Incidence of coagulase-negative staphylococcal bacteremia among ICU patients: decontamination studies as a natural experiment

Incidence of coagulase-negative staphylococcal bacteremia among ICU patients: decontamination studies as a natural experiment:

Abstract

The epidemiology of coagulase-negative staphylococcal (CNS) bacteremia among adult ICU patients remains unclear. Decontamination studies among ICU patients provide a unique opportunity to study the impacts of different diagnostic criteria, exposure to various decontamination interventions, and various other factors, on its incidence over three decades. Decontamination studies among ICU patients reporting CNS bacteremia incidence data were obtained mostly from recent systematic reviews. The CNS bacteremia incidence within component (control and intervention) groups of decontamination studies was benchmarked versus studies without intervention (observational groups). The impacts of antibiotic versus chlorhexidine decontamination interventions, control group concurrency, publication year, and diagnostic criteria were examined in meta-regression models. Among non-intervention (observational) studies which did versus did not specify stringent (≥ 2 positive blood cultures) diagnostic criteria, the mean CNS bacteremia incidence per 100 patients (and 95% CI; n) is 1.3 (0.9–2.0; n = 23) versus 3.6 (1.8–6.9; n = 8), respectively, giving an overall benchmark of 1.8 (1.2–2.4; n = 31). Versus the benchmark incidence, the mean incidence is high among concurrent control (5.7; 3.6–9.1%) and intervention (5.2; 3.6–6.9%), but not non-concurrent control (1.0; 0.4–3.9%) groups of 21 antibiotic studies, nor among eleven component groups of chlorhexidine studies. This high incidence remained apparent (p < 0.01) in meta-regression models adjusting for group wide factors such as diagnostic criteria and publication year. The incidence of CNS bacteremia within both intervention and concurrent (but not non-concurrent) control groups of antibiotic-based decontamination studies are unusually high even accounting for variable diagnostic criteria and other factors.