Predictors of post-stroke fever and infections: a systematic review and meta-analysis

BACKGROUND: Fever after stroke is common, and often caused by infections. In the current study, we aimed to test the hypothesis that pneumonia, urinary tract infection and all-cause fever (thought to include at least some proportion of endogenous fever) have different predicting factors, since they differ regarding etiology. METHODS: PubMed was searched systematically for articles describing predictors for post-stroke pneumonia, urinary tract infection and all-cause fever. A total of 5294 articles were manually assessed; first by title, then by abstract and finally by full text. Data was extracted from each study, and for variables reported in 3 or more articles, a meta-analysis was performed using a random effects model. RESULTS: Fifty-nine articles met the inclusion criteria. It was found that post-stroke pneumonia is predicted by age OR 1.07 (1.04–1.11), male sex OR 1.42 (1.17–1.74), National Institutes of Health Stroke Scale (NIHSS) OR 1.07 (1.05–1.09), dysphagia OR 3.53 (2.69–4.64), nasogastric tube OR 5.29 (3.01–9.32), diabetes OR 1.15 (1.08–1.23), mechanical ventilation OR 4.65 (2.50–8.65), smoking OR 1.16 (1.08–1.26), Chronic Obstructive Pulmonary Disease (COPD) OR 4.48 (1.82–11.00) and atrial fibrillation OR 1.37 (1.22–1.55). An opposite relation to sex may exist for UTI, which seems to be more common in women. CONCLUSIONS: The lack of studies simultaneously studying a wide range of predictors for UTI or all-cause fever calls for future research in this area. The importance of new research would be to improve our understanding of fever complications to facilitate greater vigilance, monitoring, prevention, diagnosis and treatment.