Presentations

Abstract

Introduction: Blood cultures (BCx) are a frequently ordered laboratory test to evaluate for bacteremia. After an initial positive BCx, follow-up cultures are often ordered to assess for clearance of infection. While indications for follow-up cultures, timing of follow-up culture collection, predictors of follow-up culture positivity, and the association between obtaining a follow-up culture and mortality have been described, the appropriate number of follow-up cultures to collect is unknown. Given the recent national shortage of BCx bottles, it is important to determine optimal use of this resource for patient care. There is a recommendation to obtain two rather than one culture when BCx are drawn to optimize identification of possible pathogens. However, when follow-up cultures are ordered, it is unclear whether multiple are always necessary. Collecting excess cultures has disadvantages. Each BCx set consists of an aerobic and anaerobic bottle, each requiring 10-15 mL of blood; thus, two sets would require removing up to 60 mL of blood from the patient. Obtaining BCx is also painful for the patient, requires time for specimen procurement, and consumes laboratory resources. While ordering only a single follow-up set could reduce clinical waste, the risk of failing to detect ongoing bacteremia is unclear. The balance between practicing diagnostic stewardship and maximizing patient safety by reliably identifying continuing bacteremia warrants investigation. We sought to assess the frequency at which the second follow-up set is positive after the first follow-up set is negative to determine how many follow-up sets are needed after an initial positive BCx.

Methods: We conducted a retrospective descriptive study of all BCx submitted to Sentara Health microbiology laboratory from 1/1/18-11/1/23. The study included all patients ≥18 years old within the Sentara Health system who had at least two follow-up BCx drawn 24-72 hours after an initial positive culture. Cultures obtained within two hours of each other were counted as one set. Data collected from the electronic medical record system (Epic) included patient demographics, length of stay, source of infection, BCx dates/times and results, relevant workup performed, and relevant antibiotics received. Patients were divided into four groups based on BCx positivity, with a focus on the cohort with an initial positive culture, a negative follow-up set, and then a positive follow-up set (PNP group). Different strains of an organism were considered to be different organisms.

Results: There were 28,875 patients with an initial positive BCx, and of these, 2,636 had at least two follow-up cultures drawn in the selected timeframe. Within this group, 585 (22.2%) had two positive follow-up sets, 1500 (56.9%) had two negative, 431 (16.4%) had a positive followed by a negative, and 120 (4.6%) had a negative followed by a positive (PNP). Of the PNP cohort, 71 (2.7%) grew the same organism in the initial and second follow-up cultures, while 49 (1.9%) did not. In the same-organism subset, the most commonly identified classes of organisms were coagulase-negative staphylococci (n=21; 0.8%), gram-negative bacteria (n=17; 0.6%), methicillin-sensitive Staphylococcus aureus (n=13; 0.5%), and methicillin-resistant S. aureus (n=7; 0.3%). The most frequently isolated organisms in this subset were SA (n=20, 0.8%), Staphylococcus epidermidis (n=16, 0.6%), and Escherichia coli (n=11, 0.4%). In the different-organism subgroup, 35 (1.3%) of the second follow-up sets had suspected contamination, though true bacteremia from skin/soft tissue (n=4; 0.2%), central line (n=4; 0.2%), unknown (n=3; 0.1%), and other sources was also observed, often due to S. aureus (n=4; 0.2%), E. coli (n=2; 0.1%), and Candida (n=2; 0.1%).

Conclusion: The number of patients who had ongoing bacteremia that would have been missed with one follow-up BCx was small. Having a positive initial BCx followed by a negative follow-up set and then a positive set growing the same organism has been described for S. aureus, but we found this "skip phenomenon" occurred with gram-negative organisms as well. There were also second follow-up BCx that were positive for a new organism, usually a contaminant. Further data are needed to determine when two follow-up sets should be obtained and when a single one may be sufficient.