Antibiotic prophylaxis for cirrhotic patients with upper gastrointestinal bleeding

Norberto C Chavez‐Tapia; Tonatiuh Barrientos‐Gutierrez; Felix I Tellez‐Avila; Karla Soares‐Weiser; Misael Uribe

doi:10.1002/14651858.CD002907.pub2

. 2010 Sep 8;2010(9):CD002907. doi: 10.1002/14651858.CD002907.pub2

Antibiotic prophylaxis for cirrhotic patients with upper gastrointestinal bleeding

Norberto C Chavez‐Tapia ^1,^✉, Tonatiuh Barrientos‐Gutierrez ², Felix I Tellez‐Avila ³, Karla Soares‐Weiser ⁴, Misael Uribe ³

Editor: Cochrane Hepato‐Biliary Group

PMCID: PMC7138054 PMID: 20824832

Abstract

Background

Bacterial infections are a frequent complication in patients with cirrhosis and upper gastrointestinal bleeding. Antibiotic prophylaxis seems to decrease the incidence of bacterial infections. Oral antibiotics, active against enteric bacteria, have been commonly used as antibiotic prophylaxis in patients with cirrhosis and upper gastrointestinal bleeding. This is an update of a Cochrane review first published in 2002.

Objectives

To assess the benefits and harms of antibiotic prophylaxis in cirrhotic patients with upper gastrointestinal bleeding.

Search methods

We searched The Cochrane Hepato‐Biliary Group Controlled Trials Register, The Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library, MEDLINE, EMBASE, and Science Citation Index EXPANDED until June 2010. In addition, we handsearched the references of all identified studies.

Selection criteria

Randomised clinical trials comparing different types of antibiotic prophylaxis with no intervention, placebo, or another antibiotic to prevent bacterial infections in cirrhotic patients with upper gastrointestinal bleeding.

Data collection and analysis

Three authors independently assessed trial quality, risk of bias, and extracted data. We contacted study authors for additional information. Association measures were relative risk (RR) for dichotomous outcomes and mean difference (MD) for continuous outcomes.

Main results

Twelve trials (1241 patients) evaluated antibiotic prophylaxis compared with placebo or no antibiotic prophylaxis. All trials were at risk of bias. Antibiotic prophylaxis compared with no intervention or placebo was associated with beneficial effects on mortality (RR 0.79, 95% CI 0.63 to 0.98), mortality from bacterial infections (RR 0.43, 95% CI 0.19 to 0.97), bacterial infections (RR 0.36, 95% CI 0.27 to 0.49), rebleeding (RR 0.53, 95% CI 0.38 to 0.74), days of hospitalisation (MD ‐1.91, 95% CI ‐3.80 to ‐0.02), bacteraemia (RR 0.25, 95% CI 0.15 to 0.40), pneumonia (RR 0.45, 95% CI 0.27 to 0.75), spontaneous bacterial peritonitis (RR 0.29, 95% CI 0.15 to 0.57), and urinary tract infections (RR 0.23, 95% CI 0.12 to 0.41). No serious adverse events were reported. The trials showed no significant heterogeneity of effects. Another five trials (650 patients) compared different antibiotic regimens. Data could not be combined as each trial used different antibiotic regimen. None of the examined antibiotic regimen was superior to the control regimen regarding mortality or bacterial infections.

Authors' conclusions

Prophylactic antibiotic use in patients with cirrhosis and upper gastrointestinal bleeding significantly reduced bacterial infections, and seems to have reduced all‐cause mortality, bacterial infection mortality, rebleeding events, and hospitalisation length. These benefits were observed independently of the type of antibiotic used; thus, no specific antibiotic can be preferred. Therefore, antibiotic selection should be made considering local conditions such as bacterial resistance profile and treatment cost.

Plain language summary

Antibiotic prophylaxis for prevention of bacterial infections and death in cirrhotic patients with upper gastrointestinal bleeding

Patients with liver cirrhosis have an impaired immune response. Often, liver cirrhosis patients experience complications from portal hypertension, such as gastroesophageal varices. These varices can bleed, increasing the risk of infection and death in a short period of time, despite proper endoscopic management. Patients who develop bacterial infections during hospitalisation for gastroesophageal haemorrhage are at increased risk of dying. Twelve trials (1241 patients) assessing several antibiotic prophylaxis regimens versus no intervention or placebo were analysed, showing that antibiotic prophylaxis successfully reduced the incidence of bacterial infections. Antibiotic prophylaxis was also associated with a reduction in mortality, mortality from bacterial infections, rebleeding rate, and days of hospitalisation. The prophylactic treatment was not associated with important adverse effects. Five trials (650 patients) assessed one antibiotic regimen compared with another. All antibiotic regimens provided similar benefits and none seemed superior. Thus, to this point there is no evidence to recommend one specific antibiotic regimen over the other. All trials analysed were subject to bias; thus, results should be interpreted carefully.

Background

Description of the condition

Chronic liver diseases are characterised by important changes on hepatic physiology with portal hypertension being their haemodynamic expression. The presence of clinically significant portal hypertension (portal pressure gradient ≥10 mmHg) promotes the formation of collateral portosystemic circulation, portal hypertensive gastropathy, gastric varices, and oesophageal varices (Bosch 2009).

Bleeding, secondary to oesophageal or gastro‐oesophageal varices, is observed in up to 30% of the patients with liver cirrhosis during the course of their illness, recurring in 70% of the patients and being fatal in 20% (NIEC 1988). The highest mortality peak is observed during the first six weeks after the bleeding episode (Burroughs 2009). Hepatic functional status (assessed by Child‐Pugh score), renal dysfunction (assessed by creatinine serum levels), and bacterial infections are the most important mortality risk factors (Augustin 2009). Consequently, guidelines for treatment of patients suffering variceal gastrointestinal bleeding include volume expansion, haemorrhage control, use of vasoconstrictors, and short‐term antibiotic prophylaxis (Garcia‐Tsao 2009).

Description of the intervention

The prophylactic use of oral or intravenous antibiotics has been recommended in several consensus guidelines. The recommended drugs are mainly oral quinolones (norfloxacin 400 mg b.i.d for 7 days) or intravenous cephalosporins (ceftriaxone 1 g/day for 7 days) (Garcia‐Tsao 2009). However, other groups of antibiotics have been assessed such as beta‐lactams or aminoglycosides (Bernard 1999).

How the intervention might work

Cirrhosis is characterised by cellular and humoral immune dysfunction as well as increased bacterial translocation from the gut into the bloodstream, facilitating the development of infections (Chavez‐Tapia 2007). The most common bacterial infections are caused by gram‐negative bacteria, producing spontaneous bacterial peritonitis (25%), urinary tract infections (20%), pneumonia (15%), and bacteraemia (12%) (Fernandez 2002). Considering the increased mortality associated with infections, the vulnerability of the immune system and the bacteriological profile, the use of antibiotics is recommended.

