Description of the condition
Peritoneal dialysis (PD) is an effective form of renal replacement therapy for people with advanced kidney disease. However, peritonitis continues to represent a significant complication of PD (Voinescu 2002) despite the introduction of effective prevention strategies such as disconnect and double bag systems (Bazzato 1980; Monteon 1998; Strippoli 2004). The reported incidence of peritonitis episodes varies from one in nine patient-months to one in 53 patient-months (Grunberg 2005; Kawaguchi 1999). Risk factors for peritonitis include diabetes mellitus (Oxton 1994), race (Juergensen 2002; Lim 2005), obesity (McDonald 2004), temperate climates (Alves 1993; Szeto 2003), and depression (Troidle 2003). In addition, some studies have shown that PD modality may influence peritonitis rates, although other studies have not confirmed this (Huang 2001; Oo 2005).
PD-associated peritonitis results in significant morbidity, and in some cases, mortality. Catheter removal becomes necessary in cases not responding to antibiotic therapy. This may be temporary and followed by a return to PD, or permanent, resulting in technique failure. Ultrafiltration failure can occur both acutely due to increases in capillary permeability (Ates 2000; Smit 2004) and in the longer term result in technique failure (Coles 2000; Davies 1996). In many countries, peritonitis is a leading cause of permanent transfer to haemodialysis. Peritonitis is prevalent among patients with encapsulating sclerosing peritonitis and may be a causal factor (Kawanishi 2005; Rigby 1998). In some patient groups peritonitis is thought to increase overall mortality rates (Fried 1996). It is estimated that PD-associated peritonitis results in death in 6% of affected patients (Troidle 2006).
Description of the intervention
Early and effective management of peritonitis is important to reduce the risk of adverse outcomes such as catheter removal (Choi 2004; Heaf 2004). The mainstay of treatment is antimicrobial therapy, although adjunctive therapies have been employed including the use of fibrinolytic agents (Innes 1994; Pickering 1989), peritoneal lavage (Ejlersen 1991) and routine early catheter removal.
How the intervention might work
Current guidelines recommend the use of antibiotics which cover gram positive and gram negative organisms in cases of peritonitis (CARI 2005; Piraino 2005). However, several questions about the optimal treatment of PD-associated peritonitis remain unanswered, particularly with respect to choice, route of administration (Passadakis 2001) and duration of antimicrobial therapy. Many treatment regimens are based on continuous ambulatory PD (CAPD) and their applicability to automated PD (APD) is untested (Fielding 2002). The optimal total duration of antimicrobial therapy, and the duration of systemic (intraperitoneal (IP) or intravenous (IV)) treatment is also unclear, as are the roles of peritoneal lavage and urokinase. The majority of studies performed have focused on the outcomes of empirical antibiotic therapy, with little consideration of treatment initiated once organism identification and sensitivities are available.
Why it is important to do this review
To address existing uncertainties, we performed a systematic review of randomised controlled trials (RCT) evidence examining the effectiveness of different treatment options for PD-associated peritonitis.
To evaluate the benefits and harms of treatments for PD-associated peritonitis.
Criteria for considering studies for this review
Types of studies
All RCTs and quasi-RCTs (RCTs in which allocation to treatment was obtained by alternation, use of alternate medical records, date of birth or other predictable methods) on the effect of any interventions, including anti-infective agents, fibrinolytic agents, peritoneal lavage and early catheter removal, for the treatment of peritonitis in PD patients were included.
Types of participants
Adult and paediatric patients who were receiving PD (CAPD or APD) and developed PD-associated peritonitis.
Types of interventions
Studies looking at the use of any antimicrobial agent, fibrinolytic agent, peritoneal lavage, IP immunoglobulin or early catheter removal were included. Interventions could be tested directly against each other or compared to placebo/no treatment. The following could be included:
- Studies of the same antibiotic agent administered by different routes (e.g. IP versus oral, IP versus IV).
- Studies comparing the same antibiotic agent administered at different doses.
- Studies comparing different schedules of administration of antimicrobial agents (in particular regimens involving single daily dosing versus more than one daily dose).
- Comparisons of different regimens of antimicrobial agents.
- Studies comparing different treatment durations with the same antimicrobial agents.
- Studies comparing any other intervention including fibrinolytic agents, peritoneal lavage, IP immunoglobulin administration, and early catheter removal.
Types of outcome measures
- Primary peritonitis treatment failure (failure to achieve a clinical response, defined as resolution of symptoms and signs, by day 4 to 6)
- Complete cure (clinical or microbiological improvement or both with no subsequent relapse)
- Peritonitis relapse (reoccurrence of peritonitis due to the same organism with the same antibiotic sensitivities within 28 days of completing treatment)
- Death due to peritonitis (all-cause mortality data were also collected)
- Toxicity of antibiotic treatments (ototoxicity, decline in residual kidney function, rash, nausea and vomiting, convulsions, other).
- Time to peritonitis relapse
- Need to change antibiotic following culture results
- Catheter removal or replacement or both
- Hospitalisation (duration of hospital stay) and hospitalisation rate (number of patients hospitalised)
- Technique failure (transfer from PD to haemodialysis or transplantation due to peritonitis).
