A incidence of bacterial and fungal infections,

Transcription

1 Prospective Study Comparing the Efficacy of Prophylactic Parenteral Antimicrobials, With or Without Enteral Decontamination, in Patients With Acute liver Failure Nancy Rolando, *Jeremy J. Wade, t Arie Stangou, * Alexander E.S. Gimson, * Julia Wendon, *John Philpott-Howard, t Mark W. Casewell, t and Roger Williams * The efficacy of prophylactic parenteral antibacterials, with or without selective decontamination of the digestive tract, was compared in patients with acute liver failure (ALF) or severe acetaminophen hepatotoxicity. One hundred eight patients were randomized on admission to receive intravenous ceftazidime and flucloxacillin, plus either oral and enteral decontamination with colistin, tobramycin, and amphotericin B (group I), or enteral amphotericin B alone (group ). The two groups were comparable with respect to age, gender, etiology, coma grade on admission, international normalization ratio, presence of renal failure, Acute Physiology and Chronic Health Evaluation I score, and indicators of poor prognosis. Patients were monitored for clinical and microbiological evidence of infection. There were 5 episodes of infection in 0 of 47 patients ( %) in group and 7 episodes in of 6 patients (0%) in group. No differences in incidence, site, and causative organisms of infection were observed between the two groups. Overall, the incidence of infection was significantly higher in patients who developed encephalopathy than in those who did not. In patients who on arrival were not encephalopathic, the development of infection was associated with progression to coma. Duration of Liver Intensive Care Unit (LICU) stay was an independent risk factor for the development of infection. Parenteral antibiotics are effective at reducing the risk of infection in patients with ALF; enteral decontamination provided no additional benefit. Copyright by the American Association for the Study of Liver Diseases cute liver failure (ALF) is associated with a high A incidence of bacterial and fungal infections, with an infection-related mortality of 30% to 50% in most Early recognition of infection can be difficult in these patients. Clinical parameters such as the peripheral white blood cell count (WCC) and temperature may be normal, even in the presence of microbiologically confirmed infection. This has encouraged the use of antimicrobial prophylaxis and selective decontamination of the digestive tract (SDD) in this group of patient^.^ In the intensive care unit setting, SDD has been shown to eliminate potentially pathogenic aerobic bacteria and fungi from the gastrointestinal tract, while preserving anerobic flora and reducing the incidence of nosocomial pneumo- From the *Institute of Liver Studies and the tdulwich Public Health Laboratory and Medical Microbiology, King s College School of Medicine and Dentistry, London SE5 9PJ, England. Address reprint requests to: Professor Roger Williams, CBE, Institute of Liver Studies, King s College Hospital, Denmark Hill, London SE5 9RS, England. Copyright by the American Associution for the Study of Liver Diseases /96/00-000$3.00/0 nia in high-risk patient^.^-^ Other workers have used systemic parenteral and enteral antimicrobial regimens (SPEAR), in which systemic antibacterials are given as prophylaxis against infection during the first few days of SDD, until gut decontamination is established. In a previous study from this unit,4 patients with ALE who were administered SPEAR had a lower incidence of infection compared with the control group. In that study, cefuroxime rather than ceftazidime was used to provide improved Gram-positive antimicrobial cover as Staphylococcus aureus was shown to be an important pathogen at that time. The clinical benefit of SPEAR seemed to be related to the early administration of antimicrobials rather than to the administration of the full regimen of SPEAR per se, because there was no significant difference in infection rates between patients receiving the SPEAR and patients receiving only early antibacterials; however, the numbers of patients in each group were small. Therefore, the prospective study reported here was undertaken to compare the efficacy of the early use of parenteral antibacterials, with or without enteral decontamination, in ALF patients. 8 Liver Transplantation and Surgq, Vol, No (January), 996: pp 8-3

