Another lovely JAMA Rational Clinical Examination article relevant to the Emergency Department – this time regarding the utility of blood cultures. Blood cultures are frequently requested for febrile inpatients, however, the incidence of false positive ranges between 2.5% and 8%. This leads, unfortunately, to additional patient harms from additional treatment or observation.
This article is a systematic review of several studies gathering clinical features of patients for whom blood cultures were requested, as well as the clinical outcomes of the cultures, in an attempt to identify features predictive of positive or negative cultures. They also examine a couple validated clinical decision instruments to determine their potential utility in stratifying the appropriateness of cultures.
Essentially, based on a few pieces of decent evidence and a few pieces of poor evidence, the authors determine a few general categories of infectious etiology with varying pretest probability for bacteremia. These are:
• Cellulitis, community-acquired pneumonia, community-acquired fever: low (<14%) probability
• Pyelonephritis: mid (19-25%)
• Severe sepsis, septic shock, bacterial meningitis: high (38-69%)
In general, however, no individual clinical feature had a positive or negative likelihood ratio of sufficient magnitude to guide testing. Combinations of clinical features – such as patients with SIRS – were capable of excellent sensitivity & negative likelihood ratios, but only had specificities of 0.27 to 0.47.
However, the more important clinical aspect of blood cultures and bacteremia is not addressed in this article, which is how frequently the true positives even change clinical management.