Search methods for identification of studies
For this update we searched the Cochrane Renal Group's Specialised Register and EMBASE to 5 March 2014 without language restriction.
The Cochrane Renal Group’s Specialised Register contains studies identified from:
- Monthly searches of the Cochrane Central Register of Controlled Trials (CENTRAL)
- Weekly searches of MEDLINE OVID SP
- Handsearching of renal-related journals and the proceedings of major renal conferences
- Searching of the current year of EMBASE OVID SP
- Weekly current awareness alerts for selected renal journals
- Searches of the International Clinical Trials Register (ICTRP) Search Portal and ClinicalTrials.gov.
Studies contained in the Specialised Register are identified through search strategies for CENTRAL, MEDLINE, and EMBASE based on the scope of the Cochrane Renal Group. Details of these strategies, as well as a list of handsearched journals, conference proceedings and current awareness alerts, are available in the Specialised Register section of information about the Cochrane Renal Group.
Please refer to our review published in 2008 for the original search strategies used (Wiggins 2008).
Searching other resources
- Reference lists of clinical practice guidelines, review articles and relevant studies.
- Letters seeking information about unpublished or incomplete studies to investigators known to be involved in previous studies.
Data collection and analysis
Selection of studies
The original review was undertaken by four authors in 2008 and seven authors in 2014. The search strategies described were used to obtain titles and abstracts of studies that might be relevant to the review. The titles and abstracts were screened independently by multiple authors, who discarded studies that were not eligible based on the inclusion criteria for this review; however studies and reviews that might include relevant data or information on additional published or unpublished studies were retained initially and their full-text version was analysed.
Data extraction and management
Four authors independently assessed the retrieved abstracts, and if necessary, the full text of these studies to determine eligibility. Data extraction was carried out independently by the same authors using standard data extraction forms. Studies reported in non-English language journals were translated before assessment. Where more than one publication of one study existed, reports were grouped together and the publication with the most complete data was used in the analyses. Where relevant outcomes were only published in earlier versions these data were used Disagreements were resolved in consultation among authors.
Assessment of risk of bias in included studies
- Was there adequate sequence generation (selection bias)?
- Was allocation adequately concealed (selection bias)?
- Was knowledge of the allocated interventions adequately prevented during the study (detection bias)?
- Participants and personnel
- Outcome assessors
- Were incomplete outcome data adequately addressed (attrition bias)?
- Are reports of the study free of suggestion of selective outcome reporting (reporting bias)?
- Was the study apparently free of other problems that could put it at a risk of bias?
Measures of treatment effect
Results were expressed as risk ratio (RR) with 95% confidence intervals (CI) for all categorical outcomes of the individual studies.
Dealing with missing data
Any further information or clarification required from the authors was requested by written or electronic correspondence and relevant information obtained in this manner was included in the review. Evaluation of important numerical data such as screened, randomised patients as well as intention-to-treat (ITT), as-treated and per-protocol (PP) population was performed. Attrition rates, for example drop-outs, losses to follow-up and withdrawals were investigated. Issues of missing data and imputation methods (for example, last-observation-carried-forward (LOCF)) were critically appraised (Higgins 2011).
Assessment of heterogeneity
Heterogeneity was analysed using a Chi² test on N-1 degrees of freedom, with an alpha of 0.05 used for statistical significance and with the I² statistic (Higgins 2003). I² values of 25%, 50% and 75% correspond to low, medium and high levels of heterogeneity.
Assessment of reporting biases
It was planned that if sufficient RCTs were identified an attempt would be made to assess funnel plot asymmetry due to small study effect, as this may be indicative of publication bias (Egger 1997). There were however too few included studies to construct meaningful funnel plots.
Treatment effects were summarised using a random-effects model. For each analysis, the fixed-effect model was also evaluated to ensure robustness of the model chosen and susceptibility to outliers. Where continuous scales of measurement were used to assess the effects of treatment (time to peritonitis relapse, days of hospitalisation, measures of residual kidney function) the mean difference (MD) was used.
Subgroup analysis and investigation of heterogeneity
Subgroup analysis was planned to explore how possible sources of heterogeneity (paediatric versus adult population, age, gender, cause of end-stage kidney disease, body mass index, diabetes mellitus, duration of dialysis, PD modality (CAPD versus APD), previous peritonitis episodes, type of dialysate and micro-organism isolated) might influence treatment effect.
We aimed to perform sensitivity analyses to explore the influence of the following factors on effect size:
- repeating the analysis excluding unpublished studies
- repeating the analysis taking account of risk of bias
- repeating the analysis excluding any very long or large studies to establish how much they dominate the results
- repeating the analysis excluding studies using the following filters: diagnostic criteria, language of publication, source of funding (industry versus other), country.
Data were however insufficient to perform these analyses.