2 Antimicrobial Prophylaxis in ALF 9 Materials and Methods One hundred eight consecutive patients admitted to the Liver Intensive Care Unit (LICU) at King s College Hospital, from January 99 to September 993, with ALF or severe acetaminophen hepatotoxicity, were enrolled into the study. Patients were excluded if pregnant, or under 8 years of age. Informed written consent was obtained from the patient or next of kin in each case. Patients were randomized on admission to the LICU to receive either the group or group treatment regimen. Group patients received the following: ceftazidime, intravenous (IV) g every 8 hours plus flucloxacillin (an isoxazolyl penicillin), IV 500 mg every 6 hours for 5 days; and for SDD a liquid suspension of colistin (00 mg), tobramycin (80 mg), and amphotericin B (500 mg) was administered through a nasogastric tube every 6 hours. For decontamination of the oropharynx, a paste (Orabase; Bristol- Myers Squibb, New York, NY) containing % of each of the SDD antimicrobials was applied to the mouth every 6 hours. Group patients received the same V antibacterial regimen and, in addition, amphotericin B in suspension, administered through a nasogastric tube every 6 hours. They did not receive enteral colistin and tobramycin, or oropharyngeal decontamination. SDD was administered until discharge from the Liver Failure Unit or until liver transplantation. In both groups, females received clotrimazole (5 g, as 0% vaginal cream) on admission and once weekly thereafter. Clinical and microbiological monitoring included daily blood, sputum, and urine cultures. The distal tips of all intravascular catheters and, when available, ascitic, wound, and drainage fluids were submitted for examination using standard microbiological techniques. To avoid nonessential manipulations, which can provoke an increase in intracranial pressure in patients with cerebral edema, stool specimens and rectal swabs were not performed. If patients developed overwhelming sepsis and did not improve after eight doses of the IV antibacterials, the systemic antibacterial regimen was changed as indicated after a clinical and microbiological assessment. For each patient, the APACHE scorelo was calculated on admission and during each episode of infection. Definitions and Trial End Points Infections were defined and categorized according to the site and the type of pathogen isolated. Infections were considered early if they occurred before day 4 after admission and late if they occurred on day 4 or thereafter. Bacteremia was defined as the isolation of S aureus, Candida spp or aerobic Gram-negative rods from at least one set of blood cultures. Isolation from two sets ofblood cultures, or one set in the presence of a positive culture from a known site ot infection, were required for other isolates to be considered significant. Chest infection was diagnosed when new infiltrates appeared on the chest radiograph in the presence of hypoxemia (Po < 0 KPa), and a pathogen was isolated in heavy growth from purulent sputum (ie, sputum with more than five pus cells per high-power field) or endotracheal aspirates. Clinical episode of chest infection was defined as the presence of new infiltrates in the chest radiograph with concomitant pyrexia more than 38 C, WCC greater than 5 X 09/L, purulent sputum, and hypoxemia in the absence of positive cultures. Urinary tract infection, in the absence of a urinary catheter, was defined by the isolation of a pure growth of greater than lo4 bacteria/ml on one occasion or greater than lo4 yeast/ml on two occasions. When a urinary catheter was in situ, infection was diagnosed on the basis of a pure growth of greater than lo5 organisms/ml plus one or more clinical indicators of infection (temperature > 37.5 C or a WCC > 5 X 09/L) or a blood culture yielding the same organism. Renal failure was defined as () urine output less than 479 ml/4 h or 5 59 ml/8 h, () blood urea 36 mmol, or (3) serum creatinine greater than 30 Fmol/L. Line-associated infection was defined as a yield of 5 or more colonies from an intravascular catheter tip, according to the method and criteria of Maki et al. Fungal infection was diagnosed when fungi were cultured from a significant site (blood, bronchial washings, urine) or when fungi were isolated from two or more superficial (mucosal) sites in the presence of clinical evidence of infection, during or after the second week of hospitalization. Clinical evidence of fungal infection included pyrexia unresponsive to antibacterial therapy, increased WCC, deterioration in coma grade, and increase of the INR after an initial impr~vernent.~ End points of the trial were death, discharge from the LICU, and in those patient undergoing liver transplantation, survival at 30 days after enrollment into the study. Ethical Approval This study was conducted after approval by the Research Ethics Committee of King s College Hospital. Statistical Analysis x analysis was performed using Yates correction for continuity for small numbers. Means were compared using Student s t-test, and medians, using the Wilcoxon Signed-Rank test. Univariate analyses and multiple logistic regression analyses were performed, and relative risks determined, using the STATA statistical package (Stata Corporation, College Station, TX). Results One hundred eight patients were randomized into the two groups. The groups were comparable in age, gender, etiology, coma grade on admission, maximum coma grade, presence of renal failure, intemational normalized ratio (INR), APACHE I score, and indicators of poor prognosis. There were 5 males and 56 females with a mean age of 9 years (range, 8 to 69). ALF was caused by acetaminophen hepatotoxicity in 86 cases, viral infection in 5, drug toxicity in 3, and other causes in 4 (Table ). The maximum

