Recurrent and Relapsing Peritonitis: Causative Organisms and Response to Treatment

Recurrent and Relapsing Peritonitis: Causative Organisms and Response to Treatment Cheuk-Chun Szeto, MD, FRCP, Bonnie Ching-Ha Kwan, MBBS, MCP(UK), Kai-Ming Chow, MBChB, MRCP(UK), Man-Ching Law, BN, RN, Wing-Fai Pang, MBChB, MRCP(UK), Kwok-Yi Chung, MBChB, MRCP(UK), Chi-Bon Leung, MBChB, FRCP(Edin), and Philip Kam-Tao Li, MD, FRCP Background: The clinical behavior and optimal treatment of relapsing and recurrent peritonitis episodes in patients undergoing long-term peritoneal dialysis are poorly understood. Study Design: Retrospective study over 14 years. Setting & Participants: University dialysis unit; 157 relapsing episodes (same organism or culturenegative episode occurring within 4 weeks of completion of therapy for a prior episode), 125 recurrent episodes (different organism, occurs within 4 weeks of completion of therapy for a prior episode), and 764 control episodes (first peritonitis episode without relapse or recurrence). Predictors: Exit-site infection, empirical antibiotics. Outcome Measures: Primary response (resolution of abdominal pain, clearing of dialysate, and peritoneal dialysis effluent neutrophil count 100 cells/ml after 10 days of antibiotic therapy), complete cure (resolution by using antibiotics without relapse/recurrence), catheter removal (for any cause while on antibiotic therapy), and mortality. Results: Compared with the control group, more relapsing episodes were caused by Pseudomonas species (16.6% versus 9.4%) and were culture negative (29.9% versus 16.4%); recurrent infections commonly were caused by Enterococcus species (3.2% versus 1.2%) or other Gram-negative organisms (27.2% versus 11.1%) or had mixed bacterial growth (17.6% versus 12.7%). There were significant differences in primary response, complete cure, and mortality rates among groups (P 0.001 for all comparisons). Compared with the control and relapsing groups, post hoc analysis showed that the recurrent group had a significantly lower primary response rate (86.4%, 88.5%, and 71.2%, respectively), lower complete cure rate (72.3%, 62.4%, and 42.4%, respectively), and higher mortality rate (7.7%, 7.0%, and 20.8%, respectively). Limitations: Retrospective analysis. Conclusion: Relapsing and recurrent peritonitis episodes are caused by different spectra of bacteria and probably represent 2 distinct clinical entities. Recurrent peritonitis episodes had a worse prognosis than relapsing ones. Am J Kidney Dis 54:702-710. 2009 by the National Kidney Foundation, Inc. INDEX WORDS: Renal failure; nosocomial infection; antibiotics. From the Department of Medicine and Therapeutics, Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong, China. Received December 1, 2008. Accepted in revised form April 3, 2009. Originally published online as doi: 10.1053/ j.ajkd.2009.04.032 on July 6, 2009. Address correspondence to Cheuk-Chun Szeto, MD, FRCP, Department of Medicine & Therapeutics, Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong, China. E-mail: ccszeto@cuhk.edu.hk 2009 by the National Kidney Foundation, Inc. 0272-6386/09/5404-0015$36.00/0 doi:10.1053/j.ajkd.2009.04.032 Peritonitis is a serious complication of peritoneal dialysis (PD) and probably the most common cause of technique failure in PD. 1-5 In the United States, 18% of the infection-related mortality in PD patients is the result of peritonitis. 6 Although less than 4% of peritonitis episodes result in death, 7 peritonitis is a contributing factor to death in 16% of deaths in patients on PD therapy. 8 In Hong Kong, peritonitis is the direct cause of death in more than 16% of PD patients. 9 In the latest recommendations for the management of PD-related infections published by the International Society for Peritoneal Dialysis (ISPD) in 2005, 10 2 similar terms, relapsing and recurrent peritonitis, are defined differently. In essence, peritonitis that is treated with appropriate antibiotic therapy, appears to resolve, and yet returns with the same organism or as sterile peritonitis within 4 weeks (relapsing) is different from an episode of peritonitis that occurs within 4 weeks of a prior episode, but with a different organism (defined as recurrent). 10 The choice of terms and a difference in definition are meant to have clinical implication. Theoretically, antimi- 702 American Journal of Kidney Diseases, Vol 54, No 4 (October), 2009: pp 702-710

