EASL Clinical Practice Guidelines

Spontaneous bacterial peritonitis

SBP is a very common bacterial infection in patients with cirrhosis and ascites [[10], [105], [106], [107]]. When first described, its mortality exceeded 90% but it has been reduced to approximately 20% with early diagnosis and treatment [[6], [108]].

Diagnosis of spontaneous bacterial peritonitis

Diagnostic paracentesis: in whom and when

The diagnosis of SBP is based on diagnostic paracentesis [10]. All patients with cirrhosis and ascites are at risk of SBP and the prevalence of SBP in outpatients is 1.5–3.5% [[109], [110]] and ∼10% in hospitalized patients [109]. Half the episodes of SBP are present at the time of hospital admission while the rest are acquired during hospitalization [10].

Patients with SBP may have one of the following [[10], [109], [111]]: (1) local symptoms and/or signs of peritonitis: abdominal pain, abdominal tenderness, vomiting, diarrhea, ileus; (2) signs of systemic inflammation: hyper or hypothermia, chills, altered white blood cell count, tachycardia, and/or tachypnea; (3) worsening of liver function; (4) hepatic encephalopathy; (5) shock; (6) renal failure; and (7) gastrointestinal bleeding. However, it is important to point out that SBP may be asymptomatic, particularly in outpatients [[109], [110]].

Ascitic fluid cell analysis

Peritoneal infection causes an inflammatory reaction resulting in an increased number of neutrophils in ascitic fluid. Despite the use of sensitive methods, ascites culture is negative in as many as 60% of patients with clinical manifestations suggestive of SBP and increased ascites neutrophil count [[10], [106], [107], [108]]. Ascitic fluid neutrophil count is obtained as follows: ascitic fluid is centrifuged, then a smear is stained with Giemsa and total and differential cell counts are made with an optical microscope. This can be done in less than 4 h [[10], [107], [108], [112]]. Historically, manual counts were recommended, as coulter counter determinations of neutrophil counts were inaccurate at the relatively low levels of neutrophils in ascitic fluid [10]. However, a recent study found excellent correlation between these two techniques, even at low counts, suggesting that automated counting may replace manual counts [113]. The greatest sensitivity for the diagnosis of SBP is reached with a cutoff neutrophil count of 250/mm3, although the greatest specificity is reached with a cutoff of 500 neutrophils/mm3 [[10], [66], [107]]. Since there may be some delay in obtaining an ascitic fluid cell count, the use of reagent strips (RSs) has been proposed for a rapid diagnosis of SBP (reviewed in [114]). These reagent strips, designed for use in urine, identify leukocytes by detecting their esterase activity via a colorimetric reaction [114]. However, a large, multicenter prospective study has shown that the Multistix 8 SG® RS has a low diagnostic accuracy for the diagnosis of SBP [109]. A critical review of 19 studies comparing RSs (i.e., either Multistix 8 SG®, Nephur®, Combur®, UriScan®, or Aution®) to cytobacteriological methods has shown that RSs have low sensitivity and a high risk of false negative results, in particular in patients with SBP and low neutrophil count [114]. Thus, the use of reagent strips cannot be recommended for the rapid diagnosis of SBP.

Ascitic fluid culture

When culture is positive (∼40% of cases), the most common pathogens include Gram-negative bacteria (GNB), usually Escherichia coli and Gram-positive cocci (mainly streptococcus species and enterococci) [[10], [105], [106], [107], [108]]. A recent study has shown that 30% of isolated GNB are resistant to quinolones and 30% are resistant to trimethoprim–sulfamethoxazole [106]. Seventy percent of quinolone-resistant GNB are also resistant to trimethoprim–sulfamethoxazole [106]. The incidence of SBP due to quinolone-resistant GNB is higher in patients on norfloxacin therapy than in patients 'naïve' for this treatment [106]. The rate of cephalosporin-resistant GNB is low in patients with SBP regardless of norfloxacin prophylaxis [106]. Patients on norfloxacin prophylaxis may develop SBP caused by Gram-positive cocci [[10], [106], [107], [108]]. Finally, the epidemiology of bacterial infections differs between community-acquired (in which GNB infections predominate) and nosocomial infections (in which Gram-positive infections predominate) [106].