Description of studies
Results of the search
The literature search undertaken for Wiggins 2008 retrieved 1684 reports of which 1617 were excluded. Analysis of the remainder identified 36 studies (2089 participants, 2480 peritonitis episodes) published in 42 reports. For this update the Cochrane Renal Group's Specialised Register identified 62 reports of potential studies and two ongoing studies. After full text review a further six eligible studies (344 participants, 533 peritonitis episodes) were identified taking the total number of unique RCTs to 42 (58 reports; 2433 participants, 3013 episodes of peritonitis). Search results are summarised in Figure 1.
|Figure 1. Study flow diagram.|
We identified 36 studies (1949 patients) that considered the use of antimicrobial agents. There were 14 studies that compared different routes of antibiotic administration - IP versus IV (3 studies, 156 participants: Bailie 1987; Bennett-Jones 1987; Vargemezis 1989) and IP versus oral (11 studies, 601 participants: Bennett-Jones 1990; Boeschoten 1985; Chan 1990; Cheng 1991; Cheng 1993; Cheng 1997; Cheng 1998; Gucek 1994; Lye 1993; Raman 1985; Tapson 1990).
Different IP antibiotic classes or combinations or both were tested head-to-head in 17 studies (Bowley 1988; de Fijter 2001; Drinovec 1988; Flanigan 1991; Friedland 1990; Gucek 1997; Hernandez 2004 Jiménez 1996; Khairullah 2002; Klaus 1995a; Leung 2004; Lui 2005; Lupo 1997; Merchant 1992; Wale 1992; Were 1992; Wong 2001). These included three studies (234 participants) that compared glycopeptides to first generation cephalosporins (Flanigan 1991; Khairullah 2002; Lupo 1997); and five studies (421 participants) that compared intermittent and continuous IP antibiotic dosing (Boyce 1988; Choy 2001; Lye 1995; Klaus 1995a; Velasquez-Jones 1995).
There were six studies that investigated adjunctive therapies: urokinase versus placebo (Gadallah 2000c; Innes 1994; Tong 2005a); catheter removal or replacement or both (Williams 1989), peritoneal lavage (Ejlersen 1991), and IP immunoglobulin (Coban 2004). Data for automated PD were absent.
We excluded 49 studies (67 reports). Reasons for exclusion were: not RCTs (19); not appropriate population (24); not peritonitis treatment (2); not appropriate outcomes (4).
Risk of bias in included studies
We have summarised the risk of bias of included studies in Figure 2 and Figure 3. In general, risks of bias were in high in most studies. Overall, both blinding and selective reporting were assessed at high risk of bias. Suboptimal reporting in many studies meant that risks of bias were assessed as unknown.
|Figure 2. Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies|
|Figure 3. Risk of bias summary: review authors' judgements about each risk of bias item for each included study|
Random generation sequence was assessed as low risk in four studies (10%) (Chan 1990; Cheng 1991; Lui 2005; Tapson 1990); allocation concealment was adequate in 7/42 (17%) studies (Cheng 1991; Cheng 1998; Friedland 1990; Klaus 1995a; Lui 2005; Tapson 1990; Wong 2001).
Incomplete outcome data
Percentages of participants lost to follow-up ranged from 0% to 64.5%.
Effects of interventions
There were no significant differences in the results of analyses performed using random and fixed-effects models. The results presented therefore refer to those obtained using a random-effects model. Subgroup analyses and evaluations for bias from small-study effects were not performed because the small numbers of participants and studies made the power of these analyses too small to assess. Few data were available on the pharmacokinetics of many commonly used antibiotics when administered intraperitoneally.
Intravenous (IV) versus intraperitoneal (IP) antimicrobial agents
Bennett-Jones 1987 reported a statistically significant increase in the primary treatment failure rate for IV versus IP vancomycin and tobramycin ( Analysis 1.1.2 (75 participants): RR 3.52, 95% CI 1.26 to 9.81). It is noteworthy that in the study by Bailie 1987, in which IP versus IV administration of a loading dose of vancomycin followed by an IP maintenance dose were compared, there were no primary treatment failures reported in either group. Vargemezis 1989 also looked at IP versus IV administration of vancomycin; however, reporting limitations made it difficult to interpret meaningful results.
Bailie 1987 reported no significant differences in the incidence of rash ( Analysis 1.2.1 (20 participants): RR 5.00, 95% CI 0.27 to 92.62) or infusion pain ( Analysis 1.3 (20 participants): RR 3.00, 95% CI 0.14 to 65.90) between IV and IP vancomycin. Bennett-Jones 1987 reported no significant difference in hypotension ( Analysis 1.2.2 (76 participants): RR 5.26, 95% CI 0.26 to 106.11) between IV and IP vancomycin and tobramycin.
Oral versus IP administration of the same antimicrobial agent
Oral administration of quinolone antibiotics (ciprofloxacin, ofloxacin) had uncertain effects on primary treatment failure compared to IP administration ( Analysis 2.1 (2 studies, 83 participants): RR 1.34, 95% CI 0.71 to 2.56; P = 0.37; I²= 0%) and relapse ( Analysis 2.2 (2 studies, 83 participants): RR 3.38, 95% CI 0.74 to 15.35; P = 0.11; I² = 0%). Assessment of low-quality evidence indicated that IP quinolone therapy may increase complete cure ( Analysis 2.3 (2 studies, 83 participants): RR 1.66, 95% CI 0.98 to 2.83; P = 0.06; I² = 0%) compared with oral treatment, although therapy failure rates were high in both arms of these studies (52.4% and 31.7% in the oral and IP groups, respectively).
