;on behalf of the PDOPPS-Korea and Arbor Research InvestigatorsIntroduction
Peritoneal dialysis (PD)-related peritonitis is among the most serious complications of PD, significantly impacting patient morbidity and mortality [1,2]. Ensuring the timely and successful treatment of peritonitis is critical to improving patient outcomes and reducing its negative impact on long-term patient and technique survival [3].
Following the international expert proposal by the 2012 European Society of Clinical Microbiology and Infectious Diseases (ESCMID), multidrug-resistant organisms (MDROs) are defined as resistance to at least one agent in each of three or more antibiotic categories [4]. MDRO incidence is increasing not only in healthcare facilities but also within the community [5–7]. Particularly, peritonitis caused by MDROs presents a significant challenge to successful treatment [8]. Identifying the risk factors associated with infections caused by MDROs and appropriate antibiotic selection based on the specific organism is crucial for successful peritonitis treatment in patients at risk [9].
Globally, research on MDRO-peritonitis has been extremely limited, with only one single-center study conducted to date, which reported that the proportion of MDRO-peritonitis has remained high in recent years and that MDRO infections are more likely to cause worse clinical outcomes [8]. Further investigation into the factors associated with MDRO-peritonitis, causative organisms, and antibiotic susceptibility in individuals receiving PD is warranted. Here, we utilized Korean national data from the Peritoneal Dialysis Outcomes and Practice Patterns Study (PDOPPS)-Korea to 1) describe the incidence and clinical characteristics of MDRO-peritonitis, 2) examine its association with clinical factors, and 3) analyze various factors influencing peritonitis-related outcomes.
Methods
Study population and design
The PDOPPS is a multinational, prospective cohort observational study that aims to identify the relationship between clinicians’ practice patterns, patient and treatment characteristics, and PD patient outcomes, with the goal of improving patient outcomes. This study used Korean data from the PDOPPS Phase 2, conducted from June 2019 to December 2021. Overall, 20 medical institutions in South Korea participated in the PDOPPS, collecting data on both prevalent and incident PD patients. Prevalent PD patients were defined as individuals aged 18 years or older who were already receiving PD, either self-administered or provided in a care facility, and were randomly selected by each institution. Patients who were not currently undergoing PD—such as those temporarily receiving hemodialysis (HD) or those who had started but subsequently discontinued PD—were excluded from the pool of prevalent PD patients.
Incident PD patients were defined as individuals aged 18 years or older who initiated PD during the study period and performed at least one PD dialysate exchange at home or in a care facility within 60 days following discharge. Patients who had previously paused PD and later resumed, those who had temporarily received HD before PD initiation, patients who initiated PD due to acute kidney injury, and those undergoing hybrid therapy with combination PD and HD were excluded from the pool of incident PD patients.
We analyzed data from Korean participants in the PDOPPS cohort who experienced peritonitis. Peritonitis was diagnosed in accordance with the definition from the 2016 International Society for Peritoneal Dialysis (ISPD) peritonitis guideline [10]. The following exclusion criteria were applied: 1) dialysis effluent culture was not performed; 2) culture results of the dialysis effluent were negative; 3) cultured organism data were not available; 4) the cultured organism in the dialysis effluent was fungus; 5) the cultured organism in the dialysis effluent was Mycobacterium tuberculosis; and 6) antibiotic susceptibility results were not provided.
Peritonitis episodes were categorized into two groups based on the antibiotic resistance results of the cultured organisms: the MDRO group and the non-MDRO group. MDROs were defined according to the international expert proposal published by the 2012 ESCMID. This proposal clearly defined MDROs and provided specific antibiotic resistance criteria for common healthcare associated MDROs, including Staphylococcus aureus, Enterococcus species, Enterobacteriaceae (excluding Salmonella and Shigella), Pseudomonas species, and Acinetobacter species [4]. MDROs of coagulase-negative Staphylococcus (CoNS) and Streptococcus species, both Gram-positive bacteria, were defined based on the criteria used for S. aureus.
