Cirrhotic Patients Are at Risk for Health Care Associated Bacterial Infections

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2010;8:979 985 Cirrhotic Patients Are at Risk for Health Care Associated Bacterial Infections MANUELA MERLI,* CRISTINA LUCIDI,* VALERIO GIANNELLI,* MICHELA GIUSTO,* OLIVIERO RIGGIO,* MARCO FALCONE, LORENZO RIDOLA,* ADOLFO FRANCESCO ATTILI,* and MARIO VENDITTI *Division of Gastroenterology, Department of Clinical Medicine, Department of Infectious Diseases and Tropical Medicine, Sapienza University of Rome, Rome, Italy BACKGROUND & AIMS: Bacterial infections are a frequent and serious burden among patients with cirrhosis because they can further deteriorate liver function. We assessed the epidemiology, risk factors, and clinical consequences of bacterial infections in hospitalized cirrhotic patients. METH- ODS: In a cohort of hospitalized cirrhotic patients (n 150) referred to a tertiary care setting, all episodes of bacterial infections were recorded prospectively. Infections were classified as community-acquired (CA), health care associated (HCA), or hospital-acquired (HA). Site of infection, characteristics of bacteria, and prevalence of antibiotic resistance were reported; consequences for liver function and patient survival were evaluated. RESULTS: Fiftyfour infections were observed among 50 patients (12 CA, 22 HCA, and 20 HA). Bacterial resistance was more frequent among patients with HCA or HA infections (64% of isolates). Mortality was 37% from HA, 36% from HCA, and 0% from CA infections. Independent predictors of infection included a previous infection within the past 12 months (P.0001; 95% confidence interval [CI], 2.2 10.6), model of end-stage liver disease score 15 (P.01; 95% CI, 1.3 6.1), and protein malnutrition (P.04; 95% CI, 1.5 10). Infectious episodes worsened liver function in 62% of patients. Patients with infection more frequently developed ascites, hepatic encephalopathy, hyponatremia, hepatorenal syndrome, or septic shock. Child class C (P.006; 95% CI, 1.67 23.7), sepsis (P.005; 95% CI, 1.7 21.4), and protein malnutrition (P.001; 95% CI, 2.8 38.5) increased mortality among patients in the hospital. CONCLUSIONS: In hospitalized cirrhotic patients, the most frequent infections are HCA and HA; these infections are frequently resistant to antibiotics. As infections worsen, liver function deteriorates and mortality increases. Cirrhotic patients should be monitored closely for infections. Keywords: Sepsis; Survival; Multidrug Resistance; Nutritional Status. Bacterial infections are a frequent and severe complication in cirrhotic patients. 1 Episodes of infection are reported in 40% of hospitalized cirrhotic patients. 2 Infections are associated with a longer hospital stay and a higher risk of death; infectionrelated mortality rate, in fact, ranges between 15% and 19%. 2,3 Most of the infections in cirrhotic patients are caused by enteric bacteria. 4 This suggests that the defense mechanisms of these patients fail to prevent the microorganisms present in the intestinal lumen from reaching the systemic circulation. Sepsis favors acute decompensation of cirrhosis and may lead to hepatic encephalopathy, renal insufficiency, shock, and acute on chronic liver failure. 1 The possible risk factors for infections have been scarcely evaluated in chronic liver disease and a recent review indicated that a Child Pugh C score can be considered the only independent predictor. 5 In most of the studies, bacterial infections in cirrhotic patients are caused, in large prevalence (70% 80%), by gram-negative cocci. In the past decade, however, infections induced by gram-positive bacilli have increased, owing to longterm antibiotic prophylaxis with quinolones recommended for those patients with previous spontaneous bacterial peritonitis (SBP) episodes, 6,7 which prevents infections caused by gramnegative bacilli but not those caused by gram-positive cocci. It recently has been proposed that infections occurring in patients who have had previous recent contact with the health care system can be classified as health care associated (HCA) and may have a worse prognosis. 