A Scenario-Based Survey of Clinician Recommendations for Follow-Up Blood Cultures in Patients With Varying Risk of Persistent Gram-Negative Bacteremia

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Background: The utility of follow-up blood cultures (FUBCs) in patients with gram-negative bacteremia (GNB) is unsettled. In practice, the use of FUBCs is variable. Understanding practice variation and drivers of FUBC ordering may clarify how clinicians identify patients at high risk for persistent GNB. Methods: We conducted a global survey (21 February 2024 to 9 April 2024) of healthcare workers using 10 hypothetical clinical scenarios to assess FUBC decision-making in GNB. Scenarios varied clinical factors and risk of persistence using the published AIMS scoring tool (Antibiotics, Infection source, Medical conditions, Serratia). The anonymous web-based survey was distributed via listservs, email, and social media. Interrespondent agreement was assessed using Fleiss’ kappa (κ). Mixed-effect logistic regression identified factors associated with FUBC recommendations. Results: Among 864 clinicians from 50 countries, 691 (80%) provided recommendations across 10 scenarios (6910 recommendations). Most were physicians (86.5%) and infectious disease specialists (79.7%). Follow-up blood cultures were frequently recommended, with >50% of clinicians endorsing FUBCs in 6 of 10 scenarios. Overall agreement was fair (κ = 0.26; 95% confidence interval, .10–.43). In adjusted analyses, factors associated with FUBC recommendations included female respondent sex (adjusted odds ratio [aOR] 1.40; P = .042), higher AIMS score (aOR 1.78 per point; P < .0001), older patient age (aOR 2.90; P < .0001), inadequate source control (aOR 4.77; P < .0001), and clinical instability (aOR 114.99; P < .0001). Conclusions: In this survey study, clinicians frequently recommended FUBCs, particularly in clinical vignettes with higher AIMS scores, inadequate source control, and clinical instability. Only fair agreement across risk-stratified scenarios suggests substantial practice variation and supports the need for prospective studies to understand when and in whom FUBCs should be used.
Funding Information
  • National Institutes of Health (K23 HL159275)
  • National Institutes of Health (K08 AI171183)
  • National Institutes of Health (R01 AI165671)
  • National Institutes of Health (P30AG028716)

This publication has 31 references indexed in Scilit: