Burden of Infection in Patients with End-Stage Renal Disease Requiring Long-Term Dialysis

Transcription

1 MAJOR ARTICLE Burden of Infection in Patients with End-Stage Renal Disease Requiring Long-Term Dialysis Steven J. Berman, 1,2 E. William Johnson, 2 Curtis Nakatsu, 2 Michael Alkan, 3 Randi Chen, 1 and Jean LeDuc 1 1 Department of Infection Control and Epidemiology and Renal Institute of the Pacific, St. Francis Health Care Systems of Hawaii, and 2 Department of Medicine, John A. Burns School of Medicine, Honolulu, Hawaii; and 3 Ben-Gurion University Center for Health Sciences, Beersheba, Israel Background. This study examines the spectrum of infections in a selected population of patients requiring long-term dialysis, enlarging the focus beyond infections associated with the dialysis process. Methods. Infection data were reviewed from complete archived inpatient and outpatient dialysis records of 433 patients who were treated at a single hospital-based dialysis program and its dialysis satellites over a 9-year period, from 1 January 1992 to 31 December Results. The study period included 424,700 days of dialysis experience. A total of 2412 episodes of bacterial or fungal infections were treated in 433 patients. The infection rate was 5.7 episodes per 1000 days of dialysis. Patients received 5111 courses of antibiotics over 42,627 days of treatment, which cumulatively accounted for 10% of the total days of the study. Infections associated with hemodialysis vascular access devices comprised 20.5% of the total episodes. Infections below the knee (19.3% of infection episodes), pneumonia (13%), and other skin and soft-tissue infections (9%) were also important types and sources of infection, accounting for 142% of the total episodes. Eighty-two percent of the infections (1971 episodes) were acquired in the community. Of these, 868 (44%) required hospitalization. An additional 441 episodes were nosocomial. The profile of bacteria isolated from patients with community-acquired infections mirrored that of bacteria recovered from patients with nosocomial infections. Conclusion. Patients with end-stage renal disease have an enormous burden of infection. The majority of the infections are unrelated to dialysis. Frequent and long-term antibiotic use and cohorting of patients in the dialysis unit have altered the microbiological flora of such individuals, with clinical and epidemiological implications. When long-term dialysis became a federal entitlement in 1972, the program was envisioned as a way station before transplantation [1]. Over the ensuing 30 years, the size and demographic composition of the dialysis population has shifted dramatically to become much older and much sicker. For many patients undergoing long-term dialysis, there is little hope of receiving a new kidney. In 1999, Medicare spent more than $15 billion on the End-Stage Renal Disease (ESRD) program. The resources required for this care have outstripped all projections. Because of the formula for reimbursement under the federally mandated ESRD program, dialysis Received 26 February 2004; accepted 2 June 2004; electronically published 18 November Reprints or correspondence: Dr. Steven J. Berman, 1380 Lusitana St., Honolulu, Hawaii (sberman@aloha.net). Clinical Infectious Diseases 2004; 39: by the Infectious Diseases Society of America. All rights reserved /2004/ $15.00 centers have scant resources to deal with the chronic problems of individual patients [2, 3]. Since the insertion of the first vascular-access devices (VADs) that made chronic hemodialysis practical (i.e., hemodialysis VADs [HVADs]), VAD-associated infection has been a major issue [4 6]. Infection-control programs in dialysis units have, with great success, minimized the opportunity for infectious complications from viruses and bacteria acquired through contamination of dialysis fluids and equipment [7]. Infections of HVADs continue to be a problem; patients average 3.5 episodes of HVAD infections per 100 months of hemodialysis [6]. Many of these infections are caused by gram-positive methicillin-resistant Staphylococcus aureus (MRSA) [8], and widespread vancomycin use [9] has contributed to the appearance in this population of vancomycin-resistant S. aureus (VRSA) [10] and vancomycin-resistant enterococci (VRE) [11]. Prevention of these infections and of acquisition of multidrugresistant gram-positive bacteria are, understandably, an Infections in Patients Requiring Dialysis CID 2004:39 (15 December) 1747