Recurrent Peritonitis in PD 703 crobial resistance may develop during the treatment, resulting in early relapse with an identical species, but with a different susceptibility pattern, after completion of the treatment. Alternatively, the source of the relapse may be the catheter through either biofilm or tunnel infection. Therefore, for relapsing peritonitis, most experts in the field would recommend catheter removal unless the susceptibility pattern of the organism has changed, indicating that additional antibiotic therapy may be warranted. However, in recurrent peritonitis, the patient s immunity may be impaired by the first episode, leading in some cases to an episode of peritonitis from a completely different organism, implying a different cause. 10 These episodes frequently can be treated successfully without catheter removal. Unfortunately, this definition could be confusing because the 2 terms were often used interchangeably before publication of the 2005 ISPD recommendation. 11 Although the differentiation between relapsing and recurrent peritonitis episodes logically is sound, there has been no published evidence to support this approach. In the present study, we review clinical outcomes of relapsing and recurrent peritonitis in a large cohort of PD patients. METHODS Patients All PD patients treated at our center gave written consent for reviewing their clinical data when they entered the dialysis program. All episodes of PD peritonitis in our unit from 1994 to 2007 were reviewed. Data were collected by reviewing the Hong Kong Renal Registry database and hospital records of individual patients by the investigators. The diagnosis of peritonitis was based on at least 2 of the followings conditions 12,13 : (1) abdominal pain or cloudy peritoneal dialysis effluent (PDE); (2) leukocytosis in PDE (white blood cell count 100 cells/ml); and (3) positive Gram stain or culture results from PDE. Episodes with peritoneal eosinophilia, but negative bacterial culture results, were excluded. Exit-site infection was diagnosed when there was purulent drainage with or without erythema from the exit site. 14 In this study, relapsing peritonitis was defined as an episode with either the same organism or a negative culture result that occurs within 4 weeks of completion of therapy of a prior episode. 10 Recurrent peritonitis was defined as an episode that occurs within 4 weeks of completion of therapy of a prior episode, but with a different organism. 10 In the 14 years of the study, 2,390 episodes of peritonitis were recorded; 240 episodes (10.0%) were relapsing peritonitis in 157 patients, whereas 189 (7.9%) were recurrent peritonitis in 125 patients. To avoid complex analysis that accounts for multiple events, we included the first episode from each individual for statistical analysis. Demographic characteristics, underlying medical conditions, previous peritonitis episodes, antibiotic regimen for the peritonitis episode, catheter removal, and clinical outcome were examined. Microbiological Investigations Bacterial culture of PDE was performed using BacTAlert bottles (Organon Teknika Corp, Durham, NC). Species identification was performed by using the API 20E identification system (BioMerieux, Marcy l Etolie, France). Antibiotic sensitivity was determined by using the disc-diffusion method according to the National Committee for Clinical Laboratory Standard. 15 Clinical Management As described previously, 16-19 peritonitis episodes were treated by using the standard antibiotic protocol used by our center at that time, which was changed systemically over time (Table 1). Initial antibiotic regimens for peritonitis generally were intraperitoneal administration of a third- or fourth-generation cephalosporin plus or minus intermittent vancomycin every 5 days or cefazolin as continuous administration plus an aminoglycoside or ceftazidime. 5 Vancomycin and cefazolin dosages followed the contemporary guidelines. 10 During the review period, 2 clinical trials of peritonitis monotherapy using cefepime and imipenem/cilastatin were conducted in our center 18,19 ; 281 episodes in this review therefore were treated initially with monotherapy. Antibiotic regimens for individual patients were modified when culture results were available. Table 1. Standard Protocols for the Treatment of Peritoneal Dialysis Related Peritonitis at Various Times in the Unit Time January 1994-December 1995 January 1996-June 1996 July 1996-July 1998 August 1998-October 1998 November 1998-March 1999 April 1999-February 2001 March 2001-February 2002 March 2002-present Standard Protocol* Vancomycin ceftazidime Cefepime vancomycin Vancomycin ceftazidime or cefepime Sulperazone Vancomycin cefepime Cefazolin netilmicin Imipenem/cilastatin or cefazolin ceftazidime Cefazolin ceftazidime *All antibiotics were administered intraperitoneally unless the patient had features of systemic sepsis. Imipenem/ cilastatin and all cephalosporins were administered continuously; vancomycin and aminoglycosides were administered intermittently. Randomized controlled trial comparing the 2 protocols.