Why it is important to do this review

Prophylactic use of antibiotics during an episode of upper gastrointestinal bleeding in cirrhotic patients is considered standard of care. However, differences in the antibiotics used, schedules of administration, duration of therapy, and changes in the bacteriological profile across clinical trials make the evaluation of this intervention difficult. This review systematically assesses these issues and updates the information from a previous published review (Soares‐Weiser 2002).

Objectives

To assess the benefits and harms of antibiotic prophylaxis in cirrhotic patients with upper gastrointestinal bleeding.

Specifically this review was designed to:

Compare the all‐cause mortality and infection mortality between cirrhotic patients with gastrointestinal bleeding receiving antibiotic prophylaxis or no intervention/placebo.
Compare the proportion of bacterial infections in patients with gastrointestinal bleeding receiving antibiotic prophylaxis versus no intervention/placebo.
Determine the most effective antibiotic regimen.

Methods

Criteria for considering studies for this review

Types of studies

Randomised clinical trials comparing different types of antibiotic therapy against no intervention, or placebo, or another antibiotic, in the prophylaxis of bacterial infections in cirrhotic patients with upper gastrointestinal bleeding were identified. Trials were included irrespective of publication status, language, or blinding.

Types of participants

Adult patients with cirrhosis and upper gastrointestinal bleeding were included, regardless of the aetiology of cirrhosis or severity of the disease.

Types of interventions

The following interventions, used alone or in combination, and regardless of the mode of administration (intravenous or oral), were considered.

Aminoglycosides (eg, gentamicin, neomycin, tobramycin);
Amoxicillin with or without clavulanic acid;
Cephalosporins (eg, cefotaxime, ceftriaxone, ceftazidime, cefonicid);
Quinolones (eg, ciprofloxacin, ofloxacin, norfloxacin);
Trimethoprim/sulphamethoxazole;
Non‐absorbable antibiotics (eg, colistin, nystatin);
Other antibiotics.

Control groups received no intervention, placebo, or any antibiotic.

Types of outcome measures

Primary outcome measures:

Number of deaths;
Number of patients that developed bacterial infections (bacteraemia, pneumonia, urinary tract infection, spontaneous bacterial peritonitis, and/or other bacterial infections);
Quality of life score (measured by any scale) between groups;
Adverse events (ICH‐GCP 1997):

1. Any serious adverse events that were fatal, life‐threatening, or requiring inpatient hospitalisation or prolongation of existing hospitalisation;
2. Any adverse events that resulted in significant disability or incapacity;
3. Any important medical events that might not be immediately life‐threatening or resulted in death or hospitalisation, but might jeopardise the patient or required intervention to prevent one of the above outcomes;
4. Any adverse events that required discontinuation of medication.

Secondary outcome measures:

Number of patients who developed bacterial infections after an invasive procedure to stop upper gastrointestinal bleeding;
Number of patients who developed superinfection or antibiotic resistance in at least one of the follow‐up cultures;
Number of patients who developed rebleeding during the follow‐up (overall rebleeding rate and up to seven days rate);
Number of patients who dropped out from the trial after randomisation;
Cost of different types of antibiotics used for prophylaxis;
Number of days of hospitalisation.

Search methods for identification of studies

Electronic searches

Relevant randomised trials were identified by searching The Cochrane Hepato‐Biliary Group Controlled Trials Register (Gluud 2010), the Cochrane Central Register of Controlled Trials (CENTRAL) (Issue 2, 2010) in The Cochrane Library, MEDLINE (1950 to 21 June 2010), EMBASE (1980 to 21 June 2010), and Science Citation Index EXPANDED (1945 to 21 June 2010) (Royle 2003).

Search strategies and time span of the searches are given in Appendix 1.

Searching other resources

The references of all identified studies were inspected for more trials. Additionally, the first or corresponding author of each included trial, as well as researchers active in the field, were contacted for information regarding unpublished trials and complementary information on their own trial.

References from an existing review on this topic (Bernard 1999) were also checked for any missing trials.

Data collection and analysis

Selection of studies

Three authors (NC, FT, TB) independently inspected each identified reference and applied the inclusion criteria. For potentially relevant articles, or in cases of disagreement between the three reviewers, the full text article was obtained and inspected independently. If resolving disagreement by discussion was not possible, the article was added to those 'awaiting assessment' and the authors of the original study were contacted for clarification. In the event of no reply from the authors within three months, a fourth reviewer (MU or KSW) reviewed the article to solve the disagreement. Justification for study exclusion was documented.

Data extraction and management

Two authors (NC and TB) independently extracted the data from the included trials. In case of disagreement between the two authors, a third author (FT) extracted the data. The data extraction was discussed, decisions documented, and, when necessary, the authors of the original studies were contacted for clarification. Justification for study exclusion was documented. Trials were identified with the last name of the first author and the year in which the trial was first published, and ordered chronologically.

The following data were extracted, verified, and recorded:

Characteristics of trials

Date, location, and setting of trial;
Publication status;
Case definitions used (clinical, serological, bacteriological);
Sponsor of trial (known or unknown; industry or not industry).

Characteristics of participants

Number of participants in each group;
Age, sex, nationality;
Severity of liver disease and cirrhosis according to the aetiology of liver disease, regardless of the criteria used.

Characteristics of interventions

Type of antibiotic, dose, mode of administration, schedule, length of follow‐up (in months).
Number of days that antibiotic prophylaxis was provided.

Characteristics of outcome measures

Whenever possible, the number of events previously listed under Types of outcome measures were recorded in each group of the randomised trials.

Assessment of risk of bias in included studies

Two authors (NC and TB) independently assessed bias risk of the trials, without masking the trial names. For this purpose, instructions given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2009) and the Cochrane Hepato‐Biliary Group Module (Gluud 2010) were followed. The risk of overestimation of intervention effects in randomised trials due to inadequate methodological quality (Schulz 1995; Moher 1998; Jüni 2001; Kjaergard 2001; Wood 2008) was assessed using the domains below. Whenever information was not available in the published trial, authors were contacted directly.

Sequence generation

Low risk of bias, if the allocation sequence was generated by a computer or random number table. Drawing of lots, tossing of a coin, shuffling of cards, or throwing dice were also considered as adequate if a person who was not otherwise involved in the recruitment of participants performed the procedure.
Unclear, if the trial was described as randomised, but the method used for the allocation sequence generation was not described or insufficient to permit judgement.
High risk of bias, if a system involving dates, names, or admittance numbers were used for the allocation of patients.

Allocation concealment

Low risk of bias, if the allocation of patients involved a central independent unit, on‐site locked computer, identically appearing numbered drug bottles or containers prepared by an independent pharmacist or investigator, or sealed envelopes.
Unclear, if the trial was described as randomised, but the method used to conceal the allocation was not described or insufficient to permit judgement.
High risk of bias, if the allocation sequence was known to the investigators who assigned participants or if the study was quasi‐randomised.

