Risk factors for resistance to ceftriaxone and its impact on mortality in community, healthcare and nosocomial spontaneous bacterial peritonitis

Transcription

1 Risk factors for resistance to ceftriaxone and its impact on mortality in community, healthcare and nosocomial spontaneous bacterial peritonitis Xavier Ariza 1, José Castellote 1,, Jaime Lora-Tamayo 2, Anna Girbau 1, Sílvia Salord 1, Rosa Rota 1, Javier Ariza 2, Xavier Xiol 1 1 Hepatology Unit, Gastroenterology Department, Hospital Universitari de Bellvitge, IDIBELL, Universitat de Barcelona, Barcelona, Spain; 2 Infectious Diseases Department, Hospital Universitari de Bellvitge, IDIBELL, Universitat de Barcelona, Barcelona, Spain See Editorial, pages Background & Aims: The recent emergence of third-generation cephalosporin resistance in spontaneous bacterial peritonitis is of great concern, although neither the risk factors for resistance nor its real impact on mortality have been well defined. Methods: We conducted a retrospective study of all spontaneous bacterial peritonitis episodes with positive blood and/or ascitic culture at our center ( ). Episodes were classified according to the place of acquisition: community, healthcare system, or nosocomial. Results: Two hundred and forty-six episodes were analyzed in 200 patients (150 males, 57.3 years): 34.6% community-acquired, 38.6% healthcare system-acquired and 26.8% nosocomiallyacquired. Third-generation cephalosporin resistance occurred in 21.5% (7.1% community-acquired, 21.1% healthcare systemacquired, 40.9% nosocomially-acquired). These resistant cases were categorized as extended-spectrum b-lactamase-producing Gram-negative bacilli, other resistant Gram-negative bacilli, and Enterococci. Risk factors for resistance were previous use of cephalosporins, diabetes mellitus, upper gastrointestinal bleeding, nosocomial acquisition, and a low polymorphonuclear count in ascites. Regarding third-generation cephalosporin resistance, adequate empirical treatment was 80.7%. Independent predictors of mortality were nosocomial acquisition, poor hepato-renal Keywords: Cirrhosis; Third-generation cephalosporins; Spontaneous bacterial peritonitis; Healthcare system; Cephalosporins resistance, Mortality. Received 18 July 2011; received in revised form 23 November 2011; accepted 25 November 2011; available online 13 December 2011 q DOI of original article: /j.jhep Corresponding author. Address: Hepatology Unit, Gastroenterology Department, Hospital Universitari de Bellvitge, c/ Feixa Llarga s/n, L Hospitalet de Llobregat, Barcelona, Spain. Tel.: ; fax: address: jcastellote@bellvitgehospital.cat (J. Castellote). Abbreviations: SBP, spontaneous bacterial peritonitis; 3rdGC, third-generation cephalosporins; MR-Cef, microorganisms resistant to third-generation cephalosporins; ESBL, Enterobacteriaceae producing extended-spectrum b-lactamases; PMN, polymorphonuclear leukocytes; MELD, Model for End-Stage Liver Disease; MELD-Na, Model for End-Stage Liver Disease including sodium; CA, communityacquired; HCR, episodes related to the healthcare system; NA, nosocomial acquisition; UGB, upper gastrointestinal bleeding; GNB, Gram-negative bacilli; IQR, interquartile range; CI95, confidence interval 95%; GPC, Gram-positive cocci; AT, appropriate empirical antibiotic treatment. function, immunosuppressive therapy, a marked inflammatory response during the episode and either third-generation cephalosporin-resistance or low rates of adequate empirical treatment. Conclusions: The risk of third-generation cephalosporin resistance was particularly high in nosocomially-acquired episodes of spontaneous bacterial peritonitis, but also occurred in healthcare system-acquired cases. The extent of resistance and the adequacy of empirical antibiotics had a significant effect on mortality along with the patient s hepato-renal function. These data can help determine the most suitable empirical antimicrobial treatments in these patients. Ó 2011 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. Introduction Spontaneous bacterial peritonitis (SBP) is a frequent infection in cirrhotic patients with ascites. Current mortality rates are around 20% [1 3]. Most cases are caused by the Enterobacteriaceae and Streptococcus sp. that are part of the normal commensal gut flora [4]. Third-generation cephalosporins (3rdGC) are commonly recommended in empirical treatment guidelines and offer success rates of 80 90% [1 3]. However, recent publications have warned of an increasing incidence of cases of SBP caused by microorganisms resistant to 3rdGC (MR-Cef) [5 7]. Among community-acquired cases, the emergence of Enterobacteriaceae producing extendedspectrum b-lactamases (ESBL) [8] has challenged the validity of this empirical approach. In addition, the widespread use of quinolones for SBP prophylaxis may have caused a change in the usual intestinal flora of cirrhotic patients, as well as in their antibiotic susceptibility patterns [9 11]. Among hospital-acquired cases, changes in the patient s flora during hospitalization, the previous use of antibiotics, and invasive procedures may account for the resistant pathogens that cause SBP [6,12,13]. Furthermore, modern forms of external hospitalization and treatment, such as day hospital care or frequent visits to the emergency department, may have created a new Journal of Hepatology 2012 vol. 56 j