704 Primary response was defined as resolution of abdominal pain, clearing of dialysate, and PDE neutrophil count less than 100 cells/ml on day 10 by using antibiotics alone. Complete cure was defined as complete resolution of peritonitis by using antibiotics alone without relapse or recurrence within 4 weeks of completion of therapy. In general, patients received the effective antibiotic for 14 days, whereas the effective antibiotic would be continued for a total of 21 days for episodes caused by Staphylococcus aureus or Pseudomonas species. 17,20 When the PDE did not clear up after 5 days of effective antibiotic therapy, the Tenckhoff catheter was removed irrespective of the in vitro sensitivity of the bacterial strain, and effective antibiotic therapy was continued for another 2 weeks. Tenckhoff catheters were removed and patients were put on temporary hemodialysis therapy when peritonitis failed to resolve by using antibiotics. Tenckhoff catheter reinsertion was attempted in all cases. In our locality, as described in our previous study, 4 patients were switched to long-term hemodialysis therapy only when attempts at Tenckhoff catheter reinsertion failed because of peritoneal adhesion or there was ultrafiltration failure caused by peritoneal sclerosis. All patients were followed up for at least 3 months after their treatment completed. Patient mortality was defined as death from any cause during antibiotic treatment (generally 2 to 3 weeks, depending on the specific organism) or temporary hemodialysis therapy (generally 4 weeks after catheter removal). Statistical Analysis Statistical analysis was performed using SPSS (SPSS Inc, Chicago, IL) for Windows (Microsoft Corp, Redmond, WA) software, version 15.0. All data are expressed as mean SD unless otherwise specified. Baseline data were compared by using 2 test, Fisher exact test, Student t test, or 1-way Szeto et al analysis of variance, as appropriate. Clinical outcomes were compared among groups and choices of empirical antibiotic treatment by using logistic regression to adjust for the effect of age and Charlson Comorbidity Index score, which were significantly different among groups. Interactions between concomitant exit-site infection and patient group and between the causative organism of the previous peritonitis episode and choice of empirical antibiotic treatment on clinical outcome also were tested by using logistic regression models. Post hoc analysis was performed by using 2 test; P values were adjusted by using the Bonferroni method to allow for the effect of multiple comparisons. P 0.05 is considered significant. All probabilities were 2 tailed. RESULTS From 1994 to 2007, a total of 2,390 episodes of PD-related peritonitis were recorded in our unit. The overall peritonitis rate was 1 episode/ 20.9 patient-months of follow-up. We analyzed 157 episodes of relapsing peritonitis, 125 episodes of recurrent peritonitis, and 764 control episodes. Baseline clinical characteristics of patients at the time of peritonitis are compared among groups and listed in Table 2. Patients with relapsing peritonitis were significantly younger (1-way analysis of variance, P 0.005) and had a significantly lower Charlson Comorbidity Index score (P 0.006) compared with the other 2 groups. There was no significant difference in baseline clinical data between patients in the recurrent and control groups. Table 2. Baseline Characteristics of Patients at the Time of the First Peritonitis Episode Relapsing Group Recurrent Group Control Group No. of patients 157 125 764 Men/women 79/78 70/55 392/372 Age (y) 54.6 13.6 58.0 12.7 57.0 13.6 Duration of dialysis (mo) 37.4 33.0 34.8 35.6 24.2 26.2 Body height (cm) 160.0 8.5 159.9 8.4 159.7 8.3 Body weight (kg) 59.3 11.6 58.9 10.3 59.5 11.2 Diagnosis Glomerulonephritis 50 (31.8) 34 (27.2) 216 (28.3) Diabetes 31 (19.7) 27 (21.6) 237 (31.0) Hypertension 14 (8.9) 14 (11.2) 56 (7.3) Polycystic 5 (3.2) 5 (4.0) 25 (3.3) Obstruction 12 (7.6) 9 (7.2) 45 (5.9) Others/unknown 45 (28.7) 36 (28.8) 185 (24.2) Major comorbidity Coronary heart disease 25 (15.9) 23 (18.4) 153 (20.0) Cerebrovascular disease 28 (17.8) 25 (20.0) 155 (20.3) Peripheral vascular disease 10 (6.4) 7 (5.6) 48 (6.3) Diabetes 49 (31.2) 36 (28.8) 293 (38.4) Charlson Comorbidity Index score 5.1 2.3 5.5 2.3 5.6 2.4 Note: Values expressed as mean SD or number (percent).