Patients with an ascitic fluid neutrophil count ⩾250 cells/mm3 and negative culture have culture-negative SBP [[10], [115]]. Their clinical presentation is similar to that of patients with culture-positive SBP [[10], [116]] and should be treated in a similar manner.

Some patients have 'bacterascites' in which cultures are positive but there is normal ascitic neutrophil count (<250/mm3) [10]. In some patients bacterascites is the result of secondary bacterial colonization of ascites from an extraperitoneal infection. These patients usually have general symptoms and signs of infection. In other patients, 'bacterascites' is due to the spontaneous colonization of ascites, and they can either be clinically asymptomatic or have abdominal pain or fever. While in some patients, particularly in those who are asymptomatic, bacterascites represents a transient and spontaneously reversible colonization of ascites, in other patients, mainly those who are symptomatic, bacterascites may represent the first step in the development of SBP [10].

Spontaneous bacterial pleural empyema

Infection of a pre-existing hydrothorax, known as spontaneous bacterial pleural empyema, is uncommon although the exact prevalence is unknown [112]. The diagnosis is based on pleural fluid analysis obtained by diagnostic thoracocentesis. In the largest observational study reported so far, the diagnosis of spontaneous bacterial empyema was established when the pleural fluid analysis showed a positive culture and more than 250 neutrophils/mm3 or a negative culture and more than 500 neutrophils/mm3, in the absence of lung infection [117]. Pleural fluid culture in blood culture bottles was positive in 75% of cases [117]. Spontaneous bacterial pleural empyema was associated with SBP in ∼50% of cases [117].

Secondary bacterial peritonitis

A small proportion of patients with cirrhosis may develop peritonitis due to perforation or inflammation of an intra-abdominal organ, a condition known as secondary bacterial peritonitis. The differentiation of this condition from SBP is important. Secondary bacterial peritonitis should be suspected in patients who have localized abdominal symptoms or signs, presence of multiple organisms on ascitic culture, very high ascitic neutrophil count and/or high ascitic protein concentration, or in those patients with an inadequate response to therapy [112]. Patients with suspected secondary bacterial peritonitis should undergo appropriate radiological investigation such as CT scanning [112]. The use of other tests such as measurement of glucose or lactate dehydrogenase in ascitic fluid has been suggested to help with the diagnosis of secondary bacterial peritonitis [112]. However, there are very limited data on the specificity and sensitivity of these tests in this setting.

Recommendations A diagnostic paracentesis should be carried out in all patients with cirrhosis and ascites at hospital admission to rule out SBP. A diagnostic paracentesis should also be performed in patients with gastrointestinal bleeding, shock, fever, or other signs of systemic inflammation, gastrointestinal symptoms, as well as in patients with worsening liver and/or renal function, and hepatic encephalopathy (Level A1).

The diagnosis of SBP is based on neutrophil count in ascitic fluid of >250/mm3 as determined by microscopy (Level A1). At present there are insufficient data to recommend the use of automated cell counters or reagent strips for the rapid diagnosis of SBP.

Ascitic fluid culture is frequently negative even if performed in blood culture bottles and is not necessary for the diagnosis of SBP, but it is important to guide antibiotic therapy (Level A1). Blood cultures should be performed in all patients with suspected SBP before starting antibiotic treatment (Level A1).

Some patients may have an ascitic neutrophil count less than 250/mm3 but with a positive ascitic fluid culture. This condition is known as bacterascites. If the patient exhibits signs of systemic inflammation or infection, the patient should be treated with antibiotics (Level A1). Otherwise, the patient should undergo a second paracentesis when culture results come back positive. Patients in whom the repeat ascitic neutrophil count is >250/mm3 should be treated for SBP, and the remaining patients (i.e., neutrophils <250/mm3) should be followed up (Level B1).