2 group of patients with healthcare-related acquisition [14] in whom SBP may present with a different etiological and clinical pattern, midway between community and nosocomial episodes [6,15]. The clinical impact of these epidemiological changes is unknown. Higher mortality among episodes of SBP by MR-Cef has been suggested, and the need for a broader antibiotic policy has also been raised [6,7,13]. However, the adverse impact on antibiotic resistance that could follow the implementation of an alternative therapeutic protocol should be considered. The aim of this study was to determine the incidence and risk factors for SBP caused by MR-Cef, as well as the clinical impact of inappropriate empirical antibiotic treatments. We use a stratified approach based on the place of acquisition of the infection: community, healthcare facilities or the nosocomial environment. Materials and methods Setting and study design The study was conducted at the Hospital Universitari de Bellvitge, an 850-bed teaching hospital for adult patients in Barcelona, Spain. The center provides medical care to a population of 1 million. We conducted a retrospective observational study on a cohort of patients diagnosed with liver cirrhosis and culture-positive SBP, between 2001 and Clinical protocol Patients underwent diagnostic paracentesis with a differential leukocyte count and microbiological culture. Blood cultures were performed by standard methods. Ceftriaxone 1 g/24 h (iv) was normally used as the initial empirical therapy; although not supported by current guidelines, it was based on our previous results [16]. It was subsequently maintained or replaced depending on the clinical course and the in vitro susceptibility of isolated microorganisms, up to a planned period of 5 10 days in a hospitalized basis [2,3]. Intravenous expansion was performed with albumin 30 g daily for 3 days, based on assistant physician criteria. Definitions The diagnosis of liver cirrhosis was based on clinical, biochemical, histological and/or radiological findings. A culture-positive SBP was defined as the presence of P250 polymorphonuclear leukocytes (PMN)/mm 3 in the ascitic fluid (automatic leukocyte count with ABX Pentra DF120, Horiba, Japan; leukocyte differentiation by microscope, 100 magnification, May-Grunwald Giemsa staining) and culture positivity of ascites and/or simultaneous blood cultures. We excluded patients with culture-negative SBP, bacterascites, polymicrobial infection, and cases of secondary peritonitis [2,17]. Patients with HIV infection or under immunosuppressive therapy were included in this analysis. The state of cirrhosis was assessed at the time of the SBP diagnosis using the Child-Pugh classification, the Model for End-Stage Liver Disease (MELD) score and the modified model including sodium (MELD-Na) [18]. Cases were classified following modified criteria of Friedman or Merli [14,19,20] as follows. (1) Community-acquired (CA): the diagnosis was made within 48 h of hospitalization in patients with no contact with the healthcare system, defined as hemodialysis or chemotherapy in the 30 days previous to the episode or more than 2 days at the day hospital, visits to the emergency department and/or previous admission to hospital in the 3 months previous to the episode. (2) Episodes related to the healthcare system (HCR): the diagnosis was made within 48 h of hospitalization in patients with prior contact with the healthcare system. (3) Nosocomial acquisition (NA): the diagnosis was made after more than 48 h of hospitalization. Information was collected about comorbid medical conditions, the use of immunosuppressive therapy, HIV infection, the etiology of liver cirrhosis, and the presence of hepatocellular carcinoma. Factors predisposing to infection, such as upper gastrointestinal bleeding (UGB) during the 10 days prior to SBP and refractory ascites [2,3], were also logged. All previous antibiotics administered for more than 3 days during the 3 months prior to the episode were registered. Data on the clinical presentation of every episode of SBP and laboratory data were gathered. Antimicrobial treatment before the result of cultures and the antibiogram were considered empirical, and became definite once this information was available. Empirical antibiotic treatment was considered appropriate on the basis of in vitro susceptibility. Microbiological methods Blood cultures were inoculated into bottles of BacT (Bactec