: Publication 18248

Publication 18248

Title:	The role of frailty and comorbidities in severe infections and the risk of dementia: a prospective, multicohort, observational study
Journal:	The Lancet Healthy Longevity
Published:	1 Mar 2026
Pubmed:	https://pubmed.ncbi.nlm.nih.gov/41969011/
DOI:	https://doi.org/10.1016/j.lanhl.2026.100826

Abstract

BACKGROUND: Severe infections have been linked to an increased risk of dementia, but whether this association reflects pre-existing conditions that predispose individuals to infections and dementia remains unclear. We assessed whether the association between severe infections and dementia risk is explained by pre-existing frailty, comorbidities, or other age-related physical diseases.

METHODS: This prospective, multicohort, observational study included participants without severe infections or dementia (between Dec 19, 2006, and Oct 1, 2010, from the UK Biobank study and between Oct 8, 2002, and Nov 30, 2004, from the Whitehall II prospective cohort study). The primary analysis was conducted in the UK Biobank cohort. Frailty was assessed at baseline using the widely used phenotypic Fried Frailty Scale and comorbidities were assessed using the Charlson Comorbidity Index and 78 age-related diseases. Incident severe (hospital-treated) infections were identified from linked hospital discharge diagnoses. Follow-up included a 5-year exposure period from baseline, a 2-year washout period; and a final period extending to the end of follow-up. Incident dementia occurring after the exposure and washout periods was ascertained from hospital records and death certificates until 2022 (Oct 31 in England, Sept 30 in Scotland, and May 31 in Wales). Participants with a record of dementia during the exposure period or subsequent washout period were excluded. We repeated the main analyses in the Whitehall II study and conducted several supplementary analyses with follow-up until March 1, 2023.

FINDINGS: We included 449 223 participants without severe infection or dementia from the UK Biobank study and 6106 from the Whitehall II study. Most participants were White (424 405 [94·5%] in UK Biobank and 5616 [92·0%] in the Whitehall II study). In UK Biobank, the mean age of participants was 56·5 years (SD 8·1), of whom 245 139 (54·6%) were female and 204 084 (45·4%) were male. Of 435 957 participants with dementia follow-up, 23 860 (5·5%) had a record of severe infection during the 5-year exposure period. After a 2-year washout period, the remaining follow-up for dementia occurred for a mean of 6·5 years (SD 1·2). 6756 participants developed incident dementia, corresponding to an incidence of 2·37 (95% CI 2·31-2·43) per 1000 person-years. Adjustment for frailty resulted in no statistically significant attenuation of the infection and dementia association (adjusted hazard ratio [HR] 1·54 [95% CI 1·43-1·67] for presence vs absence of severe infections before adjustment for frailty and 1·49 [1·37-1·61] after adjustment). The association was also evident within a subgroup of non-frail participants (1·34 [1·18-1·53]). Pre-frail and frail participants had a higher corresponding adjusted HR (1·62 [1·47-1·79]). In the Whitehall II study, the mean age of participants was 60·6 years (SD 6·0), of whom 1772 (29·0%) of 6106 were female and 4334 (71·0%) were male. Of 5824 participants available for dementia follow-up, 237 (4·1%) had a record of infection during the 5-year exposure period. After a 2-year washout period, the remaining mean follow-up for dementia was 11·3 years (SD 2·9). 545 incident cases of dementia were recorded (incidence 8·3 per 1000 person-years). Findings in the Whitehall II study, using an alternative proteomic measure of frailty and cognitive decline as a surrogate outcome, were consistent but less precise.

INTERPRETATION: Frailty, comorbidities, and other age-related diseases were associated with a higher risk of dementia but did not account for the elevated dementia risk observed after severe infections. Interventional research to test whether targeting infections lowers dementia incidence is needed.

FUNDING: Finnish Medical Foundation, Päivikki and Sakari Sohlberg Foundation, Research Council of Finland, Wellcome Trust, UK Medical Research Council, and National Institutes of Health.

Abstract

14 Keywords

8 Authors