Cheng 1993 reported no significant differences in catheter removal rates ( Analysis 2.4 (48 participants): RR 2.00, 95% CI 0.19 to 20.61), hospitalisation rates ( Analysis 2.5 (48 participants): RR 1.00, 95% CI 0.51 to 1.95), or nausea and vomiting ( Analysis 2.6 (48 participants): RR 0.50, 95% CI 0.05 to 5.15) between oral and IP cephalosporin (cephradine) therapy.
Oral versus IP administration of different antimicrobial agents
Comparison of oral versus antibiotic regimens had uncertain effects on risks of failure to achieve complete cure ( Analysis 3.1 (8 studies, 510 participants): RR 1.06, 95% CI 0.80 to 1.40; P = 0.69; I² = 0%). Subgroup analysis showed this was similar for oral quinolones versus IP aminoglycoside/glycopeptide combinations ( Analysis 3.1.1 (5 studies, 304 participants): RR 1.19, 95% CI 0.83 to 1.72), oral quinolones versus IP cephalosporins ( Analysis 3.1.2 (2 studies, 148 participants): RR 1.00, 95% CI 0.55 to 1.81), and oral cephradine versus IP cefuroxime ( Analysis 3.1.3 (58 participants): RR 0.77, 95% CI 0.40 to 1.46). Similarly, primary treatment failure ( Analysis 3.2 (7 studies, 472 participants): RR 1.04, 95% CI 0.64 to 2.15; P = 0.86; I² = 0%), relapse ( Analysis 3.3(5 studies, 304 participants): RR 1.17, 95% CI 0.64 to 2.15; P = 0.61; I² = 2%), catheter removal ( Analysis 3.4 (2 studies, 170 participants): RR 1.18, 95% CI 0.49 to 2.87; P = 0.71; I² = 0%), hospitalisation rate ( Analysis 3.5 (1 study, 45 participants): RR 0.70, 95% CI 0.30 to 1.63), all-cause mortality ( Analysis 3.6 (1 study, 46 participants): RR 0.36, 95% CI 0.02 to 8.46), and microbiological eradication (not defined by investigators) ( Analysis 3.7 (1 study, 39 participants): RR 1.26, 95% CI 0.46 to 3.46) were equivalent in both groups. There was an increased risk of nausea and vomiting with oral antibiotics compared to IP antibiotics ( Analysis 3.8.1 (3 studies, 158 participants): RR 9.91, 95% CI 1.89 to 51.99; P = 0.007; I² = 0%).
Oral versus IP administration of the same or different antimicrobial agent(s)
When all studies that compared oral versus IP administration of an antimicrobial agent were combined, treatment failure rates of oral versus IP administration were similar without evidence for between-trial heterogeneity ( Analysis 4.1 (9 studies, 555 participants): RR 1.12, 95% CI 0.79 to 1.60; P = 0.52; I² = 0%).
Low versus high dose antibiotic
Merchant 1992 reported low dose imipenem (total 1 g IP daily) was associated with a significant increase in failure to achieve complete cure ( Analysis 5.1 (30 participants): RR 4.38, 95% CI 1.27 to 15.06) and the number relapsing ( Analysis 5.2 (28 participants): RR 12.00, 95% CI 1.60 to 90.23) compared with high dose imipenem (total 2 g IP daily). High dose imipenem had an uncertain effect on seizures ( Analysis 5.3 (30 participants): RR 0.60, 95% CI 0.03 to 11.23). However this study was not powered to detect seizures and the protocol was changed mid-study from high dose to low dose imipenem because two participants in the imipenem group had seizures.
Intermittent versus continuous IP antimicrobial agents
The effects of intermittent compared with continuous dosing on complete cure ( Analysis 6.1 (4 studies, 338 participants): RR 0.92, 95% CI 0.64 to 1.33; P = 0.65; I² = 0%), primary treatment failure ( Analysis 6.2 (5 studies, 522 participants): RR 1.11, 95% CI 0.77 to 1.62; P = 0.57; I² = 0%) and risk of relapse ( Analysis 6.3 (4 studies, 338 participants): RR 0.76, 95% CI 0.45 to 1.28; P = 0.31; I² = 0%) were uncertain. Choy 2001 reported no significant difference in catheter removal rates between groups ( Analysis 6.4 (20 participants): 0.98, 95% CI 0.43 to 2.24). The only side-effect evaluated was vancomycin-induced rash (Boyce 1988); effects of continuous compared with intermittent dosing were uncertain ( Analysis 6.5 (51 participants): RR 0.70, 95% CI 0.05 to 10.57).
