ICU-acquired candidemia within selective digestive decontamination studies: a meta-analysis.

Abstract

To estimate the direct and indirect (contextual) effects of the factorized constituents of selective digestive decontamination and selective oropharyngeal decontamination (SDD/SOD), being topical antibiotic (TA) and protocolized antifungal prophylaxis (PAFP), on ICU-acquired candidemia. A broad range of ICU candidemia incidence studies were sourced to serve as points of reference. The candidemia incidence was extracted from component (control and intervention) groups decanted from studies of various designs (concurrent or non-concurrent) and whether investigating SDD/SOD versus non-TA methods of ICU infection prevention. The candidemia incidences were summarized in regression models using generalized estimating equation (GEE) methods. Groups derived from observational studies (no prevention method under study) provided an overarching external benchmark candidemia incidence for calibration. Within studies investigating SDD/SOD, the mean (and 95% confidence interval) candidemia incidence among concurrent component groups (40 control; 2.4%; 1.7-3.2% and 43 intervention groups; 2.4%; 1.6-3.1%), but not non-concurrent control groups (11 groups; 1.6%; 0.1-2.7%), is higher than that of the benchmark candidemia incidence derived from 54 observational groups (1.5%; 1.2-1.9%). The TA constituent within SDD/SOD has significant direct and indirect (contextual) effects in GEE models even after adjusting for the publication year and the group-wide presence of either candidemia risk factors or PAFP use. The TA constituent of SDD/SOD is associated with a contextual effect on candidemia incidence which is similar in magnitude to that of the conventional candidemia risk factors and against which PAFP partially attenuates. This increase is inapparent within individual SDD/SOD studies examined in isolation.