Clinical characteristics—including age, sex, body mass index, PD vintage, continuous ambulatory PD (CAPD) or automated PD (APD) modality, prior hospitalization, presence of diabetes mellitus (DM), management at a tertiary referral hospital, nature of peritonitis, exit-site care method, and relevant laboratory findings—were compared between the two groups. Peritonitis episodes were categorized by causative organisms, and for each organism, MDRO rates, empiric antibiotic resistance rates, antibiotic susceptibility results, and clinical outcomes (peritonitis cure rate and overall mortality) were presented.
Ethical approval
The study complied with the Declaration of Helsinki and received approval from the Medical Ethics Committee of each participating center of PDOPPS-Korea and Arbor Research. The Institutional Review Board approval numbers for the participating centers are as follows: Hallym University Kangnam Sacred Heart Hospital (HKS2019-07-005), Konyang University Hospital (KYUH 2019-07-013), Korea University Guro Hospital (2019GR0307), Pusan National University Hospital (H-1908-014-082), Seoul St. Mary’s Hospital (KC19OEDI0603), Yonsei University Hospital (4-2017-0817), Ajou University Hospital (AJIRB-MED-SUR-19-232), Wonju Severance Christian Hospital (CR3190634), Ewha Womans University Seoul Hospital (SEUMC2019-07-006), Inha University Hospital (INHAUH 2019-07-028), Chonnam National University Hospital (CNUH-2019-223), Chosun University Hospital (CHOSUN 2019-06-010), Soonchunhyang University Cheonan Hospital (SCHCA 2019-07-045), Cheju Halla General Hospital (2019-L04), Inje University Haeundae Paik Hospital (HPIRB2017-10-005), Nowon Eulji Medical Center (EMCS2019-06-016), Bongseng Memorial Hospital (BSIRB-2019-004), National Health Insurance Service Ilsan Hospital (NHIMC 2019-07-031), Daegu Fatima Hospital (DFH19OROO378), Hallym University Chuncheon Sacred Heart Hospital (CHUNCHEON 2019-07-002), Kyungpook National University Hospital (KNUH2018-06-012-004), and Seoul National University Hospital (H-1707-170-873). Written informed consent was obtained from all patients prior to their participation.
Data collection
All data were collected using uniform and standardized tools established across the PDOPPS program. Demographic and clinical data, including age, sex, height, weight, PD vintage, whether the patient was performing CAPD or APD, prior hospitalization, the presence of DM, and whether the episode was from a tertiary referral hospital, were collected. Blood test results from samples collected within 2 months prior to the peritonitis events were used in the analysis. Exit-site care methods were identified based on patient-reported surveys about their routine exit-site care practices. The white blood cell (WBC) count in the dialysis effluent was obtained from initial samples collected at the time of the peritonitis event. For each cultured organism isolated from the dialysis effluent, antibiotic resistance was identified using reported results, with intermediate resistance classified as resistant.
Definitions and outcomes
Because it has been reported as a risk factor for increased MDRO occurrence, prior hospitalization was defined as hospitalization within 90 days before the peritonitis event [11]. Relapsing, recurrent, and repeat peritonitis were defined per the ISPD guideline [10]. Empiric antibiotic resistance was identified as resistance of the cultured organism to the initial antibiotics administered before culture and susceptibility results became available. Peritonitis-related mortality was defined as death occurring within 50 days of the peritonitis event. Peritonitis cure was determined as the absence of peritonitis-related mortality, subsequent peritonitis event (relapse or recurrence), PD catheter removal, or HD transfer (defined as deemed permanent transfer to HD, or temporary transfer to HD with failure to return to PD within 84 days) [12]. Overall mortality was identified as death from any cause during the follow-up period. The primary outcome was peritonitis cure, with specific causes of treatment failure contributing to this composite outcome detailed separately. The secondary outcome was overall mortality.