Recurrent Peritonitis in PD 705 Table 3. Microbiological Cause of the Second Episode of Peritonitis Organisms Identified Causative Organism Relapsing Group Recurrent Group Control Group Gram-positive organisms Staphylococcus aureus 12 (7.6) 6 (4.8) 104 (13.6) Coagulase-negative Staphylococcus species 20 (12.7) 12 (9.6) 92 (12.0) Enterococcus species 1 (0.6) 4 (3.2) 9 (1.2) Other Streptococcus species 2 (1.3) 4 (3.2) 81 (10.6) Others 6 (3.8) 7 (5.6) 15 (2.0) Gram-negative organisms Pseudomonas species 26 (16.6) 14 (11.2) 72 (9.4) Escherichia coli 12 (7.6) 10 (8.0) 52 (6.8) Others 20 (12.7) 34 (27.2) 85 (11.1) Fungi 1 (0.6) 10 (8.0) 17 (2.2) Mycobacterium 0 (0) 2 (1.6) 15 (2.0) Polymicrobial growth 10 (6.4) 22 (17.6) 97 (12.7) Culture negative 47 (29.9) 0 (0) 125 (16.4) Total 157 125 764 Note: Values expressed as number (percent). Microbiological causes of the second episodes of peritonitis are listed in Table 3. There was a significant difference in the distribution of causative organisms among groups (overall 2 test, P 0.001). Specifically, compared with the control group, there was a trend to a greater percentage of peritonitis episodes in the relapsing group that were caused by Pseudomonas species and that were culture negative, whereas the proportion of episodes caused by S aureus was slightly lower (Table 3). In contrast, in the recurrent group, there was a trend to a greater percentage of peritonitis episodes that were caused by Enterococcus species and Gramnegative organisms and that had mixed bacterial growth (Table 3). Concomitant exit-site infections were present in 10.2%, 10.4%, and 14.1% of the relapsing, recurrent, and control groups, respectively (P 0.3); exit-site infections with the same organism as the peritonitis episode were present in 7.0%, 4.0%, and 9.3% of the relapsing, recurrent and control groups, respectively (P 0.1). In either case, the incidence was not different among patient groups. Clinical Response Major clinical outcomes are shown in Fig 1. After adjusting for patient age and Charlson comorbidity score, there were significant differences in primary response, complete cure, and mortality rates among groups (P 0.001 for all comparisons), whereas the catheter removal rate was not significantly different among groups. On further post hoc analysis, the recurrent group had a significantly lower primary response rate (71.2% versus 86.4%; P 0.001), lower complete cure rate (42.4% versus 72.3%; P 0.001), and higher mortality rate (20.8% versus 7.7%; P 0.001) than the control group; as well as a lower primary response rate (71.2% versus 88.5%; P 0.001), lower complete cure rate (42.4% versus 62.4%; P 0.003), and higher mortality rate (20.8% versus 7.0%; P 0.002) than the relapsing group (Fig 1). There were no significant differences in primary response, catheter removal, or mortality rates between the relapsing and control groups. However, the relapsing group had a slightly, but significantly, lower complete cure rate than the control group (62.4% versus 72.3%; P 0.04). We also found a significant interaction between the effects of concomitant exit-site infec- Figure 1. Comparison of clinical outcomes among groups. See text for the P value for the overall comparison by means of logistic regression. P values in parentheses represent overall comparison by means of logistic regression after adjusting for age and Charlson comorbidity score; other P values in the figure are the results of post hoc analysis adjusted by using the Bonferroni method for 3 comparisons.

