial surgical management of the 57 cases of HGI consisted of partial excision of the infected synthetic material and placement of a PTFE interposition graft in two (4%),
partial excision of the PTFE alone in 11 (19%), and complete excision of all synthetic material in the remaining 43 (75%). One patient (2%) who had an exposed and presumably infected graft was treated with brachioradialis muscle flap coverage. The arterial and venous anastomotic sites were managed with ligation, primary repair, or patch closure and in one instance with construction of an arteriovenous anastomosis. The wounds were treated with appropriate drainage and localized care. Specific antibiotic therapy was employed in 93% (53/57) of the HGIs. Four patients, all of w h o m were eventually treated with complete excision of the graft material, received no supplemental antibiotics. Three patients were initially treated with total excision. A fourth patient, w h o had recurrent infection after a partial excision at another institution, underwent total excision at our hospital but was subsequently lost to followup. There were no documented recurrent infections in this small subgroup, presumably because the septic focus was completely removed. None of the 43 complete excisions was associated with early recurrence of infection at the same site (Fig. 1), whereas this complication did occur following six (43%) of the 14 procedures during which residual prosthetic material was left in place (p = 0.00008). None of the infected grafts could be salvaged without removal of at least part of the original graft. The single patient whose exposed graft was treated with brachioradialis flap coverage alone had a recurrence of sinus drainage 6 weeks later but refused another operation. He subsequently underwent reoperation at another hospital l0 weeks after the recurrence. In one of the two cases of HGI treated with partial excision and interposition grafting, Pseudomonas infection recurred 6 months later and resulted in death from sepsis. The other patient, infected with S. aureus, has remained free of infection for 18 months. There have been f o u r , recurrent infections among the 11 patients w h o underwent partial graft excision without direct reconstruction. Two of these were caused by S. aureus and the third by Pseudornonas. The fourth patient, w h o was lost to late follow-up, did not have culture data available for review. The patient with Pseudomonas died of persistent sepsis 3 weeks later despite complete excision of all residual graft material. One of the two remaining patients with staphylococcal sepsis required complete graft excision 2 months later for refractory sepsis, and the other was treated with local w o u n d dressings and died 2 years later of a stroke but had no documented recurrent sepsis.
Vol. 9, No. 4 1995
Infected PTFE hemodialysis grafts 381
43 45 40 35 30 INFECTION EPISODES
25 20 15 10 5 0
COMPLETE (0)* EXCISION
METHOD
PARTIAL (43%)* (% RECURRENCE)
Fig. 1. Incidence of recurrent graft infection for all 57 episodes of graft sepsis (52 patients). There were significantly fewer recurrent graft infections among patients who underwent complete excision of infected synthetic material. (*p = 0.00008, Fisher's exact test.)
T a b l e IV. Follow-up s u m m a r y for all 52 patients (57 HGIs) Excision method Complete (38 patients/43 HGIs) Deaths < = 30 days > 30 days Late survivors Lost to late follow-up Total
Partial (14 patients/14 HGIs)
Entire series (52 patients/57 HGIs)
No.
Rec.
No.
Rec.
No.
Rec.
25 4 21 12 1 38
0 0 0 0 0 0
8 2 6 5 1 14
5 2 3 0 1 6
33 6 27 17 2 52
5 2 3 0 1 6
Rec. = recurrent infection.
N o n e of the 38 patients (43 episodes of HGI) w h o were initially treated w i t h complete excision of the infected graft material developed r e c u r r e n t sepsis at the graft site, a l t h o u g h four of t h e m died w i t h i n 30 days of graft removal, three f r o m systemic sepsis (one of w h o m also h a d a m y o c a r d i a l infarction) and one f r o m myocardial i n f a r c t i o n alone. T w e n t y - o n e others w h o u n d e r w e n t complete excision are k n o w n to have died during the late follow-up period; causes of d e a t h included u r e m i a in 11, stroke in two, a n d cardiac disease in four. Four others died of sepsis. In three of t h e m the cause of d e a t h was clearly p n e u m o n i a , colon perforation, or u r i n a r y sepsis, respectively, w h e r e a s the f o u r t h d e a t h was related to either u n d e t e r m i n e d sepsis or p u l m o n a r y embolism. The last of these patients, w h o initially p r e s e n t e d w i t h Staphylococcus bacteremia, died of Candida sepsis.
