al a

aspontaneous bacterial peritonitis

Simrn , Ulrika Broom , Staffan Wahlin ,

Abstract

Background: Spontaneous bacterial peritonitis (SBP), which has been reported to be present in 1030% of patients with cirrhotic ascites, may

European Journal of Internal Medicieasily be overlooked. An important aim of our study was to determine whether there are any clinical signs which, in clinical practice, maypredict or exclude SBP.Methods: We studied 133 patients with cirrhotic ascites from medical units at nine Swedish university hospitals where there had been at leastone diagnostic ascites tap with analysis of polymorphonuclear leukocytes in the ascites fluid. The patients had initially been questioned aboutbackground factors and physically examined according to a standardized case record form. Samples of blood, urine, and ascites were thendrawn for analysis according to a structured schedule.Results: SBP could be excluded in 80% of all the cases and was confirmed in 8% of the 133 patients in the final analysis. Abdominal painand abdominal tenderness were more common in patients with SBP (pb0.01), but no other physical sign or laboratory test could separateSBP cases from the others.Conclusions: SBP was present in about one-tenth of the hospitalized patients with cirrhotic ascites in this cohort. Performing repeated physicalexaminations and paying particular attention to abdominal tenderness may be the best way to become aware of the possible development of thiscomplication. 2006 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.

Keywords: Cirrhosis; Diagnosis; Signs; Spontaneous bacterial peritonitis; Symptoms

1. Introduction

Although spontaneous bacterial peritonitis (SBP) is animportant complication of cirrhotic ascites, having been

and clinical picture are somewhat obscure. There are severalexplanations for this, due to changes in the diagnostic criteriathat have been established for SBP in the years since thisclinical entity was first described in 1964 [2]. At that time,Lars Lf d, Rolf Hultcrantz e, Klas Sjberg f, Hanna Sandberg Gertzn g,Hanne Prytz h, Sven Almer i

a Department of Medicine, Sahlgrenska University Hospital/stra, SE-416 85, Gteborg, Swedenb Department of Medicine, Sahlgrenska University Hospital/Sahlgrenska, Gteborg, Swedenc Department of Medicine, Karolinska University Hospital/Huddinge, Huddinge, Sweden

d Department of Medicine, Akademiska University Hospital, Uppsala, Swedene Department of Medicine, Karolinska University Hospital/Solna, Solna, Sweden

f Department of Medicine, University Hospital MAS, Malm, Swedeng Department of Medicine, University Hospital, rebro, Swedenh Department of Medicine, University Hospital, Lund, Sweden

i Department of Medicine, University Hospital, Linkping, Sweden

Received 13 February 2006; received in revised form 1 June 2006; accepted 11 July 2006Sven Wallerstedt a,, Rolf Olsson b, Magnus b c cOrigin

Abdominal tenderness inestimated to be present in 1030% of cases [1], its prevalence

Corresponding author. Tel.: +46 31 3434000; fax: +46 31 259254.E-mail address: sven.wallerstedt@medic.gu.se (S. Wallerstedt).

0953-6205/$ - see front matter 2006 European Federation of Internal Medicindoi:10.1016/j.ejim.2006.07.022rticle

scites patients indicates

ne 18 (2007) 4447www.elsevier.com/locate/ejimcultures positive for ascites and blood were obligatory for thediagnosis. Two decades later, Runyon and Hoefs described anascites culture-negative variant of SBP [3]. One reason for anegative ascites culture could be the low bacterial content in

e. Published by Elsevier B.V. All rights reserved.

(Table 1) before the ascites tap. The ascites volume wasdetermined as the fluid volume received by total paracentesiswith complete ascites mobilization as the goal (maximum 21 Lin a non-SBP patient). The results of routine analyses of blood,serum, ascites, and urine were also recorded in accordancewith a structured schedule (Table 2), and the patients were thenclassified according to their ChildPugh classification [7].Body mass index (BMI), calculated after ascites mobilization,was used to reflect malnutrition. The local ethics committeesapproved the study.

For the statistical analysis of possible differences betweenSBP and non-SBP cases with regard to background factors,symptoms, physical findings, and laboratory results, we usedFisher's non-parametric permutation test [10]. A p-valuebelow 0.05 was regarded as significant.

3. Results

SBP could be excluded in 123 of the original cohort of 154patients (80%). A positive SBP diagnosis was established in10 of the 133 patients included in the final analysis (7.5%). Apositive bacterial ascites culture was found in two of these tenpatients (Escherichia coli and alpha streptococci, respective-ly). In two of the eight culture-negative cases, antibiotics hadbeen given before the paracentesis.

