Impact of species and antibiotic therapy of enterococcal peritonitis on 30-day mortality in critical care—an analysis of the OUTCOMEREA database

Intra-abdominal infections (IAI) represent the second most common cause of infection in the ICU [1]. Indeed, they are complicated with septic shock in 40% of cases [2]. Despite improvements in sepsis management, mortality remains high up to 40% in nosocomial IAI [3, 4]. The primary treatment of IAI combines early source control and adequate antimicrobial therapy.

The incidence of Enterococci in IAI is 5 to 20% in community-acquired IAI and 30 to 40% in nosocomial IAI [5]; however, the pathogenicity of Enterococci in IAI is debatable. According to Dupont, Koch, and Fisher, Enterococcus might express virulence factors and might synergize with other bacteria like Escherichia coli and anaerobes [6‐8]. It has been clearly demonstrated that Enteroccoci are associated with proinflammatory responses, greater clinical disease burden, and shock [9‐11]. So far, all authors agree with an increase in morbidity (septic shock, higher APACHE 2, and Sequential Organ Failure Assessment (SOFA) scores, higher postoperative infection scores, longer duration of mechanical ventilation and vasopressors, more relaparotomies), but the impact of Enterococcus on mortality is unclear [12‐16]. Some studies found that the presence of Enterococci on peritoneal samples is a predictive factor for death [17‐19] whereas others did not [20, 21].

Currently, the impact on prognosis of early antimicrobial therapy against Enterococcus spp. is not known and the indication of an initial empiric anti-enterococcal therapy differs among recommendations. Thus, the aim of our study was to compare the role of appropriate versus inappropriate antimicrobial therapy on 30-day mortality in ICU patients with IAI positive for Enterococcus.

This study was a retrospective data analysis from the OUTCOMEREA database (OutcomeRea®).

Data were prospectively collected daily by senior physicians with research assistants in the participating ICUs. All codes and definitions were established prior to study initiation and have been previously described [22]. We collected delay between hospitalization, diagnosis, and surgery.

Patients’ age, sex, and McCabe score were recorded. Severity of illness was evaluated on the first ICU day using the Simplified Acute Physiology Score (SAPS II), Sequential Organ Failure Assessment (SOFA) score, and Glasgow Coma Scale (GCS) score. Knaus’ scale definitions were used to record preexisting chronic organ failures including respiratory, cardiac, hepatic, renal, and immune system failures [23, 24]. Admission category (medical, scheduled surgery, or unscheduled surgery), admission diagnosis (cardiac, respiratory, or neurological failure, infection, and other), invasive procedures (arterial or venous central catheter, Swan-Ganz catheter, or endotracheal intubation), and treatment of organ failures (inotropic support, hemodialysis, and mechanical ventilation) and ICU-acquired infections, bacteriological samples, and daily antimicrobial therapy were also collected.

Data collected from hospitalization records were as follows: risk factors for healthcare-associated infections, antimicrobial therapy within the 3 months prior to ICU admission (particularly with cephalosporins), date of diagnosis of IAI according to clinical, biological and radiological findings, anatomical origin of IAI, localized or generalized type of IAI, community-acquired or nosocomial infection, and pathophysiological mechanisms. We also collected delay between diagnosis according to clinical, biological, and radiological findings and surgery, initial and adapted antimicrobial therapy, Enterococcus species, sensibility to antimicrobial therapy, appropriate or inappropriate type of antimicrobial therapy against Enterococcus species, day of appropriate antimicrobial therapy for Enterococcus species and other microorganisms, surgical complications, need for redo laparotomy or percutaneous drainage, and development of tertiary peritonitis.

Between 1997 and 2016, data of 1017 patients with IAI were analyzed. Only 76 IAI with Enterococcus were included (Additional file 1). Incidence of Enterococcus differed between centers (Additional file 2). Median [IQR] age was 72 [59–78]. Fifty-seven percent of patients were male. Median SAPS II score at day 0 was 52 [41–64]. Median time between admission in the hospital or in the ICU and diagnosis of IAI was 8 [5‐10] and 1 [1‐4] days, respectively. Eight (10.5%) patients had a community-acquired IAI. Sixty-eight (89.5%) patients had a nosocomial IAI, among them 2 were healthcare-associated IAI and 66 were hospital-acquired IAI. Two patients had a percutaneous drainage as an initial treatment, and 7 had a percutaneous drainage secondarily after an initial surgical treatment. IAI characteristics are described in Table 1.