Blinding

Low risk of bias, if the trial was described as double blind and the method of blinding involved identical placebo or active drugs.
Unclear, if the trial was described as double blind, but the method of blinding was not described.
High risk of bias, if the trial was not double blind.

Incomplete outcome data

Low risk of bias, if there were no post‐randomisation drop‐outs or withdrawals.
Unclear, if it is not clear whether there are any drop‐outs or withdrawals or if the reasons for these drop‐outs are not clear.
High risk of bias, if the reasons for missing data are likely to be related to true outcomes.

Selective outcome reporting

Low risk of bias, considering that most of the included trials were made before of the obligatory registration on randomised controlled trials databases, and the pre‐specified outcomes are not available. The following outcomes were considered fundamental as outcome to avoid selective reporting; a) mortality, b) response rate, and c) adverse events.
Unclear, there is insufficient information to assess whether the magnitude and direction of the observed effect is related to selective outcome reporting.
High risk of bias, not all of the trial's pre‐specified primary outcomes have been reported or similar.

Other sources of bias

Low risk of bias, the trial appears to be free of other sources of bias, considering; a) baseline imbalance, b) source of funding, c) early stopping, and d) interim analysis.
Unclear, there is insufficient information to assess whether other sources of bias are present.
High risk of bias, it is likely that potential sources of bias.

Following the definitions of the above domains, an included trial was judged as a trial with a low risk of bias when the risk of bias was evaluated as 'low' in all domains. If the risk of bias was judged as 'uncertain' or 'high', then the trial was judged as having 'high risk of bias'.

Furthermore, we registered whether or not the randomised clinical trials had used 'intention‐to‐treat' analysis (Gluud 2001), the length of follow‐up, and sample size calculation. Any disagreement was resolved by discussion and settled by a third author (FT). We contacted the trial author for clarification as necessary.

Measures of treatment effect

Dichotomous data were analysed calculating the relative risk (RR) for each trial, expressing the uncertainty with 95% confidence intervals (CI). Continuous data were analysed calculating mean differences between groups of each trial and its 95% CI. Comparisons were made between trials evaluating antibiotic prophylaxis against no intervention or placebo, and trials comparing different antibiotic regimens.

Assessment of heterogeneity

We checked the heterogeneity of effects across trials by visual inspection of the forest plots and Chi² and I² tests for heterogeneity (Higgins 2009). Statistical heterogeneity was defined as a P value ≤ 0.10 (Chi²) or I² > 25%. When heterogeneity existed, subgroup analyses were performed in order to assess the impact of potential sources of heterogeneity over the main results.

Assessment of reporting biases

A funnel plot estimating the precision of trials (plot of logarithm of the RR against the sample size) was examined in order to estimate potential asymmetry. In addition, the standard normal deviate (SND), defined as the RR divided by its standard error, was regressed against the estimate's precision (regression equation: SND = a + b x precision) in order to facilitate the prediction of potential heterogeneity or data irregularities in the meta‐analyses (Egger 1997). In this equation, the SND reflects the degree of funnel plot asymmetry as measured by the intercept from the regression analysis.

Data synthesis

For the statistical analyses, we used RevMan Analyses (RevMan 2010). Dichotomous data were synthesised poling the RR and 95% CI from all trials to estimate the global effect of the intervention. Continous data were synthesised pooling mean differences and 95% CI from each trial to calculate the average mean difference.

In order to compare the RR from different antibiotic groups and detect differences among the antibiotics tested versus no intervention or placebo, a test for interaction was calculated (Altman 2003).

Sensitivity analysis

We analysed data using both fixed‐effect and random‐effects models. When both models produced similar estimates, the fixed‐effect result was reported; otherwise, we reported the results from both analyses. Outcomes were analysed as reported in the trial, that is, either per protocol or as intention‐to‐treat analysis. In order to examine the influence of drop‐outs, we performed both worst‐best‐case (assigning bad outcomes to all of the missing experimental group patients and good outcomes to all of the missing control group patients) and best‐worst‐case (assigning good outcomes to all of the missing experimental group patients and bad outcomes to all of the missing control group patients) analyses.

To assess the reliability of the meta‐analyses on mortality, mortality from bacterial infections, and bacterial infections, the required information size (RIS) was calculated by trial sequential analysis (TSA). We respectively assumed an average event proportion of 22%, 5%, and 36% in the control group of the three meta‐analyses; a 20% relative risk reduction of the experimental intervention, and statistical error levels of 5% alpha and 20% beta (80% power). Whenever the cumulative information size in the meta‐analysis was smaller than the RIS, the threshold to maintain statistical significance was calculated with the O’Brien‐Fleming boundaries (Brok 2008; Wetterslev 2008; Thorlund 2009).

Results

Description of studies

Results of the search

Forty‐one relevant references were initially identified and screened for retrieval. Subsequently, reviews (fourteen references), comments and editorials (four references) were excluded. Twenty‐six studies were considered suitable to be included, but nine studies were excluded because they were not randomised trials (six studies), did not include patients with upper gastrointestinal bleeding (two studies), or did not include any of the outcomes assessed in this review (one study) (See 'Excluded studies' for additional information). Finally, seventeen trials (twenty‐one references) were included for analyses (see 'Included studies' for additional information).

Included studies

Seventeen trials evaluating th effectiveness of antibiotic prophylaxis against bacterial infections in 1891 patients were included in this review (Rimola 1985; Soriano 1992; Rolando 1993; Blaise 1994; Selby 1994; Pauwels 1996; Zacharof 1997; Hsieh 1998; Sabat 1998; Spanish Group 1998; Gulberg 1999; Hong 2002; Lin 2002; Hou 2004; Lata 2005; Fernandez 2006; Jun 2006), in twelve trials (fourteen references) comparison with no intervention or placebo was conducted (Rimola 1985; Soriano 1992; Rolando 1993; Blaise 1994; Selby 1994; Pauwels 1996; Zacharof 1997; Hsieh 1998; Hong 2002; Lin 2002; Hou 2004; Jun 2006).

Five trials (eight references) were head‐to‐head antibiotic comparisons (Sabat 1998; Spanish Group 1998; Gulberg 1999; Lata 2005; Fernandez 2006).

Most trials were retrieved as complete manuscript and two trials as abstracts (Spanish Group 1998; Zacharof 1997). The only non‐English article was written in Korean (Hong 2002).

Trials were conducted in Australia (Selby 1994), Czech Republic (Lata 2005), France (Blaise 1994; Pauwels 1996), Germany (Gulberg 1999), Greece (Zacharof 1997), Korea (Hong 2002; Jun 2006), Spain (Rimola 1985; Soriano 1992; Sabat 1998; Spanish Group 1998; Fernandez 2006), Taiwan (Hsieh 1998; Lin 2002; Hou 2004), and the United Kingdom (Rolando 1993) (see 'Characteristics of included studies' for details).