706 Szeto et al infection and patient group on primary response, complete cure, and catheter removal rates (details not shown). Figure 2. Effect of concomitant exit-site infection on the mortality of patient groups. P values are the results of post hoc analysis adjusted by using the Bonferroni method for 3 comparisons. tion and patient group on mortality. Subgroup analysis showed that in patients without exit-site infection, the mortality rate in the recurrent group (22.3%) was significantly greater than those for the control and relapsing groups (8.1% and 5.7%, respectively; P 0.001 for either compared with the recurrent group; Fig 2). In contrast, for patients with concomitant exit-site infection, the relapsing group had a marginally greater mortality rate than the control group (18.8% versus 5.6%; P 0.2); however, the result was not statistically significant. There were no interactions between the effects of concomitant exit-site Choice of Antibiotics Because relapsing and recurrent peritonitis episodes could not be differentiated at their initial presentation, we examined the effect of choice of empirical antibiotic therapy on clinical outcome for both conditions as a group. The result is shown in Fig 3. After adjusting for age and Charlson comorbidity score, empirical use of vancomycin as Gram-positive coverage compared with cefazolin was associated with a significantly higher primary response rate (87.9% versus 72.7%; P 0.003), significantly lower mortality rate (6.8% versus 20.0%; P 0.002), and marginally lower catheter removal rate (6.1% versus 12.7%; P 0.07), although the latter did not reach statistical significance (Fig 3). However, complete cure rates were similar between patients treated with vancomycin and cefazolin (56.1% versus 49.1%; P 0.3). After adjusting for age and Charlson comorbidity score, empirical use of ceftazidime as Gram-negative coverage compared with an aminoglycoside (either netilmicin or gentamicin) was associated with a significantly higher primary response rate (82.3% versus 62.5%; P 0.02), significantly lower cath- Figure 3. Relation between clinical outcome and choice of initial empirical antibiotics: (A) Gram-positive coverage, vancomycin versus cefazolin, and (B) Gram-negative coverage, ceftazidime versus aminoglycoside (netilmicin or gentamicin).

Recurrent Peritonitis in PD 707 eter removal rate (7.5% versus 25.0%; P 0.006), higher complete cure rate (54.3% versus 41.7%; P 0.06), and lower mortality rate (11.8% versus 25.0%; P 0.07), although the latter 2 comparisons did not reach statistical significance (Fig 3). We further tested the interaction between choice of initial empirical antibiotics and the causative organism of the previous peritonitis episode on clinical outcome. In brief, there was a significant interaction between empirical Gram-positive antibiotic coverage and causative organism of the previous peritonitis episode on primary response rate (P 0.05) and rate of catheter removal (P 0.04) after adjusting for 4 comparisons by using the Bonferroni method. Subgroup analysis showed that the benefit of vancomycin is confined largely to patients whose previous peritonitis episode was caused by Gram-positive organisms (Fig 4), in which case empirical vancomycin had a significantly higher primary response rate (96.1% versus 70.6%; P 0.001) and significantly lower catheter removal rate (2.0% versus 17.6%; P 0.01) than cefazolin treatment. The causative organism of the previous peritonitis episode did not have a significant interaction with choice of antibiotic coverage for Grampositive organisms on the complete cure or mortality rate and did not have a significant interaction with choice of antibiotic coverage for Gram-negative organisms on any clinical outcome (details not shown). DISCUSSION In this retrospective study, we show that relapsing peritonitis commonly is caused by coagulasenegative Staphylococcus and Pseudomonas species, whereas recurrent peritonitis often is caused by non-pseudomonas Gram-negative organisms and mixed bacterial growth. Our findings suggest that relapsing and recurrent peritonitis are 2 distinct clinical entities and provide support to the current terminology proposed in the ISPD recommendation. 