Spontaneous bacterial pleural empyema may complicate hepatic hydrothorax. Diagnostic thoracocentesis should be performed in patients with pleural effusion and suspected infection with inoculation of fluid into blood culture bottles (Level A1). The diagnosis is based on positive pleural fluid culture and increased neutrophil count of >250/mm3 or negative pleural fluid culture and >500 neutrophils/mm3 in the absence of pneumonia (Level B1).

Patients with suspected secondary bacterial peritonitis should undergo appropriate radiological investigation such as CT scanning (Level A1). The use of other tests such as measurement of glucose or lactate dehydrogenase in ascitic fluid cannot be recommended for the diagnosis of secondary bacterial peritonitis (Level B1).

Management of spontaneous bacterial peritonitis

Empirical antibiotic therapy

Empirical antibiotic therapy must be initiated immediately after the diagnosis of SBP, without the results of ascitic fluid culture [[10], [107]]. Potentially nephrotoxic antibiotics (i.e., aminoglycosides) should not be used as empirical therapy [10]. Cefotaxime, a third-generation cephalosporin, has been extensively investigated in patients with SBP because it covers most causative organisms and because of its high ascitic fluid concentrations during therapy [[118], [119], [120], [121], [122]]. Infection resolution is obtained in 77–98% of patients. A dose of 4 g/day is as effective as a dose of 8 g/day [119]. A 5-day therapy is as effective as a 10-day treatment [123] (Table 6).

Table 6
Antibiotic therapy for spontaneous bacterial peritonitis in patients with cirrhosis.

Reference Treatments Number of patients Infection resolution (%) In-hospital survival (%)
Felisart, 1985 [ 118] Tobramycin (1.75 mg/kg/8h IV)

plus ampicillin (2 g/4h IV)

versus cefotaxime (2 g/4h IV)
36 56 61
37 85 73
Rimola, 1995 [ 119] Cefotaxime (2 g/6h IV)

versus cefotaxime (2 g/12h IV)
71 77 69
72 79 79
Navasa, 1996 [ 120] Ofloxacin (400 mg/12h PO)

versus cefotaxime (2 g/6h IV)
64 84 81
59 85 81
Sort, 1999 [ 121] Cefotaxime (2 g/6h IV)

versus cefotaxime (2 g/6h IV) plus IV albumin
63 94 71
63 98 90
Ricart, 2000 [ 122] Amoxicillin/clavulanic acid (1/0.2 g/8h)

IV followed by 0.5/0.125 g/8h PO

versus cefotaxime (1 g/6h IV)
24 87 87
24 83 79
Terg, 2000 [ 124] Ciprofloxacin (200 mg/12h IV for 7 days) versus ciprofloxacin (200 mg/12h for 2 days, followed by 500 mg/12h PO for 5 days) 40 76 77
40 78 77
*p <0.02 versus tobramycin plus ampicillin.

**p = 0.01 versus cefotaxime alone.

Alternatively, amoxicillin/clavulanic acid, first given intravenously then orally, has similar results with respect to SBP resolution and mortality, compared with cefotaxime [122] and with a much lower cost. However, there is only one comparative study with a small sample size and results should be confirmed in larger trials. Ciprofloxacin, given either for 7 days intravenously or for 2 days intravenously followed by 5 days orally, results in a similar SBP resolution rate and hospital survival compared with cefotaxime, but with a significantly higher cost [124]. However, switch therapy (i.e., use of intravenous antibiotic initially, followed by oral step-down administration) with ciprofloxacin is more cost-effective than intravenous cefotaxime [125]. Oral ofloxacin has given similar results as intravenous cefotaxime in uncomplicated SBP, without renal failure, hepatic encephalopathy, gastrointestinal bleeding, ileus, or shock [120]. Cefotaxime or amoxicillin/clavulanic acid are effective in patients who develop SBP while on norfloxacin prophylaxis [10].

