127Topical antibiotics used to prevent infection in the ICU: an accidental natural herd exposure experiment

Abstract

\n \n \n Multiple studies of topical antibiotics applied to ICU patients appear to show potent infection prevention effects versus studies of other interventions. However, the effect is less apparent for studies using non-concurrent (NCC) versus concurrent controls (CC) implying the possibility of a herd effect.\n \n \n \n 206 studies of infection prevention among ICU patients, sourced from 15 systematic reviews were stratified into those using topical antibiotics with NCC versus with CC versus studies of other prevention methods. The event rates were summarised using generalized estimating equations and compared to other studies without an intervention (literature benchmark).\n \n \n \n The summary effect sizes for pneumonia and mortality prevention derived in the systematic reviews were replicated.\n \n \n \n The mean ICU mortality incidence for topical antibiotic study CC control groups (28.5%; 95% CI, 25.0-32.3; n\u2009=\u200941) is higher versus a literature benchmark (23.7%; 19.2%-28.5%; n\u2009=\u200934), versus NCC control groups (23.5%; 19.3-28.3; n\u2009=\u200914) and versus topical antibiotic intervention groups (24.4%; 22.1 – 26.9; n\u2009=\u200962). In meta-regression models adjusted for group mean age and publication year, CC group membership within a topical antibiotic study remains associated with higher mortality (p\u2009=\u20090.027).\n \n \n \n Within topical antibiotic studies, the CC control group mortality incidences are inexplicably high, whereas the intervention group incidences are paradoxically similar to a literature-derived benchmark.\n \n \n \n An adverse herd effect is apparent for topical antibiotics used to prevent infection among ICU patients.\n