Statistical analysis
All continuous variables are expressed as mean ± standard deviation. The Student t tests were used to compare continuous variables. Nominal variables are expressed as proportions. The chi-square test or Fisher exact test was used for the comparison of nominal variables, as appropriate. To evaluate the independent factors associated with the occurrence of MDRO-peritonitis and the cure of peritonitis, two separate mixed-effects logistic regression analyses were performed. Given that multiple peritonitis episodes could occur within the same individual, patient-specific random intercepts were incorporated into each model to account for intraindividual correlation, using patient identifiers (Patient IDs) to cluster repeated episodes belonging to the same subject. Each outcome variable was modelled as a binary response, and a logistic function was applied to model the probability of the event. Odds ratios and corresponding 95% confidence intervals were estimated for each independent variable. All statistical analyses were conducted using SPSS software version 24 (IBM Corp.) and SAS software version 9.4 (SAS Institute), with statistical significance defined as a two-sided p-value of <0.05.
Results
Assembly of the study cohort and baseline patient characteristics
Overall, 292 peritonitis episodes occurred in 186 patients. Of these, 145 episodes in 85 patients were excluded (based on the study’s exclusion criteria), leaving 147 episodes in 101 patients for the subsequent comparative analysis (Fig. 1). Among the total 292 peritonitis episodes, 41 were due to MDRO infections, accounting for approximately one in seven episodes (14.0%). Within the 147 episodes included in the analysis, these 41 episodes and 106 episodes were classified as the MDRO and non-MDRO groups, respectively.
Baseline characteristics between the MDRO and non-MDRO groups are shown in Table 1. There were no significant differences between the two groups in terms of age, sex, body mass index, PD vintage, and CAPD/APD modality. However, prior hospitalization (22.0% in the MDRO group vs. 1.9% in the non-MDRO group) and the proportion of patients with DM (61.0% in the MDRO group vs. 39.6% in the non-MDRO group) were significantly higher in the MDRO group than in the non-MDRO group. Additionally, the proportion of episodes from tertiary referral hospitals (29.3% in the MDRO group vs. 50.0% in the non-MDRO group) was significantly lower in the MDRO group than in the non-MDRO group. Although there were no significant differences in laboratory results between the two groups, hemoglobin levels tended to be lower in the MDRO group than in the non-MDRO group. Moreover, the proportion of recurrent peritonitis was significantly higher in the MDRO group than in the non-MDRO group (14.6% in the MDRO group vs. 0.9% in the non-MDRO group), whereas the rates of relapsing or repeat peritonitis did not differ between the groups. Additionally, there were no significant differences in the exit-site care methods or the WBC count of the dialysis effluent at the time of peritonitis onset between the two groups.
Distribution of causative organisms and their antibiotic resistance profiles
Fig. 2 presents the distribution of causative organisms, while Fig. 3 depicts the antibiotic resistance profiles by each causative organism category. Gram-positive organisms accounted for 114 episodes (77.6%), whereas Gram-negative organisms were identified in 33 episodes (22.4%). Among the causative organisms, Staphylococcus species were most frequently identified (36.1%, 53 episodes), including S. aureus (21.1%, 31 episodes) and CoNS (15.0%, 22 episodes), followed by Streptococcus species (30.6%, 45 episodes) and Enterobacteriaceae (14.3%, 21 episodes) (Fig. 2). Among S. aureus, CoNS, and Streptococcus species, the highest resistance rates were observed for penicillin, followed by erythromycin. Notably, 21.4% of S. aureus isolates were identified as methicillin-resistant S. aureus (MRSA) based on their resistance to oxacillin. Similarly, 57.9% of CoNS isolates were classified as methicillin-resistant CoNS (MR-CoNS) according to their resistance to oxacillin. Nonetheless, no vancomycin resistance was observed in any of the S. aureus, CoNS, or Streptococcus species. For Enterococcus species, erythromycin exhibited the highest resistance rate, with vancomycin resistance observed in 14.3% of episodes (Fig. 3A). In Enterobacteriaceae, ampicillin exhibited the highest resistance rate, and Serratia, Providencia, indole-positive Proteus, Citrobacter, and Enterobacter organisms were observed in 23.8% of cases. In Pseudomonas species, resistance to ciprofloxacin was most prevalent, while in Acinetobacter species, the highest resistance was observed for cefoxitin. Among carbapenem antibiotics, imipenem demonstrated a resistance rate of 6.7% in Enterobacteriaceae, while no resistance was observed in Pseudomonas species. Other categories, including Moraxella, Sphingomonas, and Brevibacterium species, demonstrated varying degrees of antibiotic resistance (Fig. 3B).