If ascitic fluid neutrophil count fails to decrease to less than 25% of the pre-treatment value after 2 days of antibiotic treatment, there is a high likelihood of failure to respond to therapy [[10], [112]]. This should raise the suspicion of an infection caused by bacteria resistant to antibiotic therapy, indicating the need for modification of antibiotic treatment according to in vitro sensitivity or on empiric basis or the presence of 'secondary peritonitis'.

Recommendations. Empirical antibiotics should be started immediately following the diagnosis of SBP (Level A1).

Since the most common causative organisms of SBP are Gram-negative aerobic bacteria, such as E. coli, the first line antibiotic treatment are third-generation cephalosporins (Level A1). Alternative options include amoxycillin/clavulanic acid and quinolones such as ciprofloxacin or ofloxacin. However, the use of quinolones should not be considered in patients who are taking these drugs for prophylaxis against SBP, in areas where there is a high prevalence of quinolone-resistant bacteria or in nosocomial SBP (Level B1).

SBP resolves with antibiotic therapy in approximately 90% of patients. Resolution of SBP should be proven by demonstrating a decrease of ascitic neutrophil count to <250/mm3 and sterile cultures of ascitic fluid, if positive at diagnosis (Level A1). A second paracentesis after 48 h of start of treatment may help guide the effect of antibiotic therapy.

Failure of antibiotic therapy should be suspected if there is worsening of clinical signs and symptoms and/or no marked reduction or increase in ascitic fluid neutrophil count compared to levels at diagnosis. Failure of antibiotic therapy is usually due to resistant bacteria or secondary bacterial peritonitis. Once secondary bacterial peritonitis has been excluded, antibiotics should be changed according to in vitro susceptibility of isolated organisms, or modified to alternative empiric broad spectrum agents (Level A1).

Spontaneous bacterial empyema should be managed similarly as SBP

Intravenous albumin in patients with spontaneous bacterial peritonitis without septic shock

SBP without septic shock may precipitate deterioration of circulatory function with severe hepatic insufficiency, hepatic encephalopathy, and type 1 hepatorenal syndrome (HRS) [[121], [126], [127]] and has approximately a 20% hospital mortality rate despite infection resolution [[121], [126]].

A randomized, controlled study in patients with SBP treated with cefotaxime showed that albumin (1.5 g/kg body weight at diagnosis, followed by 1 g/kg on day 3) significantly decreased the incidence of type 1 HRS (from 30% to 10%) and reduced mortality from 29% to 10% compared with cefotaxime alone. Treatment with albumin was particularly effective in patients with baseline serum bilirubin ⩾68 μmol/L (4 mg/dl) or serum creatinine ⩾88 μmol/L (1 mg/dl). It is unclear whether intravenous albumin is useful in patients with baseline bilirubin <68 μmol/L and creatinine <88 μmol/L, as the incidence of type 1 HRS was very low in the two treatment groups (7% without albumin and 0% with albumin) [121]. Non-randomized studies in patients with SBP also show that the incidence of renal failure and death are very low in patients with moderate liver failure and without renal dysfunction at diagnosis of SBP [[128], [129], [130]]. It is not known whether crystalloids or artificial colloids could replace albumin in the prevention of HRS in patients with SBP. Albumin improves circulatory function in patients with SBP while equivalent doses of hydroxyethyl starch have no such beneficial effect [131]. Clearly, further studies are needed to assess the efficacy of albumin as well as other expanders in the management of SBP. Until further trials are completed, albumin infusion appears a valuable adjunction to the treatment of SBP.

Recommendations HRS occurs in approximately 30% of patients with SBP treated with antibiotics alone, and is associated with a poor survival. The administration of albumin (1.5 g/kg at diagnosis and 1g/kg on day 3) decreases the frequency of HRS and improves survival (Level A1). It is unclear whether albumin is useful in the subgroup of patients with baseline serum bilirubin <68 μmol/L and creatinine <88 μmol/L (Level B2). Until more information is available, we recommend that all patients who develop SBP should be treated with broad spectrum antibiotics and intravenous albumin (Level A2).