Germs associated with Enterococcus were as follows: Gram-positive cocci (22%), Gram-negative bacilli (74%), anaerobes (20%), and yeasts (24%). IAI were associated with Enterococcus bacteremia in 4 (5%) of cases. Eleven (14%) of the 76 IAI were growing with only Enterococcus species. Empirical antimicrobial therapies were piperacillin-tazobactam (49%), carbapenems (33%), vancomycin (30%), and third-generation cephalosporins (9%) in combination with aminoglycosides in 70% of the cases. Initial empirical antimicrobial therapy was inappropriate against Enterococcus species isolated from peritoneal sample in 18 (23.7%) of cases and against other germs in 12 (15.8%) of cases. Antimicrobial therapy was modified to cover the recovered Enterococci in 13 (72%) patients. Sensitivity to amoxicillin and vancomycin was always available. There were 3 ESBL and no VRE.

Table 2 compares patients who received adequate versus inadequate empirical therapy. The two groups were similar for year of inclusion, age, gender, causes, and origins of IAI. Thirty-day mortality was significantly higher in the group who received inadequate empiric antimicrobial therapy against Enterococcus species identified on peritoneal sample, but there was no difference in postoperative complications. In this group, Enterococcus spp. other than Enterococcus faecalis were more frequently identified. Survival curves according to adequacy of IAT against Enterococcus species identified on peritoneal sample are shown in Fig. 1.

Comparisons between patients with IAI growing with Enterococcus spp. other than Enterococcus faecalis and Enterococcus faecalis alone are described in Table 2. Day 30 mortality was significantly higher in IAI growing with Enterococcus spp. other than E. faecalis alone but there was no difference in postoperative complications. Survival curves according to Enterococcus species are shown in Fig. 2.

Impact on 30-day survival of the enterococcal species and the adequacy of antibiotic therapy on enterococci is displayed on Table 3.

Univariate analysis demonstrated that the identification of species other than E. faecalis alone was associated with death and the mortality rate was greater if the antibiotic therapy was inadequate against Enterococcus spp. identified on peritoneal samples.

However, after adjusting for confounders (i.e., SAPS II and septic shock at IAI diagnosis, ICU-acquired peritonitis, and adequacy of antibiotic therapy for other germs), the impact of the adequacy of antimicrobial therapy was no longer significant (Table 3). The impact of culturing enterococci other than E. faecalis remained a poor prognosis (HR = 2.283 [0.730–7.141], p = 0.156). A septic shock at IAI diagnosis and an ICU-acquired IAI were associated with death regardless of the adequacy of IAT and Enterococcus species.

Survival curves according to adequacy of IAT on germs other than Enterococcus species identified on peritoneal sample are shown in Additional file 3.

In a large cohort of severe peritonitis with enterococci admitted in the ICU, we found that inadequate IAT against Enterococcus spp. was associated with increased 30-day mortality. We also found that Enterococcus spp. other than E. faecalis alone were more frequent in cases of previous therapy with third-generation cephalosporins in the past 3 months and in ICU-acquired peritonitis. It was associated with inadequate IAT and a poorer prognosis.

Intraabdominal infections growing with Enterococcus are associated with a worse prognosis. Theunissen et al. showed that presence of Enterococci is a predictive factor for death in both nosocomial and community-acquired IAI and is independently associated with mortality (OR 3.88 (1.05–14.28) p = 0.044) [19]. In our study, the large majority of IAI was nosocomial (90%) and mortality was high (22% in the whole population and 39% in the inappropriate first antimicrobial therapy group). It was comparable with previously published data: 39% in Montravers et al. study (mainly postoperative IAI with multidrug-resistant bacteria), 25% in Sotto et al. (ICU IAI), and 25% in Theunissen et al. (40% of nosocomial IAI) [17, 19, 26]. Recently, Freedberg et al. found in a cohort of 301 medical ICU patients that VRE colonization and Enterococcus domination were both associated with death or all-cause infection at 30 days (aHR 1.46, 95% CI 1.06–2.00 and aHR 1.47, 95% CI 1.00–2.19, respectively) after adjusting for severity of illness [27].

We found that an inappropriate IAT against Enterococcus spp. identified on peritoneal samples was associated with a higher 30-day mortality. The only prospective randomized trial that compared treatment with antimicrobial therapy active or inactive against Enterococcus (penicillin vs cephalosporin) concluded no differences between both groups [28]. This study included only non-severe community-acquired IAI (median APACHE scores 10 and 9); the number of IAI growing with Enterococcus was very low (6 over 110 peritonitis) [28]. However, in a population of 200 postoperative IAI among which 42 were growing with Enterococcus, Sitges-Serra et al. found that mortality was higher when IAT did not cover these Enterococcus (21% vs 4%, p <  0.001) [14]. In Enterococcus bacteremia, early use of anti-Enterococcus antimicrobial therapy within 48 h is a protective factor against death [29].