The source of the haemorrhage was gastroesophageal varices in six trials (Rolando 1993; Blaise 1994; Hong 2002; Hou 2004; Lata 2005; Jun 2006), mixed (variceal and non‐variceal gastrointestinal haemorrhage) in seven trials (Rimola 1985; Soriano 1992; Pauwels 1996; Hsieh 1998; Sabat 1998; Lin 2002; Fernandez 2006), and not specified in four trials (Selby 1994; Zacharof 1997; Spanish Group 1998; Gulberg 1999).

While all trials assessed the use of antibiotic prophylaxis, the primary outcome differed between them. The most common was prevention of bacterial infections in fourteen trials (Rimola 1985; Soriano 1992; Rolando 1993; Blaise 1994; Selby 1994; Pauwels 1996; Zacharof 1997; Hsieh 1998; Sabat 1998; Spanish Group 1998; Gulberg 1999; Hong 2002; Lin 2002; Fernandez 2006), rebleeding rate in two trials (Hou 2004; Jun 2006), and early and late mortality in one trial (Lata 2005).

All trials were performed on hospitalised patients, but four specifically described critical care settings (Rimola 1985; Blaise 1994; Pauwels 1996; Lata 2005). Eleven trials reported the Child‐Pugh score as a categorical variable, from them nine included patients with Child‐Pugh score A/B/C (overall distribution 17/51/32%, respectively), two trials reported only patients with Child Pugh score B/C. Diagnostic and management of the gastrointestinal haemorrhage was done by endoscopy.

The antibiotics compared versus no intervention or placebo were: quinolones (five trials), quinolones plus beta‐lactams (two trials), cephalosporins (three trials), carbapenems (one trial) and non‐absorbable antibiotics (one trial) (Table 2).

1. Group of antibiotics compared versus no intervention or placebo.

Group of antibiotics	Reference
Quinolones	Soriano 1992; Zacharof 1997; Hsieh 1998; Hong 2002; Hou 2004
Quinolones plus beta‐lactams	Blaise 1994; Pauwels 1996
Cephalosporins	Selby 1994; Lin 2002; Jun 2006
Carbapenems	Rolando 1993
Non absorbable	Rimola 1985

Reference	Intervention 1	Intervention 2
Sabat 1998	Quinolone	Quinolone plus cephalosporin
Spanish Group 1998	Quinolone (Norfloxacin)	Quinolone (Ofloxacin)
Gulberg 1999	Cephalosporin low dose	Cephalosporin high dose
Lata 2005	Quinolone	beta‐lactams
Fernandez 2006	Quinolone	Cephalosporin

Antibiotic regimen 1	Antibiotic regimen 1	P‐value
Cephalosporins	Quinolones	0.39
Cephalosporins	Quinolones + beta‐lactams	0.17
Cephalosporins	Other antibiotics	0.03
Quinolones	Quinolones + beta‐lactams	0.29
Quinolones	Other antibiotics	0.004
Quinolones + beta‐lactams	Other antibiotics	0.19

Trial ID	Outcome	Experimental (n/N)	Control (n/N)	Relative Risk	95% CI
Sabat 1998	Mortality	Norfloxacin + ceftriaxone (1/24)	Norfloxacin (2/22)	0.46	0.04 to 4.71
Sabat 1998	Bacterial infections	Norfloxacin + ceftriaxone (3/24)	Norfloxacin (4/22)	0.69	0.17 to 2.73
Sabat 1998	Cost	Norfloxacin + ceftriaxone ‐ US $99.3 to 220.1	Norfloxacin ‐ US $1.9 to 745.5	‐	‐
Sabat 1998	Drop‐outs	Norfloxacin + ceftriaxone (4/28)	Norfloxacin (6/28)	0.67	0.21 to 2.11
Gulberg 1999	Mortality	Ceftriaxone 1g (0/40)	Ceftriaxone 2g (0/42)	Risk difference: 0.00	‐0.05 to 0.05
Gulberg 1999	Bacterial infections	Ceftriaxone 1g (1/40)	Ceftriaxone 2g (1/42)	1.05	0.11 to 9.80
Spanish Group 1998	Bacterial infections (proven)	Norfloxacin 800mg (26/183)	Ofloxacin 400mg (27/182)	0.96	0.58 to 1.58
Spanish Group 1998	Bacterial infections (suspected)	Norfloxacin 800mg (51/183)	Ofloxacin 400mg (53/182)	0.96	0.69 to 1.32
Lata 2005	Mortality	Ampicillin and sulbactam 3g (12/21)	Norfloxacin 800 mg (7/25)	2.04	0.98 to 4.23
Fernandez 2006	Mortality	Ceftriaxone 1g (8/54)	Norfloxacin 800 mg (6/57)	1.41	0.52 to 3.79
Fernandez 2006	Mortality from bacterial infections	Ceftriaxone 1g (1/54)	Norfloxacin 800 mg (1/57)	1.06	0.07 to 16.46
Fernandez 2006	Bacterial infections	Ceftriaxone 1g (6/54)	Norfloxacin 800 mg (5/57)	1.27	0.41 to 3.94

Date	Event	Description
4 August 2010	New citation required but conclusions have not changed	Six new trials were added to the review.
29 June 2010	New search has been performed	This review was updated: the list of authors was changed, six new trials were included, the Background section was changed, the Methods section was modified according the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2009), the discussion was modified. The primary outcomes were arranged to include quality of life and adverse events as primary outcomes in concordance with the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2009).

Database	Time span	Search strategy
The Cochrane Hepato‐Biliary Group Controlled Trials Register	June 2010	(antibiotic* OR antibacteri) AND (bleed OR hemorr* or haemorr) AND cirrho
Cochrane Central Register of Controlled   Trials (CENTRAL) in The Cochrane Library	Issue 2, 2010	#1 MeSH Descriptor antibiotic‐prophylaxis explode all trees in MeSH products   #2 antibiotic* in All Fields in all products   #3 antibacteri* NEAR prophyl* in All Fields in all products   #4 (#1 OR #2 OR #3)   #5 MeSH descriptor Gastrointestinal Hemorrhage explode all trees in MeSH products   #6 bleed* in All Fields in all products   #7 haemorr* in All Fields in all products   #8 hemorr* in All Fields in all products   #9 (#5 OR #6 OR #7 OR #8)   #10 MeSH Descriptor Liver Cirrhosis explode all tress in MeSH products   #11 cirrho* in All Fields in all products   #12 (#10 OR #11)   #13 (#4 AND #9 AND #12)
MEDLINE (Ovid SP)	1950 to June 2010	#1 explode antibiotic‐prophylaxis /All subheadings   #2 antibiotic* prophyl*   #3 antibiotic* pre   #4 #1 or #2 or #3   #5 explode liver‐cirrhosis /All subheadings   #6 liver cirrho   #7 hepatic cirrho*   #8 liver fibro*   #9 #5 or #6 or #7 or #8   #10 explode gastrointestinal‐hemorrhage /All subheadings   #11 gastr* hemorrhage   #12 gastr* haemorrhage   #13 gastr* bleeding   #14 #10 or #11 or #12 or #13   #15 random* or blind* or placebo* or meta‐analysis   #16 #4 and #9 and #14 and #15
EMBASE (Ovid SP)	1980 to June 2010	#1 explode “antibiotic‐prophylaxis”/all subheadings   #2 antibiotic*   #3 antibacteri* prophy*   #4 #1 or #2 or #3   #5 explode “gastrointestinal‐hemorrhage”/all subheadings   #6 hemorr*   #7 haemorr*   #8 bleed*   #9 #5 or #6 or #7 or #8   #10 explode “liver‐cirrhosis”/all subheadings   #11 cirrho*   #12 #10 or #11   #13 #4 and #9 and #12   #14 random* or blind* or placebo* or meta‐analysis   #13 #13 and #14
Science Citation Index EXPANDED   (http://apps.isiknowledge.com)	1945 to June 2010	#1 TS=(antibiotic* OR antibacteri* prophyl)   #2 TS=(bleed or hemorr* or haemorr)   #3 TS=(cirrho)   #4 #3 AND #2 AND #1   #5 TS=(random* or blind* or placebo* or meta‐analysis)   #6 #5 AND #4