Table 1Clinical data on 133 patients with and without SBP

SBP non-SBP

45nal of Internal Medicine 18 (2007) 4447SBP ascites, i.e., one to two organisms per mL [4], which, tosome extent, can be overcome using a blood culture technique[1]. Nowadays, a positive ascites culture is not mandatory forthe diagnosis of SBP, which is defined as neutrocytic asciteswith at least 0.250109 polymorphonuclear leukocytes (PMN)per liter in the absence of an abdominal infectious focus [1].Thus, the diagnosis may be overlooked if abdominal pa-racentesis and analysis of ascites are not performed. In the past,paracentesis was generally restricted to severely ill patientsbecause of an unfounded fear of complications [5]. In thesepatients, the prevalence of SBP was thought to be high. Themodern practice of paracentesis, also in less diseased patients,should result in a lower frequency of SBP. This idea issupported by the finding of SBP in 3.5% of 427 asymptomaticcases among 513 outpatients with cirrhotic ascites who under-went paracentesis [6].

Various predisposing factors, clinical signs, and prognosticmarkers for SBP development have been studied. Analyses offactors that are associated with SBP have generally shown thatthe parameters included in the ChildPugh classification [7],albumin, coagulation factors (prothrombin complex), bilirubin,and encephalopathy are the most important. In 1986, Runyonconcluded that there was an inverse relation between theprotein content in ascites and susceptibility to infections [8].This is supported, for example, by the finding that a proteinconcentration in ascites of less than 10 g/L is common in SBP,whereas carcinomatosis-related ascites with a high proteincontent very seldom becomes infected.

One aim of the present study was to evaluate the prevalenceof SBP in Swedish hospitalized patients with cirrhotic ascites.Based on our clinical experience, we hypothesized that SBP inmodern clinical practice should be infrequent. Another aimwas to study to what extent clinical and laboratory data could,in clinical practice, predict or exclude the presence of SBP inthese patients.

2. Materials and methods

Patients with a diagnosis of cirrhotic ascites and with ananalysis of ascites leukocyte count, but without secondaryperitonitis, were prospectively included in the study. Most ofthese patients had been hospitalized because of ascites. Onlythe first ascites episode during the study period was registered.Initially, 156 patients were included in the study, but twopatients were later excluded since the ascites was bloody andno correction [9] had been performed to compensate for addedblood leukocytes. Another 21 cases had to be excluded be-cause no differential leukocyte count had been performed,although the ascites leukocyte count was at least 0.250109/L.For the final analysis, 133 patients were available, and theywere divided into two groups. The first (non-SBP) groupconsisted of patients with a leukocyte or PMN count in ascitesbelow 0.250109/L. The second (SBP) group consisted ofpatients with an ascites PMN count of at least 0.250109/L.

S. Wallerstedt et al. / European JourAll patients were questioned about background factorsand physically examined according to a standardized form(n=10) (n=123)

Age (years) 61 58Sex [male:female] (n) 8:2 86:37

Background historySBP, earlier (n) 0 9 (119)Alcohol etiology (n) 6 84Hepatitis C antibodies (n) 4 (9) 25 (114)Antibiotics due to any infection inlast 6 months (n)

3 40 (121)

Cortisone treatment (n) 1 7Immunosuppressive treatment (n) 0 4 (122)Portosystemic encephalopathy, earlier (n) 2 26Variceal bleeding (n) 3 30 (117)Variceal sclerotherapy (n) 2 28 (115)

Present case historyAlcohol intake last week (n) 2 37Respiratory infection (n) 1 7Abdominal pain (n) 8 36 (121)Intravenous therapy (n) 6 61 (121)Days at hospital before puncture (median, n) 2.0 (8) 4.0 (50)

Physical findingsBody mass index [(weight in kg)/(length in m)2] 25.4 (8) 23.5 (85)ChildPugh-class [A:B:C] (n) 0:3:7 0:25:98Spiders3 (n) 6 59 (115)Fever (n) 4 29 (119)38.0 C (n) 1 7Abdominal tenderness (n) 7 28 (120)Portosystemic encephalopathy (n) 2 22 (119)In cases of missing data, the number of patients is given within parentheses.

nal ofThere were few clinical differences between the two groups(Table 1). Complaints of abdominal pain, as well as the pre-sence of abdominal tenderness, were, however, more commonin SBP than in non-SBP patients (80% vs. 30%, p=0.0050,and 70% vs. 23%, p=0.0079, respectively). With regard toabdominal tenderness as a possible sign of SBP, the positive

Table 2Laboratory tests (mean values with confidence intervals within parentheses)in 133 patients with and without SBP

SBP(n=10)

non SBP(n=123)

B-hemoglobin (g/L) 116 (102131) 114 (110117)B-leukocyte count (109/L) 11.5 (5.817.2) 10.5 (9.411.7)12

