In sepsis, antibiotics reduced yield of blood cultures by almost 40%
by Molly Walker, Associate Editor, MedPage Today
If you want an accurate reading from sepsis patients' blood cultures, don't start antibiotics until you've drawn the blood samples, a new study suggested.
Among patients with severe sepsis, blood cultures taken prior to antibiotic therapy were positive for one or more microbial pathogens in 31.4% of patients compared with 19.4% of patients for cultures taken after empirical antibiotic treatment had already started, reported Matthew Cheng, MD, of Brigham and Women's Hospital in Boston, and colleagues, in Annals of Internal Medicine.
Co-author Murtaza Akhter, MD, of the University of Arizona College of Medicine in Phoenix, said in a statement that this is a "constant debate in the medical field," specifically between emergency medicine physicians and internal medicine physicians.
"Emergency medicine physicians want to administer antibiotics as soon as possible because it prevents mortality, whereas internal medicine physicians want two sets of blood cultures before antibiotics are administered so they can more reliably diagnose the organism," Akhter said.
"It was unknown how quickly blood cultures lose their ability to grow organisms after antibiotic administration. Now we know that even if blood cultures are performed very rapidly after antibiotics, they lose a significant amount of their diagnostic ability," he added.
Cheng and colleagues cited the Surviving Sepsis Campaign guidelines as stating that blood cultures should be drawn prior to starting antimicrobial therapy, with 45-minute delays considered acceptable. However, they characterized prompt initiation of antimicrobial therapy as "a critical determinant of survival."
In the new study, called FABLED (eFfect of Antimicrobial administration on BLood culture positivity in patients with severe manifestations of sepsis in the Emergency Department) study, the researchers examined adult patients presenting to seven emergency departments in the U.S. and Canada. Patients had two blood cultures drawn prior to starting antibiotic therapy and additional sets drawn within 2 hours of administration of antibiotics (i.e., patients served as their own controls).
Severe manifestations of sepsis included systolic blood pressure <90 mm Hg or a serum lactate level of ≥4 mmol/L, the authors said.
They examined data from 325 patients from November 2013 to September 2018. Patients' mean age was 66 and over 60% were men. Empirical treatment for about 60% of patients included a β-lactam antibiotic, while around a quarter included a third-generation cephalosporin. While baseline characteristics were similar, the authors noted that a higher portion of patients with positive pre-antimicrobial blood cultures had respiratory failure (19.6% vs 8.9%, respectively). The most commonly isolated pathogens in pre-antimicrobial blood cultures were Escherichia coli (22.5%), followed by Staphylococcus aureus (15.7%).
There was an absolute difference of 12% (95% CI 5.4%-18.6%, P<0.001), and the authors noted that sensitivity of post-antimicrobial blood cultures was 52.9% (95% CI 42.8%-62.9%).
In an accompanying editorial, Jacqueline Geer, MD, and Mark Siegel, MD, of Yale School of Medicine in New Haven, Connecticut, acknowledged the diagnostic value of blood cultures prior to administration of antibiotics, but cautioned against "unintentional treatment delays," given the higher risk for death associated with delaying treatment.
"Future studies should explore methods to ensure that blood is drawn expeditiously, without imposing treatment delays," the editorialists wrote. "In addition, given that up to 50% of patients with sepsis are ultimately found to be 'culture negative,' it would be useful to create a practical diagnostic tool to identify patients who can be treated immediately without waiting to obtain blood cultures if they are unlikely to provide useful diagnostic information."
Limitations to the data include that a proportion of patients had repeated blood cultures that were drawn outside the window specified in the study protocol, though the authors noted that results were consistent within this per protocol population. In addition, 31.4% of patients were bacteremic, which the authors characterize as "slightly lower than expected," though another large randomized trial of comparable patients had similar results.
Source: Annals of Internal Medicine