Comparison of empiric antibiotic resistance rates and clinical outcomes according to multidrug-resistant organism status across causative organisms
Tables 2 and 3 compare empiric antibiotic resistance rates and clinical outcomes between MDRO and non-MDRO groups for each causative organism, stratified by Gram-positive and Gram-negative organisms. Among Gram-positive organisms, empiric antibiotic resistance was significantly higher in the MDRO group for S. aureus (71.4% in the MDRO group vs. 0% in the non-MDRO group) and CoNS (70.0% in the MDRO group vs. 8.3% in the non-MDRO group) than in the non-MDRO group. Peritonitis cure rates and overall mortality did not differ significantly between the MDRO and non-MDRO groups for any of the Gram-positive organisms (Table 2). For Gram-negative organisms, there were no significant differences in empiric antibiotic resistance rates or clinical outcomes between the MDRO and non-MDRO groups (Table 3). However, when stratified by Gram stain, peritonitis cure rates were significantly lower for Gram-negative organisms than for Gram-positive organisms, regardless of MDRO status (89.5% [102/114] in Gram-positive organisms vs. 51.5% [17/33] in Gram-negative organisms, p < 0.001).
Multivariate logistic regression analyses for factors influencing multidrug-resistant organism-peritonitis occurrence and peritonitis cure
Multivariate logistic regression analyses were conducted to identify factors associated with MDRO-peritonitis occurrence and peritonitis cure, clustering individual peritonitis episodes by patient (Tables 4, 5). For MDRO-peritonitis occurrence, model 1 included variables that exhibited significant differences in baseline characteristics (prior hospitalization, DM, whether the episode was from a tertiary referral hospital, and recurrent peritonitis), along with clinically relevant factors (age, sex, PD vintage, relapsing and repeat peritonitis, and exit-site care method). Moreover, model 2 incorporated the same variables as those of model 1, with the addition of significant laboratory findings (hemoglobin and serum albumin levels). Across all models, prior hospitalization, DM presence, and recurrent peritonitis were consistently identified as significant factors for MDRO-peritonitis (Table 3).
For peritonitis cure, model 1 included variables significant in univariate analysis (management at a tertiary referral hospital, and whether the causative organism was Gram-negative) along with clinically relevant factors (age, sex, PD vintage, CAPD/APD modality, DM, MDRO status, and prior peritonitis history). Furthermore, model 2 added other laboratory findings (hemoglobin and serum albumin levels) to the variables included in model 1. PD vintage, management at a tertiary referral hospital, whether the causative organism was Gram-negative, and serum albumin level were found to be significant factors for peritonitis cure. However, MDRO status did not significantly affect peritonitis cure (Table 5).
Discussion
This study utilized nationally representative, multicenter cohort data from the PDOPPS-Korea to investigate factors influencing MDRO-peritonitis and peritonitis-related outcomes, and to describe antibiotic resistance patterns and treatment outcomes for individual causative organisms. Prior hospitalization, DM, and recurrent peritonitis were significant risk factors for MDRO-peritonitis. Nonetheless, MDRO status itself was not a significant factor influencing peritonitis cure; rather, the type of causative organism—specifically whether it was Gram-positive or Gram-negative—was a more critical factor.
Here, the MDRO group’s baseline characteristics demonstrated a significantly higher prevalence of prior hospitalization, DM, and recurrent peritonitis compared to those of the non-MDRO group, which is consistent with the theoretical background suggesting that recurrent infections and repeated exposure to healthcare facilities contribute to an increased risk of MDRO infections [11,13–15]. Interestingly, the proportion of episodes from tertiary referral hospitals in the MDRO group was significantly lower than that in non-tertiary referral hospitals. Specifically, the proportion of MDRO in peritonitis episodes was 18.5% (12/65) and 35.4% (29/82) in tertiary and non-tertiary referral hospitals, respectively. This difference could potentially be explained by stricter infection control measures in tertiary referral hospitals compared to non-tertiary referral hospitals [16]. Nevertheless, due to the multifactorial and complex nature of the factors involved, a definitive explanation cannot be provided. Among the causative organisms of peritonitis, Staphylococcus was the most common, accounting for 36.1% (31 episodes of S. aureus and 22 episodes of CoNS), followed by Streptococcus at 30.6% (45 episodes of Streptococcus species). These findings align relatively well with those of a previous single-center study in South Korea, which reported that Gram-positive organisms were the most frequent causative organisms of peritonitis between 2009 and 2015, with a prevalence of 53.8% [17]. This observation is also consistent with the widely recognized findings that Staphylococcus epidermidis, S. aureus, and various Streptococcus species are commonly reported in peritonitis episodes [18–20].