First generation cephalosporin versus glycopeptide-based regimens
Achievement of complete cure was significantly more likely with a glycopeptide-based regimen than one based on cephalosporins ( Analysis 7.1 (3 studies, 370 participants): RR 1.66, 95% CI 1.01 to 2.72; P = 0.04; I² = 41%). This was true for both vancomycin ( Analysis 7.1.1 (2 studies, 305 participants): RR 1.51, 95% CI 1.03 to 2.22; P = 0.26; I² = 20%) and teicoplanin-based regimens ( Analysis 7.1.2 (1 study, 65 participants): RR 9.65, 95% CI 1.04 to 20.58). Despite the overall advantage of glycopeptides on complete cure, effects on primary treatment failure ( Analysis 6.2 (2 studies, 305 participants): RR 1.14, 95% CI 0.69 to 1.87; P = 0.38; I² = 0%), relapse ( Analysis 7.3 (3 studies, 350 participants): RR 1.68, 95% CI 0.84 to 3.36; P = 0.14; I² = 0%), catheter removal ( Analysis 7.4 (2 studies, 305 participants): RR 0.95, 95% CI 0.41 to 2.19; P = 0.90; I² = 52%) and microbiological eradication ( Analysis 7.5 (1 study, 45 participants): RR 0.83, 95% CI 0.62 to 1.13) were uncertain likely due to the lack of power within the meta-analysis.
It is noteworthy that these results were largely influenced by Flanigan 1991 in which the cephazolin dose used was 50 mg/L, which is below the dose currently recommended in the International Society for Peritoneal Dialysis (ISPD) guidelines of 125 mg/L (Li 2010). In contrast, Khairullah 2002 found no difference in cure rates for vancomycin and cephazolin (50% and 40% complete cure for glycopeptides and cephalosporins respectively) when a higher cephalosporin dose was used.
Teicoplanin versus vancomycin-based IP antibiotic regimens
Primary treatment failure was less likely with teicoplanin than vancomycin ( Analysis 8.1 (2 studies, 178 participants): RR 0.36, 95% CI 0.13 to 0.96; P = 0.04), however, effects on complete cure were uncertain ( Analysis 8.2 (2 studies, 178 participants): RR 0.67, 95% CI 0.40 to 1.15; P = 0.14; I² = 0%). The risk of relapse rates was also similar for both agents ( Analysis 8.3 (2 studies, 178 participants): RR 1.01, 95% CI 0.49 to 2.11; P = 0.97; I² = 0%). There was no significant heterogeneity associated with either outcome.
Different regimens of oral antibiotics
Effects of oral rifampicin and ofloxacin (regimen 2) compared with oral ofloxacin alone (regimen 1) (Chan 1990) in achieving a complete cure ( Analysis 9.1 (74 participants): RR 0.88, 95% CI 0.35 to 2.17) and catheter removal ( Analysis 9.2 (74 participants): RR 2.00, 95% CI 0.19 to 21.11) were uncertain. Chan 1990 also reported there was no differences in the need to change antibiotics following culture results ( Analysis 9.3 (74 participants): RR 0.33, 95% CI 0.04 to 3.06), nausea and vomiting ( Analysis 9.4.1 (74 participants): RR 3.00, 95% CI 0.13 to 71.34) and rash ( Analysis 9.4.2 (74 participants): RR 3.00, 95% CI 0.13 to 71.34) between the two regimens.
Fibrinolytic agents versus non-urokinase or placebo
Studies of IP urokinase found uncertain effects for benefit of urokinase versus placebo on complete cure in persistent peritonitis ( Analysis 10.1 (88 participants): RR 1.23, 95% CI 0.84 to 1.79), or primary response to treatment in the setting of resistant peritonitis ( Analysis 10.2 (2 studies, 99 participants): RR 0.63, 95% CI 0.32 to 1.26; P = 0.19; I² = 33%). Resistant CAPD peritonitis was defined as either persistent infection or recurrent infection (Innes 1994) or as persistence of symptoms and signs of peritonitis together with turbid peritoneal dialysate 48 hours after the initiation of antibiotic treatment (Tong 2005a). Persistent infection was defined as no resolution of peritonitis within four days of treatment with antibiotics active against the bacteria isolated (Innes 1994). Relapse and catheter removal were uncertain with urokinase, either in the setting of persistent peritonitis ( Analysis 10.3.1; Analysis 10.4.1) or initiation of fibrinolytic therapy at the time peritonitis was diagnosed ( Analysis 10.3.2; Analysis 10.4.2). Tong 2005a reported no significant difference in all-cause mortality ( Analysis 10.5).
Urokinase versus simultaneous catheter removal and replacement
In Williams 1989, a study of participants presenting with a second recurrence of peritonitis, simultaneous catheter removal and replacement was better than urokinase in reducing recurrent episodes of peritonitis ( Analysis 11.1 (37 participants): RR 2.35, 95% CI 1.13 to 4.91).
Ejlersen 1991 investigated the effects of peritoneal lavage (2 L exchanges during the initial 24 hours, no dwell time with dialysis fluid (60 L) containing vancomycin 20 mg/L and netilmicin) plus nine further days antibiotic treatment compared with continued prolonged exchanges (2 rapid exchanges containing vancomycin (40 mg/L) followed by routine CAPD schedule including antibiotics for 10 days. This study reported no significant differences between lavage and usual care for complete cure ( Analysis 12.1 (36 participants): RR 2.50, 95% CI 0.56 to 11.25), relapse of peritonitis with subsequent laparotomy and colostomy ( Analysis 12.2 (36 participants): RR 2.50, 95% CI 0.56 to 11.25), technical failure ( Analysis 12.3 (36 participants): RR 3.00, 95% CI 0.13 to 69.09), or adverse events ( Analysis 12.4 (36 participants): 3.00, 95% CI 0.13 to 69.09).