2 obvious focus of and priority for infection control in all dialysis programs [12]. Nephrologists and infectious diseases specialists familiar with patients receiving long-term dialysis recognize that dialysis-related infections are only part of the problem. This article calls attention to the remaining bacterial infections in the dialysis population, the pervasive exposure to antibiotics, and the bacterial flora that colonize and infect these patients. MATERIALS AND METHODS Patient population and source documents. This is a retrospective review of the medical records of 433 patients who had undergone dialysis between 1 January 1992 and 31 December A total of 341 patients started dialysis for the first time after 1 January The inpatient hospital and outpatient dialysis records were the source documents. Patients received dialysis at 1 of 4 neighborhood hemodialysis centers and were hospitalized at a single facility under the umbrella of the St. Francis Renal Institute of the Pacific (SFRIP), a not-for-profit hospital-based program in Honolulu, Hawaii. For patients who are currently undergoing dialysis, the outpatient medical records are kept at the neighborhood dialysis units and are collated with the inpatient records after the patient leaves the program. Thus, the review was limited to patients who were not undergoing dialysis at the time of data collection. Of the patients in the study, 417 patients died, and 16 patients left the SFRIP dialysis program. Patients were excluded from the analysis if they underwent successful kidney transplantation or were alive and undergoing dialysis at the end of the study period, if they received hemodialysis for!90 days, or if they were hospitalized elsewhere (unless there was sufficient documentation for the reason for hospitalization and complete information on antibiotic therapy and culture results). Statistical analysis. Personnel who are experienced with dialysis reviewed the inpatient and outpatient records for all study patients. One of the investigators (C.N.) performed a second review. The data were entered into a relational database (4th Dimension) and were analyzed using a statistical package (SPSS). The x 2 test was used to compare the proportion of gram-positive and gram-negative bacterial isolates recovered during episodes of community-acquired infection with the proportion recovered during nosocomial infections, as well as to compare the infection-related mortality rate in the study cohort with that reported in the US Renal Data System (USRDS) [14]. Fisher s z test was used to test the differences between the study population at SFRIP and data from the USRDS 1999 and 2002 reports. The infection rate was defined for each patient as the number of infections per 1000 days of life during the study period. Student s t test was used to compare the infection rates for patients whose primary cause of ESRD was diabetes mellitus with infection rates for patients with ESRD due to other causes. All P values are based on results of the 2-sided test. R.C. was responsible for the statistical analysis. Definition of infection. An episode of infection was defined as the use of antibiotic treatment for 3 days in the context of criteria of signs and symptoms, as outlined in the Centers for Disease Control and Prevention guidelines for the definition of nosocomial infections [13]. The half-life of some antibiotics is prolonged in patients with severe renal failure. Some protocols call for the administration of antibiotics only at the end of each hemodialysis session. A course of treatment was acceptable if at least 2 antibiotic doses, separated by a nonhemodialysis day, were administered. Treatment with 1 dose of vancomycin was included as an episode of infection if supporting clinical signs or microbiological data were present by the third day of treatment, because 1 dose of vancomycin may sustain adequate serum concentration for at least 7 days in patients who are dialyzed with low-flux dialyzers. Because fever during hemodialysis is common and may arise from a variety of causes, including bacteremia and use of an HVAD, nephrologists may order blood cultures and an empirical dose of antibiotics when notified about the fever. There is usually no continuation of treatment at the next dialysis session, unless there is additional evidence of infection. Thus, patients who received a single dose of antibiotics or treatment courses for a duration 13 days without microbiological or clinical evidence of infection at the time of the next dialysis were excluded from the analysis. Infection at a previous site was counted as a new episode if no antibiotics were given for a minimum of 120 days before the current course of treatment. All below-the-knee infections (BKIs), with the exception of postoperative wound infection, were grouped together. All other skin and soft-tissue infections (SSTIs) were similarly grouped. Sepsis syndrome was diagnosed if a patient had fever, tachycardia, and leukocytosis without localizing signs of infection and received antibiotics for at least 3 days. The diagnosis of septicemia required bacteremia in the presence of fever, no local signs of infection that suggested the source of the bacteremia, and no relapse of bacteremia for 120 days after completion of the antibiotic course. Patients who had received prophylactic courses of antibiotics, treatment for tuberculosis and leprosy, and treatment for viral infections were not included in the analysis. Infection leading to death required not only the clinical diagnosis of infection noted above, but also documentation by the attending physician or the infectious diseases consultant that the infection was responsible for the patient s death. versus community-acquired infection. infections were defined according to standard criteria [13]. Infections in patients who were directly admitted from a skilled nursing or long-term care facility were also considered to be nosocomial. If the onset of signs and symptoms of in CID 2004:39 (15 December) Berman et al.