8 Patients with advanced liver disease frequently need to be hospitalized and therefore may be included in this risk category. No information is available about the prevalence and consequences of HCA infections in cirrhotic patients. The current study consists of a prospective investigation aimed at determining, in a large cohort of hospitalized cirrhotic patients, the following: (1) the prevalence and etiology of community-acquired (CA), HCA, and hospital-acquired (HA) bacterial infections; (2) the risk factors associated with the development of bacterial infections; (3) the short-term clinical consequences of infection; and (4) patient survival 6 months after hospital discharge. Patients and Methods Patients From October 2008 to June 2009, we consecutively enrolled all cirrhotic patients hospitalized at our University Hospital (a tertiary referral center). Diagnosis of cirrhosis was based on liver biopsy, when available, or on obvious clinical, biochemical, or ultrasonographic and endoscopic features. Exclusion criteria were a concomitant human immunodeficiency virus infection, high-dose corticosteroid treatment, and immunosuppressive therapy. The study was approved by the local Abbreviations used in this paper: CA, community-acquired; CI, confidence interval; HA, hospital-acquired; HCA, health care associated; HCC, hepatocellular carcinoma; HRS, hepatorenal syndrome; MDR, multidrug resistant; OR, odds ratio; SBP, spontaneous bacterial peritonitis; SIRS, systemic inflammatory response syndrome. 2010 by the AGA Institute 1542-3565/$36.00 doi:10.1016/j.cgh.2010.06.024

980 MERLI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 11 Ethical Committee Review Board, and written informed consent was signed by all the participants. Clinical Evaluation and Management Demographic, clinical, and biochemical parameters were recorded for each patient in a preformed digitalized dataset. At admission, previous relevant clinical data, such as origin of liver disease, history of alcohol abuse, ascites, encephalopathy, gastrointestinal bleeding, acute or chronic renal failure, and coexistence of other diseases, were recorded. Previous hospitalizations (within 6 mo), episodes of infections diagnosed within the past 12 months, and therapy administered during the past 4 weeks also was noted. The main cause of hospitalization was identified and basal clinical and biochemical parameters were assessed to define the severity of liver disease, renal function, and electrolyte imbalance. Mean arterial pressure, heart rate, respiratory rate, and body temperature were measured and the presence of systemic inflammatory response syndrome (SIRS) was investigated. 9 The severity of liver disease was assessed using the Child Pugh score 10 and the model of end-stage liver disease (MELD) score. 11 Hepatorenal syndrome (HRS) was diagnosed according to international criteria. 12 The nutritional status was assessed in all the patients by anthropometric measurements. 13 Complications of cirrhosis were treated according to recent guidelines. 14 Antibiotic prophylaxis was administered in patients with previous SBP or recent gastrointestinal bleeding. 15,16 Overt hepatic encephalopathy was diagnosed according to the West Haven criteria 17 and treated only when the symptoms were clinically evident. Diagnosis and Management of Infection Bacterial infections were actively looked for in all patients based on the following assessment: (1) medical history reporting symptoms of infection; (2) physical examination focused on symptoms and signs suggestive of infection; (3) laboratory tests including erythrocyte sedimentation rate, C-reactive protein level, polymorphonucleater cell count, hepatic and renal function tests, and urinary sediment; (4) analysis of the ascitic fluid when present; (5) chest radiograph; and (6) abdominal ultrasound. When a bacterial infection was suspected, further investigations (cultures of blood, urine, sputum, ascitic fluid, or purulent secretions) were performed. SBP was defined as a polymorphonucleater cell count greater than 250/mm 3 in the ascitic fluid a positive culture; pneumonia was defined as the presence of radiologic evidence of consolidation plus at least 2 of the following criteria: fever higher than 38 C or hypothermia less than 35 C, dyspnea, cough and purulent sputum, pleuritic chest pain, or signs of consolidation on physical examination. Urinary tract infections, biliary tract infections, cellulitis, and gastroenteritis were all diagnosed according to congruent symptoms and biochemical and imaging parameters following standard criteria. 18 The evidence of a positive blood culture without a recognized site of infection was defined as spontaneous bacteremia. Study Design, Follow-Up Evaluation, and Outcomes A complete patient assessment was performed at hospital admission and at discharge. If a diagnosis of infection was made during hospitalization, the patient assessment was repeated at that time. Main outcomes were modifications of the liver function, development of kidney dysfunction, development of hepatic encephalopathy, gastrointestinal bleeding, ascites, hyponatremia, length of hospital stay, and in-hospital survival. Patients were re-evaluated as outpatients at 1, 3, and 6 months or until death or liver transplant. The study was closed when the last patient enrolled had completed at least the 6-month follow-up evaluation. Definitions Infections were classified as follows. Infections were classified as HA if the diagnosis of infection was made after more than 48 hours of hospital stay. Infections were classified as HCA if the diagnosis was made within 48 hours of hospitalization in patients with any of the following criteria: (1) had attended a hospital or a hemodialysis clinic, or had received intravenous chemotherapy during the 30 days before infection; or (2) were hospitalized for at least 2 days, or had undergone surgery during the 180 days before infection; or (3) had resided in a nursing home or a long-term care facility. 8 Infections were classified as CA if the diagnosis of infection was made within 48 hours of hospitalization and the patient did not fulfill the criteria for HCA infection 8 (ie, had no recent contact with the health care system and was not hospitalized in the past 6 months). Patients with infections were treated immediately with empiric large-spectrum antibiotics, based on the site of infection and according to standard guidelines and local epidemiology. The antibiotic treatment then was modified according to the results of cultures (if available) and in case of treatment failure. All patients with infection underwent a consultation by an infectious diseases specialist with expertise in nosocomial infections. Patients were considered to have SIRS when they fulfilled the criteria established by the most recent international guidelines. Sepsis was diagnosed in the presence of SIRS and a known or highly suspected infection. Septic shock was defined as sepsis with hypotension refractory to intravascular volume loading, associated with perfusion abnormalities that required the use of inotropes. 9 The definition of a multidrug resistant (MDR) pathogen was used to describe a methicillin-resistant Staphylococcus aureus, an Acinetobacter baumannii, an extended-spectrum -lactamases producing gram-negative strain, or any bacterial isolate resistant to at least 3 classes of antimicrobial agents. 19,20 Protein malnutrition was diagnosed when the midarm muscle circumference was below the fifth percentile of the referral standard. 13 Statistical Analysis Each patient was considered only once during the first hospitalization. To evaluate liver function modifications induced by the infection, the tests performed at the time of diagnosis were compared either with recent previous results (when available) or with the liver function test at the first outpatient control after discharge (1 month), when the infection had completely resolved. All the values are reported as means standard deviations; P values less than.05 were considered significant. Data were analyzed as continuous or categoric by using the Student t test for parametric data and the Mann Whitney U test or the Wil-

November 2010 BACTERIAL INFECTIONS IN CIRRHOTIC PATIENTS 981 coxon test for nonparametric data; the chi-square test was used for the comparison of dichotomous data. The logistic regression analysis was used to identify possible predictors of infection and in-hospital mortality. Univariate analysis (log rank) was used to identify prognostic factors of survival, and the variables selected by this analysis were included in a multivariate analysis according to a Cox regression model. The software used for the analysis was NCSS (Number Cruncher Statistical System, Kaysville, UT) 2007. Results Patients Four patients were excluded, 1 for high-dose steroid therapy and 3 because of a recurrence of cirrhosis after liver transplantation. A total of 150 patients was enrolled: 48 women and 102 men, with a mean age of 64 13 years. The origin of cirrhosis was hepatitis C in 51% of cases, alcohol abuse in 19%, cryptogenetic in 15%, autoimmune in 5%, hepatitis B in 4%, and both hepatitis B and C in 3.3% of patients. Child class was A in 30%, B in 39%, and C in 31% of patients. The mean MELD score at admission was 13.5 (range, 5 40). The main reasons for hospital admission were as follows: ascites in 37 patients (25%), elective procedures in 33 (22%), suspected infection in 21 (14%), encephalopathy in 20 (13%), gastrointestinal bleeding in 12 (8%), renal failure in 5 (3%), and other reasons in the remaining 22 patients (15%). Forty-nine patients had clinical evidence of ascites at the time of hospitalization, and 82 were known to have esophageal varices (48 small and 34 large varices). Fiftytwo patients reported a previous documented infection in the past 12 months (19 urinary tract