NR-860 system, Johnson Laboratories, Towson, MD, USA) and incubated for 7 days (35 C). Ascites samples were inoculated into bottles of BacT at the patient s bedside [21]. Microorganisms and their antibiotic susceptibility were identified using the MicroScan system (Dade Behring, West Sacramento, CA, USA). Antimicrobial susceptibility was defined according to CLSI criteria [22]. ESBL production was detected by the double-disc synergy test [23] and by the E-test method (AB Biodisk, Solna, Sweden). For the purposes of this study, Gram-negative bacilli (GNB) with intermediate in vitro susceptibility to 3rdGC or quinolones or ESBL-producing GNB were considered resistant to these antibiotics. Statistical analysis Categorical variables were compared using the X 2 test or Fisher s exact test, while continuous variables were compared with the Student s t-test or the Mann Whitney U test. Multivariate logistic regression was performed to determine the variables that were independently associated with the risk of SBP by MR-Cef. A Cox proportional hazard model analysis was carried out to identify independent predictors of mortality for every episode, defining death as the main event. Loss of follow-up, liver transplantation, a new episode of SBP and patients alive at the end of follow-up were defined as censored cases. All tests were two-tailed and p <0.05 was indicative of statistical significance. SPSS 15.0 was used for statistical analyses. Results Spontaneous bacterial peritonitis (SBP) with microbiological isolate/10 6 inhabitants Episodes by ceftriaxone-resistant microorganisms Episodes by ceftriaxone-susceptible microorganisms Year Fig. 1. Incidence of 3rd-generation resistant- and susceptible-episodes of SBP. General characteristics and clinical presentation A microbiological isolate was found in 261 (51.5%) out of 506 episodes of SBP. Fifteen (5.7%) of these were excluded because of polymicrobial etiology. The present analysis therefore focuses on the remaining 246 episodes, which correspond to 200 patients. Of these, 150 were male (75.0%) and the median age was 57 years (interquartile range [IQR] 47 67). Thirty-seven patients (18.5%) had more than one episode. In terms of the place of acquisition, there were 85 CA (34.6%), 95 HCR (38.6%), and 66 NA (26.8%) episodes of SBP (8 of these had previously been admitted to an intensive care unit). No increase in the incidence of SBP with any microbiological isolate or by MR-Cef was observed during this time (Fig. 1): there were episodes/ 826 Journal of Hepatology 2012 vol. 56 j

3 JOURNAL OF HEPATOLOGY Table 1. Basal characteristics and clinical presentation. Community acquired (CA) (n = 85) Health-care related (HCR) (n = 95) Nosocomial acquisition (NA) (n = 66) All (n = 246) Sex (M) 64 (75.3%) 66 (69.5%) 51 (77.3%) 181 (73.6%) Age (yr) 55.9 ± ± ± ± 12.4 Diabetes mellitus 22 (25.9%) 26 (27.4%) 17 (25.8%) 65 (26.4%) Immunosuppressive-therapy^ 4 (4.7%) 7 (7.4%) 12 (18.2%) 23 (9.3%) HIV Infection 11 (12.9%) 12 (12.6%) 5 (7.6%) 28 (11.4%) Hepatocellular carcinoma 11 (12.9%) 21 (22.1%) 14 (21.2%) 46 (18.7%) Refractory ascites* 9 (10.6%) 38 (40.0%) 19 (28.8%) 66 (26.8%) Upper gastrointestinal bleeding^ 3 (3.5%) 3 (3.2%) 11 (16.7%) 17 (6.9%) Child-Pugh C^ 61 (71.8%) 60 (63.2%) 57 (86.4%) 178 (72.4%) MELD-Na score^ 26.6 ± ± ± ± 7.5 Previous antibiotics Any antibiotic* 18 (21.2%) 66 (69.5%) 46 (69.7%) 116 (47.2%) Cephalosporins* 0 (0%) 43 (45.3%) 31 (46.9%) 74 (30.1%) Penicillins* 3 (3.5%) 25 (26.3%) 22 (33.3%) 50 (20.3%) Carbapenems* 0 (0%) 7 (7.4%) 6 (9.1%) 13 (5.3%) Quinolones * 14 (16.5%) 43 (45.3%) 32 (48.5%) 89 (36.2%) Systolic blood pressure <90 mmhg 10 (11.8%) 12 (12.6%) 7 (10.6%) 29 (11.8%) SIRS 53 (62.4%) 55 (57.9%) 29 (43.9%) 137 (55.7%) Temperature 37 ºC * 67 (78.8%) 59 (62.1%) 37 (56.1%) 163 (66.3%) Abdominal pain^ 54 (63.5%) 59 (62.1%) 26 (39.4%) 139 (56.5%) Serum PMN (x cells/l) 9.1 ± ± ± ± 6.6 Renal failure* 46 (54.1%) 68 (71.6%) 51 (77.3%) 165 (67.1%) Ascites PMN (x cells/l)* 3.9 ( ) 2.3 ( ) 2.0 ( ) 2.8 ( ) Protein (g/l) 11.8 ± ± ± ± 6.9 Microbiology Ascites (+) culture 68 (80.0%) 77 (81.1%) 56 (84.8%) 201 (81.7%) Bacteremia 54 (63.5%) 46 (48.4%) 35 (53.0%) 135 (54.9%) Antibiotic Cefotaxime*^ 79 (92.9%) 75 (78.9%) 39 (59.1%) 193 (78.5%) susceptibility Amoxicillin-clavulanate^ 76 (89.4%) 80 (84.2%) 42 (63.6%) 198 (80.5%) Piperacillin-tazobactam*^ 85 (100.0%) 87 (91.6%) 52 (78.8%) 224 (91.1%) Imipenem 82 (96.5%) 93 (97.9%) 61 (92.4%) 236 (95.9%) Quinolones 65 (76.5%) 59 (62.1%) 41 (62.1%) 165 (67.1%) Outcome Appropriate treatment^ 79 (92.9%) 81 (85.3%) 46 (69.7%) 206 (83.7%) 30-day mortality Ø^ 24 (28.2%) 24/93 (25.8%) 32/59 (54.2%) 80/237 (33.8%) Categorical variables: number (percentage). Continuous variables: mean ± standard deviation or median (interquartile range). p <0.05 for comparisons between CA and HCR; ^HCR and NA; and CA and NA. Renal failure: creatinine >133 lmol/l and/or urea >11 