3 Table 1. Characteristics of study patients with end-stage renal disease (ESRD) from the St. Francis Renal Institute of the Pacific (SFRIP; Honolulu, HI) and the 2002 US Renal Data Survey (USRDS). Characteristic SFRIP (n p 433) USRDS (n p 92,635) a P Age at onset of dialysis, years One-year survival rate NS Race b White African American 1 30 Asian 54 3 Pacific Islander 30 Female sex NS Cause of ESRD Diabetes mellitus Hypertension Chronic glomerulonephritis Comorbidity Congestive heart failure Coronary artery disease Peripheral vascular disease Chronic obstructive pulmonary disease Death due to infection c NS Hospitalization Duration, days per patient-year No. of admissions per patient-year NS Reason Vascular-access procedure NS Pneumonia Septicemia No. of VAD infections per 100 months d NS Kt/V ratio 11.2 e 82 NOTE. Data are percentage of patients, unless otherwise indicated. NS, not significant. a Data are from USRDS 2002 [2], unless otherwise indicated. All patients were new recipients of dialysis in b Hawaii is a multiethnic society. c 1999 USRDS report [14], by x 2 analysis. d As reported by Tokars et al. [7]. e Adequacy of urea removal, where K is the urea clearance rate, t is the treatment duration, and V is the volume of water in the patient s body. fection and the first treatment occurred in the outpatient setting or within 48 h after hospital admission, the infection was considered to be community acquired. Microbiological analysis. Bacterial isolates were included if cultures were performed within 2 days after starting antibiotic therapy and if culture specimens were obtained from sites compatible with the clinical diagnosis. All isolates recovered from cultures of blood and other normally sterile body fluids were included, except for coagulase-negative staphylococci (CONS), which required isolation from 11 blood culture. Specimens from central catheters, including those providing access for hemodialysis, were cultured by the roll plate technique and were included in the analysis if 115 colonies were present [14]. RESULTS Patient characteristics. The archived medical records of 433 patients requiring long-term dialysis between 1 January 1992 and 31 December 2000 were reviewed. There were 424,700 dialysis days reported for these patients. The mean age at the onset of dialysis was 62.3 years, and the mean duration ( SD) of dialysis before death was days. Diabetes mellitus was the most common cause of ESRD (56% of patients), and 80% of patients were of Asian or Pacific Islander race. Other than these differences, the demographic and clinical profiles of patients in the study population were similar to those of Medicare patients undergoing dialysis, as reported by the USRDS Infections in Patients Requiring Dialysis CID 2004:39 (15 December) 1749

4 Figure 1. Number and types of community-acquired and nosocomial infections diagnosed in patients with end-stage renal disease. BKI, belowthe-knee infection; DIARRHEA, enteric bacterial infection; ENT, ears/nose/throat; G/U, genital/urinary tract; HVAD, hemodialysis vascular-access device; MISC, miscellaneous; NHVAD, non hemodialysis vascular-access device; PD, peritoneal dialysis related infection; POSTOP WOUND, postoperative wound; SSTI, skin/soft-tissue infection. [15]. Similar profile findings included the percentage of patients who were alive after 1 year of dialysis, the rate of HVAD infection, the rate of death due to infection, and/or the rate of hospitalization for any reason (table 1). Clinical infection. A total of 2412 episodes of infection were identified, with a rate of 5.7 episodes per 1000 days of dialysis. Community-acquired infections accounted for 1971 episodes (82%); of these, 868 (44%) required hospitalization. There were an additional 441 nosocomial infections treated in the hospital. Dialysis-related infections were responsible for 24% of the episodes, including 495 HVAD infections