There are two problems with blood cultures. The first question is with regard to the likelihood you’ll get a true positive result. That question is covered by this JAMA Rational Clinical Examination.
The second question regards whether the true positive result is clinically meaningful. This retrospective review of 639 cellulitis patients – 325 without medical comorbidities and 314 with – evaluated for changes in therapy as a result of positive cultures. 46 cultures returned positive – with half being judged due to contaminants. Of the 23 true positives, 5 resulted in a change of antibiotic therapy – only 2 of which expanded the initial antibiotic choice to include coverage for a new pathogen. Both changes in therapy occurred in the immunosuppressed group.
Yet another example of the incredibly low yield of an expensive test. We’re clearly simply asking a question for which we already have the answer.
“Blood culture results do not affect treatment in complicated cellulitis”
www.ncbi.nlm.nih.gov/pubmed/23588078
Hi Ryan.
I'm not sure a change in antibiotic group is a relevant endpoint. Blood cultures are primarily not used to direct antimicrobial therapy in cellulitis but rather to guide further clinical decisions. A staphyloccocus bacteremia has the potential for more complications than a simple cellulitis and as such requires extra investigations and usually longer treatment. These endpoints were not assessed in the study.
You are correct the endpoint of change in antibiotic therapy is rather narrow – probably chosen because it's a relatively easy endpoint to measure and abstract retrospectively. If there are, as you say, additional investigations to be performed by knowing precisely what organism caused, for example, a soft-tissue infection, that would be interesting to track. However, blood cultures are still staggeringly overused in clinical scenarios where they are of greater harm or cost than benefit.