Coban 2004 reported the use of IP immunoglobulin was associated with a statistically significant reduction in numbers of exchanges executed for the dialysate white cell count to fall below 100/mL ( Analysis 13.1 (24 participants): MD -7.30 exchanges, 95% CI -8.12 to -6.48). There were no treatment failures and no relapses in any participants in this study.
Of the 12 studies in which different regimens of IP antibiotics were compared head-to-head, the only statistically significant outcome was reported by de Fijter 2001: rifampicin/ciprofloxacin was better than cephradine in reducing treatment failure ( Analysis 14.2.7 (98 participants) RR 0.50, 95% CI 0.28 to 0.89).
Duration of antibiotic treatment
Altmann 1984 compared duration of antibiotic treatment (vancomycin and gentamicin for 10 versus 21 days) and reported uncertain effects on risk of relapse ( Analysis 15.1 (49 participants): RR 1.56, 95% CI 0.61 to 3.95). It is noteworthy that five participants, all of whom received more than 21 days of gentamicin, developed clinical evidence of vestibular damage versus no patients who received a 10-day course of treatment.
It was not possible to tabulate a summary of findings because there were insufficient studies for each analysis.
Summary of main results
This review found that in generally low-quality evidence, IP antibiotic therapy may lower risks of primary treatment failure compared with IV antibiotics in two small studies.
The benefits of intermittent dosing of some antibiotics (vancomycin, gentamicin, ceftazidime and teicoplanin) compared with continuous therapy in the treatment of peritonitis, and of IP versus oral antibiotics, are uncertain.
In a single small study of participants presenting with a second recurrence of peritonitis, simultaneous catheter removal and replacement was superior to urokinase to reduce risks of relapsing and remitting PD-associated peritonitis. Insufficient data were available to determine if specific antibiotic classes are most effective for reducing treatment failure and relapse, although glycopeptides may improve complete cure rates compared with first generation cephalosporins.
It was unclear if peritoneal lavage improves response to concomitant antimicrobial therapy.
We also found that IP immunoglobulin administration decreased the time for the dialysate inflammatory cell count to fall, but effects on patient-important outcomes, such as treatment failure or risk of relapse, are uncertain. Longer duration of antibiotics had unclear effects on risk of relapse compared with shorter treatment courses, and may increase adverse events.
Oral antibiotics were associated with increased risk of nausea and vomiting compared with IP administration.
Data for automated PD were scant.
Overall completeness and applicability of evidence
Our review revealed a significant paucity of evidence underlying many widely-used and accepted clinical practices in the treatment of peritonitis, a condition that is associated with significant patient morbidity and mortality. Consequently, some aspects of treatment are uncertain, such as duration of antimicrobial therapy and optimal timing of catheter removal. While valuable information was gained from this review, the few available studies at generally high risk of bias resulted in a lack of evidence in many important areas of clinical practice. Studies tended to focus on choice and route of antibiotic without consideration of other variables such as total duration of therapy, drug dose and the role of patient factors, such as comorbidities and residual kidney function. No RCTs have been conducted to determine if early catheter removal is beneficial in patients not responding to therapy. The follow-up period of most studies was 28 days or fewer; therefore, long-term outcomes, such as technique failure and mortality, were not evaluated. Loss of residual kidney function during peritonitis may be accelerated by aminoglycoside therapy (Baker 2003; Shemin 2000) although this has been refuted by a recent study (Badve 2012). As a result of these factors, there is insufficient evidence regarding several aspects of management that are clinically important and this makes the provision of definitive treatment guidelines difficult at the present time.
Quality of the evidence
The risk of bias of included studies was generally moderate-to-high. In particular, inadequate randomisation and concealment methods were common. Definitions of peritonitis, successful treatment, and relapse varied among studies, thereby reducing their comparability. Many studies were conducted in single-centre settings with small patient numbers, and were underpowered to detect short-term (treatment failure and catheter removal), medium-term (relapse and recurrence) or long-term (mortality and technique failure) effects. Similarly, studies did not systematically evaluate adverse events. Hence there was significant potential for type II statistical errors (finding no treatment effect when a treatment effect exists) in most of our analyses. Studies often predated the current era of lower peritonitis rates, newer antibiotic therapies and increased awareness of multiresistant organisms, thereby potentially reducing the applicability of our meta-analyses.
A significant issue was that there was marked heterogeneity among studies of outcome definitions. Treatment failure was variably measured by resolution of symptoms and signs, clearing of dialysate, fall in dialysate white cell count and microbiological eradication of the causative organism. The time frame in which these changes were required to occur also varied, ranging from 48 hours to 28 days. Similarly there was a large degree of variation in the time elapsed after a primary peritonitis episode for a second peritonitis episode to be considered as a relapse (Li 2010).
An additional problem was interaction of endpoints. For example, primary treatment failure often necessitates catheter removal, which is an endpoint in itself. Some studies defined treatment failure as a need to change the antimicrobial agent or catheter removal. In contrast, other studies defined primary failure as ongoing symptoms beyond 48 hours of antibiotic therapy, with catheter removal evaluated as a separate outcome. These factors reduced the comparability of studies.