and 90 peritoneal dialysis related infections. Three other types of infections caused 50% of the total episodes: BKIs (468 episodes), pneumonia (315 episodes), and SSTIs at other sites (214 episodes) (figure 1). The infection rate in patients with ESRD due to diabetes mellitus was higher than that for the rest of the study cohort (6.72 vs episodes per 1000 dialysis-days). Almost all of the difference in the infection rate (94%) between the 2 study populations was due to BKIs and SSTIs. Independent of the cause of ESRD, the final year of life was especially burdensome: 1201 (50%) of the total episodes occurred during this period. Antibiotics. Patients received at least 1 antibiotic for the treatment of infection on 42,627 days a cumulative total of 10% of the study days. Antibiotics with favorable pharmacokinetic characteristics in patients with renal failure were used in both the hospital and outpatient dialysis settings (figure 2). Ten antibiotics accounted for 80% of the 5111 courses of treatment: cefazolin (1013 courses), vancomycin (939), tobramycin (939), gentamicin (606), ceftazidime (347), ciprofloxacin (289), ceftriaxone (263), metronidazole (185), ampicillin (120), and piperacillin/tazobactam (88). Microbiological findings. In 1134 episodes of infection, 1440 organisms were isolated (table 2). Cultures were not performed (905 episodes) or were sterile (373 episodes) for the remainder. A total of 48% of the organisms were gram-positive cocci, and 35% were aerobic gram-negative rods. S. aureus (328 isolates), CONS (206), Pseudomonas aeruginosa (148), Escherichia coli (102), Klebsiella species (77), and Enterobacter species (64) were the most frequent isolates recovered. S. aureus and CONS were the most commonly isolated bacteria from patients with an HVAD. There was no significant difference between the relative proportion of gram-positive cocci and aerobic gram-negative rods associated with either nosocomial or community-acquired infections of the VAD, with pneumonia, or with BKI. Gram-negative bacteria were isolated from the primary cultures at similar rates for each of these types of infection. Most striking was the frequency of isolation of gram-negative 1750 CID 2004:39 (15 December) Berman et al.

5 Figure 2. Courses of antibiotics administered to outpatients in the dialysis unit and to inpatients in the hospital. PIP-TAZO, piperacillin-tazobactam. bacilli from patients with community-acquired pneumonia. These bacteria were present in 55% of the 113 episodes associated with positive sputum culture results, including 23 episodes in which P. aeruginosa was isolated. DISCUSSION This study reviews the inpatient and outpatient infection experience in a cohort of patients with ESRD requiring long-term dialysis. Infection was common, averaging 5.7 events per 1000 dialysis-days, and antibiotic use was extensive, occurring on 10% of the study days. Although infections at the dialysis access site were responsible for 24% of the total episodes, the rest of the infection burden was unrelated to dialysis. BKIs, other SSTIs, and pneumonia accounted for 42% of the total number of infections. The cohort was selected retrospectively on the basis of available medical records from a 9-year period, compiling data from 1425,000 days after the initiation of long-term dialysis. It was not designed to show the relationship between the natural history of infection and the vintage of dialysis, because 95% of the 433 study patients died during the study period, and patients who were alive at the end of the study were excluded from the analysis. Despite this limitation, the cohort experienced rates of 1-year survival, HVAD infection, and hospitalization and numbers of hospital-days per year that were similar to those of patients described in the national database of ESRD (USRDS) and in other published studies [6, 16]. The demographic characteristics of this cohort