mmol/l at any moment during the episode. SIRS, systemic inflammatory response syndrome; PMN, polymorphonuclear leukocytes. Quinolones also includes prophylaxis (both primary and secondary): 77 patients. à Corticosteroids in 15 patients, cyclosporine in 6, mycophenolate-mofetil in 3 and tacrolimus in 2. Ø Excluding cases by 3rd-generation cephalosporins-resistant microorganisms adequately treated with carbapenems or piperacillin tazobactam. year and 1 million inhabitants (confidence interval 95% [CI95]: to 2.153; p = 0.769) and episodes by MR-Cef/year, and 1 million inhabitants (CI95: to 1.156; p = 0.553). Table 1 summarizes the clinical data. Most patients had poor hepatocellular condition (72% Child-Pugh C). This was more evident among NA-SBP cases. Previous antibiotic intake was registered in almost half of the episodes (mainly in the HCR and NA groups). Diffuse abdominal pain was the most frequent symptom (56.5%) and a low temperature was a frequent finding (median 37.4 C, IQR ). However, 21 cases (8.5%) presented no clinical symptoms. While clinical findings in the HCR and CA groups were similar, episodes in the NA group had fewer signs of local and systemic inflammatory response, including a decreased PMN count in ascites. Microorganisms The ascites culture was positive in 201 cases (81.7%). Blood cultures were performed in 220 episodes (89.4%) and were positive Journal of Hepatology 2012 vol. 56 j

4 in 135 (61.3%). Bacteraemia was documented in 45 patients (18.3%) with a negative culture of ascites. Table 2 shows GNB were the most frequently isolated microorganisms. The NA group had a higher prevalence of GNB (72.7%) and fewer Streptococci (19.7%), while in the HCR group the frequency of Enterococci was notable (7.4%). MR-Cef were isolated in 53 cases (21.5%). The distribution of MR-Cef varied widely among groups: the rate was low in the CA group (7.1%), very high in the NA group (40.9%) and intermediate in the HCR group (21.1%) (Table 1). Among GNB, a noteworthy finding was the resistance of Escherichia coli (10%) and Klebsiella sp. (12%), linked to a mechanism of ESBL production. Among other Enterobacteriaceae, the rate of resistance was very high (55.6%), mainly due to a chromosomal b-lactamase. The group including Pseudomonas aeruginosa and other GNB with intrinsic resistance to 3rdGC was responsible for 32.1% of MR-Cef, and was particularly prevalent in the NA group. Among Gram-positive cocci (GPC), resistance was linked to the presence of Enterococci. The in vitro antibiotic susceptibility (Table 1) shows strikingly different patterns when analyzed by groups. Among CA episodes, around 90% of microorganisms were susceptible to different families of antibiotics, except for quinolones. By contrast, only imipenem was valid in more than 90% of NA cases. Risk factors associated with ceftriaxone resistance Risk factors for SBP by MR-Cef are shown in Table 3. Interestingly, resistant episodes presented a low-grade inflammatory pattern in which there were lower temperatures, less abdominal pain and a lower count of leukocytes and PMN in ascites. The use of antibiotics, particularly cephalosporins and quinolones, showed a strong association with MR-Cef. In non-hospitalized patients, the probability of MR-Cef increased with the days of contact with the healthcare system (Fig. 2). Among all episodes, independent predictors of MR-Cef were previous use of cephalosporins, UGB, diabetes mellitus and nosocomial acquisition. Also, a high count of PMN in ascitic fluid was an independent predictor for cephalosporin susceptibility. Among non-hospitalized patients, only diabetes mellitus was an independent risk factor (OR 2.826; CI95: ), and a trend to prediction of MR-Cef was seen in previous use of cephalosporins and/or quinolones (OR 2.505; CI95: ; p = 0.076). In nosocomial episodes, previous use of cephalosporins (OR 5.276; CI95: ) and days of admission before the development of SBP (OR 1.608; CI95: ) were independent predictors of MR-Cef (Fig. 2). Table 2. Microorganisms and 3rd-generation cephalosporins-resistant strains in SBP episodes. CA HCR NA Total N (%) Resistant (%)* N (%) Resistant (%)* N (%) Resistant (%)* N (%) Resistant (%)* E. coli 39 (45.9) 1 (16.7) 36 (37.9) 5 (25.0) 25 (37.9) 4 (14.8) 100 (40.7) 10 (18.9) Klebsiella spp. 6 (7.1) - 10 (10.5) 1 (5.0) 4 (6.1) 2 (7.4) 20 (8.1) 3 (5.7) Other Enterobacteriaceae 3 (3.5) 1 (16.7) (9.1) 4 (14.8) 9 (3.7) 5 (9.4) P. aeruginosa (3.2) 3 (15.0) 3 (4.5) 3 (11.1) 6 (2.4) 6 (11.3) A. baumannii (1.1) 1 (5.0) 4 (6.1) 4 (14.8) 5 (2.0) 5 (9.4) A. hydrophyla 4 (4.7) - 2 (2.1) (2.4) - C. jejuni 1 (1.2) 1 (16.7) (3.0) 2 (7.4) 3 (1.2) 3 (5.7) Other Gram-negatives 1 (1.2) - 3 (3.2) - 2 (3.0) 1 (3.7) 6 (2.4) 1 (1.9) Anaerobes (3.0) 2 (7.4) 2 (0.8) 2 (3.8) All Gram-negatives 54 (63.5) 3 (50.0) 55 (57.9) 10 (50.0) 48 (72.7) 22 (81.5) 157 (63.8) 35 (66.0) S. pneumoniae 13 (15.3) - 8 (8.4) - 3 (4.5) - 24 (9.8) - S. bovis 