infections, 13 pneumonia, 4 spontaneous bacteremia, 4 biliary tract infections, 3 gastrointestinal, 3 skin infections, 2 SBP, and 6 in other sites). Seventeen patients received antibiotics in the preceding 30 days, 9 because of a recent infectious episode and 8 because of prophylaxis with quinolones. Thirteen patients had been treated previously with transjugular intrahepatic portosystemic shunt. Twenty-six cases had a diagnosis of hepatocellular carcinoma (HCC), 18 were within the Milan criteria, 21 and 8 exceeded the Milan criteria (3 of these patients also had portal thrombosis). The 26 patients with HCC were hospitalized either for diagnostic purposes (3 cases), to receive locoregional treatment (21 cases), or for other complications of cirrhosis (2 patients). A total of 54 episodes of bacterial infection were recorded in 50 patients (33%); 4 patients presented more than one episode of infection during hospitalization. Demographic, clinical, and biochemical characteristics of the patients with and without infection are shown in Table 1. The degree of liver impairment was significantly more severe in the patients with infection. The patients who developed an infection had significantly lower albumin levels (P.001), higher bilirubin levels (P.002), higher creatinine levels (P.001), and lower sodium levels (P.008) than those who did not. Bacterial Infections, Systemic Inflammatory Response Syndrome, and Sepsis Among the 54 episodes of infection, 12 were CA, 22 were HCA, and 20 were HA. The infections most commonly recorded were urinary tract infections (37%), pneumonia (22%), and spontaneous bacteremia (13%). Table 2 shows the frequency of the different sites of infection in the 3 epidemiologic groups. Twenty-six infections were documented microbiologically. Gram-negative organisms, particularly Escherichia coli, were isolated more frequently (62%) than gram-positive ones (38%), but gram-positive bacteria were prevalent in HA infections (64% of isolates). Blood cultures were positive in 12 episodes of infection: 3 urinary infections, 4 cases of cholangitis, 2 cases of pneumonia, 1 case of SBP, and 2 cases of spontaneous bacteremia. As shown in Table 2, 16 patients had an infection caused by MDR bacteria. Six of the 16 patients (37.5%) with MDR pathogens died during their hospital stay (3 HRS, 2 septic shock, 1 variceal bleeding). Among the 17 patients undergoing antibiotic therapy in the preceding 30 days, MDR pathogens were isolated more frequently (5 of 7 infectious episodes; P.002). SIRS was diagnosed in 43 patients and sepsis was diagnosed in 31 patients. The type of infection more often leading to sepsis was pneumonia. SIRS unrelated to infection mostly occurred in those patients undergoing invasive treatments. Risk Factors for the Development of Infection The variables examined as possible risk factors for the development of infection or sepsis are reported in Table 3. At multivariate analysis, a history of previous infection in the past 12 months, a MELD score of 15 or greater, and a diagnosis of protein malnutrition were independent predictors for infections and sepsis (Table 4). Potential risk factors that could be involved in the development of HA infections also were evaluated. The number of invasive procedures during hospitalization was significantly higher (P.02) in patients who developed HA infections (Table 5). We also found that HA infections were associated with hospitalization in rooms in which there was a need to place additional beds (P.0002) (Table 5). Clinical Features A worsening of liver function frequently was observed in patients with infection, especially in those with sepsis. The Child Pugh score deteriorated in 62% of cases with infection (mean increase in Child score, 1.9 0.8; range, 1 2) and in 71% of patients with sepsis (mean increase in Child score, 2.5 0.7; range, 1 4), whereas the MELD score in the same patients showed a mean increase of 3.8 2.7 (range, 1 4) in patients with infection and of 4.9 2.7 (range, 1 12) in those with sepsis. The parameters that deteriorated more often were as follows: bilirubin ( 4.4 12 mg/dl), albumin ( 0.34 0.9 g/dl), creatinine ( 0.6 1.8 mg/dl), and serum sodium ( 3.3 6.1 meq/l) levels, respectively. Moreover, episodes of ascites, hepatic encephalopathy, hyponatremia, HRS, and septic shock were more frequent in patients with infection as compared with those who were not infected (Table 6). Outcomes Overall in-hospital mortality was 12%; 14 patients with and 4 patients without infection died (28% vs 4%; P.00007). The mortality rate was higher in the patients with sepsis (12 of 31; 38%) than in those with infection but no sepsis (2 of 19, 10%; P.00000).