B-platelet count (109/L) 164 (98230) 183 (164202)S-AST (kat/L) 4.61 (011.06) 1.92 (1.562.28)12

S-ALT (kat/L) 1.92 (03.95) 1.24 (0.432.04)12

S-alkaline phosphatases (kat/L) 13.3 (3.323.3) 8.8 (7.89.8)12

S-bilirubin (mol/L) 140 (68212) 112 (88136)12

Prothrombin complex (INR) 1.4 (1.191.70)4 1.5 (1.461.59)12

S-sodium (mmol/L) 132 (128136) 133 (132134)S-potassium (mmol/L) 4.2 (3.64.8) 4.0 (3.84.1)S-creatinine (mol/L) 157 (92222) 124 (106142)12

S-albumin (g/L) 26 (2231)4 24 (2325)11

S-immunoglobulin G (g/L) 16 (1221)2 17 (1518)10

Ascites-volume (L) 5.5 (2.28.9)1 6.7 (5.57.9)5

Ascites-protein (g/L) 17 (726)2 13 (1115)8

Ascites-albumin (g/L) 10 (416)3 8 (510)9

Ascites-amylase (kat/L) 0.6 (01.3)1 0.6 (0.40.7)6

U-sodium (mmol/L) 31 (657)1 65 (5476)7

U-potassium (mmol/L) 37 (1361)1 33 (2938)7

1n=5, 2n=6, 3n=8, 4n=9, 5n=59, 6n=65, 7n=71, 8n=72, 9n=90, 10n=96,11n=115, 12n=122. B, blood; S, serum; U, urine.

46 S. Wallerstedt et al. / European Jourpredictive value was 20% (7/35) and the negative predictivevalue 97% (92/95).

None of the SBP patients had had an earlier SBP episode,whereas this was reported in the files of 9 of the 119 non-SBPcases. The time period between admission and paracentesiswas similar in the two groups. Signs of liver dysfunction,malnutrition, the presence of an alcohol or hepatitis C viraletiology, immune-modulating therapy, suspicion of a suscep-tibility to infections, esophageal varices, and intravenouscatheterization were not more common in the SBP group. Theresults from laboratory tests did not differ between the twogroups (Table 2).

4. Discussion

We found that SBP, as a complication of cirrhotic ascites,was less frequent than what has generally been published todate [1]. Although some of the local laboratories did notperform a PMN count in cases with an ascites leukocytecount below 0.500109/L, SBP could be excluded in morethan 80% of the cases since at least the total leukocyte countwas below 0.250109/L. Thus, the occurrence of SBP in7.5% of our cases might have been higher if this com-plication had been present in some of the excluded cases. Onthe other hand, our figure may be too high since the highprevalence of abdominal tenderness in our cohort, 27%, couldreflect readiness for paracentesis, especially in such caseswhere SBP is common. Furthermore, our clinical impressionis that there is a substantial number of patients with cirrhoticascites who do not undergo a diagnostic paracentesis.

A problem when comparing various prevalence studies isthe use of different diagnostic criteria. For example, in awell-performed Danish study from 1987 [11], 4 out of 13patients with SBP would nowadays have been classified ashaving bacterial ascites [1,4] rather than SBP, reducing thereported SBP prevalence from 19% to 13%. Other factors,such as geographical differences, may explain the rather highSBP prevalence of 23.7% in a recent Slovakian study, al-though an old disease definition, i.e., a PMN count of at least0.500109/L, was used in that study [12]. This reported pre-valence figure would have been even higher if the moderndefinition had been used.

Several studies have been performed to reveal factors pre-disposing to SBP. Andreu et al. [13] found that a prothrombincomplex greater than 1.2 (INR), a serum albumin below 27 g/L,and a serum bilirubin above 42.7 mol/L had prognostic sig-nificance. The two last parameters could be used in a formula tocalculate the relative risk of a first SBP. The three parametersmentioned are also components of the formula used for themodified ChildPugh classification [7]. Thus, it is not sur-prising that ChildPugh class C has been reported to be a riskfactor for SBP [5]. As is apparent fromour study, there is a greatoverlap in these factors between cases with and without SBP.This is also apparent with other reported risk factors, such asprotein content in ascites below 10 g/L, a history of varicealbleeding, or an earlier SBP episode [5].