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mortality	12	1241	Risk Ratio (M‐H, Fixed, 95% CI)	0.79 [0.63, 0.98]
2 Mortality from bacterial infections	6	761	Risk Ratio (M‐H, Fixed, 95% CI)	0.43 [0.19, 0.97]
3 Bacterial infections	12	1241	Risk Ratio (M‐H, Random, 95% CI)	0.36 [0.27, 0.49]
4 Bacteremia	9	987	Risk Ratio (M‐H, Fixed, 95% CI)	0.25 [0.15, 0.40]
5 Pneumonia	9	1041	Risk Ratio (M‐H, Fixed, 95% CI)	0.45 [0.27, 0.75]
6 Spontaneous bacterial peritonitis	8	890	Risk Ratio (M‐H, Fixed, 95% CI)	0.29 [0.15, 0.57]
7 Urinary tract infections	9	1098	Risk Ratio (M‐H, Fixed, 95% CI)	0.23 [0.13, 0.41]
8 Other infections	3	257	Risk Ratio (M‐H, Fixed, 95% CI)	0.43 [0.12, 1.56]
9 Drop outs before end of study	8	919	Risk Ratio (M‐H, Fixed, 95% CI)	1.07 [0.76, 1.51]
10 Mortality according to trial sample size	12	1241	Risk Ratio (M‐H, Fixed, 95% CI)	0.79 [0.63, 0.98]
10.1 Sample size smaller than 101 patients	5	340	Risk Ratio (M‐H, Fixed, 95% CI)	0.71 [0.44, 1.15]
10.2 Sample size larger than 100 patients	7	901	Risk Ratio (M‐H, Fixed, 95% CI)	0.81 [0.63, 1.03]
11 Bacterial infections according to trial sample size	12	1241	Risk Ratio (M‐H, Fixed, 95% CI)	0.37 [0.30, 0.46]
11.1 Sample size smaller than 101 patients	5	340	Risk Ratio (M‐H, Fixed, 95% CI)	0.42 [0.29, 0.62]
11.2 Sample size larger than 100 patients	7	901	Risk Ratio (M‐H, Fixed, 95% CI)	0.35 [0.27, 0.46]
12 Mortality sensitivity analysis worst‐best case	8	978	Risk Ratio (M‐H, Random, 95% CI)	1.45 [1.04, 2.02]
13 Mortality sensitivity analysis best‐worst case	8	969	Risk Ratio (M‐H, Fixed, 95% CI)	0.48 [0.38, 0.60]
14 Mortality from bacterial infections sensitivity analysis worst‐best case	5	708	Risk Ratio (M‐H, Random, 95% CI)	3.30 [1.43, 7.62]
15 Mortality from bacterial infections sensitivity analysis best‐worst case	5	699	Risk Ratio (M‐H, Fixed, 95% CI)	0.14 [0.06, 0.31]
16 Bacterial infections sensitivity analysis worst‐best case	8	978	Risk Ratio (M‐H, Fixed, 95% CI)	0.77 [0.63, 0.93]
17 Bacterial infections sensitivity analysis best‐worst case	8	969	Risk Ratio (M‐H, Random, 95% CI)	0.26 [0.16, 0.43]
18 Mortality according to the antibiotic used	12	1241	Risk Ratio (M‐H, Fixed, 95% CI)	0.79 [0.63, 0.98]
18.1 Quinolones	5	473	Risk Ratio (M‐H, Fixed, 95% CI)	0.88 [0.61, 1.27]
18.2 Quinolones + beta‐lactams	2	264	Risk Ratio (M‐H, Fixed, 95% CI)	0.74 [0.42, 1.31]
18.3 Cephalosporins	3	255	Risk Ratio (M‐H, Fixed, 95% CI)	0.72 [0.46, 1.11]
18.4 Other antibiotics	2	249	Risk Ratio (M‐H, Fixed, 95% CI)	0.76 [0.50, 1.17]
19 Mortality from bacterial infections according to the antibiotic used	6	761	Risk Ratio (M‐H, Fixed, 95% CI)	0.43 [0.19, 0.97]
19.1 Quinolones	2	248	Risk Ratio (M‐H, Fixed, 95% CI)	0.76 [0.18, 3.34]
19.2 Quinolones + beta lactam	1	147	Risk Ratio (M‐H, Fixed, 95% CI)	0.67 [0.08, 5.80]
19.3 Cephalosporins	2	217	Risk Ratio (M‐H, Fixed, 95% CI)	0.44 [0.10, 1.95]
19.4 Others	1	149	Risk Ratio (M‐H, Fixed, 95% CI)	0.14 [0.02, 1.15]
20 Bacterial infections according to the antibiotic used	12	1241	Risk Ratio (M‐H, Fixed, 95% CI)	0.36 [0.29, 0.45]
20.1 Quinolones	5	473	Risk Ratio (M‐H, Fixed, 95% CI)	0.27 [0.18, 0.39]
20.2 Quinolones + beta‐lactams	2	264	Risk Ratio (M‐H, Fixed, 95% CI)	0.38 [0.23, 0.62]
20.3 Cephalosporins	3	255	Risk Ratio (M‐H, Fixed, 95% CI)	0.16 [0.05, 0.48]
20.4 Others	2	249	Risk Ratio (M‐H, Fixed, 95% CI)	0.57 [0.41, 0.81]
21 Rebleeding	3	280	Risk Ratio (M‐H, Fixed, 95% CI)	0.53 [0.38, 0.74]
22 Early rebleeding (up to 7 days)	3	280	Risk Ratio (M‐H, Fixed, 95% CI)	0.24 [0.12, 0.50]
23 Days of hospitalisation	7	769	Mean Difference (IV, Random, 95% CI)	‐0.79 [‐1.84, 0.26]
23.1 Intensive care unit	2	264	Mean Difference (IV, Random, 95% CI)	‐0.27 [‐1.55, 1.00]
23.2 Full‐lenght	5	505	Mean Difference (IV, Random, 95% CI)	‐1.91 [‐3.80, ‐0.02]