982 MERLI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 11 Table 1. Demographic, Clinical, and Biochemical Characteristics of Patients Included in the Study Classified According to the Presence of Infection During Their Hospitalization Variables Patients without infections (n 100) Patients with infections (n 50) P value Age, y 63 13 65 13.3 Male/females 64/36 38/12.1 Active alcohol abusers, n (%) 21 (21) 14 (28).6 Origin of liver disease Alcohol, n (%) 19 (19) 10 (20).8 HCV-related, n (%) 56 (56) 21 (42).1 HBV-related, n (%) 6 (6) 5 (10).37 Other, n (%) 19 (19) 14 (28).09 Main cause of admission Elective procedure, n (%) 31 (31) 2 (4).001 Ascites, n (%) 30 (30) 7 (14).001 Suspected infection, n (%) 0 (0) 21 (42).001 Encephalopathy, n (%) 11 (11) 9 (18).16 Gastrointestinal bleeding, n (%) 9 (9) 3 (6).4 Renal failure, n (%) 2 (2) 3 (6).14 Other, n (%) 17 (17) 5 (10).15 HCC, n (%) 20 (20) 7 (14).36 MELD score 11.8 4.5 17.1 7.4.001 Child Pugh score 7.6 2.0 8.7 2.1.001 Mean arterial pressure, mm Hg 88 11 84 20.08 Heart rate, beats/min 75 13 82 14.001 Respiratory rate, breaths/min 15 2 18 4.001 Temperature, 37.5 C 6 (6) 37 (74).016 White blood cells, cell/mm 3 5297 2754 9229 5973.001 Hemoglobin level, g/dl 11.6 2.3 10.9 3.6.13 Platelets, num/mm 3 10 3 120 95 123 88.9 Alanine aminotransferase level, IU/L 57 62 58 79.9 Serum bilirubin level, mg/dl 3.1 5.6 7.6 12.0.002 Serum albumin level, mg/dl 3.3 0.6 2.7 0.7.001 International normalized ratio 1.5 1.3 1.5 0.4.54 Serum creatinine level, mg/dl 0.91 0.44 1.57 1.58.001 Blood urea nitrogen level, mg/dl 23.2 18.2 37.9 30.5.001 Serum sodium level, mmol/l 135.7 5.1 130.0 19.6.008 Serum potassium level, mmol/l 4.4 0.7 4.2 0.2.7 Sedimentation rate, mm/h 21 19 37 26.001 HCV, hepatitis C virus; HBV, hepatitis B virus. Patient mortality was related to the epidemiology of infection: patients with HA infections, in fact, had a mortality rate (37%) similar to that of patients with HCA infections (36%), whereas none of the patients with CA infections died (P.046). Univariate analysis showed that sepsis (P.00000), bacterial infections (P.00002), MELD score of 15 or greater (P.03), Child class C (P.00004), protein malnutrition (P.000001), development of antibiotic resistance (P.00009), and hyponatremia (P.002) were related significantly to mortality. Multivariate analysis selected Child class C (P.006; odds ratio [OR], 6.3; 95% confidence interval [CI], 1.67 23.7), sepsis (P.0056; OR, 6.01; 95% CI, 1.7 21.46), and protein malnutrition (P.0004; OR, 10.44; 95% CI, 2.8 38.5) as independent predictors of in-hospital death. Causes of in-hospital death in patients with infections were HRS (6 patients), septic shock (5 patients), hemorrhagic shock (2 patients), severe arrhythmia causing cardiovascular arrest (1 patient). All of the 4 patients who died without infection died of end-stage liver failure, and 3 of them also had a diagnosis of advanced HCC. The overall mortality in the following 6 months also was evaluated. The mortality rate at 3 months was 34% versus 6%, and at 6 months was 50% versus 11% in patients with infection as compared with those without infection (Supplementary Figure 1). The main causes of death after hospital discharge were end-stage liver failure (7 patients), variceal bleeding (4 patients), renal failure (4 patients), septic shock (2 patients), and myocardial infarction (1 patient); 5 patients underwent liver transplantation. Discussion Cirrhosis presently may be considered one of the most common immunodeficiency-acquired conditions and previous studies reported an infection rate of 40% to 50% among hospitalized cirrhotic patients, with consequent significant morbidity and mortality rates. 2,5 In the present study, we found the following: (1) the large majority of bacterial infections in hospitalized cirrhotic patients are HCA or HA; (2) the high prevalence of these epidemiologic categories contributes to a higher