Prophylaxis of spontaneous bacterial peritonitis

Since most episodes of SBP are thought to result from the translocation of enteric GNB, the ideal prophylactic agent should be safe, affordable, and effective at decreasing the amounts of these organisms from the gut while preserving the protective anaerobic flora (selective intestinal decontamination) [108]. Given the high cost and inevitable risk of developing resistant organisms, the use of prophylactic antibiotics must be strictly restricted to patients at high risk of SBP. Three high-risk patient populations have been identified: (1) patients with acute gastrointestinal hemorrhage; (2) patients with low total protein content in ascitic fluid and no prior history of SBP (primary prophylaxis); and (3) patients with a previous history of SBP (secondary prophylaxis).

Patients with acute gastrointestinal hemorrhage

Bacterial infection, including SBP, is a major problem in patients with cirrhosis and acute gastrointestinal hemorrhage, occurring in between 25% and 65% of patients with gastrointestinal bleeding [[132], [133], [134], [135], [136], [137], [138], [139], [140], [141]]. The incidence of bacterial infection is particularly high in patients with advanced cirrhosis and/or severe hemorrhage [[138], [139]]. In addition, the presence of bacterial infection in patients with variceal hemorrhage is associated with an increased rate of failure to control bleeding [[142], [143]], rebleeding [[136], [138]], and hospital mortality [[139], [143], [144], [145]]. Antibiotic prophylaxis has been shown to prevent infection in patients with gastrointestinal bleeding [[10], [107], [108]] and decrease the rate of rebleeding [144]. A meta-analysis [139] of five studies performed in patients with gastrointestinal bleeding [[132], [134], [135], [137], [140]] has shown that antibiotic prophylaxis significantly decreased both the incidence of severe infections (SBP and/or septicemia) and mortality.

Selective intestinal decontamination with norfloxacin (400 mg/12 h orally for 7 days), a quinolone with relatively poor gastrointestinal absorption, and which has antibacterial activity against GNB but not against Gram-positive cocci or anaerobic bacteria, is the most commonly used approach for the prophylaxis of bacterial infections in patients with gastrointestinal hemorrhage [[10], [107], [134]]. In recent years, the epidemiology of bacterial infections in cirrhosis has changed, with an increasing incidence of SBP and other infections caused by quinolone-resistant bacteria (see above) [[106], [146], [147]]. In addition, a substantial number of infections in patients with gastrointestinal hemorrhage are caused by Gram-positive bacteria likely related to invasive procedures used in these patients [106].

A recent study comparing oral norfloxacin to intravenous ceftriaxone for the prophylaxis of bacterial infection in patients with gastrointestinal bleeding and advanced cirrhosis (at least 2 of the following: ascites, severe malnutrition, encephalopathy, or bilirubin >3 mg/dl) showed that ceftriaxone was more effective than norfloxacin in the prevention of infections [148].

Recommendations In patients with gastrointestinal bleeding and severe liver disease (see text) ceftriaxone is the prophylactic antibiotic of choice, whilst patients with less severe liver disease may be given oral norfloxacin or an alternative oral quinolone to prevent the development of SBP (Level A1).

Patients with low total protein content in ascitic fluid without prior history of spontaneous bacterial peritonitis