Twelve surviving patients, followed Ior a m e a n of 31 m o n t h s ( m e d i a n 11 m o n t h s , range 0.4 to 90 m o n t h s ) after graft excision, have not h a d recurrent sepsis. The r e m a i n i n g patient has b e e n lost to follow-up. Late follow-up i n f o r m a t i o n for all 52 patients is s u m m a r i z e d in Table IV. The 30-day m o r t a l i t y rate following the last operation for t r e a t m e n t of HGI was 12% (6/52). The early m o r t a l i t y rate for complete excision (11%, 4/38) was not significantly different f r o m that for partial excision (14%, 2/14). The c u m u lative 5-year survival rate for the entire series was 29% (Fig. 2). Five of the early deaths were related to sepsis a n d e a c h of these patients h a d positive blood cultures. The remaining early d e a t h was caused by a ~m y o c a r d i a l infarction. The causes of early a n d late d e a t h a m o n g all 52 patients are s u m m a r i z e d in Table V.
Annals of Vascular Surgery
382 Tabbara et al.
100 90
~
80 70
ACTUARIAL 6o SURVIVAL so ( ~176 )
~
40
" ~ 1 3 ( 1 3 )
30
~
- ~
-~;,~(6)
20
~-'o
v,W
(:)~;c (2)
10 0 0
I
I
I
I
I
I
I
I
I
~0
30 days
1
2
3
4
5
6
7
8
9
POSTOPERATIVE YEAR Fig. 2. Actuarial survival of 52 patients who were treated for 57 episodes of PTFE angioaccess graft infection. Ctanulative 5-year survival rate was 29% for the entire series of patients. The number of patients still eligible for consideration at each interval is in parentheses.
Table V. Causes of all 33 deaths among the 52 patients (57 HGIs) in the entire series Cause of death
Complete excision (n = 38)
Sepsis Myocardial infarction
Early deaths 50% (3/6) 17% (1/6)
Early deaths 33% (2/6) 0
Early deaths 83% (5/6) 17% (1/6)
Uremia Sepsis Cardiac Stroke
Late deaths 41% (11/27) 15% (4/27) 15% (4/27) 7% (2/27)
Late deaths 4% (1/27) 11% (2/27) 0 7% (2/27)
Late deaths 44% (12/27) 26% (7/27) 15% (4/27) 15% (4/27)
DISCUSSION Synthetic HGI is a serious problem for several reasons. The infection itself m a y be fatal if septicemia occurs and at best it usually results in termination of the use of a scarce angioaccess site. Furthermore, this infection can be costly to treat. In a recent report HGI accounted for 27% of all patients admitted for construction or revision of a hemodialysis vascular access site who required a hospital stay of more than 14 days. ~6This consideration is becoming increasingly important as the population ages, since it is estimated that more than 300,000 Medicare patients will require treatment for end-stage renal disease (ESRD) by the year 2000, an increase of nearly 50%) y AI-
excision (n = 14)
Partial
Entire series (n = 52)
though it is clear that HGI is best avoided, rapid eradication and prevention of recurrences are imperative once it develops. According to our departmental registry, 700 angioaccess grafts were placed in 335 (48%) men and 365 (52%) w o m e n at The Cleveland Clinic from January 1989 through May 1993. Forty-four percent (146/335) of the m e n and 63% (231/365) of the w o m e n required synthetic grafts because of inadequate or unavailable autogenous vein. Women m a y require synthetic grafts for angioaccess more frequently than m e n because they possess smaller veins. This observation may explain the larger proportion of women (66%) with HGI compared to men (34%) in the current series.