3 0 Rolando et al Table. Patient Characteristics on Admission to the Liver Failure Unit Gender M/F* Age meanlrange (yr) Etiology* Acetaminophen toxicity Non-A non-6 hepatitis Hepatitis B Hepatitis A Other drug toxicity Others Indicators of poor prognosis* Renal failure* on admission INR on admission7 APACHE II scoret Coma grade* 0 I II Ill IV Group Group (n = 47) (n = 6) / *Data expressed as number of patients. tdata expressed as mean INRon admission was 5.3, mean 6.4; the maximum INR reached was 5.3, mean 7.9. Of the 47 patients in group the maximum coma grades were I,,, and IV, in 5, 5, 0, and 7 patients, respectively. Of the 6 patients in group, the maximum coma grades reached were I,,, and IV in, 3, 3, and 33 patients, respectively. Fortyeight of 08 patients were not encephalopathic on admission; 30 of these did not progress to coma, 0 in group and 0 in group. Twenty patients developed renal failure after admission, 0 in each group. Sixty patients required mechanical ventilation; 9 of 47 (40%) in group and 4 of 6 (39%) in group. Overall there were 3 episodes of microbiologically confirmed infection in 08 patients; 5 in 0 of 47 (%) patients in group, and 7 in of 6 (0"h) patients in group (P = NS) (Table ). There were 0 bacteremias; two in group and eight in group (P = NS). Both of the bacteremias in group and two of the eight in group were early. The main organisms causing bacteremia were Entevococcus spp. and coagulase-negative staphylococci (Table ). There were episodes of chest infection; 7 in group and 5 in group (P = NS). Eighty-three percent of chest infections were bacterial; of these, 70% were caused by Gram-negatives and 30% by S aureus. Two episodes of chest infection, one in each group, were caused by Candida albicans. There were no significant differences between the groups in the numbers, or causative organism of, chest infection. There were no significant differences in the incidence of clinical episodes of chest infection: 8 of 47 ( 7%) in group and 6 of 6 (6%) in group. The main causes of IV line infections were Entevococcus faecium and Klebsiella spp, which caused four and two episodes, respectively. A patient in group developed an intra-abdominal collection; blood cultures and aspirated pus yielded Entevococcus faecalis. The timing of infections did not differ in the two groups. Microbio- Table. Sites and Causative Organisms of 3 Episodes of Microbiologically Confirmed Infection in Patients 7- ~ ~ Bacteremia Chest Infection* UT Infection IV Line Infection Other Infections CN staphylococci o/o 00 Staphylococcus aureus o/ Enterococcus faecium Enterococcus faecalis 'viridans' streptococci Escherichia coli Klebsiella spp /I 00 Pseudomonas spp Xanthomonas maltophilia Candida albicans 0 /o NOTE. Data expressed as number of patients (group /group ) with infection. Abbreviations: CN, coagulase-negative; UT, urinary tract; IV, intravascular. *There were 4 clinical episodes of chest infection.

4 Antimicrobial Pr-ophylnxis in ALF logically confirmed infection occurred early ( 5 3 days) in 7 patients (9 episodes) in group, and in 4 patients (4 episodes) in group ; late infections (> 3 days) occurred in 5 patients (6 episodes) in group and 9 patients (3 episodes) in group. Ten episodes of infection were caused by multiply resistant bacteria: 3 episodes in patients in group and 7 episodes in 5 patients in group. Five episodes were caused by vancomycin-resistant Efaecium and caused by Stenotrophomonas (Xanthomonas) maltophilia; 4 episodes were caused by Klebsiella pneumonilze resistant to all aminoglycosides (including tobramycin), all cephalosporins (including ceftazidime) and sensitive only to the carbapenem imipenem. These were shown to be producers of the cephalosporinase SHV-5 (Desai et al, unpublished observations, 995). All S aureus and half the coagulase-negative staphylococci isolated were sensitive to methicillin, and therefore were considered to be sensitive to flucloxacillin. The relationship between infection and the APACHE I scores was assessed. For the patients who became infected, the median APACHE I score on admission was 4, and when infected was 9 (P =.00). A univariate analysis was performed to compare the patients who developed confirmed infections with the 86 who did not. Patients who developed infection were significantly older (median ages, 40.5 v 7 years; P =,048, were more likely to be receiving inotropes on admission (relative risk [RR],.54; 95%confidenceinterval [CI],.7 to5.55; P =,04, and significantly more required incubation (RR, 6.0; 95% C,.88-9.; P =,00). The median maximum coma grade for the infected patients was significantly higher than for noninfected patients (4 v.5; P =,00). Infected patients had significantly longer ITU admissions (median, 9.5 v 3 days; P =,000). Multiple logistic regression analysis identified only the duration of LICU stay as independently predictive of the development of proven infection (odds ratio,.5; 95% CI.06 to.399; P <,000). A further univariate analysis including both microbiologically confirmed and clinically suspected