November 2010 BACTERIAL INFECTIONS IN CIRRHOTIC PATIENTS 983 Table 2. Characteristics of the 54 Episodes of Infections According to the Epidemiology Classification CA (n 12) HCA (n 22) HA (n 20) Sites of infection Urinary tract infections, n (%) 4 (33) 4 (18) 9 (45) Pneumonia, n (%) 3 (25) 5 (23) 4 (20) Spontaneous bacteriemia, n (%) 1 (8) 4 (18) 2 (10) Spontaneous bacterial 1 (8) 2 (9) 1 (5) peritonitis, n (%) Biliary tract infections, n (%) 1 (8) 2 (9) 1 (5) Other infections, n (%) (eg, 2 (16) 4 (18) 2 (10) skin, gastrointestinal, lymphangitis, bursitis) Microbiologically documented 4 (33) 11 (50) 11 (55) infections Isolated pathogens Staphylococcus aureus 0 1 4 Coagulase-negative 1 1 3 Staphylococci Enterococcus faecalis 1 0 2 Enterobacteriaceae (E coli, 2 9 2 Klebsiella, Proteus) MDR 2 (16) 9 (41) 5 (23) prevalence of antibiotic-resistant strains and a worse outcome; (3) in addition to the severity of the liver disease, protein malnutrition is an important independent risk factor for the development of infections in cirrhosis; and (4) infection and sepsis induce liver deterioration and increase the risk of hepatic encephalopathy, renal failure, hyponatremia, and mortality in cirrhotic patients. In our experience, the most common sites of infection were the urinary (35%) and the lower respiratory (22%) tracts. These latter infections were those more frequently leading to a sepsis syndrome. In all the patients hospitalized with clinical evidence of ascites (49 cases), the peritoneal fluid was analyzed for neutrophil cell count and bacterial culture, and 4 (8%) patients Table 3. Variables Associated With Infection at Univariate Analysis in the Patients Included in the Study Variables Patients without infections (100) Patients with infections (50) P value Previous hospitalizations 37 (37) 33 (66).006 (past 6 months), n (%) Previous infections (past 26 (26) 26 (52).002 12 months), n (%) Child Pugh class C, n (%) 24 (24) 22 (44).012 MELD score, 15,n(%) 25 (25) 25 (50).002 Protein malnutrition, n (%) 19 (19) 26 (52).001 Transjugular intrahepatic 5 (5) 12 (24).02 portosystemic shunt, n(%) Recent gastrointestinal 6 (6) 4 (8).64 bleeding ( 7 d),n(%) Antibiotic therapy in the 9 (9) 8 (16).2 preceding 30 days Diabetes, n (%) 29 (29) 16 (32).8 Table 4. Variables Independently Associated With Infection and Sepsis at Multivariate Analysis Previous infections (past 12 months) Infections OR, 4.7 95% CI, 2.2 10.6 P.000 MELD score, 15 OR, 2.8 95% CI, 1.3 6.1 P.001 Protein malnutrition OR, 4 95% CI, 1.5 10 P.004 Sepsis OR, 3.4 95% CI, 1.3 8.1 P.007 OR, 4.4 95% CI, 1.8 10 P.001 OR, 4 95% CI, 1.5 10 P.004 had a diagnosis of SBP, a rate similar to that reported previously in other Italian studies. 3 Enteric gram-negative cocci were the microorganisms more frequently isolated, whereas gram-positive ones were more frequent in the HA infection group. It has recently been proposed that antibiotic resistances are increasing in hospitalized cirrhotic patients and this has been attributed to the larger diffusion of quinolone prophylaxis for SBP. 6 In the present study we found a consistent (64%) number of MDR pathogens among the isolates. This high prevalence was probably related to the great number of HCA and HA infections that were diagnosed in our series, because the patients with advanced liver disease frequently are in need of hospital care. Previous studies, performed on noncirrhotic patients, showed that HA and HCA infections are caused more frequently by antibiotic-resistant bacteria and are associated with a more severe clinical course than CA infections. 8,22 A different empiric treatment for this new epidemiologic category has been suggested. The present prospective and observational study reports data on HCA infections in cirrhotic patients. We confirm that, as in other categories, in cirrhotic patients HCA infections also are similar to HA infections with regard to the prevalence of antibioticresistant pathogens, severe clinical course, and unfavorable outcome (in-hospital mortality in our series was 37% in HCA, 36% in HA, and none of the patients with CA infections died). In a recent meta-analysis, only the severity of liver disease has been Table 5. Prevalence of Factors That May Be Involved in the Development of HA Infections Variables Patients without infections (100) Patients with HA infections (16) P value Patients undergoing 66 (74) 14 (75).08 invasive procedures, n(%) Invasive procedures, 1.77 1.1 2.4 1.7.02 number per patient a Patients hospitalized in a room with an additional bed, n (%) 19 (19) 10 (62).001 a Invasive procedures were as follows: paracentesis, dialysis, urinary catheter, endoscopic variceal ligation, sclerotherapy, endoscopic retrograde cholangiopancreatography, percutaneous transhepatic cholangiography, locoregional treatments of HCC, and placement of a transjugular intrahepatic portosystemic shunt.

