Risk of dementia after bloodstream infection—a nationwide propensity score matched cohort study

BACKGROUND: Dementia is a global public health threat. Current treatments have limited efficacy, making identification of modifiable risk factors of paramount importance. Observational studies link severe infections to increased dementia risk but often suffer confounding. Here, we address this by determining dementia risk following bloodstream infection (BSI) in a nationwide study. METHODS: We conducted a cohort study within the Secure Anonymised Information Linkage Databank, containing anonymised population-scale electronic health record data for the population of Wales, UK. Patients with microbiologically confirmed BSI were propensity score matched 1:1 to controls (n = 2.5 million without dementia at baseline). We created two comparative models to estimate the true effect of BSI on dementia risk: modelling hospitalisation with sterile inflammation, uncomplicated total knee replacement (TKR) replaced BSI as the exposure; assessing residual confounding, lung cancer replaced dementia as outcome. RESULTS: We included 26 792 people with BSI and 26 792 matched controls, all without dementia at cohort entry. BSI was associated with an increased cumulative hazard of dementia corresponding to 160 (128-182) additional cases per 1000 person-years 10 years after exposure. TKR was not associated with an increased risk of dementia. BSI was potentially associated with a small excess hazard of lung cancer [12.4 (5.4-19.5) additional cases per 1000 person-years 10 years after exposure]. CONCLUSIONS: BSI was associated with incident dementia in excess of that expected by hospitalisation or residual confounding. These findings suggest that treatment of BSI and other severe infections merits further investigation as a potentially modifiable risk factor for dementia.