The primary objective of this study was to identify clinical risk factors for MDRO-peritonitis, with the aim of providing critical information that can aid in optimizing interventions for managing patients with peritonitis. Prior hospitalization, DM, and recurrent peritonitis were independent risk factors for MDRO-peritonitis. Specifically, DM likely contributes as a risk factor for MDRO infections due to its effect on causing immunocompromised status and increasing patients’ susceptibility to repeated infections [21,22]. Moreover, prior hospitalization and recurrent peritonitis were observed as independent risk factors for MDRO-peritonitis, presumably due to repeated infections and exposure to healthcare facilities, along with prior antibiotic exposure [11,14,15,20]. However, both prior hospitalization and recurrent peritonitis exhibited wide confidence intervals, likely a reflection of the small number of cases in the multivariate model, which warrants caution in interpreting the magnitude of their effects. Moreover, due to data limitations, the specific causes of prior hospitalization could not be analyzed, which limits our ability to fully characterize the heterogeneity of this risk factor. Although evidence supporting its effectiveness in peritonitis prevention remains limited, the ISPD guideline recommends the use of topical ointment for exit-site care [20]. Nonetheless, concerns have been raised regarding the potential for repeated exposure to topical antibiotics to contribute to the emergence of MDROs, leading to their infrequent use in Japan [12]. Interestingly, here, the use of topical ointment at the exit site was not associated with the occurrence of MDRO-peritonitis.
In our study, analysis of factors influencing peritonitis cure revealed that longer PD vintage and peritonitis due to Gram-negative organisms were significantly associated with lower cure rates, whereas higher serum albumin level and management at a tertiary referral hospital significantly correlated with improved outcomes. Infection with Gram-negative organisms was significantly associated with poorer peritonitis-related outcomes, consistent with prior findings from large-scale international prospective cohort data from PDOPPS Phase 1 [12]. To the best of our knowledge, only a single-center study conducted in China has specifically investigated MDRO-peritonitis [8], which reported a significantly higher treatment failure rate in the MDRO group than it did in the non-MDRO group (17.1% vs. 6.5% in the non-MDRO group, p < 0.001). Conversely, our study found that the type of causative organism—specifically whether it was Gram-negative or Gram-positive—was more strongly associated with peritonitis cure than MDRO status itself. This discrepancy is likely attributable to differences in the proportion of causative organisms: in our cohort, 80.5% (33/41) and 19.5% (8/41) of MDRO episodes were attributable to Gram-positive and Gram-negative organisms, respectively. This high proportion of Gram-positive MDROs in our study is consistent with findings from earlier Korean studies on PD-related peritonitis, which, although not specifically focused on MDROs, showed Gram-positive organisms to be two times more prevalent than Gram-negative organisms and exhibited lower antibiotic susceptibility rates (60%−70% in Gram-positive organisms vs. 80%−90% in Gram-negative organisms) [23,24]. In contrast, in the previous Chinese study, Gram-positive and Gram-negative organisms accounted for 24.7% (36/146) and 75.3% (110/146) of MDRO episodes, respectively. Gram-positive organisms are generally known to have more favorable outcomes in peritonitis compared to Gram-negative organisms [12,25]. Furthermore, although the frequently isolated MDRO organisms—S. aureus, CoNS, and Streptococcus species—exhibited high rates of resistance to empiric antibiotics, all isolates remained susceptible to vancomycin. This allowed for the timely administration of appropriate subsequent antibiotics in most cases, which may have contributed to favorable clinical outcomes (Supplementary Table 1, available online). However, despite the high prevalence of MRSA (21.4%) and MR-CoNS (57.9%) among S. aureus and CoNS, respectively, vancomycin was included in the initial empiric regimen in only 16.1% and 18.2% of these cases, respectively (Supplementary Table 1, available online). This difference highlights a potential gap in empiric antibiotic coverage and underscores the need to re-evaluate treatment strategies for PD-related peritonitis in South Korea, particularly in settings with a high prevalence of methicillin-resistant organisms.