Vol. 9, No. 4 1995
There is uncertainty regarding the optimal management of synthetic graft infections even with respect to traditional vascular reconstructions for aneurysmal or occlusive disease. ~3'~8-2~ Some of this controversy may reflect the lack of a suitable severity index that could be universally applied to all such infections. Traditionally, complete resection of the involved graft material with either direct autogenous or extra-anatomic synthetic bypass has been advocated to minimize the risk of recurrent local infection. ~3 Recently, innovative reports have suggested that it may be possible to entirely debride infected synthetic grafts in conjunction with in situ replacement under selected circumstances. ~8'22 This technique may be more likely to be successful in the treatment of localized graft infections caused by Staphylococcus epiderrnidis confined to a glycoprotein biofilm that is completely removed with the graft itself. 22 Consequently, it is not a reasonable consideration for most of the patients in this series. In the presence of bacteremia or a large inoculum of bacteria capable of producing proteolytic enzymes (e.g., Pseudornonas or S. aureus), however, contamination of the newly placed synthetic graft still seems likely. Furthermore, unlike the synthetic graft material used for conventional arterial or venous reconstructions, dialysis grafts are routinely placed through skin puncture, presumably increasing the opportunities for primary and recurrent graft inoculation with microorganisms. Because late follow-up information for all patients who have had synthetic angioaccess grafts placed during the span of this series is unavailable, the precise incidence of synthetic graft sepsis at our center remains unknown. Patients with ESRD are often debilitated and immunosuppressed and their projected life expectancy is shorter than that of the general population. The actuarial 2-year survival rate for all Medicare patients with ESRD who are 65 to 69 years of age is only 52%. The adjusted 1-year actuarial survival rate from 3 months after the onset of ESRD for Medicare patients requiring hemodialysis is 70% among those with diabetes and 79% in those without diabetes. Sepsis is the second most common cause of death, exceeded only by fatal complications of cardiac disease) In the current series the 1- and 2-year actuarial survival rates are only 54% and 43%, respectively, despite a mean age of only 57 years. This finding, which is undoubtedly influenced by the 30-day mortality rate (12%) associated with HGI in our experience, probably reflects the underlying debilitation of our patient population as well as the
Infected PTFE hemodialysis grafts
383
advanced stage of infection present at the time of its detection. All but one of the early deaths in our series were caused by uncontrolled sepsis. Nearly two thirds of the patients in this study presented with documented bacteremia, indicating serious systemic involvement at the outset. All of the deaths related to sepsis were among patients who had positive blood cultures. The most common organism isolated was Staphylococcus, either alone or in combination, a finding consistent with other reports 23-25 The advanced stage of the infection by the time it was detected clinically reflects the difficulty in early diagnosis of graft infection among patients who often experience septic events unrelated to HGI. Early signs and symptoms are often nonspecific and therefore HGI is frequently a diagnosis of exclusion unless there is obvious clinical evidence localizing the infection to the synthetic graft. Some patients may have multiple occluded synthetic grafts, all of which may be potential sources of sepsis. Moreover, some cases of HGI in this series were likely caused by earlier episodes of bacteremia from another source. Clinically it may be impossible to distinguish primary from secondary HGI in its late stages. As a practical matter, early death is commonly related to failure to control sepsis and treatment should be directed toward achieving this objective. In addition to specific antibiotic therapy and nutritional support, early and complete excision of the infected synthetic material with appropriate debridement and drainage is usually required. In this series none of the infected synthetic grafts could be salvaged without removal of at least part of the original graft, a finding consistent with other reports. 12-~4 In fact the risk of recurrent sepsis was significantly increased w h e n the PTFE graft was not completely excised. The native artery and vein may be managed by ligation, primary repair, or autogenous patching w h e n necessary. Immediate construction of an autogenous arteriovenous fistula with the native artery and vein was feasible in one of our patients and has been reported previously, x~ An aggressive search for other treatable sources of sepsis should also be carried out if prompt improvement does not occur.
It was not possible from this study to determine the minimum safe interval before reimplantation of another synthetic angioaccess graft at a previously infected site. To ensure that the original source ~of sepsis has been eliminated, the treatment interval should be lengthened as much as the clinical situation will permit. Alternative
384
Tabbara et al.
dialysis options such as the use of peritoneal lavage, repeated vascular cannulation, or indwelling catheters may be required while control of sepsis is achieved, but each of these methods has limitations. If another permanent vascular access site is needed, construction of an autogenous arteriovenous fistula is the most desirable alternative provided it is feasible. In this respect intraoperative venography is a useful adjunct to determine the availability of autogenous vein.
CONCLUSION Prosthetic HGI is a serious complication that is optimally treated by early diagnosis and excision of all infected PTFE. Complete removal of synthetic material offers a significantly reduced risk of recurrent graft sepsis at the same site. Prosthetic HGI is best avoided by the preferential use of autogenous primary procedures for hemodialysis access. The statistical analysis was performed by Mariane Piedmont, Department of Biostatistics and Epidemiology.
