A meta-analysis can be useful when looking at a topic that has been studied by several different groups of investigators. The pooling of data from different published papers can sometimes bolster a conclusion about the effectiveness of a treatment. However, a meta-analysis is only as good as the studies it includes, and the biases of those performing the meta-analysis can color the results.Last month, a meta-analysis concerning antibiotics vs. surgery for the treatment of uncomplicated acute appendicitis by investigators from Nottingham University Hospitals was published in the World Journal of Surgery. The authors concluded that “antibiotic therapy represents a safe, efficacious and viable treatment option for the treatment of uncomplicated acute appendicitis.” I disagree. Five randomized trials involving 1430 subjects were included in the meta-analysis. After one year of follow-up, the efficacy of treatment for those receiving antibiotics was 62.2% compared with those undergoing appendectomy whose treatment efficacy was 88%.
Depending on the inclusion or exclusion of a particularly weak study there was said to be a 39-52% risk reduction for complications in the antibiotic group. This meta-analysis has so many problems that it is hard to know where to start. Eriksson] and included only 40 patients. The method of randomization was not described. The 20 appendectomy patients all presumably underwent open surgery which is irrelevant in 2016 when more than 80% of appendectomies in the US are done laparoscopically. After a mean follow-up of 17 months, 7/20 (35%) of those who were treated initially with antibiotics required appendectomy for recurrent appendicitis.
Another Swedish study [Hansson 2009] comprised 202 patients who were given antibiotics and 167 patients who had appendectomies. The randomization scheme was by odd or even date of birth, a notoriously poor method. Since the investigators could easily know what group the patients were to be “randomized” to, manipulation of enrollment by the treating physicians may have confounded the results. The diagnosis of appendicitis was made on clinical grounds for more than 70% of the patients. It is possible that some patients who were treated with antibiotics never had appendicitis at all.
The results of this study are difficult to interpret because patients were allowed to cross over from one group to the other based on their preference, the clinical judgment of the surgeon, or for “unspecified clinical judgment.” Since 47.5% of those assigned to antibiotic treatment crossed over and had surgery, the efficacy of antibiotics is difficult to ascertain. This paper claimed that major complications were three times higher in patients who initially had appendectomies. Two of these major complications involved patients who underwent hemicolectomies because a malignancy of the appendix or colon was found during the appendectomy. I would hardly call those complications.
A third Swedish study [Styrud 2006] involved only male patients; 124 underwent appendectomy and 128 were given antibiotics only. The diagnosis of acute appendicitis was based on clinical findings and laboratory studies. Despite being published in 2006, only 8 patients (6%) underwent laparoscopic appendectomy. This was probably because the study was conducted from 1996 to 1999. One wonders why it took seven years to be published. 15 of the 128 patients in the antibiotic group had appendectomies within 24 hours because they failed to improve on antibiotics, and seven of them had perforations. Of those treated with antibiotics successfully during the initial hospitalization, another 16 required appendectomy during the one year of follow-up.
The total failure rate for antibiotics was 24%. Hospital stays, sick leave days, and time off from work were similar in both groups. It is not clear whether the second hospitalization for the 16 patients in the antibiotic group requiring subsequent appendectomy was counted in the hospital stay and time off from work data. The fourth study was from France [Vons 2011]. Antibiotics were given to 120 patients compared to 119 who underwent appendectomy. Diagnosis of uncomplicated acute appendicitis was made by CT scan in all patients. Of note was that 18% of those who underwent appendectomy as the initial treatment had complicated appendicitis found at surgery despite their CT scan interpretations.
Combining those who failed antibiotic therapy during the index hospitalization and those who required appendectomy later within the year of follow-up, the failure rate for antibiotics was 36.6%. The fifth study was the recent randomized trial from Finland which I dissected soon after it was published [here and here]. Briefly, 27% of patients treated with antibiotics required appendectomy within one year, only 5.5% of appendectomies were done laparoscopically, the antibiotic used was not a first line drug in the US, and the length of follow-up was only one year.
All of the randomized trials in the meta-analysis above have serious methodological flaws. Comparing antibiotics to open appendectomy is not valid in the 21st century. Their follow-ups were too short, and they had failure rates that were, in my opinion, unacceptably high. I remain thoroughly unconvinced about the efficacy of antibiotics for the treatment of appendicitis.