5 (5.9) - 7 (7.4) - 1 (1.5) - 13 (5.3) - S. agalactiae 3 (3.5) - 1 (1.1) - 1 (1.5) - 5 (2.0) - S. mitis 3 (3.5) - 5 (5.3) - 5 (7.6) 1 (3.7) 13 (5.3) 1 (1.9) S. sanguis 2 (2.4) - 2 (2.1) - 1 (1.5) - 5 (2.0) - S. salivarius (4.2) (1.6) - Other S. viridans 2 (2.4) - 3 (3.2) - 2 (3.0) - 7 (2.8) - E. faecalis (4.2) 4 (20.0) 1 (1.5) 1 (3.7) 5 (2.0) 5 (9.4) E. faecium (1.1) 1 (5.0) 1 (1.5) 1 (3.7) 2 (0.8) 2 (3.8) Other Enterococci 1 (1.2) 1 (16.7) 2 (2.1) 2 (10.0) (1.2) 3 (5.7) S. aureus 1 (1.2) 1 (16.7) 1 (1.1) 1 (5.0) 2 (3.0) 2 (7.4) 4 (1.6) 4 (7.5) Other Gram-positives 1 (1.2) 1 (16.7) 1 (1.1) 1 (5.0) 1 (1.5) - 3 (1.2) 2 (3.8) All Gram-positives 31 (36.5) 3 (50.0) 39 (41.1) 9 (45.0) 18 (27.3) 5 (18.5) 88 (35.8) 17 (32.1) Other Candida albicans (1.1) 1 (5.0) (0.4) 1 (1.9) TOTAL 85 (34.6) 6 (7.1) 95 (38.6) 20 (21.1) 66 (26.8) 27 (40.9) 246 (100) 53 (21.5) Gram-negative bacteria Gram-positive bacteria CA, community-acquired; HCR, health-care related; NA, nosocomially-acquired. Resistant strains: number (percentage of the total number of resistant microorganism for every population described). 828 Journal of Hepatology 2012 vol. 56 j

5 Table 3. Univariate and multivariate analysis for risk factors for SBP caused by a 3rd-generation cephalosporin-resistant microorganism. Background Episode of SBP Resistant cases (n = 53) Susceptible cases (n = 193) Adjusted odds ratio λ (95% CI) Age (yr) 58.8 ± ± Sex (M) 36 (67.9%) 145 (75.1%) - Hepatocellular carcinoma 8 (15.1%) 38 (19.7%) - Immunosuppressive therapy 8 (15.1%) 15 (7.8%) - HIV infection 8 (15.1%) 20 (10.4%) - Diabetes mellitus 21 (39.6%) 44 (22.8%) ( ) Upper gastrointestinal bleeding 11 (20.8%) 6 (3.1%) ( ) MELD-Na score 26.3 ± ± Place of acquisition Community 6 (11.3%) 79 (40.9%) - Health-care 20 (37.7%) 75 (38.9%) - Nosocomial 27 (50.9%) 39 (20.2%) ( ) Previous antibiotics Any antibiotic 39 (73.6%) 71 (36.8%) - Penicillins 17 (32.1%) 33 (17.1%) ( ) Cephalosporins 33 (62.3%) 41 (21.2%) ( ) Carbapenems 7 (13.2%) 6 (3.1%) - Quinolones* 31 (58.5%) 58 (30.1%) ( ) Bacteremia 26 (49.1%) 109 (56.5%) - SIRS 30 (56.6%) 107 (55.4%) - Systolic blood pressure <90 mmhg 7 (13.2%) 22 (11.4%) - Temperature 37 ºC 29 (54.7%) 134 (69.4%) ( ) Serum PMN (x cells/l) 9.3 ± ± Renal failure 36 (67.9%) 129 (66.8%) - Ascites PMN (x cells/l) 1.4 ( ) 3.2 ( ) ( ) Protein (g/l) 9.3 ± ± Positive culture 42 (79.2%) 159 (82.4%) - Categorical variables: number (percentage). Continuous variables: mean ± standard deviation or median (interquartile range). Bold entries: p <0.05. SBP, spontaneous bacterial peritonitis. SIRS, systemic inflammatory response syndrome. PMN, polymorphonuclear leukocytes. Quinolones also include primary or secondary prophylaxis. à Renal failure: creatinine >133 lmol/l and/or urea >11 mmol/l at any moment during the episode. k Area under the ROC-curve for this multivariate model: (CI95: ). JOURNAL OF HEPATOLOGY Outcome Cases were treated for a median of 10 days (IQR 6 12), mostly with ceftriaxone, which was appropriate in 84.5% [92.0%, 84.7%, and 67.6% in the CA, HCR, and NA groups, respectively (p = 0.005)]. Piperacillin tazobactam or imipenem were given in 39 (15.9%) cases with severe clinical conditions, mainly in HCR and NA episodes, actually being MR-Cef 14 (36%). Overall, empirical treatment was considered appropriate (AT) in 206 cases (83.7%). If these 39 cases had been treated with 3rdGC, AT-per protocol would have been 80.7%. The following analysis of mortality excludes 9 patients who were administered AT because they were empirically treated with piperacillin tazobactam or carbapenems. Overall mortality at 30 days was 80 cases (33.8%), 138 patients were still alive (58.2%), 8 had undergone a liver transplant (3.4%), 9 developed a new episode of SBP (3.8%) and 2 were lost to follow-up (0.8%). While mortality in the CA and HCR groups was similar, it was higher among NA episodes (Table 1). Table 4 shows the analysis of mortality at 30 days. Overall mortality at 5 and 90 days was 15.2% and 43.9%, respectively. Predictive factors of mortality were similar to those described for 30 days (data not shown). Multivariate analysis identified nosocomial acquisition of the episode, immunosuppressive therapy, variables related to liver and renal function, a high count of PMN in ascites and either the adequacy of empirical treatment or resistance to 3rdGC as independent predictors for mortality. When non-hospitalized and NA populations were analyzed separately, most of these factors were similar, with the exception of resistance to 3rdGC. Discussion This paper provides a detailed assessment of the problem of MR-Cef in cirrhotic patients with SBP. Our results show high rates of resistance (21.5%) and are broadly in agreement with other recently reported studies [5 7,13]. The recent worldwide emergence of ESBL-producing GNB in the field of infectious diseases [24] is relevant to SBP [1,3]. Some reports have highlighted the presence of ESBL-producing GNB in SBP, and have warned of their potential impact on mortality in relation to inappropriate empirical antibiotic treatment [8,25]. However, their role in SBP has not been well defined. Here, they were found to be present in around 10% of Enterobacteriaceae in the HCR and NA groups. Journal of Hepatology 2012 vol. 56 j