REFERENCE S 1. U.S. Renal Data System, 1991 Annual Data Report, The National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, Md., August 1991. V. Survival probabilities and causes of death. Am J Kidney Dis 1991;18:49-60. 2. Connolly JE, Brownell DA, Levine EE et al. Complications of renal dialysis access procedures. Arch Surg 1984; 119:13251328. 3. Hinsdale J, Lipkowitz G, Hoover E. Vascular access for hemodialysis in the elderly: Results a n d perspectives in a geriatric population. Dialysis Transplant 1985;14:560-565. 4. Zibari GB, Rohr MS, Landreneau MD, et al. Complications from permanent hemodialysis vascular access. Surgery 1988; 104:681-688.
5. Tellis VA, Weiss E Matas AJ, et al. Skin-flap coverage of polytetrafluqroethylene vascular access graft exposed by previous infection. Surgery 1988; 103:118-121. 6. Bhat D, Tellis V, Kohlberg W, et al. Management of sepsis involving expanded polytetrafluoroethylene grafts for hemodialysis access. Surgery 1980;87:445-450.
Annals of Vascular Surgery
7. Okadome K, watanabe T, I~na M, et al. Skin sliding closure technique is effective for management of infected prosthesis in cases of arterial reconstruction with synthetic grafts. Jpn J Surg 1982;12:349-355. 8. Elliot M, Bazzaniga A, Thomas J, et al. Use of expanded polytetrafluoroethylene grafts for vascular access in hemodialysis: Laboratory and clinical evaluation. Am Surg 1977;43: 455-459. 9. Raju S. PTFE grafts for hemodialysis access. Ann Surg 1987;206:666-673. 10, Gifford RR. Management of tunnel infections of dialysis polytetrafluoroethylene grafts. J Vasc Surg 1985;2:854-858. 11. Padberg FJ, Lee BC, Curl GR. Hemoaccess site infection. Surg Gynecol Obstet 1992;174:103-108. 12. Corry RJ, Patel NP, West JC. Surgical management of complications of vascular access for hemodialysis. Surg Gynecol Obstet 1980;151:49-54. 13. O'Hara P, Hertzer N, Beven E, et al. Surgical management of infected abdominal aortic grafts: Review of a 25-year experience. J Vasc Surg 1986;3:725-731. 14. Baker L, Johnson J, Goldfarb D. Expanded polytetrafluoroethylene (PTFE) subcutaneous arteriovenous conduit: An improved vascular access for chronic hemodialysis. Trans Am Soc Artif Intern Organs 1976;22:382-387. 15. McMullen K, Hayes D, Hussey JL, et al. Salvage of hemodialysis access in infected arteriovenous fistulas. Arch Surg 1991;126:1303-1305. 16. Mayers JD, Markell MS, Cohen LS, et al. Vascular access surgery for maintenance hemodialysis. Variables in hospital stay. ASAIO Trans 1992;38:113-115. 17. Iglehart J. Health Policy Report: The American Health Care System, The End Stage Renal Disease Program. N Engl J Med 1993;328:366-371. 18. Kieffer E, Bahnini A, Koskas F, et al. In situ allograft replacement of infected infrarenal aortic prosthetic grafts: Results in forty-three patients. J Vase Surg 1993;17:349-356. 19. Yeager RA, Porter JM. Basic data underlying clinical decision making in vascular surgery: Arterial and prosthetic graft infection. Ann Vase Surg 1992;6:485-491. 20. Francois F, Thdvenet A. Conservative treatment of prosthetic aortic graft infection with irrigation. Ann Vasc Surg 199l;5: 199-201. 21. Calligaro KD, Veith FJ. Clinical review: Diagnosis and management of infected prosthetic aortic grafts. Surgery 1991; 110:805-813. 22. Bandyk DF, Bergamini TM, Kinney EV, et al. In situ replacement of vascular prostheses infected by bacterial biofihns. J Vasc Snrg 1991;13:575-583. 23. Kirami N, Tuazon C, Murray H, et al. Staphylococcus aureus carriage rate of patients receiving long-term hemodialysis. Arch Intern Med 1978;138:1657-1659. 24. Bunt T. Synthetic vascular graft infections. I. Graft infections. Surgery 1983;93:733-746. 25. Fivush BA, Bock GH, Guzzetta PC, et al. Vancomycin prevents polytetrafluoroethylene graft infections in pediatric patients receiving chronic hemodialysis. Am J Kidney Dis 1985;5:120-123.