De Smet et al reheat the data from their 2009 NEJM study, in which they cluster-randomized and crossed-over 5,939 patients staying >48hrs in 13 Dutch ICUs to receive either 1) standard care; 2) selective oropharyngeal decontamination (topical amphotericin B, colistin, and tobramycin in the oropharynx), or 3) selective digestive tract decontamination (the same drugs in the mouth, stomach, and also IV cefotaxime for 4 days). Drugs were open-label, randomized by a blinded pharmacist. Back then, they reported a ~3% absolute mortality reduction in the treated groups.