984 MERLI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 11 Table 6. Clinical Complications in 150 Cirrhotic Patients With and Without Infections and With and Without Sepsis During Their Hospitalization Patients with infection (50) Patients without infection (100) P value Patients with sepsis (31) Patients without sepsis (119) P value HRS type 1, n (%) 7 (14) 0.001 5 (16) 2 (1.6).001 HRS type 2, n (%) 4 (8) 2 (2).05 4 (13) 2 (1.6).004 Hepatic encephalopathy, n (%) 19 (38) 15 (15).0015 11 (35) 23 (19).05 Ascites, n (%) 33 (66) 51 (51).08 23 (74) 61 (51).02 Variceal bleeding, n (%) 7 (14) 6 (6).2 5 (16) 8 (6.7).2 Hyponatremia, n (%) 19 (38) 14 (14).001 13 (42) 8 (6.7).001 Shock, n (%) 6 (12) 0.001 6 (19) 0.001 In-hospital mortality, n (%) 14 (28) 4 (4).001 12 (38) 6 (5).001 rewarded as a reliable risk factor for infections and sepsis. 5 When the factors associated with both these conditions were examined, we could confirm that the severity of the liver disease (indicated by a MELD score of 15) is a good predictor, but we also found that additional parameters (a previous episode of infection in the past 12 months and protein malnutrition) may allow identification of a subgroup of cirrhotic patients at high risk of infection. All these parameters are easy to identify at hospital admission through the clinical history, simple anthropometric measurements, and biochemical data. Malnutrition is known to be a negative prognostic factor for in-hospital morbidity and mortality in cirrhotic patients 23 and increases the rate of infection in liver recipients. 24 As previously reported, the consequences of infection/sepsis in cirrhotic patients are highly detrimental. 3 The liver function deteriorated in about two thirds of the patients who developed an infection: these patients showed a worsening of the main indicators of hepatic function (bilirubin and albumin) and a higher risk of encephalopathy, ascites, renal dysfunction, hyponatremia, and hemodynamic shock. The mechanism through which infections induce a further deterioration of liver function in cirrhotic patients has not been investigated extensively as yet. Infections, and especially sepsis, are accompanied by a downright cytokine storming, mostly interleukin-6 and tumor necrosis factor-. These latter are the major hepatic triggers for the production of acute-phase proteins. 25 It is conceivable that a cirrhotic liver facing an infection or an increased endotoxin blood level may be overwhelmed by an increased demand of acute-phase reaction protein, 26 and this may lead to an exhaustion of the hepatocytes reserve function. It is worth noting that the in-hospital mortality rate was higher in those patients who had an episode of infection during hospitalization. Moreover, survival was worse in this group even for those who were discharged from the hospital. The main causes of death in this setting were liver failure, variceal bleeding, and renal failure: the episode of infection by deteriorating liver function seems to accelerate the natural history of the disease. In conclusion, the results of our study show that the majority of infections in hospitalized cirrhotic patients are HCA. Severe liver disease, malnutrition, and infection in the past 12 months are parameters that identify patients at higher risk. Because infection and sepsis are highly detrimental in hospitalized patients with cirrhosis and frequently are caused by resistant pathogens the identification of the criteria to select a high-risk group may help in planning a different empiric antibiotic approach in this category. Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at www.cghjournal.org, and at doi:10.1016/ j.cgh.2010.06.024. References 1. Gustot T, Durand F, Lebrec D, et al. Severe sepsis in cirrhosis. Hepatology 2009;50:2022 2033. 2. Borzio M, Salerno F, Piantoni L, et al. Bacterial infection in patients with advanced cirrhosis: a multicentre prospective study. Dig Liver Dis 2001;33:41 48. 3. Fasolato S, Angeli P, Dallagnese L, et al. Renal failure and bacterial infection in patients with cirrhosis: epidemiology and clinical features. Hepatology 2007;45:223 229. 4. Wong F, Bernardi M, Balk R, et al. International Ascites Club. Sepsis in cirrhosis: report on the 7th meeting of the international ascites club. Gut 2005;54:718 725. 5. Christou L, Pappas G, Falagas ME. Bacterial infection-related morbidity and mortality in cirrhosis. Am J Gastroenterol 2007; 102:1510 1517. 