Multicenter Study Evaluating the Role of Enterococci in Secondary Bacterial Peritonitis

Abstract

A 1-year prospective multicenter study was performed to explore the significance of the presence of enterococci in cultures of peritoneal fluid from patients with secondary bacterial peritonitis in seven Spanish hospitals. The clinical records of patients with positive peritoneal fluid cultures were reviewed and distributed into cases (patients with cultures yielding enterococci) and controls (patients with cultures not yielding enterococci). Of a total of 158 records, 38 (24.1%) were cases and 120 (75.9%) were controls. The percentages or the scores (cases versus controls) for the variables included in the multivariate analysis were as follows: age of >50 years, 89.5% versus 68.3%; malignancy, 39.5% versus 18.3%; chronic obstructive pulmonary disease (COPD), 15.8% versus 4.2%; postoperative peritonitis, 55.3% versus 30.1%; nosocomial onset, 57.9% versus 34.2%; a higher Charlson comorbidity index, 3.29 ± 3.38 versus 1.84 ± 2.31; APACHE II score, 10.71 ± 4.37 versus 8.76 ± 5.49; ultimately or rapidly fatal disease, 63.2% versus 34.8%; need for surgical reintervention, 36.1% versus 15.1%; and admission to an intensive care unit, 45.9% versus 30.8%. In the multivariate analysis, enterococci were associated only with postoperative peritonitis (P = 0.009; odds ratio [OR] = 5.0; 95% confidence interval [CI] = 1.49 to 16.80), a higher Charlson comorbidity index (P = 0.002; OR = 1.30; 95% CI = 1.11 to 1.54), and COPD (P = 0.046; OR = 6.50; 95% CI = 1.04 to 40.73). The results of this study showed that enterococci were associated with comorbidity. An association with mortality could not be demonstrated.

Aggregation substances and the extracellular protein Esp have been identified as virulence factors in enterococci, and both play a role in colonization of the host (5, 10). In addition, resistance to multiple antimicrobial agents allows enterococci to proliferate in patients receiving antimicrobial chemotherapy (14). The factors enabling enterococci to cause human infections have yet to be clarified, although the development of traits that make it possible for the microorganism to occupy new niches, evade host defenses, or exploit a weakened host immune system go some way to providing an explanation (10).

It has been suggested that enterococci act synergistically with other bacteria and thus increase the rates of morbidity and mortality (15). Nevertheless, most reports of high rates of mortality in association with enterococcal bacteremia involve series that include severely debilitated patients; therefore, bacteremia could be a marker of this state and not the cause of it (15).

Although the role of enterococci in intra-abdominal infections has not been fully defined, it is clear that they can cause peritonitis (in patients with nephrotic syndrome or cirrhosis and in patients undergoing peritoneal dialysis) (17) and that perioperative treatments with active agents during abdominal surgery decrease the probability of development of subsequent enterococcal wound infections (24). More recent studies excluding postoperative cases have reported on the minor role of enterococci in secondary peritonitis in relatively healthy patients (20), although others report that the presence of this entity in cultures of peritoneal fluid significantly increases the rate of morbidity but not the rate of mortality (6).

Few recent series have reported on the rate of isolation of enterococci in secondary peritonitis, and their role in this disease remains undefined (15, 16). Therefore, we carried out a prospective multicenter study in Spain to evaluate the role of enterococci in secondary bacterial peritonitis.

MATERIALS AND METHODS

We studied all peritoneal fluid samples from patients with secondary peritonitis received in the microbiology departments of seven Spanish hospitals in 2004. Sample collection was done by the hospital-specific routines. Peritoneal fluids were sent to the laboratory in aerobic and anaerobic blood culture bottles. In those cases in which the sample volume was sufficient, an aliquot of the sample was also sent to the laboratory in a sterile container. This aliquot was directly plated (onto suitable media) after centrifugation and was incubated aerobically and anaerobically. One of the seven hospitals received samples in sterile containers only. Samples from nonsurgical departments yielding monobacterial growth were excluded to rule out the possibility that samples from patients with primary peritonitis would be studied.

Secondary peritonitis was defined as a peritoneal infection caused by bacterial contamination secondary to inflammatory, vascular, mechanical, or malignant abnormalities of the gastrointestinal, biliary, pancreatic, or genitourinary tract in patients aged >18 years. We reviewed the clinical records of patients with positive culture results to collect clinical data, data on secondary peritonitis, and outcomes data. We also collected the data necessary to calculate the values of the following prognostic indices: the Charlson index (3), the APACHE II score (8, 9), the Mannheim index (2, 19), the McCabe and Jackson index (13), and the American Society of Anesthesiologists' Physical Status Classification Scale (ASA) (1).

