|Home | About | Journals | Submit | Contact Us | Français|
Contributors: KSW formulated the problem, designed the protocol, searched the literature, participated in the data collection and analysis, and wrote the paper. MB collaborated in data collection and analysis and commented on the final draft. MP helped in the data collection and provided details from the patient. LL coordinated and supported the protocol design, data analysis, and interpretation and helped in writing the paper. KSW and LL are the guarantors.
A 55 year old woman, previously diagnosed with cirrhosis secondary to chronic hepatitis C infection, was admitted to our department with fever. She seemed well and had no focal symptoms or signs of infection. As ascites was present, she had paracentesis. This yielded a Gram negative clear fluid with a polymorphonuclear count of 700 cells/mm3. We thought that secondary peritonitis was unlikely and diagnosed spontaneous bacterial peritonitis. She had had no previous episodes or prophylactic antibiotic treatment. Empirical treatment with cefotaxime (2 g every 8 hours) was started.
When admitting the patient, the junior doctor had access to two main databases: the Cochrane Library, which contained no relevant information,1 and UpToDate, which recommended intravenous cefotaxime or oral ofloxacin for patients with uncomplicated spontaneous bacterial peritonitis.2
On the morning after her admission, there was a lively discussion at the departmental meeting. The main question was whether the patient could have started taking oral ofloxacin, given her excellent clinical condition. Other questions were raised about the strength of the evidence supporting the standard treatment with cefotaxime and the ideal dose and duration of treatment. We therefore decided to do a systematic review of the literature on antibiotic treatment for spontaneous bacterial peritonitis.
We searched Medline (1966-December 2000) and the Cochrane Library (Issue 4, 2000) for randomised trials comparing different antibiotics for spontaneous bacterial peritonitis (table (table1).1). Additionally, we inspected the references of all identified studies and a consensus document organised by the International Ascitic Club3 and contacted authors by email to ask for complementary information.
Our search strategy identified 18 trials. We excluded five studies because of inadequate concealment of allocation4–8; two because less than 10% of the patients had spontaneous bacterial peritonitis diagnosed9,10; one because it lacked an antibiotic comparison11; and another because of missing relevant information.12 We included nine randomised trials in the review (table (table22).13–21
We searched for the following outcomes in the nine included studies: mortality, antibiotic effectiveness in current episode, and life threatening adverse events. None of the trials compared similar experimental and control treatments. We therefore analysed the results of each trial separately.
No reliable evidence exists to place cefotaxime as the treatment of choice for spontaneous bacterial peritonitis, although this has been suggested by many authors.3,17,22 In one trial in which 72 patients were randomised to intravenous cefotaxime or ampicillin plus tobramycin, cefotaxime had no significant benefit on mortality or fatal adverse events, although it did increase resolution of spontaneous bacterial peritonitis (table (table22).13
What is the optimal dose and duration of cefotaxime?—Only two relevant trials were identified.19,20 The results indicate that cefotaxime 4 g/day may be as effective as cefotaxime 8 g/day19 in terms of mortality and resolution of symptoms. Treatment for 10 days was no more effective than treatment for five days.20
Are oral or intravenous antibiotics more effective?—Four trials evaluated the effects of oral and intravenous antibiotics on mortality and resolution of symptoms,14,16,17,21 but no definite conclusions could be drawn. No significant differences were found in trials of oral ofloxacin versus cefotaxime16 and of oral versus intravenous ciprofloxacin.21 However, the trials were small and should be considered inconclusive. Finally, it has been suggested that patients with moderate symptoms and a positive response to a short course of intravenous antibiotics could benefit from oral treatment with quinolones.3,21,22 The only trial investigating quinolones found no significant effect on mortality or resolution of symptoms.21
On the third day, cultures of the blood and ascitic fluid were negative. The patient was afebrile and doing well, and we offered to discharge her on oral ofloxacin. She was reluctant to switch to oral therapy and leave the hospital. Although she understood that one trial showed no benefit for cefotaxime over oral ofloxacin,16 she was alarmed that we could not rule out that the risk of a death with ofloxacin might be twice as high as with cefotaxime. We tried to balance that against the risk of a severe nosocomial infection (about 2% a day).23 The patient was unconvinced. She found the monitoring and intravenous therapy reassuring and opted for five days' treatment with cefotaxime and then oral ofloxacin. She was discharged on the sixth day after an uneventful stay.
The patient was treated in real time according to the available information.2 A later discussion led to a systematic review. Clearly, clinicians cannot do a systematic review for every question raised in their daily experience. Nevertheless, reviewing the literature in depth enables clinicians to comprehend where practices stem from, how these are founded on evidence, and sometimes, as in this case, how frail the evidence is to support routine practices.
We found no convincing evidence concerning efficacy of antibiotics for spontaneous bacterial peritonitis, and we identified several gaps that warrant future research. For example, we found no conclusive evidence to support cefotaxime as the treatment of choice3 or to recommend quinolones for patients with uncomplicated spontaneous bacterial peritonitis.16 Until large, well conducted trials have been published, antibiotic treatment for spontaneous bacterial peritonitis has to be based on clinical experience.
Randomised trials of spontaneous bacterial peritonitis need to include several hundred patients in order to have sufficient power. Recruitment of sufficient patients should be possible given the relatively high prevalence of cirrhosis complicated by ascites and the incidence of spontaneous bacterial peritonitis in these patients.3,23 Future trials should also examine long term outcomes, recurrence rate, long term survival, and the development of resistant pathogens, particularly with quinolones.
This paper was based on the results of a systematic review recently published in the Cochrane Library.24
Funding: The review was partly supported by a European Union 5th framework grant (TREAT, IST 1999-11459) and a Danish Medical Research Council grant on getting research into practice.
Competing interests: None declared.