Methods	Data collection: 09/1990 to 01/1992.   Exclusion from analysis: patients with no oesophageal varices or infected.   Follow‐up period: 14 days.
Participants	France   Hospitalised cirrhotic patients with upper gastrointestinal haemorrhage, without infection, or treatment with antibiotics in the previous two weeks.   Etiology of bleeding: oesophageal varices.   Initial infection assessment: chest X‐ray, blood count, urine/sputum/blood/ascites culture.   Endoscopy scheduled within 12 h after enrolment.   Sclerotherapy performed together with endoscopy to stop bleeding.   Child‐Pugh class A/B/C: 0/20/7.
Interventions	Experimental: intravenous + oral ofloxacin, 400 mg/day, 10 days; amoxicillin + clavulanic acid (bolus, 1g) before each endoscopy procedure. Control: no antibiotic prophylaxis.
Outcomes	Prevent bacterial infections.
Notes	Authors were contacted for the first version of this review.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	No information provided.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	Missing data are balanced in number across the intervention groups.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Low risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	High risk	The excluded participants are balanced and with a justifiable reason.
Sample calculation?	High risk	No sample calculation was done.

Methods	Data collection 02/2000 to 04/2004.   Exclusion from analysis: occult infection (positive blood cultures obtained prior to randomisation) and less than two signs of liver failure (Norfloxacin group 6/63; Ceftriaxone group 7/61).   Follow‐up period: 10 days.
Participants	Spain Multicenter trial on hospitalised patients in patients with cirrhosis and upper gastrointestinal haemorrhage. Etiology of bleeding: portal hypertension related in 77% of participants. Initial infection assessment: blood cultures, ascitic fluid polymorphonuclear count, urine sediment and culture, chest X‐ray. Emergency endoscopy within 24 h after onset of the haemorrhage, plus somatostatin or terlipressin on portal hypertension related haemorrhage. Sclerotherapy or banding was performed to stop bleeding. Chil‐Pugh class A/B/C: 0/52/59.
Interventions	Experimental: intravenous ceftriaxone 1g per day for 7 days. Control: oral norfloxacin 400 mg b.i.d. for 7 days.
Outcomes	Prevention of bacterial infections.
Notes	The authors were contacted via e‐mail and response was received (28‐June‐2010).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	The sequence was generated using a random number software.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	Missing data are balanced in number across the intervention groups.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Low risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	High risk	The excluded participants are balanced and with a justifiable reason.
Sample calculation?	Low risk	The sample was calculated considering the expected incidence of proved and possible infections.

Methods	Date of collection: no information provided.   Exclusion from analysis: no information provided.   Follow‐up period: 7 days.
Participants	Germany   Cirrhotic patients with upper gastrointestinal haemorrhage, without infection, or treatment with antibiotics in the previous week, who underwent TIPS.   Etiology of bleeding: no information provided.   Initial infection assessment: fever, white blood cells, C‐reactive protein. Child‐Pugh class A/B/C: 29/39/14.
Interventions	Experimental: intravenous ceftriaxone, 1g, single dose before TIPS. Control: intravenous ceftriaxone, 2g, single dose before TIPS.
Outcomes	Bacterial infection after TIPS procedure.
Notes	Authors were contacted for the first version of this review.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	No information provided.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	No missing outcome data.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	High risk	The report support the intervention effects on surrogates of infection.
Intention to treat analysis?	Unclear risk	No information provided.
Sample calculation?	High risk	No sample calculation procedure.

Methods	Data collection: 12/1998 to 9/2001.   Exclusion from analysis: death or surgery within 24 h.   Follow‐up period: 30 days.
Participants	Korea Hospitalised cirrhotic patients with oesophageal variceal bleeding, without infection or previous use of antibiotics. Initial infection assessment: physical examination, white blood cells count, liver function test, ascitic fluid analysis and culture. All patients were enrolled after emergency endoscopic oesophageal variceal ligation. Child‐Pugh class: 8.8 ± 1.6 vs 9.1 ± 1.4.
Interventions	Experimental: intravenous ciprofloxacin 200 mg b.i.d. for 3 days. Control: no antibiotic prophylaxis.
Outcomes	Prevention of bacterial infections.
Notes	The authors were contacted via e‐mail and no response was received (23‐June‐2010).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	No information provided.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Unclear risk	No information provided.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Unclear risk	No information provided.
Intention to treat analysis?	Unclear risk	No information provided.
Sample calculation?	Unclear risk	No information provided.

Methods	Data collection: 01/2001 to 02/2003.   Exclusion from analysis: positive initial bacteriological sample.   Follow‐up period: until death or 3 months.
Participants	Taiwan Hospitalised patients with endoscopy‐proven gastroesophageal variceal bleeding without signs of infection. Etiology of bleeding: endoscopy‐proven gastroesophageal variceal bleeding. Endoscopic procedure was completed within 24 h of admission or bleeding onset. Endoscopic variceal ligation or sclerotherapy was preceded by vasoactive agent or balloon tamponade. Child‐Pugh class (A/B/C): 20/64/27.
Interventions	Experimental: intravenous ofloxacin 200 mg b.i.d. for 2 days followed by oral ofloxacin 200 mg b.i.d. for 5 days. Control: no antibiotic prophylaxis.
Outcomes	Rebleeding rate.
Notes	The authors were contacted via e‐mail and no response was received (23‐June‐2010).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	The sequence was generated using a random number software.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	Missing data are balanced in number across the intervention groups.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Low risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	High risk	The exclusion of the patients was not based on an exclusion rule.
Sample calculation?	Low risk	The sample was calculated considering the expected rebleeding rate.

Methods	Data collection: 07/95 to 07/96.   Exclusion from analysis: no information.   Follow‐up period: 30 days.
Participants	Taiwan   Hospitalised cirrhotic patients with upper gastrointestinal haemorrhage, without infection, or treatment with antibiotics in the previous two weeks.   Etiology of bleeding: oesophageal or gastric varices, peptic ulcers, others.   Initial infection assessment: chest X‐ray, blood count, urine/blood/ascites culture.   Source of bleeding determined by endoscopy scheduled within 24 h after enrolment.   Sclerotherapy, or band ligation performed within 48 h after enrolment.   Child‐Pugh class A/B/C: 11/64/45.
Interventions	Experimental: oral ciprofloxacin, 1 g/day, 7 days. Control: placebo.
Outcomes	Prevention of bacterial infections.
Notes	Authors were contacted for the first version of this review.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	No information provided.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	Unclear risk	No information provided.
Incomplete outcome data addressed?   All outcomes	Low risk	No missing outcome data.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Low risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	Low risk	All randomised participants were included in the analysis.
Sample calculation?	High risk	No sample calculation procedure.