6 Probability of SBP by MR-Cef in hospitalized patients Days Probability of SBP by MR-Cef in non-hospitalized patients Fig. 2. Probability of developing an episode of Spontaneous Bacterial Peritonitis (SBP) by a 3rd-generation cephalosporin-resistant microorganism (MR- Cef) depending on the days after admission (circles) or days of contact with the health-care system (triangles). Some of the reported microbiological changes are related to an increasing role of GPC, due to the widespread use of quinolones for prophylaxis of SBP [9 11]. In this regard, our GPC rate of 35.8% was higher than that reported in classical studies [4], and rose to almost 50% in the group of non-hospitalized patients. The presence of Enterococci in the HCR population was particularly notable. These data contrast with other reports that have noted the prevalence of methicillin-resistant S. aureus in nosocomial SBP. This may be due to local epidemiological factors [12]. Epidemiological changes in the care of cirrhotic patients over the last two decades has led to a new scenario with regards to microbial flora and its resistance. Thus, nosocomially-acquired SBP has recently been associated with higher mortality [6,12,13] than the more traditional community-acquired SBP [4]. Similarly, many patients who become infected outside the hospital have had repeated and significant contact with the healthcare environment, and a new healthcare-related popula- Table 4. Univariate and multivariate analysis for risk factors for 30-day mortality. Risk factor Background Episode of spontaneous peritonitis Unadjusted HR (95% CI) Model 1-adjusted HR (95% CI) Model 2-adjusted HR (95% CI) Age (yr) ( ) - - Sex (M) ( ) - - Hepatocellular carcinoma ( ) - - Immunosuppressive therapy ( ) ( ) ( ) HIV infection ( ) - - Diabetes mellitus ( ) - - Upper gastrointestinal bleeding ( ) ( ) ( ) Child-Pugh score ( ) - - MELD score ( ) - - MELD-Na score ( ) ( ) ( ) Place of HCR vs. CA ( ) - - acquisition NA vs. CA ( ) ( ) ( ) Bacteremia ( ) - - SIRS ( ) - - Systolic blood pressure <90 mmhg ( ) ( ) ( ) Hepatic encephalopathy ( ) ( ) ( ) Jaundice (Bilirubin 68 μmol/l) ( ) - - Laboratory PMN (cells x 10 9 /L) ( ) - - Renal failure ( ) ( ) ( ) Albumin (g/l) ( ) - - Quick time (ratio) ( ) - - Sodium (mmol/l) ( ) - - PMN ascitic fluid (1000 cells x 10 9 /L) ( ) ( ) ( ) Gram-negative etiology ( ) ( ) ( ) Resistance to 3rd-generation cephalosporins ( ) ( ) n.a. Adequate empirical treatment ( ) n.a ( ) Expansion with seroalbumin ( ) - - Bold entries: p <0.05. SIRS, systemic inflammatory response syndrome; CA, community-acquired; HCR, health care related; NA, nosocomially-acquired; PMN, polymorphonuclear leukocytes. Renal failure: creatinine >133 lmol/l and/or urea >11 mmol/l at any moment during the episode of SBP. n.a., not applicable. 830 Journal of Hepatology 2012 vol. 56 j

7 tion with SBP has therefore emerged. A few years ago, Friedman et al. [14] demonstrated that patients with bacteraemia who had had significant contact with the healthcare system presented a microbiological profile that was similar to patients with nosocomial bacteraemia. Notably, these criteria have not been clearly incorporated into SBP, although some authors have suggested their possible significance [13,15,19,20]. The approach of separating CA and HCR enabled better patient stratification in relation to reassessing the empirical antibiotic therapy. HCR patients were halfway between the NA and CA groups with respect to the incidence of MR-Cef. The definition we used to include patients in the HCR group was based on previous reports [14,19,20], but it could be reassessed to provide a better cut-off of sensitivity/specificity. We found a close association between days of contact with the healthcare system and the likelihood of MR-Cef (Fig. 2). In our series, factors that were independently associated with an increased risk of MR-Cef were diabetes mellitus, UGB, NA, previous administration of cephalosporins and a low PMN count in ascites, as reported by others [6,12,13,26,27]. Host variables, including liver function or immunosuppressant condition, did not significantly influence this probability. In our study, the previous use of quinolones showed an association with MR-Cef in the univariate analysis, but it was not an independent predictor of resistance. Thus, we do not think that these results may question the indication of SBP prophylaxis with quinolones. The weight of diabetes mellitus in the CA/HCR groups may be explained by a tendency of this condition to MR-Cef, as it happens