Several studies have evaluated the impact on mortality of the adequacy of an initial empiric antimicrobial therapy for all peritoneal germs in general (not only for Enterococcus). Sotto, Mosdell, Montravers, and Sturkenboom did not find any difference [17, 20, 30, 31]. However, Sotto studied ICU IAI but the numbers were small in the inappropriate antimicrobial therapy group (14 patients) and the three other studies included mainly patients with community-acquired and non-severe IAI. Nevertheless, Montravers’ study on postoperative IAI growing with multiresistant bacteria (2009) and Harbarth’s study on severe sepsis and septic shock showed that an inappropriate IAT was an independent risk factor of death in IAI [26, 32]. Thus, it seems that an appropriate IAT has an impact on mortality especially in severe postoperative IAI with septic shock.

However, we did not find more redo laparotomies or percutaneous drainages or infectious complications at day 30 when IAT was inappropriate. Other studies showed that an inappropriate IAT against all microbials found on the peritoneal sample was associated with a higher morbidity both in community-acquired and nosocomial IAI in terms of length of stay, wound infection, redo laparotomies, and postoperative complications [14, 26, 30, 31, 33, 34].

An inadequate IAT against Enterococcus spp. identified on peritoneal samples was associated with postoperative IAI, time between hospital admission and IAI, antimicrobial therapy with third-generation cephalosporins within the 3 months prior to IAI, IAI with E. faecium or any Enterococcus species other than E. faecalis, and an IAT that did not include vancomycin. This implies that adequacy of IAT was associated with the use of vancomycin. WSES recommendations are to cover Enterococcus in postoperative IAI but not in community-acquired IAI [35]. For IDSA, “empiric anti-enterococcal therapy is recommended for patients with healthcare-associated IAI, particularly those with postoperative infection, those who have previously received cephalosporins or other antimicrobial agents selecting for Enterococcus species, immunocompromised patients, and those with valvular heart disease or prosthetic intravascular materials”. “Anti-enterococcal therapy should be directed against Enterococcus faecalis”, and “antibiotics that can potentially be used … include ampicillin, piperacillin-tazobactam or vancomycin” [36].

In multivariate analysis, adequacy of IAT on Enterococcus species was no longer independently associated with survival after adjustment on adequacy of IAT on other germs, time of onset of IAI (ICU-acquired IAI), SAPS score, and septic shock at time of diagnosis. However, septic shock at diagnosis and ICU-acquired IAI were risk factors for death regardless of the Enterococcus species and adequacy of IAT. It has been shown multiple times in the literature that an IAI with septic shock has a higher mortality rate than an IAI without septic shock [2, 20, 37] and our study confirms this finding.

Finally, we found a difference in 30-day mortality depending on Enterococcus species: E. faecalis alone vs Enterococcus other than E. faecalis alone (mostly E. faecium). In univariate analysis, we found a higher hazard ratio for death with an Enterococcus other than E. faecalis alone that had been initially inadequately treated compared to adequately treated; this suggests that the species of Enterococcus had a greater impact than adequacy of IAT on survival. Risk factors of having an Enterococcus other than E. faecalis alone in univariate analysis were SAPS score on day 0, Cardio SOFA score on ICU admission, time between ICU admission and IAI, ICU-acquired IAI, antimicrobial therapy within the 3 months prior to IAI especially with third-generation cephalosporins, antimicrobial therapy prior to relaparotomy in postoperative IAI, and an inadequate IAT. Previous studies on enterococcal bacteremia found that SAPS score, prior antimicrobial therapy exposure (mainly penicillin and third-generation cephalosporins, but also carbapenems, aminoglycosides, and clindamycin), hematologic malignancies and neutropenia, current corticosteroid therapy, organ dysfunction, gastrointestinal disease (vs genitourinary disease), and nosocomial acquisition were risk factors for E. faecium isolation (vs E. faecalis) [38‐42]. Some studies found a higher mortality with E. faecium than with E. faecalis bloodstream infections [38, 40, 42]. Our findings are consistent with those studies. Enterococcus spp., especially other than E. faecalis, are associated with a poorer prognosis in IAI and their presence is often associated with a previous antimicrobial therapy within the 3 months prior to IAI and especially with third-generation cephalosporins.

We found that inadequacy of IAT on Enterococcus species identified on peritoneal sample was associated with a worse 30-day survival especially when it was not an E. faecalis alone. This inadequacy was partly due to the absence of vancomycin in the empiric antibiotic regimen. These results should encourage the physician to use an antimicrobial regimen active against non-faecalis Enterococcus (like vancomycin) in severe, postoperative, ICU-acquired IAI, especially when an antimicrobial therapy with third-generation cephalosporins has been used within 3 months preceding IAI. Our study also suggests that IAI with non E. faecalis Enterococci have a poorer prognosis than IAI with E. faecalis alone. It would be interesting to conduct a prospective and larger study to confirm these results.