Patients were distributed into two groups: cases (patients with positive cultures yielding the growth of enterococci) and controls (patients with positive cultures not yielding the growth of enterococci). Data for the cases and the controls were compared by using the chi-square test or Fisher exact test for qualitative variables. As data were not normally distributed in the Kolmogorov-Smirnoff test, quantitative variables were compared by use of the Mann-Whitney test or the Wilcoxon test. Statistical significance was established at a P value of ≤0.05. A logistic regression model (stepwise procedure) was performed by using the presence of enterococci (yes or no) as a dependent variable and variables showing differences (P ≤ 0.1) in the bivariate analysis as the independent variables. Statistical analysis was performed by using SPSS (version 14) software (SPSS Inc, Chicago IL). The model showing the highest R² value was considered.

RESULTS

A total of 158 evaluable samples were processed, and the corresponding clinical records were reviewed. Of these, 38 (24.1%) yielded the growth of enterococci and were defined as “cases,” while the remaining 120 (75.9%) did not yield the growth of enterococci and were defined as “controls.” Table 1 lists the microorganisms isolated from the cases and the controls. No significant differences in isolation rates were obtained for the controls versus the cases for all bacterial types isolated.

TABLE 1.

Microorganisms recovered from samples from controls and cases

Microorganism	No. (%) of samples yielding growth
	Controls (n = 120)	Cases (n = 38)
Enterococcus spp.	0 (0.0)	35 (100.0)
Streptococcus viridans group	28 (23.3)	3 (8.6)
Escherichia coli	64 (53.3)	17 (48.6)
Other Enterobacteriaceaea	39 (32.5)	15 (42.9)
Bacteroides fragilis group	23 (19.2)	10 (28.6)
Other anaerobesb	13 (10.8)	6 (17.1)
Pseudomonas aeruginosa	12 (10)	3 (8.6)
Other aerobes/facultative bacteriac	10 (8.3)	1 (2.9)
Yeastsd	5 (4.2)	5 (14.3)

^aIncludes isolates from the genera Klebsiella, Proteus, Morganella, Citrobacter, and Serratia.

^bIncludes isolates from the genera Clostridium, Fusobacterium, Peptostreptococcus, Prevotella, Veillonella, and Leuconostoc.

^cIncludes isolates from the genera Acinetobacter, Aeromonas, and Propionibacterium.

^dIncludes isolates from the genera Candida and Torulopsis.

Table 2 shows the comorbid conditions of the controls and the cases. The cases tended to be older, with a significantly higher percentage of the case patients being over 50 years of age (89.5% versus 68.3% for the controls; P = 0.011). As for comorbid conditions, a significantly higher percentage of cases presented with malignancy (39.5% versus 18.3% for the controls; P = 0.014) and chronic obstructive pulmonary disease (COPD; 15.8% versus 4.2% for the controls; P = 0.014).

TABLE 2.

Comorbid conditions and characteristics of peritonitis

Variable	% of patients		P valuea
	Controls (n = 120)	Cases (n = 38)
Gender (females)	45.0	47.4	0.798
Age > 50 yr	68.3	89.5	0.011
Cardiovascular disease	24.2	26.3	0.789
Malignancy (solid tumor, leukemia, or lymphoma)	18.3	39.5	0.014
Diabetes	9.2	13.2	0.487
Chronic liver disease	6.7	7.9	0.727
Renal disease	6.7	13.2	0.204
Chronic obstructive pulmonary disease	4.2	15.8	0.014
Cause of peritonitis
Preoperative	54.2	18.4
Postoperative	30.1	55.3
Other	15.7	26.3	0.005
Origin of peritonitis
Gastroduodenal	8.4	5.4
Small intestine	37.0	32.4
Colon/rectosigmoid	31.9	35.1
Biliopancreatic	15.1	21.6
Other	7.6	5.5	0.230
Peritonitis duration < 24 h	41.7	21.1	0.022
Community onset	65.8	42.1	0.009
Generalized peritonitis	40.8	50.0	0.320
Previous antibiotic treatment	38.7	52.6	0.128

^aBoldface data represent significantly significant differences between the cases and the controls.

Table 2 also shows the characteristics of secondary peritonitis. While no differences in the origin of peritonitis were found between the study groups, for the cases significantly higher percentages of postoperative peritonitis (55.3% versus 30.1% for the controls: P = 0.005) and nosocomial onset (57.9% versus 34.2% for the controls; P = 0.009) were found.

Table 3 shows the prognoses and the outcomes for both the controls and the cases. The incidence of comorbidities was significantly higher in the cases than in the controls according to both the Charlson comorbidity index (3.29 ± 3.38 and 1.84 ± 2.31, respectively; P = 0.011) and the prognoses of the underlying diseases (McCabe and Jackson score) (P = 0.003). The prognosis of mortality assessed by use of the APACHE II score also showed differences between the groups, with a higher score being achieved for the cases than for the controls (10.71 ± 4.37 and 8.76 ± 5.49, respectively; P = 0.011). There were no differences between the cases and the controls for preoperative physical health in the global analysis (Table 3), although a significantly (P = 0.036) lower percentage of controls presented ASA I (18.2% and 40.2%, respectively).

TABLE 3.