to its predisposition to cutaneous and urinary tract infections by Staphylococcus, Enterococcus and ESBL-producing GNB. The influence of UGB in the likelihood of SBP by MR-Cef may be explained by the invasive procedures and antibiotic pressure to which these patients are submitted in the nosocomial setting. The observed association between SBP by MR-Cef and a decreased expression of the inflammatory response, as seen in both the clinical presentation and a lower increase of the leukocyte count in ascites, was of interest. This might be explained by the loss of their virulence or fitness [28] observed in some multiresistant microorganisms. The literature contains very few reports on cellularity levels in ascites according to microbial etiology. Campillo et al. [12] reported higher cellularity in cases due to GNB than in those caused by GPC. The presence of MR-Cef was often associated with a low rate of AT. The classification of episodes used here revealed a progressive decrease in AT. The clinical impact of AT on survival is a controversial issue in many serious infections due to multi-resistant microorganisms, as well as in SBP. Most studies have shown a positive trend when an AT was applied [6,13,25]. Cheong et al. [13] in a large series, including many patients with NA-SBP, found that MR-Cef, carrying with it low rates of AT, was an independent predictor of mortality. In the series of Song et al. [8], focused on SBP by ESLB E. coli and Klebsiella sp., ineffective initial therapy was responsible for a higher rate of treatment failure using 3rdGC. However, other well-conducted studies have failed to confirm this relationship and have suggested that factors related to host condition are what really determines mortality [26,29,30]. In our study, independent factors associated with higher mortality were the patient s hepato-renal functional status, development of a marked inflammatory response during the episode of SBP, nosocomial acquisition and MR-Cef or low rate of AT. Separate analysis of mortality in NA and HCR groups showed similar results, although MR-Cef was not identified as a significant parameter. Our data do warrant a change in policy for the empirical use of ceftriaxone/cefotaxime in the NA group, in which the percentage of resistance and mortality are very high. Among non-hospitalized cases, this change should be targeted at the group of patients with a higher risk of MR-Cef, based on the coexistence of diabetes mellitus and long contact with the healthcare system. Taken together, our data suggest that the development of SBP by MR-Cef may be proportionally followed by a clinical impact on mortality rates, although the outcome seems very dependent on the patient s hepato-renal function. Indiscriminate antibiotic escalation may have potentially deleterious consequences, and therefore a broader antimicrobial therapy should be carefully targeted at selected patients with known risk factors for SBP by MR- Cef. Further prospective studies should assess this problem, to better identify the subgroup of patients for whom appropriate empirical treatment is crucial. Conflict of interest The authors who have taken part in this study declared that they do not have anything to disclose regarding funding or conflict of interest with respect to this manuscript. Financial support Xavier Ariza was supported by a grant from the Hospital Universitari de Bellvitge. Jaime Lora-Tamayo was supported by a grant from the Instituto de Salud Carlos III (FI 09/00943). Acknowledgment We thank Michael Maudsley for revising the English manuscript. References JOURNAL OF HEPATOLOGY [1] Rimola A, Garcia-Tsao G, Navasa M, Piddock LJ, Planas R, Bernard B, et al. Diagnosis, treatment and prophylaxis of spontaneous bacterial peritonitis: a consensus document. International Ascites Club. J Hepatol 2000;32: [2] Runyon BAAASLD Practice Guidelines Committee. Management of adult patients with ascites due to cirrhosis: an update. Hepatology 2009;49: [3] European Association for the Study of the Liver, Ginès P, Angeli P, Lenz K, Møller S, Moore K, et al. EASL clinical practice guidelines on the management of ascites, spontaneous bacterial peritonitis, and hepatorenal syndrome in cirrhosis. J Hepatol 2010;53: [4] Garcia-Tsao G. Spontaneous bacterial peritonitis: a historical perspective. J Hepatol 2004;41: [5] Dupeyron C, Campillo B, Mangeney N, Richardet JP, Leluan G. Changes in nature and antibiotic resistance of bacteria causing peritonitis in cirrhotic patients over a 20 year period. J Clin Pathol 1998;51: [6] Umgelter A, Reindl W, Miedaner M, Schmid RM, Huber W. Failure of current antibiotic first-line regimens and mortality in hospitalized patients with spontaneous bacterial peritonitis. Infection 2009;37:2 8. [7] Acevedo JG, Fernández J, Castro M, García O, Rodríguez de Lope C, Navasa M, et al. Current efficacy of recommended empirical antibiotic therapy in patients with cirrhosis and bacterial infection. J Hepatol 2009;50:S5. [8] Song KH, Jeon JH, Park WB, Park SW, Kim HB, Oh MD, et al. Clinical outcomes of spontaneous bacterial peritonitis due to extended-spectrum beta-lactamase-producing Escherichia coli and Klebsiella species: a retrospective matched case control study. BMC Infect Dis 2009;9:41. Journal of Hepatology 2012 vol. 56 j

