Severe infections, domain-specific cognitive vulnerability, and future infection risk in older adults

Severe infections have been implicated in dementia risk, but their associations with specific cognitive domains and whether poorer cognition predicts subsequent infection remain unclear. We examined these associations in both temporal directions among older adults.

We analysed data from the English Longitudinal Study of Ageing Harmonised Cognitive Assessment Protocol (ELSA-HCAP), conducted in 2018 and linked to national inpatient records. Pre-HCAP hospital-treated infections were identified from 1997 to 2018; post-HCAP incident infections were ascertained from 2018 to 2024. Cognitive performance was assessed at HCAP using 21 standardized neuropsychological tests summarised into general and four domain-specific scores (executive function, memory, language, and visuospatial ability). Linear regression assessed associations between pre-HCAP hospital-treated infections and cognitive performance at HCAP; Cox models estimated associations between cognition at HCAP and risk of incident hospital-treated infections. All models were adjusted for sociodemographic, lifestyle, and health covariates.

Of 1159 participants aged ≥65 at HCAP (631 [54.1%] female; mean [SD] age, 75.6 [7.2] years), 351 (30.3%) had a hospital-treated infection before HCAP. Prior hospitalisation for any infection was associated with lower general cognition (β = -0.11 SD, 95% CI -0.21 to -0.02) and poorer executive function (β = -0.19, -0.28 to -0.09), with similar patterns across infection types. Lower respiratory tract infections were additionally associated with poorer memory (β = -0.20, -0.36 to -0.04). Cognitive scores were progressively lower among individuals with more frequent or prolonged infection-related hospitalisations, sepsis, or cardiovascular disease. Prospectively, over a mean (SD) 4.8 (1.9) years of follow-up, 271 incident hospital-treated infections occurred. Each 1-SD higher general cognition was associated with a 36% lower risk of any subsequent hospital-treated infection (HR 0.64, 0.53 to 0.78). Poorer performance across cognitive domains was associated with all-cause and bacterial infections, whereas only executive function was associated with viral infections, largely reflecting COVID-19 (HR 0.54, 0.34 to 0.85).

Severe infections were primarily associated with poorer executive function. Conversely, cognitive vulnerability across multiple domains was associated with subsequent infections requiring hospital care. These findings support a reinforcing infection-cognition cycle in later life and cognitively tailored infection-prevention strategies.