differed from those of patients in the USRDS database with respect to race and the primary cause of ESRD. Diabetes mellitus was the cause of ESRD in 56% of the patients in our cohort, compared with 43% in patients described in the 2002 USRDS. Significantly more infections occurred in our diabetic patient group, especially infections of the lower extremity (BKIs) and other SSTIs. We predict that, by 2007, diabetes mellitus will be the cause of ESRD for 142% of new dialysis patients in the United States and anticipate that the number of infections will increase accordingly. Bacteria isolated from cultures of samples obtained from patients with nosocomial or community-acquired infections were similar. Thirty-five percent of the primary isolates were aerobic gram-negative rods. This is of interest because gramnegative bacilli comprise a small percentage of the normal flora of the skin and respiratory tracts of healthy individuals [17]. Historically, colonization and infection with these organisms have been associated with the hospital setting, bedridden patients, indwelling Foley catheters, or patients requiring mechanical ventilation [18 20]. Though patients with ESRD are ambulatory and live at home, they share the 3 characteristics that support the presence of organisms associated with noso- Infections in Patients Requiring Dialysis CID 2004:39 (15 December) 1751

6 Table 2. Microbiological data on infections in patients with end-stage renal disease undergoing dialysis. All sites Pneumonia HVAD Below-knee infection Bacterial isolate (n p 370) (n p 1070) (n p 73) (n p 113) (n p 36) (n p 201) (n p 32) (n p 275) Gram-positive cocci Total 158 (43) 539 (50) 16 (22) 29 (26) 30 (83) 149 (74) 23 (72) 147 (54) Coagulase-negative staphylococci 53 (14) 153 (14) 0 1 (1) a 16 (44) 46 (23) 7 (22) 42 (15) Methicillin-susceptible Staphylococcus aureus 19 (5) 194 (18) 0 12 (11) 7 (19) 70 (35) 3 (9) 49 (18) Methicillin-resistant S. aureus 45 (6) 70 (7) 16 (22) 13 (12) 4 (11) 20 (10) 4 (13) 13 (5) Enterococcus species 23 (5) 70 (7) 0 1 (1) a 2 (6) 9 (5) 6 (19) 27 (10) Vancomycin-resistant enterococci 12 (8) 1 (0) (6) 0 Other 59 (16) 6 (1) (47) 1 (1) 1 (3) 16 (6) Aerobic gram-negative rods 126 (34) 384 (36) 47 (64) 62 (55) 5 (14) 37 (18) 9 (28) 92 (34) Pseudomonas aeruginosa 31 (8) 117 (11) 16 (2) 24 (21) 0 9 (5) 5 (16) 25 (9) Escherichia coli 25 (3) 77 (7) 2 (3) 4 (4) 1 (3) 5 (3) 2 (6) 20 (7) Klebsiella species 20 (3) 57 (5) 10 (14) 12 (11) 0 7 (4) 0 9 (3) Enterobacter species 22 (3) 42 (4) 9 (12) 5 (4) 0 10 (5) 0 0 Other 28 (3) 91 (9) 10 (14) 17 (15) 3 (9) 6 (4) 2 (6) 38 (14) Respiratory pathogens Moraxella species 4 (1) 5 (1) 4 (6) 2 (2) Haemophilus influenzae 5 (1) 18 (2) 2 (3) 11 (12) Pneumococcus species 2 (1) 9 (1) 2 (3) 7 (6) Miscellaneous 75 (20) 115 (11) 2 (3) 2 (2) 1 (3) 15 (8) 0 36 (13) NOTE. Data are no. (%) of bacterial isolates and represent a total of 1440 isolates from 1134 episodes in 433 patients., community acquired; HVAD, hemodialysis vascular-access device. a Isolated from an empyema. comial infections: antibiotic use [21], clustering in a common environment (hemodialysis units), and the presence of indwelling medical devices [22]. Cefazolin, with its favorable pharmacokinetic profile in patients undergoing hemodialysis, was the most frequently prescribed antibiotic. Consistent with the medical literature, S. aureus and CONS were the most common bacteria isolated from HVAD infections [23]. Vancomycin use was less than that reported in other centers [24], because of a lower percentage of methicillin resistant isolates from cultures yielding S. aureus [9, 12] and antibiotic policy. SFRIP made a 20-year effort to use cefazolin for all suspected VAD infections and reserved vancomycin for verified MRSA or MRSE infection or for initial empirical treatment for sepsis syndrome due to a suspected VAD source. The policy has become Hobson s choice, because the