infections was performed to identify any further associations. The 4 patients who developed either confirmed or clinical episodes of infection were compared with the 66 who did not. Encephalopathy on admission was more frequent in patients who developed infection than in those who did not (79% v 4% patients; RR,.9; 95% CI,.6 to 5.5; P <,000). Patients who subsequently developed infections had significant lower mean arterial pressures on admission than those who did not (median, 8.5 v 95 mm Hg; P =,005). Seventy-eight patients out of the 08 enrolled developed encephalopathy, 37 in group and 4 in group (P = NS). There were 36 males and 4 females, with a mean age of 30 years (range, 8 to 69 years). ALF was caused by acetaminophen toxicity in 56, viral infection in 5 and other causes in 7. The maximum INR reached was 5.3, mean, 8.. The maximum coma grades reached were I,,, and IV in 7, 8, 3, and 60 patients, respectively. Fifty-eight of the 78 patients had renal failure, and 60 required mechanical ventilation (Table 3). Within this subgroup of 78 patients, there were no significant differences between those allocated to group and in relation to gender, age, etiology, INR, maximum coma grade developed, or outcome. The incidences, sites, and causative organisms of infection did not differ significantly between the groups. Twenty-nine episodes of microbiologically-confirmed infection occurred in 0 of these 78 patients, 0 of 37 (7%) in group, and 0 of 4 (4%) in group (P = NS). Twenty-four patients had clinical episodes of chest infection, 8 of 37 (%) in group, and 6 of 4 (39%) in group (P = NS). To assess the influence of encephalopathy on infection and outcome a univariate analysis compared the 78 patients who developed encephalopathy with the 30 who did not. Patients with encephalopathy had significant more episodes of microbiologically confirmed infection (6% v 7% patients; RR, 3.8; 95% CI, 0.96 to 5.5; P =.08). Combining all confirmed and clinical episodes of Table 3. Characteristics of 78 Patients With ALF Age mean range (yr) Gender F/M* Acetaminophen toxicity* lnrt Mean APACHE II score Renal failure* Ventilated* Transplanted* Overall mortality* *Data expressed as number of patients. tdata expressed as mean (range). Group I Group II (n = 37) (n = 4) 33/ (8-6) 36/(8-69) / (.7-5) 8.3 (.7-5)

5 Rofando et a infection encephalopathic patients had significantly more infections than nonencephalopathic (50% v 0% patients; RR, 5.0; 95% CI,.7 to 4.9; P <.000), and significantly more episodes of microbiologically confirmed or clinically suspected chest infection (4% v 3Oh patients; RR,.7; 95% CI,.8 to 88; P <,000). Thirty-six (46%) of the 78 encephalopathic patients but none of the 30 nonencephalopathic patients died (P <.OOOl) and 5 of 78 (9%) encephalopathic and 0 of 30 nonencephalopathic patients underwent transplantation (P =,0. Of the 08 patients admitted to this study 48 were not encephalopathic on admission; 30 of these remained nonencephalopathic and 8 patients developed encephalopathy. These groups were compared to determine whether infection was related to the development of coma. Six of the 8 who progressed to coma developed either a confirmed or clinical episode of infection compared with 3 of 30 who did not (RR, 3.3; 95% CI rate 0.95 to.7; P =,045). The overall mortality rate was 33% (36 of 08 patients). There was no significant difference in mortality between the two groups; all 36 deaths were ALF patients, 8 in each group. Significant positive cultures during the 4 hours before death, or a clinical picture of sepsis with acidosis and hemodynamic decompensation, was considered to have contributed to 8 of 36 deaths: 7 of 8 (Oh) of patients in group and of 8 (7%) in group (P = NS). There were 34 patients with indicators of a poor prognosis. Of these, 5 underwent transplantation, 5 of 3 (38%) in group and 0/ (47%) in group (P = NS); 4 of these 5 (7%) died, in each group. Nineteen patients with poor prognostic indicators on admission did not undergo transplantation. Of these, were listed for emergency liver transplantation; 5 were withdrawn because of the development of sepsis, 4 had a cerebral death before a liver became available, and were withdrawn because of clinical improvement (these two patients survived). Eight patients were not listed for OLT: 5 because of multiple organ failure on admission (all of whom died in the first 4 hours) and three because of an adverse psychiatric history and previous overdoses (one of these patients survived). Discussion Bacterial and fungal infections are common in patients with ALF and are associated with considerable morbidity and mortality rates., A previous study of 04 ALF patients from this centre indicated that SPEAR reduced the incidence of infection, but that this effect was probably related to the early administration of antibacterials, rather than to SDD per ~ e. ~ The present study has shown that the administration of oral nonabsorbable antibacterials did not confer any additional benefit on ALF patients receiving prophylactic systemic antibacterials; neither progression to