Methods	Data collection: 01/2010 to 12/2004.   Exclusion from the analysis: past history of previous gastroesophageal variceal bleeding, or surgical or endoscopic treatment of gastroesophageal varices; or antibiotics use in the previous two weeks; or terminal illness (or non hepatic malignancy); or other causes of gastrointestinal bleeding.   Follow‐up period: 22 ± 14 months.
Participants	Korea Hospitalised cirrhotic patients with endoscopy‐proven bleeding from oesophageal or gastric varices, with no signs of infection at admission. Etiology of the bleeding: oesophageal or gastric varices. Initial infection assessment: complete blood cell count, chest X‐ray, urine analysis and culture, blood culture, and ascitic fluid neutrophil count with culture. Endoscopy procedure was performed within 12h of admission to emergency room. Before endoscopy octreotide was used, followed by endoscopic treatment (variceal ligation or sclerotherapy). Child‐Pugh class A/B/C: 8.7 ± 1.9 and 8.3 ± 2.1.
Interventions	Experimental: intravenous cefotaxime 2 g t.i.d for 7 days. Control: no antibiotic prophylaxis.
Outcomes	Rebleeding rate.
Notes	The authors were contacted via e‐mail and no response was received (23‐June‐2010).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	The sequence was generated using a table of random numbers.
Allocation concealment?	Low risk	The concealment was done using numbered envelopes.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	High risk	The proportion of missing outcomes is higher in the no prophylaxis group.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Low risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	High risk	The exclusion of the patients was not based on an exclusion rule.
Sample calculation?	Low risk	The sample was calculated considering the expected rebleeding rate.

Methods	Data collection: 05/2001 to 11/2003. Exclusion from analysis: Past history of previous variceal bleeding within 3 months. Follow‐up period: 42 days.
Participants	Czech Republic. Hospitalised (intensive care unit) patients with endoscopy proven oesophageal varices. Etiology of bleeding: oesophageal varices. Initial infection assessment: aerobic blood culture, anaerobic blood culture, perianal smear, urine examination, throat smear, ascites examination (>250 neutrophils/mL). The endoscopic treatment was performed within 3 h after admission to intensive care unit and was preceded by the administration of terlipressin. Schlerotherapy performed together with endoscopy to stop bleeding. Child‐Pugh class A/B/C: 4/19/23.
Interventions	Experimental:intravenous ampicillin/sulbactam 1.5 g b.i.d. for 7 days. Control: oral or through nasogastric tube norfloxacin 400 mg b.i.d. for 7 days.
Outcomes	Early and late mortality.
Notes	The authors were contacted via e‐mail and no response was received (23‐June‐2010).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	The sequence was generated using a system of accidental numbers.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	No missing outcome data.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Low risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	Unclear risk	No information provided.
Sample calculation?	High risk	No sample calculation procedure.

Methods	Data collection: 7/1999 to 8/2000.   Exclusion from analysis: life expectancy less than 7 days, fever or infection on entry, positive bacterial culture on entry, and previous use of antibiotics within 2 weeks prior to admission.   Initial infection assessment: blood culture, urine culture cell, chest X‐ray, sputum culture, and ascitic fluid culture.   Follow‐up period: 7 days.
Participants	Taiwan Hospitalised cirrhotic patients with endoscopy proven upper gastrointestinal bleeding, without infection or previous use of antibiotics. Endoscopy was performed 12 h within of hospitalisation. The haemostatic procedure include band ligation plus somastotatin, sandostatin or terlipressin (variceal bleeding) or endoscopic injection of water or diluted epinephrine (peptic ulcer bleeding). Child‐Pugh class A/B/C: 27/50/20.
Interventions	Experimental: intravenous cefazolin 1 g tid during 3 days and then shift to oral cephalexin 500 mg qid for 4 days. Control: no antibiotic prophylaxis.
Outcomes	Prevention of bacterial infections.
Notes	The authors were contacted via e‐mail and no response was received (23‐June‐2010).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	No information provided.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	Missing data are balanced in number across the intervention groups.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Low risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	High risk	The exclusion of the patients was based an exclusion rule.
Sample calculation?	High risk	No sample calculation procedure.

Methods	Data collection: 12/89 to 03/92.   Exclusion from analysis: infection on admission, undergoing surgery within 24 hs after admission, or death in the first 12 h.   Follow‐up period: up to 10 days after stopping bleeding (four weeks).
Participants	France   Hospitalised cirrhotic patients with upper gastrointestinal haemorrhage, without infection, or treatment with antibiotics in the previous week.   Etiology of bleeding: oesophageal or gastric varices, gastric or duodenal ulceration, and others. Initial infection assessment: chest X‐ray, white blood cell count, urine/blood/ascites culture.   Source of bleeding determined by endoscopy scheduled within 12 hs after enrolment.   Emergency sclerotherapy performed for patients bleeding from varices.   Child‐Pugh class A/B/C: 16/54/49.
Interventions	Experimental (Group III): intravenous + oral ciprofloxacin 400mg per day, amoxicillin‐clavulanic acid 3g per day, until three days after cessation of haemorrhage. Control (Group I and II): no antibiotic prophylaxis.
Outcomes	Prevention of bacterial infections.
Notes	Two group of patients received no intervention and were combined for the purpose of this review. Authors were contacted for the first version of this review.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	The sequence was generated using a table of random numbers.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	Missing data are balanced in number across the intervention groups.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Low risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	High risk	The exclusion of the patients was based on an exclusion rule.
Sample calculation?	High risk	No sample calculation procedure.

Methods	Data collection: no information provided.   Exclusion from analysis: underwent surgery, or died within 24 hs after admission.   Follow‐up period: up to 10 days after stopping bleeding (four weeks).
Participants	Spain   Hospitalised cirrhotic patients with upper gastrointestinal haemorrhage, without infection, or treatment with antibiotics in the previous two weeks.   Etiology of bleeding: oesophageal varices (128 patients), gastric haemorrhage ( 10 patients), peptic ulcers (8 patients).   Initial infection assessment: chest X‐ray, white blood cell count, urine/blood/ascites culture.   Source of bleeding determined by endoscopy scheduled within 24 hs after enrolment.   Emergency sclerotherapy performed for patients bleeding from varices.   Child‐Pugh class A/B/C: not reported.
Interventions	Experimental (non‐absorbable antibiotics):   Group Ia ‐ oral gentamicin (200mg) + vancomycin (500 mg) + nystatin (10^6 UI) every six hs, until two days after cessation of haemorrhage.   Group Ib ‐ neomycin (1 gm) + colistin (1.5 x 10^6 UI) + nystatin (10^6 UI) every six hs, until two days after cessation of haemorrhage. Control (Group II): no antibiotic prophylaxis.
Outcomes	Prevent bacterial infections.
Notes	The antibiotic therapy regimen was modified after inclusion of the first 40 patients in the experimental group because of budget constraints.   The two group of patients treated with antibiotics were combined for the purpose of this review. Authors were contacted for the first version of this review, and additional information was received.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	The sequence was generated using random number table.
Allocation concealment?	Low risk	Adequate, this item keeps as the first review.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	High risk	The proportion of missing outcomes is not balanced among the groups.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	High risk	The antibiotic therapy regimen was modified after inclusion of the first 40 patients in the experimental group because of budget constraints.
Intention to treat analysis?	High risk	The exclusion of the patients was based on an exclusion rule.
Sample calculation?	High risk	No sample calculation procedure.