Cirrhotic patients with low ascitic fluid protein concentration (<10 g/L) and/or high serum bilirubin levels are at high risk of developing a first episode of SBP [[10], [149], [150], [151], [152]]. Several studies have evaluated prophylaxis with norfloxacin in patients without prior history of SBP (Table 7) [[153], [154], [155], [156], [157]]. One pilot, randomized, open-label trial was performed comparing primary continuous prophylaxis with norfloxacin to inpatient-only prophylaxis in 109 patients with cirrhosis and ascitic fluid total protein level ⩽15 g/L or serum bilirubin level >2.5 mg/dl [154]. SBP was reduced in the continuous treatment group at the expense of more resistance of gut flora to norfloxacin in that group. In another study, 107 patients with ascitic fluid total protein level <15 g/L were randomized in a double-blind manner to receive norfloxacin (400 mg/day for 6 months) or placebo [155]. Of note, the existence of severe liver failure was not an inclusion criterion. The primary endpoint was the occurrence of GNB infections. Norfloxacin significantly decreased the probability of developing GNB infections, but had no significant effect on the probability of developing SBP or survival. However, in this trial, the sample size was not calculated to detect differences in survival. In a third investigation, 68 patients with cirrhosis and low ascites protein levels (<15 g/L) with advanced liver failure [Child-Pugh score ⩾9 points with serum bilirubin level ⩾3 mg/dl or impaired renal function (serum creatinine level ⩾1.2 mg/dl, blood urea nitrogen level ⩾25 mg/dl, or serum sodium level ⩽130 mEq/L)] were randomized in a double-blind, placebo-controlled trial, to receive norfloxacin (400 mg/day for 12 months) or placebo [156]. The primary endpoints of the trial were 3-month and 1-year survival. Norfloxacin significantly improved the 3-month probability of survival (94% versus 62%; p = 0.03) but at 1 year the difference in survival was not significant (60% versus 48%; p = 0.05). Norfloxacin administration significantly reduced the 1-year probability of developing SBP (7% versus 61%) and HRS (28% versus 41%). In a fourth study, 100 patients with ascitic fluid total protein level <15 g/L were randomized in double-blind, placebo-controlled trial to ciprofloxacin (500 mg/day for 12 months) or placebo [157]. Enrolled patients had moderate liver failure (the Child-Pugh scores were 8.3 ± 1.3 and 8.5 ± 1.5, in the placebo and ciprofloxacin group, respectively). The primary endpoint was the occurrence of SBP. Although SBP occurred in 2 (4%) patients of the ciprofloxacin group and in 7 (14%) patients of the placebo group, this difference was not significant. Moreover, the probability of being free of SBP was not significant (p = 0.076). The probability of remaining free of bacterial infections was higher in patients receiving ciprofloxacin (80% versus 55%; p = 0.05). The probability of survival at 1 year was higher in patients receiving ciprofloxacin (86% versus 66%; p < 0.04). Nevertheless, a type II error cannot be ruled out as the sample size was not calculated to detect differences in survival. The duration of primary antibiotic prophylaxis has not been established.

Table 7
Antibiotic therapy for prophylaxis of spontaneous bacterial peritonitis (SBP) in patients with cirrhosis.a

Reference Type of prophylaxis Treatments Number of patients Number of GNB b infections p-value Incidence of SBP n (%) p-value
Ginès, 1990 [ 158] Enrolled only patients

with prior SBP c
Norfloxacin

versus placebo
40 1 5 (12) 0.02
40 10 14 (35)
Soriano, 1991 [ 153] Enrolled patients without prior SBP and patients

with prior SBP d
Norfloxacin

versus no treatment
32 0 <0.001 0 (0) <0.02
31 9 7 (22.5)
Singh, 1995 [ 161] Enrolled patients without prior SBP and patients

with prior SBP d
Trimethoprim–sulfamethoxazole versus no treatment 30 9 1 (3) 0.03
30 0 8 (27) e
Rolachon, 1995 [ 160] Enrolled patients without prior SBP and patients

with prior SBP c
Ciprofloxacin

versus placebo
28 1 1 (4) <0.05
32 0 7 (22)
Novella, 1997 [ 154] Enrolled only patients

without prior SBP d
Continuous norfloxacin

versus in patient-only prophylaxis
56 11 1 (1.8) <0.01
53 13 9 (16.9)
Grangé, 1998 [ 155] Enrolled only patients

