Background: following a stroke, morbidity and mortality is high, with aspiration pneumonia being a common complication.

Objectives: to determine the levels of colonisation by and isolation of aerobic Gram-negative bacteria (AGNB) in acute stroke patients and determine the effect of selective decontamination of the digestive tract (SDD) on oral flora and whether it reduces both morbidity and mortality after an acute stroke.

Design: a prospective, randomised, placebo-controlled double blind trial.

Setting: acute stroke assessment units of three hospitals in the northwest of England.

Subjects: 203 patients admitted to hospital following a first acute stroke.

Methods: participants were randomised to SDD oral gel or placebo. Swallow was assessed on admission to hospital, and oral swabs were obtained thrice weekly. Demographic and clinical data were recorded.

Results: 203 patients (106 males and 97 females) participated, of whom 20 died during their hospitalisation, 19 withdrew and full follow-up was obtained for the remaining 164. A total of 122 AGNB were isolated in 105 samples from 48 patients. Abnormal swallow on admission was found in 58 patients (29%). A total of 34 patients carried a single gram-negative micro-organism that was present on one or more occasions. More than one AGNB was carried in 14 patients, and organisms were significantly more likely to be isolated from the placebo group than the active group during weeks 2 and 3 of treatment (P = 0.034, chi-squared). Seven patients in the placebo group and one in the treatment group developed pneumonia (P = 0.029, Fisher's exact test).

Conclusions: high carriage of and colonisation by AGNB was found within this study, which was reduced by the addition of SDD. Although SDD reduced the presence of both organisms and documented episodes of pneumonia, mortality remained unchanged.