BACKGROUND

Post-stroke pneumonia is associated with increased mortality and poor functional outcomes. This study assessed the effectiveness of antibiotic prophylaxis for reducing pneumonia in patients with dysphagia after acute stroke.

METHODS

We did a prospective, multicentre, cluster-randomised, open-label controlled trial with masked endpoint assessment of patients older than 18 years with dysphagia after new stroke recruited from 48 stroke units in the UK, accredited and included in the UK National Stroke Audit. We excluded patients with contraindications to antibiotics, pre-existing dysphagia, or known infections, or who were not expected to survive beyond 14 days. We randomly assigned the units (1:1) by computer to give either prophylactic antibiotics for 7 days plus standard stroke unit care or standard stroke unit care only to patients clustered in the units within 48 h of stroke onset. We did the randomisation with minimisation to stratify for number of admissions and access to specialist care. Patient and staff who did the assessments and analyses were masked to stroke unit allocation. The primary outcome was post-stroke pneumonia in the first 14 days, assessed with both a criteria-based, hierarchical algorithm and by physician diagnosis in the intention-to-treat population. Safety was also analysed by intention to treat. This trial is closed to new participants and is registered with isrctn.com, number ISRCTN37118456.

FINDINGS

Between April 21, 2008, and May 17, 2014, we randomly assigned 48 stroke units (and 1224 patients clustered within the units) to the two treatment groups: 24 to antibiotics and 24 to standard care alone (control). 11 units and seven patients withdrew after randomisation before 14 days, leaving 1217 patients in 37 units for the intention-to-treat analysis (615 patients in the antibiotics group, 602 in control). Prophylactic antibiotics did not affect the incidence of algorithm-defined post-stroke pneumonia (71 [13%] of 564 patients in antibiotics group vs 52 [10%] of 524 in control group; marginal adjusted odds ratio [OR] 1·21 [95% CI 0·71-2·08], p=0·489, intraclass correlation coefficient [ICC] 0·06 [95% CI 0·02-0·17]. Algorithm-defined post-stroke pneumonia could not be established in 129 (10%) patients because of missing data. Additionally, we noted no differences in physician-diagnosed post-stroke pneumonia between groups (101 [16%] of 615 patients vs 91 [15%] of 602, adjusted OR 1·01 [95% CI 0·61-1·68], p=0·957, ICC 0·08 [95% CI 0·03-0·21]). The most common adverse events were infections unrelated to post-stroke pneumonia (mainly urinary tract infections), which were less frequent in the antibiotics group (22 [4%] of 615 vs 45 [7%] of 602; OR 0·55 [0·32-0·92], p=0·02). Diarrhoea positive for Clostridium difficile occurred in two patients (<1%) in the antibiotics group and four (<1%) in the control group, and meticillin-resistant Staphylococcus aureus colonisation occurred in 11 patients (2%) in the antibiotics group and 14 (2%) in the control group.

INTERPRETATION

Antibiotic prophylaxis cannot be recommended for prevention of post-stroke pneumonia in patients with dysphagia after stroke managed in stroke units.

FUNDING

UK National Institute for Health Research.