There is a lack of evidence from relevant randomized controlled trials to support the effectiveness of oral antibiotics in the treatment of acute cellulitis. Therefore the choice of antibiotic used should be based on first principles; that is selecting an antibiotic with appropriate pharmacokinetic and pharmacodynamic properties to treat the likely causative pathogen.

• CKS did not identify any systematic reviews that investigated the use of antibiotics or other drugs for cellulitis. However, one UK-focussed narrative review [Jones, 2002], and a BMJ Clinical Evidence review [Morris, 2008], have described the available published trials on the treatment of cellulitis.
• No trials were identified that compared the effectiveness of antibiotics with placebo.
• Both reviews identified several comparative head-to-head trials.
• The purpose of most of these trials was to show equivalence of newer antibiotics with older antibiotics, for licensing purposes [Jones, 2002]. However, the lack of placebo-controlled trials and the identification of a suitable comparator of known efficacy makes interpretation of these trials difficult.
• None of the trials described in the BMJ Clinical Evidence review could be generalized to people being treated in primary care (for instance, most investigated the use of at least one intravenous drug). Also, the heterogeneous nature of the trials (e.g. different populations) combined with poor or vague reporting of outcomes ruled out the use of meta-analyses [Morris, 2008].
• Despite the limitations of the available trial data, the microbiological cure rate was estimated to be more than 85% when the causative pathogen was known. However, this rate was lower when intent-to-treat analysis of empirical treatment was measured [Jones, 2002]. The BMJ Clinical Evidence review estimated the rate of cure as being 50–100%, but commented that it is not known which regimen is most successful [Morris, 2008].
• In conclusion, in the absence of good-quality, robust, and generalizable trial data, and without sensitivity data to guide antibiotic selection, the choice of antibiotic should be made according to the known pharmacological qualities of the antibiotic in response to the likely causative pathogen. In most circumstances, this will be flucloxacillin to treat staphylococcal or streptococcal species.