Potential biases in the review process
While the review was completed according to standardised Cochrane methodology, potential biases in the review process should be considered. We completed a formal search designed by a specialist information manager including data from a wide range of sources, including handsearching, to limit the potential for omission of potentially relevant studies, although there is a possibility that relevant studies may have not been included. There were insufficient studies included in most meta-analyses to enable us to examine for potential publication bias. Most studies did not systematically report all relevant patient-centred outcomes, suggesting that data for mortality and other clinical outcomes were incomplete, and summary estimates were potentially unreliable.
Agreements and disagreements with other studies or reviews
As far as we are aware, this remains the only published systematic review of RCTs of all PD-associated peritonitis treatment. A review of antimicrobial treatment of PD-associated peritonitis published in 1991 concluded that the optimal empirical treatment was weekly vancomycin in combination with ceftazidime (Millikin 1991). However, this review predated many of the studies included in this study, and was not confined to RCTs.
The mainstay of peritonitis treatment is timely administration of empirical antimicrobial agents that are likely to eradicate the most common causative agents. This is endorsed by guidelines of the International Society of Peritoneal Dialysis (ISPD) (Li 2010) and the Australian and New Zealand Society of Nephrology (Caring for Australians with Renal Impairment - CARI, CARI 2005), both of which state that broad spectrum antibiotic agents designed to cover both gram negative and gram positive organisms should be initiated at the time a diagnosis of peritonitis is suspected. There is, however, insufficient evidence to suggest more specific agents. This has been demonstrated by this review in which we found that in 21 studies comparing different antibiotic classes, the treatment failure rates were generally in the range of 10% to 30%, with only three studies showing a difference between treatment arms (de Fijter 2001; Flanigan 1991; Lupo 1997). In each of these cases the applicability to current practice is low. de Fijter 2001 found IP ciprofloxacin and rifampicin to be superior to IP cephradine. However, monotherapy with a first generation cephalosporin is uncommon, and in this case, was associated with a low initial response rate of 50%. Furthermore, the broad spectrum of action of both ciprofloxacin and rifampicin predisposes to emergence of multiresistant organisms thereby reducing their desirability as first line agents. In our meta-analysis of two studies comparing IP cephazolin and vancomycin we found vancomycin to be superior. However, this result was strongly influenced by a larger number of patients in Flanigan 1991, in which the cephazolin dose of 50 mg/L was two and a half times less than that recommended in the current ISPD guidelines (Li 2010).
Similar efficacy rates amongst several antibiotic regimens facilitate consideration of logistical factors and adverse effect profiles when selecting antibiotics (Kan 2003). Current ISPD guidelines state that there should be centre-specific selection of agents according to local causative micro-organism and resistance patterns (Li 2010). The impact of local microbial resistance on peritonitis outcomes was apparent in two studies comparing oral and IP quinolone use (Cheng 1993; Cheng 1997). In these studies, response rates were low for both treatment arms (41.7% and 55.6% in the oral groups and 66.7% and 70.6% in the IP groups respectively). Micro-organism resistance to quinolones was the major cause of treatment failure, and previous exposure to quinolones was a risk factor for infection with resistant micro-organisms. The emergence of vancomycin-resistant enterococcus is also associated with use of broad spectrum antibiotics (Carmeli 2002; Oprea 2004). Of note increasing prevalence of methicillin-resistant Staphylococci (both S. aureus and coagulase negative species) is a relatively recent phenomenon hence limiting the ability of early studies to evaluate this problem.
In this review, we found that studies in which antibiotics (ciprofloxacin, ofloxacin and cephradine) were administered either orally or IP showed no difference in outcomes for the two routes of administration. However, initial antibiotic therapy is commonly administered intraperitoneally as this theoretically achieves higher dialysate antibiotic levels than permitted with other routes. Evidence about the relative importance of dialysate antibiotic levels was unclear (reviewed in Johnson 2011). In the study of oral versus IP ciprofloxacin included in this review, dialysate antibiotic levels were lower in the IP group but this did not affect patient outcomes (Cheng 1993). Booranalertpaisarn 2003 reported that daily dosing of ceftazidime in patients with peritonitis led to serum levels that were above the recommended minimum inhibitory concentration (MIC) throughout 24 hours, whereas dialysate levels were below the MIC for several hours on days one and four. Despite this, the response rate was 90%, suggesting that achieving therapeutic dialysate levels may not be necessary for treatment to be effective.
Benefits of intermittent (daily) dosing of antibiotics include facilitation of outpatient management and continuation of APD. In the general population, daily dosing with aminoglycosides reduces the risk of ototoxicity compared with intermittent dosing (Deamer 1996). In this review, intermittent and continuous antibiotic dosing had similar outcomes. Adequate duration of antibiotic activity with daily dosing is facilitated by long drug half-lives. Studies of CAPD patients without peritonitis have shown that serum and dialysate levels of several antibiotics remain above the mean inhibitory concentration for up to 48 hours (Grabe 1999; Manley 1999). Many drugs have peak serum levels six hours after administration suggesting that this should be the minimum dwell time. Post-antibiotic effects of drugs may also contribute to the efficacy of intermittent dosing. The applicability of results from studies of intermittent drug therapy in CAPD to APD is however unclear as drug half-lives are greater and clearances more rapid in cycler dwells compared to non-cycler dwells (Manley 2002).
