Post-report antibiotic initiation following community non-sterile-site microbiology results: an opportunity for labs to lead stewardship?

Positive culture results from non-sterile sites (NSSs) are poorly predictive of clinical infection. Despite this, these results are often interpreted as an indication for antibiotics, even in patients with limited signs of infection. We sought to quantify the influence of NSS culture results on post-report antibiotic initiation (PRAI) in patients who had not been started on antibiotics pre-report. All community wound/skin swab and sputum cultures were matched to antibiotic dispensing records from February 2017 to July 2022. Prescribing behaviour was assessed pre- and post-report. Sampling without treatment pre-report was termed 'test-and-wait' (TaW). Following TaW, PRAI was identified if antibiotics were started within 5 days post-report. There were 65 480 wound/skin swabs and 8126 sputum samples, with TaW occurring in 21 740 (35.1%) and 4185 (54.4%), respectively. Following a TaW approach PRAI occurred in 43.3% when an organism was reported, versus 10.8% (P < 0.01) for a 'no growth' report for wound/skin swabs. For the same comparison with sputum, PRAI occurred in 47.9% versus 10.8% (P < 0.01). On multivariate analysis reporting an organism remained strongly associated with PRAI. Reporting an organism in those not already on antibiotics was strongly associated with PRAI. We hypothesize that for many patients TaW suggests limited evidence of infection (i.e. insufficient to justify antibiotic treatment at time of sampling), meaning positive NSS results may be driving a considerable volume of potentially unnecessary antibiotic use. Further study on this topic is required, but strategies to reduce PRAI may offer laboratories an opportunity to meaningfully impact antimicrobial stewardship efforts.

Informations de copyright

© The Author(s) 2023. Published by Oxford University Press on behalf of British Society for Antimicrobial Chemotherapy. All rights reserved. For permissions, please e-mail: [e-mail].