10 Because recurrent peritonitis often is caused by non-pseudomonas Gramnegative organisms and mixed bacterial growth, the underlying bowel pathological state may be an important, but unrecognized, cause of recurrent peritonitis episodes. However, recent antibiotic therapy may disturb gastrointestinal flora and provoke transmural migration of bowel organisms to the peritoneal cavity. 21 In the present study, we show that recurrent peritonitis episodes had the worst primary response rate, greatest incidence of catheter removal, and greatest mortality. The clinical outcome of relapsing and recurrent episodes is contrary to what is traditionally believed. Specifically, it generally is believed that relapsing episodes are more severe than recurrent ones 11 ; Figure 4. Effect of the causative organism of the previous peritonitis episode on (A) primary response rate and (B) rate of catheter removal in relation to the choice of initial antibiotic coverage for Gram-positive organisms.

708 most experts would recommend catheter removal for relapsing peritonitis, whereas recurrent peritonitis frequently is treated without catheter removal. 11 It certainly is possible that peritonitis episodes that are prone to relapse (ie, S aureus and Pseudomonas species) are more aggressively treated nowadays compared with 10 or 15 years ago, and the clinical behavior of relapsing peritonitis episodes under the contemporary treatment guideline may be different from that observed in patients treated a decade or more ago. However, to the best of our knowledge, there are no published reports comparing clinical characteristics and therapeutic responses between relapsing and recurrent peritonitis episodes. It is unexpected to find that concomitant exitsite infection had very little impact on clinical outcome. We are not sure of the reason for this phenomenon. It is possible that prescription of antibiotics is liberal in Hong Kong, and most catheter infections are treated aggressively. However, it also is possible that most bacterial biofilm that causes relapsing peritonitis comes from the catheter lumen without overt exit-site infection. We found that for patients who present with relapsing or recurrent peritonitis, vancomycin is superior to cefazolin as empirical antibiotic treatment, especially in patients with a previous peritonitis episode caused by Gram-positive organisms. The result is consistent with our previous studies 17,22,23 and other reports 24,25 : the risk of having a methicillin-resistant isolate is increased substantially in patients with recent antibiotic therapy. We also observed that ceftazidime has certain advantage over an aminoglycoside as empirical treatment in patients with a previous peritonitis episode caused by Gram-negative organisms. The finding is consistent with our previous observation in Pseudomonas peritonitis, 20 which is particularly prone to relapse. We are not sure why aminoglycosides have an inferior treatment response, but it seems possible that the currently recommended dose of netilmicin or gentamicin could not achieve therapeutic plasma levels, especially in patients with substantial residual renal function, and therefore has minimal systemic bactericidal activity. There are a number of limitations to our present study. Although the sample size seems large, results are retrospective and uncontrolled. For Szeto et al the effect of empirical antibiotic therapy on clinical outcome, we could not exclude the possibility of selection bias (although 90% of episodes were managed according to a predefined protocol). Our result also does not exclude the possibility that with prolonged used of a certain antibiotic as empirical therapy, the pattern of antibiotic resistance would change, so that the satisfactory response initially observed would not be sustained. In theory, there could be a periodic variation in the pattern of bacteriological cause of peritonitis in our center that may confound the response to antibiotic. Nonetheless, our previous study showed that the change in bacteriological pattern was somewhat irregular, 5 and systemic bias therefore is unlikely. Second, we did not analyze the interaction between antibiotic resistance and choice of empirical therapy because the sample size of each group would be small. Similar analyses had been performed in our previous reports on peritonitis caused by S aureus, 17 coagulase-negative Staphylococcus species, 23 and Enterobacteriaceae species 16 ; however, all of them did not distinguish relapsing, recurrent, and de novo episodes. Because prophylactic antibiotic protocols are not used in our center (except for mupirocin ointment for nasal carriers of S aureus and gentamicin cream for exit-site infection), our study could not provide information about the efficacy of various prophylactic protocols in reducing the risk of peritonitis. Finally, we performed a considerable number of statistical comparisons, with the goal of generating hypotheses. Although P values are adjusted for multiple comparisons, type I statistical error remains possible. Additional studies are needed to confirm our observations. Given the possibility of a bowel origin that caused recurrent peritonitis episodes, it would be particularly interesting to test whether there is a beneficial effect of bowel decontamination for the prevention of recurrent peritonitis. Our previous study of Enterobacteriaceae peritonitis showed that treatment with 2 antibiotics compared with monotherapy may reduce the risk of relapse and recurrence. 16 It is unknown whether empirical coverage with 2 antibiotics (for example, ceftazidime plus an aminoglycoside) would result in a superior outcome in patients who present with cloudy PDE soon after completing treatment for an Enterobacteriaceae peritonitis episode. In this study, we

Recurrent Peritonitis in PD 709 included each individual only once for statistical analysis to avoid complex analysis that accounts for multiple events. However, the result of analysis would be similar if all episodes were included (details not shown). Despite the limitations, one advantage of this study is that it is from a single center, meaning that data are more complete and accuracy could be checked more readily compared with registry data. Realistically, large adequately powered studies of PD peritonitis are unlikely to be done, and clinical practice must rely heavily on large retrospective studies. In addition, our result provides important insight about the clinical management of peritonitis. Specifically, in patients who develop cloudy PDE soon after completing treatment for Gram-positive peritonitis, vancomycin would be a better choice of empirical antibiotic coverage than cefazolin, at least until information about antibiotic sensitivity is available. Similarly, in those who had a recent Gram-negative peritonitis episode, ceftazidime seems a better empirical treatment than an aminoglycoside. This policy would have an additional advantage of avoiding repeated exposure to aminoglycosides, with potential ototoxicity and vestibular toxicity. More importantly, recurrent peritonitis episodes had a poor therapeutic response and high mortality; they deserve a treatment equally, if not more, aggressive as for relapsing episodes. In summary, we show that relapsing and recurrent peritonitis episodes are caused by a different spectrum of bacteria and probably represent 2 distinct clinical entities. Recurrent peritonitis episodes had a lower primary response rate, more catheter removal, and greater mortality than relapsing and control episodes. Our findings suggest that for patients who present with relapsing or recurrent peritonitis, vancomycin and ceftazidime as the empirical treatment are associated with a better outcome than cefazolin and an aminoglycoside, respectively. ACKNOWLEDGEMENTS We thank Ms Lau Miu Fong for clerical assistance. Support: This study was supported in part by the Chinese University of Hong Kong research account 6901031 and Richard Yu Peritoneal Dialysis Research Fund. Financial Disclosure: None. REFERENCES 1. Piraino B: Peritonitis as a complication of peritoneal dialysis. J Am Soc Nephrol 9:1956-1964, 1998 2. Oreopoulos DG, Tzamaloukas AH: Peritoneal dialysis in the next millennium. Adv Ren Replace Ther 7:338-346, 2000 3. Szeto CC, Wong TY, Leung CB, et al: Importance of dialysis adequacy in mortality and morbidity of Chinese CAPD patients. Kidney Int 58:400-407, 2000 4. Szeto CC, Chow KM, Wong TY, et al: Feasibility of resuming peritoneal dialysis after severe peritonitis and Tenckhoff catheter removal. J Am Soc Nephrol 13:1040-1045, 2002 5. Szeto CC, Leung CB, Chow KM, et al: Change in bacterial aetiology of peritoneal-dialysis-related peritonitis over ten years: Experience from a center in South-East Asia. Clin Microbiol Infect 10:837-839, 2005 6. Bloembergen WE, Port FK: Epidemiological perspective on infections in dialysis patients. Adv Ren Replace Ther 3:201-207, 1996 7. Mujais S: Microbiology and outcomes of peritonitis in North America. Kidney Int Suppl 103:S55-S62, 2006 8. Fried LF, Bernardini J, Johnston JR, Piraino B: Peritonitis influences mortality in peritoneal dialysis patients. J Am Soc Nephrol 7:2176-2182, 1996 9. Szeto CC, Wong TY, Chow KM, Leung CB, Li PK: Are peritoneal dialysis patients with and without residual renal function equivalent for survival study? Insight from a retrospective review of the cause of death. Nephrol Dial Transplant 18:977-982, 2003 10. Piraino B, Bailie GR, Bernardini J, et al; ISPD Ad Hoc Advisory Committee: Peritoneal dialysis-related infections recommendations: 2005 Update. Perit Dial Int 25:107-131, 2005 11. Bonadio TL, Diaz-Buxo JA: Definition of recurrent peritonitis. Perit Dial Int 27:716-719, 2007 12. Vas SI: Peritonitis during CAPD. A mixed bag. Perit Dial Bull 1:47-49, 1981 13. Keane WF, Alexander SR, Bailie GR, et al: Peritoneal dialysis-related peritonitis treatment recommendations: 1996 Update. Perit Dial Int 16:557-573, 1996 14. Flanigan MJ, Hochstetler LA, Langholdt D, Lim VS: Continuous ambulatory peritoneal dialysis catheter infections: Diagnosis and management. Perit Dial Int 14:248-254, 1994 15. National Committee for Clinical Laboratory Standards (NCCLS): Performance Standards for Antimicrobial Susceptibility Testing, 9th Informational Supplement. NCCLS document M100-S9. Villanova, PA, NCCLS, 1999 16. Szeto CC, Chow VC, Chow KM, et al: Enterobacteriaceae peritonitis complicating peritoneal dialysis: A review of 210 consecutive cases. Kidney Int 69:1245-1252, 2006 17. Szeto CC, Chow KM, Kwan BC, et al: Staphylococcus aureus peritonitis complicating peritoneal dialysis: Review of 245 consecutive cases. Clin J Am Soc Nephrol 2:245-251, 2007 18. Li PK, Ip M, Law MC, et al: Use of intra-peritoneal (IP) cefepime as monotherapy in the treatment of CAPD peritonitis. Perit Dial Int 20:232-234, 2000

710 19. Leung CB, Szeto CC, Chow KM, et al: Cefazolin plus ceftazidime versus imipenem/cilastatin monotherapy for treatment of CAPD peritonitis A randomized controlled trial. Perit Dial Int 24:440-446, 2004 20. Szeto CC, Chow KM, Leung CB, et al: Clinical course of peritonitis due to Pseudomonas species complicating peritoneal dialysis: A review of 104 cases. Kidney Int 59:2309-2315, 2001 21. Berg RD, Wommack E, Deitch EA: Immunosuppression and intestinal bacterial overgrowth synergistically promote bacterial translocation. Arch Surg 123:1359-1364, 1988 22. Szeto CC, Chow KM, Chung KY, Kwan BC, Leung CB, Li PK: The clinical course of peritoneal dialysis-related Szeto et al peritonitis caused by Corynebacterium species. Nephrol Dial Transplant 20:2793-2796, 2005 23. Szeto CC, Kwan BC, Chow KM, et al: Coagulase negative staphylococcal peritonitis in peritoneal dialysis patients: Review of 232 consecutive cases. Clin J Am Soc Nephrol 3:91-97, 2008 24. Zelenitsky S, Barns L, Findlay I, et al: Analysis of microbiological trends in peritoneal dialysis-related peritonitis from 1991 to 1998. Am J Kidney Dis 36:1009-1013, 2000 25. Pujol M, Peña C, Pallares R, Ayats J, Ariza J, Gudiol F: Risk factors for nosocomial bacteremia due to methicillinresistant Staphylococcus aureus. Eur J Clin Microbiol Infect Dis 13:96-102, 1994