6. Fernández J, Navasa M, Gómez J, et al. Bacterial Infection in cirrhosis: epidemiological changes with invasive procedures and Norfloxacin prophylaxis. Hepatology 2002;35:140 148. 7. Fernández J, Ruiz del Arbol L, Gómez C, et al. Norfloxacin vs ceftriaxone in the prophylaxis of infections in patients with advanced cirrhosis and haemorrhage. Gastroenterology 2006;131: 1049 1456. 8. Venditti M, Falcone M, Corrao S, et al. Study Group of the Italian Society of Internal Medicine. Outcomes of patients hospitalized with community-acquired, health care-associated, and hospitalacquired pneumonia. Ann Intern Med 2009;150:19 26. 9. Dellinger RP, Levy MM, Carlet JM, et al. Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock: 2008. Crit Care Med 2008;36:296 327. 10. Pugh RN, Murray-Lyon IM, Dawson JL, et al. Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 1973; 60:646 649. 11. Kamath PS, Wiesner RH, Malinchoc M, et al. A model to predict survival in patients with end-stage liver disease. Hepatology 2001;33:464 470. 12. Salerno F, Gerbes A, Ginès P, et al. Diagnosis, prevention and

November 2010 BACTERIAL INFECTIONS IN CIRRHOTIC PATIENTS 985 treatment of hepatorenal syndrome in cirrhosis. Gut 2007;56:1310 1318. 13. Merli M, Riggio O, Dally L. Policentrica Italiana Nutrizione Cirrosi. Does malnutrition affect survival in cirrhosis? Hepatology 1996; 23:1041 1046. 14. Garcia-Tsao G, Lim JK. Members of Veterans Affairs Hepatitis C Resource Center Program. Management and treatment of patients with cirrhosis and portal hypertension. Am J Gastroenterol 2009;104:1802 1829. 15. Runyon BA. Management of adult patients with ascites due to cirrhosis: an update. Hepatology 2009;49:2087 2107. 16. Garcia-Tsao G, Sanyal AJ, Grace ND, et al. Prevention and management of gastroesophageal varices and variceal hemorrhage in cirrhosis. Hepatology 2007;43:922 933. 17. Conn HO, Leevy CM, Vlahcevic ZR, et al. Comparison of lactulose and neomycin in the treatment of chronic portal systemic encephalopathy. A double blind controlled trial. Gastroenterology 1977; 72:573 583. 18. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health care-associated infection and criteria for specific types of infections in the acute care setting. Am J Infect Control 2008;36:309 332. 19. Falagas ME, Koletsi PK, Bliziotis IA. The diversity of definitions of multidrug-resistant (MDR) and pandrug-resistant (PDR) Acinetobacter baumannii and Pseudomonas aeruginosa. J Med Microbiol 2006;55:1619 1629. 20. Rupp ME, Fey PD. Extended spectrum beta-lactamase (ESBL)- producing Enterobacteriaceae: considerations for diagnosis, prevention and drug treatment. Drugs 2003;63:353 365. 21. Mazzaferro V, Regalio E, Doci R, et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. N Engl J Med 1996;334:693 699. 22. Friedman ND, Kaye KS, Stout JE, et al. Health care associated bloodstream infections in adults: a reason to change the accepted definition of community-acquired infections. Ann Intern Med 2002;137:791 797. 23. Sam J, Nguyen GC. Protein-calorie malnutrition as a prognostic indicator of mortality among patients hospitalized with cirrhosis and portal hypertension. Liver Int 2009;29:1396 1402. 24. Merli M, Giusto M, Gentili F, et al. Nutritional status: its influence on the outcome of patients undergoing liver transplantation. Liver International 2010;30:208 214. 25. Trautwein C, Boker K, Manns MP. Hepatocyte and immune system: acute phase reaction as a contribution to early defence mechanism. Gut 1994;35:1163 1166. 26. Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl J Med 1999;340:448 454. Reprint requests Address requests for reprints to: Professor Manuela Merli, MD, Gastroenterologia II, Dipartimento di Medicina Clinica, Sapienza Università di Roma, Viale dell Università 37, 00185 Rome, Italy. e-mail: manuela.merli@uniroma1.it; fax: (39) 064453319. Acknowledgments The authors thank Mrs Maria Teresa Barbieri for assistance with the English revision. Conflicts of interest The authors disclose no conflicts. Funding Supported by a research grant C26FO9BMAN from Sapienza University of Rome.

November 2010 BACTERIAL INFECTIONS IN CIRRHOTIC PATIENTS 985.e1 Supplementary Figure 1. Survival in 150 hospitalized cirrhotic patients according to the presence of infection. Curves were estimated using the Kaplan Meier method. A period of 6-months follow-up was examined. Patients with diagnosis of infection at the time of enrollment showed a higher rate of mortality (P.0001).