The high rate of complications arising from peritonitis despite rapid institution of antibiotic therapy suggests a need exists for adjuvant treatment strategies. One such treatment is administration of IP urokinase, the rationale being to dissolve fibrin and enable access of antibiotics to entrapped bacteria (Pickering 1989). Williams 1989 showed that urokinase was inferior to simultaneous catheter removal and replacement. However, catheter removal could in itself be considered treatment failure. Meta-analysis of three other studies showed no statistically significant difference in outcomes between urokinase and catheter removal. However, it is noteworthy that in Tong 2005a, the actual number of patients achieving a primary response was five more in the urokinase than the control group, and there were three fewer catheter removals. Furthermore, adequately powered, studies in this area may be beneficial, in which the optimal outcome would be permanent transfer to haemodialysis.
Peritoneal lavage is performed at many centres because it has the potential to remove inflammatory cells and micro-organisms from the peritoneal cavity while providing symptomatic relief, and has been used successfully in abdominal surgery (O'Brien 1987). It has however been the subject of only one RCT (Ejlersen 1991), in which patients with hypotension and shock, the same group in which lavage has been used in surgical settings, were excluded. In this study, peritoneal lavage did not improve response rates. This may be a true effect due to inadvertent removal of macrophages and other components of the immune system thereby a reduction of local host defences against infection. However, further studies to evaluate this therapy further may be useful.
A novel strategy is administration of IP immunoglobulin in conjunction with antibiotics with the aim of improving local host defences (Carozzi 1988). In a study of 24 patients (Coban 2004) biochemical and clinical parameters of improvement were achieved sooner, and the duration of antibiotic therapy was shorter in the immunoglobulin treatment group. The response rate of 100% was unusually high and there were no relapses during three months of follow-up. In a larger population, a difference in response rates may have become apparent.
Implications for practice
In conclusion, currently available evidence from RCTs is not robust and does not identify an optimal antibiotic regimen for the treatment of PD-associated peritonitis.
Currently available evidence from RCTs is inadequate in many areas of clinical practice important in the management of PD-associated peritonitis and is a limiting factor in the provision of definitive treatment guidelines.
Implications for research
Future research should be adequately powered to assess outcomes such as catheter removal and mortality, and should include long-term follow-up of parameters such as ultrafiltration failure, loss of residual kidney function and technique failure.
Further studies are needed to:
In addition, future research should be conducted using standard definitions, with inclusion of information about factors that may influence the response to therapy such as prophylaxis regimens and dialysis solutions used. Current ISPD guidelines provide a comprehensive list of requirements for future studies that should be referred to when designing studies.
We would like to thank the referees for their comments and feedback during the preparation of this review.
We would like to thank Narelle Willis and Ruth Mitchell from the Cochrane Renal Group for their assistance to update this review.
We would also like to thank Dr Martin Searle from the Department of Nephrology at Christchurch Hospital for his contribution to the updated review.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Appendix 1. Electronic search strategies
Appendix 2. Risk of bias assessment tool
Last assessed as up-to-date: 5 March 2014.
Protocol first published: Issue 2, 2005
Review first published: Issue 1, 2008
Contributions of authors
Screening of titles and abstracts: AB, KW, GFMS, SP
Study eligibility: AB, KW, GFMS, SP
Risk of bias assessment, data extraction, data analysis: AB, KW, GFMS, SP
Writing of review: AB, KW, GFMS, SP, DJ, JC
Disagreement resolution: DJ, JC
Declarations of interest
Professor David Johnson is a consultant for Baxter Healthcare Pty Ltd and has previously received research funds from this company. He has also received speakers' honoraria and research grants from Fresenius Medical Care. Angela Ballinger received a student stipend for a summer studentship 2011 to 2012 from the University of Otago to assist with completing this research. Suetonia Palmer received a fellowship administered by the Consorzio Mario Negri Sud from Amgen Dompe for assistance with travel for collaboration and supervision. The other authors had no known conflicts of interest.
Sources of support
- The research team acknowledges the support received from the Cochrane Renal Group in the conduct of this review, Australia.
- Suetonia Palmer, New Zealand.received an unrestricted fellowship from Amgen Dompe, administered by the Consorzio Mario Negri Sud
- Angela Ballinger, New Zealand.Received a summer student scholarship from the Division of Medical Sciences, Unversity of Otago to complete the update of this review
Medical Subject Headings (MeSH)
Administration, Oral; Anti-Bacterial Agents [administration & dosage]; Fibrinolytic Agents [therapeutic use]; Immunoglobulins [therapeutic use]; Infusions, Parenteral; Injections, Intravenous; Peritoneal Dialysis [*adverse effects]; Peritoneal Lavage; Peritonitis [drug therapy; etiology; *therapy]; Randomized Controlled Trials as Topic; Urokinase-Type Plasminogen Activator [therapeutic use]
MeSH check words
* Indicates the major publication for the study