extensive use of cefazolin, more than likely, was a major reason for the unusually high frequency of isolation of gramnegative bacteria, such as P. aeruginosa in cultures of sputum specimens obtained from patients with pneumonia, in cases of community-acquired infection. Although it may be difficult to distinguish between pathogenicity and colonization in a retrospective study, the recovery of these organisms from the initial cultures performed for patients with infection is relevant. In the future, guidelines for empirical therapy for many community-acquired infections in patients undergoing dialysis may recommend broad-spectrum antibiotics, which will further contribute to the problem of antibiotic pressure. Because 80% of all infectious episodes occurred at home, the risk of spreading these and other antibioticresistant bacteria to members of the patient s household, although not yet quantified, is real. The infection burden associated with patients undergoing dialysis and its impact on society will only increase with the increasing number of such patients, which estimated to double to 600,000 by What can be done to alter this paradigm? In our cohort, two-thirds of the episodes of infection were in the catheter-access device, the skin and soft tissues (including the ischemic and insensitive foot), or the lung. Many of these infections are preventable with standard treatment programs and protocols. With an HVAD infection rate of 3.5 cases per 1000 days of use, the federally mandated ESRD program has made the prevention of access-site infections a priority. One of the factors that determine the HVAD infection rate is the type of vascular access. The highest rate occurs with the temporary noncuffed dialysis catheter, which is usually placed in the internal jugular vein, and the lowest rate occurs with the simple fistula, which is generally placed in the forearm. Dialysis programs have a comprehensive infrastructure for monitoring dialysis devices. They are required to report the prevalence of each type of access 1752 CID 2004:39 (15 December) Berman et al.

7 and are mandated to use simple fistulas first to decrease the infection rate (Jared Sugihara, personal communication). Standard protocols are followed to prepare the access site before the initiation of dialysis. The staff is trained to inspect the VAD for problems that may lead to infection-associated thrombosis, aneurysm, hematoma, or nonhealing puncture sites. There is no similar required structure in the dialysis unit for the prevention of infections unrelated to dialysis. Patients in frail health frequently delay seeking early medical attention the cornerstone of the personal physician s approach to preventive medical care. High-risk, infection-prone patients constitute the following groups: those with multisystem disease (including peripheral vascular disease or an insensitive foot), those with severe functional limitations [25], and those who are malnourished [26]. There are evidence-based programs to improve the health and prevent complications of these conditions [27 29]. We hope that this article will encourage the implementation of such programs, which can help the patient reduce their infection burden and limit the appearance and spread of multidrug-resistant bacteria. Acknowledgments We wish to thank Joann Tsark and Kathryn Braun for advice on study design, Michael McNiel (Tsunami Dataworks) for developing the database, Julie D. Benedetto and Jane Doll for supervising data extraction, Daniel F. Brandt and Sarah K. Kemble for conducting a pilot test of the database, and Jared Sugihara for review of the manuscript. Financial support. Centers for Disease Control and Prevention (grant H75/CCH ), the Hawaii Medical Services Association Foundation, and the Velin and Julia Kung Foundation. Potential conflicts of interest. All authors: no conflicts. References 1. Rennie D. Home dialysis and