coma nor infection were prevented. However, in this study 6% of ALF patients became infected. This is lower than the infection rate of 4% most recently reported from this center4 and markedly lower than the infection rate of 8% that we first reported. Meta-analysis of randomized trials of SDD in critically ill patients has shown significant reductions in respiratory tract infections.3 Because ALF is rare, large numbers of patients are not available for trials. However, in this study of 08 patients, we found that SDD did not influence significantly the incidence of clinical or microbiologically proven episodes of chest infection, indeed the incidence of chest infection was greater in those receiving SDD. However, as SDD was administered throughout each patient s admission, the duration of administration was effectively determined by the patient s outcome. It is of interest to note that older patients were more prone to infection, an observation that we have not previously made. The other factor associated with infection, such as inotropes on admission, intubation, and the presence of encephalopathy, are all associated with severe disease on admission. The finding that prolonged ITU admission was independently predictive of the development of infection should be treated with caution, because prolonged ITU stay predisposes to hospital-acquired infection. A high maximum coma grade is recognized to be associated with high infection rates. In addition, in this study we have found that in nonencephalopathic patients, progression to coma was associated with the development of infection. The significantly higher APACHE I score documented during episodes of infection reflect the adverse outcome of infection in this group of patients. Acetaminophen overdose patients, referred because premonitory clinical signs or biochemical tests predicted a severe course, were enrolled whether encephalopathic or not, so as to confer as early as possible any benefit of prophylaxis. It is of interest that out of 48 nonencephalopathic patients on

6 Antimicrobial Pi,ophyluxis in ALF 3 admission, 30 patients did not progress to coma. In a subanalysis of these patients we recognized the development of infection as a factor associated with the progression to coma; no other clinical, biochemical, or hematologic parameters were identified. Towards the end of the study, we encountered multiresistant bacteria, in particular gentamicinresistant Klebsiella spp and highly vancomycinresistant Efaecium. The third generation cephalosporin in our regimen may have selected for these bacteria, and we consider that prophylaxis regimens without a cephalosporin should be investigated. Such regimens might maintain the benefits of early antibacterial prophylaxis without selecting for resistant bacteria. References. Rolando N, Harvey F, Brahm J, Philpott-Howard J, Alexander G, Gimson A, et al. Prospective study of bacterial infection in acute liver failure: An analysis of fifty patients. Hepatology 990;: Rolando N, Harvey F, Brahm J, Philpott-Howard J, Alexander G, Casewell M, et al. Fungal infection: A common, unrecognised complication of acute liver failure. J Hepatol 99;: Trey C, Davidson LS. The management of fulminant hepatic failure. In: Popper H, Schaffner F, eds. Progress in liver disease. Vol. New York: Grune & Stratton, 970: Rolando N, Gimson A, Wade J, Philpott-Howard, Casewell M, Williams R. Prospective controlled trial of selective parenteral and enteral antimicrobial regimen in fulminant liver failure. Hepatology 993;7: van Saene HKF, Stoutenbeek CP. Selective decontamination. J Antimicrob Chemother 987;0: Stoutenbeek CP, van Saene HKF, Miranda DR, Zandstra DF. The effect of selective decontamination of the digestive tract on colonisation and infection rate in multiple trauma patients. Int Care Med 984; 0: Stoutenbeek CF, van Saene HKF, Miranda DR, Zandstra DF, Langreh D. The effect of oropharyngeal decontamination using topical non-absorbable antibiotics on the incidence of nosocomial respiratory tract infections in multiple trauma patients. Trauma 987;7: Unertl K, Ruckdeschel G, Selbmann HK, Jensen U, Forst H, Lenharl FP, et al. Prevention of colonization and respiratory infections in long-term ventilated patients by local antimicrobial prophylaxis. Int Care Med 987;3: Kerver AHJ, Rommes JH, Mevissen-Verhage EAE, Hulstaert PF, Vos A, Verhoef J, et al. Prevention of colonization and infection in critically ill patients: A prospective, randomized study. Crit Care Med 988;6: Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: A severity of disease classification. Crit Care Med 985;3: Maki DG, Weise CE, Saraffin HW. A semiquantitative culture method for identifying intravenous catheterrelated infections. N Engl J Med 977;96: O Grady JG, Alexander GJM, Hayllar KM, Williams R. Early indicators of prognosis in fulminant hepatic failure. Gastroenterology 989;97: Selective Decontamination of the Digestive Tract Trialists Collaborative Group. Meta-analysis of randomised controlled trials of selective decontamination of the digestive tract. Br Med J 993;307:55-53.