8 [9] Campillo B, Dupeyron C, Richardet JP, Mangeney N, Leluan G. Epidemiology of severe hospital-acquired infections in patients with liver cirrhosis: effect of long-term administration of norfloxacin. Clin Infect Dis 1998;26: [10] Fernández J, Navasa M, Gómez J, Colmenero J, Vila J, Arroyo V, et al. Bacterial infections in cirrhosis: epidemiological changes with invasive procedures and norfloxacin prophylaxis. Hepatology 2002;35: [11] Runyon BA. Changing flora of bacterial infections in patients with cirrhosis. Liver Int 2010;30: [12] Campillo B, Richardet JP, Kheo T, Dupeyron C. Nosocomial spontaneous bacterial peritonitis and bacteremia in cirrhotic patients: impact of isolate type on prognosis and characteristics of infection. Clin Infect Dis 2002;35: [13] Cheong HS, Kang CI, Lee JA, Moon SY, Joung MK, Chung DR, et al. Clinical significance and outcome of nosocomial acquisition of spontaneous bacterial peritonitis in patients with liver cirrhosis. Clin Infect Dis 2009;48: [14] Friedman ND, Kaye KS, Stout JE, McGarry SA, Trivette SL, Briggs JP, et al. Health care-associated bloodstream infections in adults: a reason to change the accepted definition of community-acquired infections. Ann Intern Med 2002;137: [15] Angeloni S, Leboffe C, Parente A, Venditti M, Giordano A, Merli M, et al. Efficacy of current guidelines for the treatment of spontaneous bacterial peritonitis in the clinical practice. World J Gastroenterol 2008;14: [16] Ariza J, Xiol X, Esteve M, Fernández Bañares F, Liñares J, Alonso T, et al. Aztreonam vs. cefotaxime in the treatment of Gram-negative spontaneous peritonitis in cirrhotic patients. Hepatology 1991;14: [17] Soriano G, Castellote J, Alvarez C, Girbau A, Gordillo J, Baliellas C, et al. Secondary bacterial peritonitis in cirrhosis: a retrospective study of clinical and analytical characteristics, diagnosis and management. J Hepatol 2010;52: [18] Biggins SW, Kim WR, Terrault NA, Saab S, Balan V, Schiano T, et al. Evidencebased incorporation of serum sodium concentration into MELD. Gastroenterology 2006;130: [19] Merli M, Lucidi C, Giannelli V, Giusto M, Riggio O, Falcone M, et al. Cirrhotic patients are at risk for health care-associated bacterial infections. Clin Gastroenterol Hepatol 2010;8: [20] Venditti M, Falcone M, Corrao S, Licata G, Serra P. Outcomes of patients hospitalized with community-acquired, health care-associated, and hospital-acquired pneumonia. Ann Intern Med 2009;150: [21] Castellote J, Xiol X, Verdaguer R, Ribes J, Guardiola J, Gimenez A, et al. Comparison of two ascitic fluid culture methods in cirrhotic patients with spontaneous bacterial peritonitis. Am J Gastroenterol 1990;85: [22] Clinical and Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing: fifteenth informational supplement M100-S15. Wayne, PA, USA: CLSI; [23] Legrand P, Fournier G, Buré A, Jarlier V, Nicolas MH, Decré D, et al. Detection of extended broad-spectrum beta-lactamases in Enterobacteriaceae in four French hospitals. Eur J Clin Microbiol Infect Dis 1989;8: [24] Ben-Ami R, Rodríguez-Baño J, Arslan H, Pitout JD, Quentin C, Calbo ES, et al. A multinational survey of risk factors for infection with extended-spectrum beta-lactamase-producing enterobacteriaceae in nonhospitalized patients. Clin Infect Dis 2009;49: [25] Park YH, Lee HC, Song HG, Jung S, Ryu SH, Shin JW, et al. Recent increase in antibiotic-resistant microorganisms in patients with spontaneous bacterial peritonitis adversely affects the clinical outcome in Korea. J Gastroenterol Hepatol 2003;18: [26] Song JY, Jung SJ, Park CW, Sohn JW, Kim WJ, Kim MJ, et al. Prognostic significance of infection acquisition sites in spontaneous bacterial peritonitis: nosocomial versus community acquired. J Korean Med Sci 2006;21: [27] Bert F, Andreu M, Durand F, Degos F, Galdbart JO, Moreau R, et al. Nosocomial and community-acquired spontaneous bacterial peritonitis: comparative microbiology and therapeutic implications. Eur J Clin Microbiol Infect Dis 2003;22: [28] Andersson DI, Hughes D. Antibiotic resistance and its cost: is it possible to reverse resistance? Nat Rev Microbiol 2010;8: [29] Singh N, Wagener MM, Gayowski T. Changing epidemiology and predictors of mortality in patients with spontaneous bacterial peritonitis at a liver transplant unit. Clin Microbiol Infect 2003;9: [30] Tandon P, Garcia-Tsao G. Renal dysfunction is the most important independent predictor of mortality in cirrhotic patients with spontaneous bacterial peritonitis. Clin Gastroenterol Hepatol 2011;9: Journal of Hepatology 2012 vol. 56 j