Although SBP may be asymptomatic, this complicationmust be considered if a patient with cirrhotic ascites developssymptoms or signs of local and/or systemic infection or anunexplained clinical deterioration. Many clinical signs havebeen studied, but in our study abdominal tenderness, presentin 70%, was the only one that was strongly related to SBP.This is consistent with what has been reported by Runyon,who claimed that 3944% of patients with a diagnosis ofSBP showed this physical sign [14]. Abdominal tendernesswas, along with jaundice, the most prominent physical sign,54.5%, in a recent study of symptomatic SBP cases [15].Also, other symptoms and signs, such as abdominal pain,clinically relevant alterations of gastrointestinal motility,fever, mental deterioration, and renal impairment, may re-flect the development of SBP [1].

Irrespective of the reported SBP risk factors, paracentesisshould be performed in every hospitalized patient with cirrhoticascites. This is also true for patients admitted for reasons otherthan ascites. It is important to get a PMNcount in order to initiateantibiotic treatment and to use an adequate culture technique [1].According to our results, repeated paracentesis must be con-sidered for patients developing abdominal pain and/or abdom-inal tenderness, even though a minority of these patients haveSBP. On the other hand, the absence of abdominal tenderness

Internal Medicine 18 (2007) 4447very strongly contradicts the presence of this complication, notonly in outpatients [6] but also in hospitalized patients.

In conclusion, our study indicates that performing repeatedclinical examinations of patients with cirrhotic ascites, payingparticular attention to abdominal tenderness, is essential indeciding whether diagnostic paracentesis should be carried outto reveal possible SBP.

5. Learning points

Although infrequent, SBP should be considered in cirrhoticpatients with ascites and abdominal tenderness.

The absence of abdominal tenderness, on the other hand,strongly contradicts SBP.

Acknowledgements

The study was initiated by SILK, the Swedish InternalMedicine Liver Club, which is supported by Meda. We alsothank Anders Odn, PhD, for the statistical analyses.

References

[1] Rimola A, Garcia-Tsao G, Navasa M, Piddock LJV, Planas R, BernardB, et al, and the International Ascites Club. Diagnosis, treatment andprophylaxis of spontaneous bacterial peritonitis: a consensus docu-ment. J Hepatol 2000;32:14253.

[2] Conn HO. Spontaneous peritonitis and bacteremia in Laennec'scirrhosis caused by enteric organisms. A relatively common but rarely

[3] Runyon BA, Hoefs JC. Culture-negative neutrocytic ascites: a variantof spontaneous bacterial peritonitis. Hepatology 1984;4:120911.

[4] BhuvaM,GangerD, JensenD. Spontaneous bacterial peritonitis: an updateon evaluation, management, and prevention. Am J Med 1994;97:16975.

[5] Such J, Runyon BA. Spontaneous bacterial peritonitis. Clin Infect Dis1998;27:66976.

[6] Evans LT, Kim WR, Poterucha JJ, Kamath PS. Spontaneous bacterialperitonitis in asymptomatic outpatients with cirrhotic ascites. Hepatol-ogy 2003;37:897901.

[7] Albers I, Hartmann H, Bircher J, Creutzfeldt W. Superiority of the ChildPugh classification to quantitative liver function tests for assessingprognosis of liver cirrhosis. Scand J Gastroenterol 1989;24:26976.

[8] Runyon BA. Low-protein-concentration ascitic fluid is predisposed tospontaneous bacterial peritonitis. Gastroenterology 1986;91:13436.

[9] Hoefs JC. Increase in ascites white blood cell and protein concentra-tions during diuresis in patients with chronic liver disease. Hepatology1981;1:24954.

[10] Good P. Permutation tests. New York: Springer-Verlag; 2000.[11] Almdal TP, Skinhoej P. Spontaneous bacterial peritonitis in cirrhosis.

Scand J Gastroenterol 1987;22:295300.[12] Jarcuska P, Veseliny E, Orolin M, Takacova V, Hancova M.

Spontaneous bacterial peritonitis in the patients with liver cirrhosis.Klin Mikrobiol Infekc Lek 2004;10:26570.

[13] Andreu M, Sola R, Sitges-Serra A, Alia C, Gallen M, Vila MC, et al.Risk factors for spontaneous bacterial peritonitis in cirrhotic patientswith ascites. Gastroenterology 1993;104:11338.

[14] RunyonBA.Ascites. In: Schiff L, Schiff ER, editors.Diseases of theLiver.7th edition. Philadelphia: JB Lippincott Company; 1993. p. 9901015.

[15] Filik L, Unal S. Clinical and laboratory features of spontaneousbacterial peritonitis. East Afr Med J 2004;81:4749.

47S. Wallerstedt et al. / European Journal of Internal Medicine 18 (2007) 4447recognized syndrome. Ann Intern Med 1964;60:56880.

Abdominal tenderness in ascites patients indicates spontaneous bacterial peritonitisIntroductionMaterials and methodsResultsDiscussionLearning pointsAcknowledgementsReferences