Methods	Data collection: 06/93 to 06/95.   Exclusion from analysis: occult infection, or underwent surgery, or died within 24 hs after admission.   Follow‐up period: up to three weeks.
Participants	Spain   Hospitalised cirrhotic patients with upper gastrointestinal haemorrhage and high risk of infection, without infection, or treatment with antibiotics in the previous two weeks.   Etiology of bleeding: no information.   Initial infection assessment: chest X‐ray, white blood cell count, urine/blood/ascites culture.   Source of bleeding determined by endoscopy scheduled within four hs after enrolment.   Emergency sclerotherapy performed for patients bleeding from varices.   Child‐Pugh class A/B/C: 4/31/11.
Interventions	Experimental: oral norfloxacin 800 mg/day, during seven days plus intravenous ceftriaxone 2g/day the first three days. Control: oral norfloxacin 800 mg/day, seven days.
Outcomes	Prevent bacterial infections.
Notes	The initial sample size was calculated to be 152 patients. An interim analysis was done after 1/3 of patients were included, a statistical significant result was found and the trial was stopped. Authors were contacted for the first version of this review.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	No information provided.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	Reassons for missing data are based on the violation of the protocol.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	High risk	The trials was stopped due to economical differences and not due to primary outcome differences.
Intention to treat analysis?	High risk	The exclusion of the patients was based on an exclusion rule.
Sample calculation?	Unclear risk	The was calculated considering the primary outcome, but the trial was stopped considering differences in secondary outcomes.

Methods	Data collection: 08/89 to 12/91.   Exclusion from analysis: no information provided.   Follow‐up period: up to 24 hs.
Participants	Australia   Hospitalised cirrhotic patients with upper gastrointestinal haemorrhage and high risk of infection, without infection, or treatment with antibiotics in the previous two weeks.   Etiology of bleeding: no information provided.   All patients underwent emergency section of sclerotherapy, followed by further sections in weekly intervals.   Initial infection assessment: chest X‐ray, white blood cell count, urine/blood/ascites culture.   Child‐Pugh class A/B/C: 8/18/13.
Interventions	Experimental: intravenous cefotaxime, 1 g immediately before sclerotherapy. Control: no antibiotic prophylaxis.
Outcomes	Prevent bacterial infections.
Notes	Authors were contacted for the first version of this review.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	The sequence was generated using a table of random numbers.
Allocation concealment?	Low risk	The treatment was contained in sealed envelopes.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	Reassons for missing data are based on the violation of the protocol.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Unclear risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	High risk	The exclusion of the patients was based on an exclusion rule.
Sample calculation?	High risk	No sample calculation procedure.

Methods	Data collection: 08/89 to 06/91.   Exclusion from analysis: underwent surgery, or died within 24 hs after admission.   Follow‐up period: up to four weeks.
Participants	Spain   Hospitalised cirrhotic patients with upper gastrointestinal haemorrhage, without infection or treatment with antibiotics in the previous two weeks.   Etiology of bleeding: oesophageal varices, gastric haemorrhage, peptic ulcers, and others.   Initial infection assessment: chest X‐ray, white blood cell count, urine/blood/ascites culture.   Source of bleeding determined by endoscopy scheduled within four hs after enrolment.   Emergency sclerotherapy performed for patients bleeding from varices.   Child‐Pugh class A/B/C: 40/55/24.
Interventions	Experimental: oral norfloxacin 800 mg/day during seven days. Control: no antibiotic prophylaxis.
Outcomes	Prevent bacterial infection.
Notes	Authors were contacted for the first version of this review.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	The sequence was generated using a table of random numbers.
Allocation concealment?	Unclear risk	No information provided.
Blinding?   All outcomes	High risk	Not a blinded trial.
Incomplete outcome data addressed?   All outcomes	Low risk	Reassons for missing data are based on the violation of the protocol and are balanced across the groups.
Free of selective reporting?	Low risk	The report include all the expected outcomes.
Free of other bias?	Low risk	The trial appears to be free of other sources of bias.
Intention to treat analysis?	High risk	The exclusion of the patients was based on an exclusion rule.
Sample calculation?	High risk	No sample calculation procedure.

PERMALINK

Antibiotic prophylaxis for cirrhotic patients with upper gastrointestinal bleeding

Norberto C Chavez‐Tapia

Tonatiuh Barrientos‐Gutierrez

Felix I Tellez‐Avila

Karla Soares‐Weiser

Misael Uribe

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Characteristics of trials

Characteristics of participants

Characteristics of interventions

Characteristics of outcome measures

Assessment of risk of bias in included studies

Sequence generation

Allocation concealment

Blinding

Incomplete outcome data

Selective outcome reporting

Other sources of bias

Measures of treatment effect

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Sensitivity analysis

Results

Description of studies

Results of the search

Included studies

1. Group of antibiotics compared versus no intervention or placebo.

2. Trials comparing different antibiotic regimens.

Excluded studies

Risk of bias in included studies

1.

2.

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Mortality

3.

Mortality from bacterial infections

4.

Bacterial infections

5.

Drop‐outs and sensitivity analysis

Adverse events

Quality of life

Antibiotic regimens

3. Test of interaction for antibiotic regimens (assessing bacterial infections).

Rebleeding

Days of hospitalisation

Antibiotic regimens versus other antibiotic regimens

Study	Reason for exclusion
Gines 1990	Spain   Randomised clinical trial.   Through communication with the first author it was clarified that the patients did not have upper gastrointestinal bleeding.
Goulis 1998	UK   Observational study. Not a randomised trial.
Henrion 1992	Belgium   Controlled clinical study.   Inadequate allocation concealment, alternate method.
Husova 2005	Czech Republic   Observational study. Not a randomised trial.
Novella 1997	Spain   Randomised clinical trial.   Through communication with the first author it was clarified that the patients did not have upper gastrointestinal bleeding.
Pohl 2004	Germany   Retrospective cohort study.
Pulanic 1989	Yugoslavia   Randomised clinical trial.   No clinical outcomes, only laboratory variables after antibiotic prophylaxis.
Wilbur 2005	Canada   Retrospective cohort study.
Zhao 2002	China   Not a randomised trial.