the costs of uremia. New Engl J Med 1978; 298: Levinsky NG. The organization of medical care: lessons from the Medicare end stage renal disease program. N Engl J Med 1993; 329: Hull AR. The legislative and regulatory process in the end-stage renal disease (ESRD) program, 1973 through Semin Nephrol 1997; 17: Altman LK. Obituary for Dr. Belding H. Scribner, medical pioneer. New York Times. National ed. 22 June 2003: Ralston AJ, Harlow GR, Jones DM, Davis P. Infections of Scribner and Brescia arteriovenous shunts. Br Med J 1971; 2: Tokars JI, Light P, Anderson J, et al. A prospective study of vascular access infections at seven outpatient hemodialysis centers. Am J Kidney Dis 2001; 37: Tokars JI, Arduino MJ, Alter MJ. Infection control in hemodialysis units. Infect Dis Clin North Am 2001; 15: Golper TA, Schulman G, D Agata EM. Indications for vancomycin in dialysis patients. Semin Dial 2000; 13: Tokars JI, Frank M, Alter MJ, Arduino MJ. National surveillance of dialysis-associated diseases in the United States, Semin Dial 2002; 15: Chang S, Sievert DM, Hagemean JC, et al. Infection with vancomycinresistant Staphylococcus aureus containing the vana resistance gene. N Engl J Med 2003; 348: Atta MG, Eustace JA, Song X, Perl TM, Scheel PJ. Outpatient vancomycin use and vancomycin-resistant enterococci colonization in maintenance dialysis patients. Kidney Int 2001; 59: D Agata EM. Antimicrobial-resistant, gram-positive bacteria among patients undergoing chronic hemodialysis. Clin Infect Dis 2002; 35: Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, Am J Infect Control 1988; 16: Maki DG, Jarrett F, Sarafin HW. A semiquantitative culture method for identification of catheter-related infection in the burn patient. J Surg Res 1977; 22: US Renal Data Survey Annual data report. Bethesda, MD: National Institute of Diabetes and Digestive Diseases and Kidney Diseases, US Renal Data Survey Annual data report. Bethesda, MD: National Institute of Diabetes and Digestive Diseases and Kidney Diseases, Mackowiak PA. The normal flora. N Engl J Med 1982; 307: Valenti MD, Trudell BS, Bentley DW. Factors predisposing to oropharyngeal colonization with gram-negative bacilli in the aged. N Engl J Med 1978; 298: Turck M, Stamm WE. infection of the urinary tract. Am J Med 1981; 70: Sadar N, Maki DG. The commonality of risk factors for nosocomial colonization and infection with antimicrobial-resistant Staphylococcus aureus, enterococcus, gram-negative bacilli, Clostridium difficile, and Candida. Ann Intern Med 2002; 136: Finland M. Changing ecology of bacterial infections as related to antibacterial therapy. J Infect Dis 1970; 122: Maki DG. bacteremia: an epidemiologic overview. Am J Med 1981; 70: Dopirak M, Hill C, Oleksiw M, et al. Surveillance of hemodialysisassociated primary bloodstream infections: the experience of ten hospital-based centers. Infect Control Hosp Epidemiol 2002; 23: Green K, Schulman G, Haas DW, Schaffner W, D Agata EM. Vancomycin prescribing practices in hospitalized chronic hemodialysis patients. Am J Kidney Dis 2000; 35: Muder RR, Brennen C, Swenson DL, Wagener M. Pneumonia in a long-term care facility: a prospective study of outcome. Arch Intern Med 1996; 156: Jaar BG, Hermann JA, Furth SL, Briggs W, Powe NR. Septicemia in diabetic hemodialysis patients: comparison of incidence, risk factors, and mortality with nondiabetic hemodialysis patients. Am J Kidney Dis 2000; 35: Bodenheimer T, Wagner EH, Grumbach K. Improving primary care for patients with chronic illness: the chronic care model, part 2. JAMA 2002; 288: Nissenson AR, Collins AJ, Dickmeyer J, et al. Evaluation of diseasestate management of dialysis patients. Am J Kidney Dis 2001; 37: Oppenheimer CC, Shapiro JR, Beronja N, et al. Evaluation of the ESRD managed care demonstration operations. Health Care Financ Rev 2003; 24:7 29. Infections in Patients Requiring Dialysis CID 2004:39 (15 December) 1753