Although MDRO status itself was not identified as a significant factor affecting peritonitis cure, additional multivariate Cox regression survival analysis revealed that resistance to empiric antibiotics was an independent risk factor for overall mortality (hazard ratio, 2.52; 95% confidence interval, 1.07–5.94) (Supplementary Fig. 1, available online). Additionally, empiric antibiotic resistance was significantly more frequent in the MDRO group than in the non-MDRO group (48.8% [20/41] vs. 21.1% [31/106], p < 0.001). Therefore, while MDRO status may not directly impact peritonitis cure, caution is warranted in the context of empiric antibiotic resistance. Furthermore, considering that the Gram-stain characteristics of causative organisms significantly affected peritonitis cure, these findings highlight the importance of incorporating both MDRO status and Gram-stain profiles when selecting empiric antibiotics. This study contributes to the field by providing detailed information on antibiotic resistance profiles of individual causative organisms, along with data on empiric and subsequent antibiotic use (Fig. 2; Supplementary Table 1, available online). By presenting both clinical risk factors for MDRO-peritonitis and organism-specific resistance data within a nationwide cohort, our findings may support clinicians in selecting more appropriate initial antibiotic regimens for patients with PD-related peritonitis.
This study, utilizing PDOPPS Phase 2 data, has the limitation of a relatively short observation period of 30 months. Specifically, the incidence of peritonitis has declined over time due to advancements in PD techniques. Recent analyses of various national registries report an average peritonitis incidence of 0.303 episodes per patient-year in 2019, showing a relatively low incidence [26]. Moreover, we excluded episodes with fungal peritonitis (four episodes) and M. tuberculosis peritonitis (two episodes) as their highly heterogeneous antibiotic resistance patterns make it inappropriate to define MDRO in the same way as for bacterial peritonitis caused by usual pathogens. Additionally, a total of 85 culture-negative episodes (29.1%) were excluded from the analysis, which was higher than the 15% rate recommended by the ISPD guideline [20]. Specifically, 145 out of 292 total peritonitis episodes were excluded in accordance with the study’s exclusion criteria, thereby reducing the sample size for comparative analysis and warranting cautious interpretation of the results due to the potential risk of selection bias. Although no significant differences in baseline characteristics were observed between the 147 included episodes and the 145 excluded episodes (Supplementary Table 2, available online), the possibility of bias arising from differences in microbiological characteristics (such as the culture-negative rate) cannot be ruled out. Nevertheless, this study holds significant value as it represents the first multicenter, prospective study on MDRO-peritonitis, utilizing consistent and reliable data collected from 20 medical institutions across South Korea through participation in PDOPPS Phase 2. PDOPPS Phase 3 is currently in the preparation phase, with plans to collect follow-up data on existing patients and register new patients, and data collection is scheduled beyond 2026. This extended observation period is expected to yield more objective and reliable findings in future research.
In conclusion, this multicenter prospective observational study, based on high-quality and reliable data from PDOPPS-Korea, identified prior hospitalization, DM, and recurrent peritonitis as significant risk factors for MDRO-peritonitis. Furthermore, the provision of organism-specific data—including antibiotic resistance profiles, peritonitis-related outcomes, and the antibiotics used (both empiric and subsequent)—adds substantial clinical value to our findings. Specifically, we found that even among MDROs, most Gram-positive organisms—mainly Staphylococcus and Streptococcus species—remained susceptible to vancomycin, and most Gram-negative organisms remained susceptible to carbapenems. This detailed information, along with the identified risk factors for MDROs, is expected to help clinicians make more informed and timely decisions regarding the initial management of PD-related peritonitis.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement table 1
Print