without prior SBP c
Norfloxacin

versus placebo
53 0 <0.04 0 (0) NA
54 6 5 (9)
Fernández, 2007 [ 156] Enrolled only patients

without prior SBP c
Norfloxacin

versus placebo
35 13 2 (6) 0.02
33 6 10 (30)
Terg, 2008 [ 157] Enrolled only patients

without prior SBP c
Ciprofloxacin

versus placebo
50 2 (4) 0.076
50 7 (14)
NA, not available.
a. Studies appear in chronological order.
b. GNB means Gram-negative bacteria.
c. Randomized, double-blind, placebo-controlled trial.
d. Randomized, unblinded trial.
e. Including one patient with spontaneous bacteremia due to Klebsiella pneumonia.

Recommendations One double-blind, placebo-controlled, randomized trial performed in patients with severe liver disease (see text) with ascitic fluid protein lower than 15 g/L and without prior SBP showed that norfloxacin (400 mg/day) reduced the risk of SBP and improved survival. Therefore, these patients should be considered for long-term prophylaxis with norfloxacin (Level A1).

In patients with moderate liver disease, ascites protein concentration lower than 15 g/L, and without prior history of SBP, the efficacy of quinolones in preventing SBP or improving survival is not clearly established. Studies are needed in this field.

Patients with prior spontaneous bacterial peritonitis

In patients who survive an episode of SBP, the cumulative recurrence rate at 1 year is approximately 70% [108]. The probability of survival at 1 year after an episode of SBP is 30–50% and falls to 25–30% at 2 years. Therefore, patients recovering from an episode of SBP should be considered for liver transplantation. There is only one randomized, double-blind, placebo-controlled trial of norfloxacin (400 mg/day orally) in patients who had a previous episode of SBP [158] (Table 7). Treatment with norfloxacin reduced the probability of recurrence of SBP from 68% to 20% and the probability of SBP due to GNB from 60% to 3%. Survival was not an endpoint of this study. In an open-label, randomized study comparing norfloxacin 400 mg/day to rufloxacin 400 mg/week in the prevention of SBP recurrence, 1-year probability of SBP recurrence was 26% and 36%, respectively (p = 0.16) [159]. Norfloxacin was more effective in the prevention of SBP recurrence due to Enterobacteriaceae (0% versus 22%, p = 0.01). Three other studies assessed the effects of ciprofloxacin, trimethoprim–sulfamethoxazole, and norfloxacin, but they included patients with and without previous episodes of SBP [[153], [160], [161]] (Table 7). All studies showed a reduced incidence of SBP with antibiotic prophylaxis.

It is uncertain whether prophylaxis should be continued without interruption until liver transplantation or death in all patients with prior SBP or if treatment could be discontinued in patients showing an improvement of liver disease.

Recommendations Patients who recover from an episode of SBP have a high risk of developing recurrent SBP. In these patients, the administration of prophylactic antibiotics reduces the risk of recurrent SBP. Norfloxacin (400 mg/day, orally) is the treatment of choice (Level A1). Alternative antibiotics include ciprofloxacin (750 mg once weekly, orally) or co-trimoxazole (800 mg sulfamethoxazole and 160 mg trimethoprim daily, orally), but evidence is not as strong as that with norfloxacin (Level A2).

Patients who recover from SBP have a poor long-term survival and should be considered for liver transplantation (Level A1).

Issues with prolonged antibiotic prophylaxis

As mentioned earlier, prolonged antibiotic prophylaxis (primary or secondary) has led to the emergence of GNB resistant to quinolones and even to trimethoprim/sulfamethoxazole [106]. In addition, there is an increased likelihood of infections from Gram-positive bacteria in patients who have received long-term SBP prophylaxis [[156], [162]]. This underlines the need to restrict the use of prophylactic antibiotics to patients with the greatest risk of SBP. Common sense would suggest that quinolone prophylaxis should be discontinued in patients who develop infection due to quinolone-resistant bacteria. However, there are no data to support this.