Prognoses and outcomes

Prognostic index	Controls (n = 120)	Cases (n = 38)	P valuea
Charlson comorbidity index (mean ± SD)	1.84 ± 2.31	3.29 ± 3.38	0.011
APACHE II score (mean ± SD)	8.76 ± 5.49	10.71 ± 4.37	0.011
Mannheim peritonitis index (mean ± SD)	19.11 ± 8.22	21.32 ± 6.90	0.078
% of patients with the following McCabe and Jackson group score:
Nonfatal	65.3	36.8
Ultimately fatal	28.0	57.9
Rapidly fatal	6.8	5.3	0.003
% of patients with the following ASA classification:
I	40.2	18.2
II	17.8	24.2
III	26.2	30.3
IV	15.9	27.3	0.114
% of patients with the following outcome:
Bacterial superinfection	15.6	27.8	0.111
Fungal superinfection	5.8	21.6	0.011
Need for surgical reintervention	15.1	36.1	0.006
Septic shock	17.5	27.0	0.203
Multiple organ failure	13.4	24.3	0.120
Admission to intensive care unit	30.8	45.9	0.091
Death	15.0	26.3	0.111

^aBoldface data represent significant differences between the cases and the controls.

With respect to outcome, values for surgical reintervention were significantly higher for the cases than for the controls (36.1% and 15.1%, respectively; P = 0.006), as was fungal superinfection (21.6% and 5.8%, respectively; P = 0.011).

In the multivariate analysis, the logistic regression was statistically significant (P < 0.001; Cox R² = 0.191), and the presence of enterococci was associated with the following: (i) postoperative peritonitis (P = 0.009; odds ratio [OR] = 5.0; 95% confidence interval [CI] = 1.49 to 16.80) when preoperative peritonitis was used as a reference (postoperative peritonitis increased the risk of isolating enterococci fivefold); (ii) the Charlson comorbidity index, in which each additional point in the index increased the risk of isolating enterococci by 30% (P = 0.002; OR = 1.30; 95% CI = 1.11 to 1.54); and (iii) COPD (P = 0.046; OR = 6.50; 95% CI = 1.04 to 40.73).

DISCUSSION

Enterococci have rarely been isolated in samples from patients with community-acquired secondary peritonitis and no prior use of antibiotics, but they are frequently present in patients with nosocomially acquired secondary peritonitis, particularly after the administration of antibacterial agents. However, the exact role of enterococci in mixed intra-abdominal infections remains undefined (15) and has been questioned because enterococci causes bacteremia less frequently than Escherichia coli and Bacteroides fragilis when they are present in mixed intra-abdominal infections (15, 18), despite the fact that enterococci are the third most common pathogen isolated from bloodstream infections in the United States (7, 10). It is important to determine whether this pathogen should be covered in the empirical treatment of secondary peritonitis, since several studies have found a significant relationship between inappropriate therapy and lower clinical success rates (11, 23) and have concluded that the selection of inappropriate antibiotics for treatment increases the rate of clinical failure 3.4-fold and the mortality rate by about 30% (4, 12).

The bivariate analysis showed that enterococci were isolated from patients with significantly higher ages, comorbid conditions (according to the Charlson and the McCabe and Jackson scores), a worse prognosis (according to the APACHE II score), and the need for surgical reintervention. However, in the multivariate analysis, only comorbid conditions (as determined by the Charlson score), postoperative peritonitis, and the presence of COPD were significantly associated with the presence of enterococci, although the coefficient of determination was low (Cox R² = 0.191). Therefore, it seems that enterococci are associated with comorbidity, but we could not demonstrate an association with mortality. In a previous study, the presence of enterococci in peritoneal fluid cultures significantly increased the rate of morbidity but not that of mortality (6), while other authors established an association between polymicrobial infections involving enterococci and increased mortality (3, 20, 21).

In the view of the presence of enterococci in more severely ill patients with secondary peritonitis (mainly postoperative peritonitis), our results suggest, but do not demonstrate, the need for coverage in this group. In this sense, the guidelines of the Infectious Diseases Society of America for the selection of anti-infective agents for complicated intra-abdominal infections do not recommend routine coverage of enterococci in patients with community-acquired intra-abdominal infections, although they do recommend coverage (with a moderate quality of evidence for the recommendation, based on clinical experience and expert committee opinions) when enterococci are recovered from health care-associated intra-abdominal infections (22). Although clinical success depends on many factors, the choice of effective empirical therapy is important, as indicated in a pharmacodynamic analysis of empirical therapy for secondary peritonitis carried out by using a Monte Carlo simulation, in which the probability of target attainment and the cumulative fraction of the response decreased for β-lactams when enterococci were included in the model (4).

As in previous studies in which the presence of enterococci in peritoneal fluid cultures significantly increased the rate of morbidity but not the rate of mortality (6), our analysis shows that the presence of enterococci might be a biomarker of a poor prognosis. Further studies are needed to fully and definitively clarify the exact role of enterococci in secondary peritonitis.