Risk factors for dementia and cognitive impairment within 5 years after stroke: a prospective multicentre cohort study

Abstract

Background: Stroke survivors frequently experience subsequent cognitive impairment or dementia. We aimed to identify risk factors for post-stroke dementia (PSD) and cognitive impairment (PSCI) within 5 years after stroke.

Methods: The DEMDAS (German Center for Neurological Diseases (DZNE) mechanisms of dementia after stroke) study is a prospective cohort of stroke patients admitted to six German tertiary stroke centres between May 1, 2011 and January 31, 2019. Eligible dementia-free patients with ischaemic or haemorrhagic stroke underwent baseline examinations and regular clinical, neuropsychological, and neuroimaging follow-ups over 5 years, with the last follow-ups completed in January 2024. PSD was the primary outcome, determined by comprehensive cognitive testing, patient and informant interviews, and review of medical records. The secondary outcomes were early-onset PSD (3-6 months), delayed-onset PSD (>6 months), and PSCI. Associations between baseline risk factors and PSD were assessed using Cox regression models adjusted for age, sex, education, and stroke severity.

Findings: Of 736 patients (245 [33%] female, mean age 68·0 years [SD 11·2], median admission National Institutes of Health Stroke Scale (NIHSS) 3 [IQR 1-5]), 557 (76%) were followed up until death or the end of the study, and 706 (96%) contributed to the PSD analysis. During a median of 5·0 years [IQR 3·3-5·1] of follow-up, 55 new dementia cases were diagnosed (6-month incidence: 3·1% [1·8-4·5], 5-year incidence: 8·8% [6·5-11·1]), of which 21 (38%) were classified as early-onset PSD. The 5-year risk of PSD was associated with older age (HR 1·13 [95% CI 1·08-1·18] per year), higher stroke severity (1·08 [1·03-1·13] per point on NIHSS), lower educational attainment (1·16 [1·05-1·28] per year), acute phase cognitive impairment (5·86 [2·21-15·58]), lower Barthel Index (1·10 [1·05-1·16] per 5 points less), atrial fibrillation (1·91 [1·10-3·30]), metabolic syndrome (MetS, 2·05 [1·15-3·64]), particularly reduced high-density lipoprotein cholesterol (HDL-C, 2·61 [1·50-4·52]) and pre-/diabetes mellitus (2·13 [1·13-4·00]), imaging markers of small vessel disease, and stroke recurrence during follow-up (2·36 [1·16-4·83]). Patients who received acute reperfusion treatment had a 65% lower risk of PSD than those who did not (0·35 [0·16-0·77]). While factors related to the severity of the index stroke were more strongly associated with early-onset PSD, MetS showed a stronger association with delayed-onset PSD. The association between MetS and PSD was independent of stroke recurrence and consistent across age subgroups, with 5-year cumulative incidence ranging from 1·7% (0·0-4·0) in patients ≤65 years without MetS to 24·5% (14·3-33·4) in patients ≥74 years with MetS.

Interpretation: The risk of dementia after stroke is multifactorial, with differing risk profiles for early-onset and delayed-onset PSD. Metabolic syndrome, including reduced HDL-C, emerged as a novel risk factor and potential target for PSD prevention.

Funding: German Center for Neurodegenerative Diseases (DZNE).

Keywords: Brain ischaemia; Cognitive decline; Dementia; Dementia epidemiology; Diabetes; Metabolic syndrome; Post-stroke cognitive impairment; Post-stroke dementia; Risk factors; Small vessel disease; Stroke; Stroke epidemiology; Stroke outcomes; Vascular dementia.

Fig. 1

Participant flow chart for the 5-year follow-up period. Follow-ups via telephone at 3, 24, and 48 months exist but are not shown here. ∗Five deaths were recorded after participants were lost to follow-up.

Fig. 2

Cumulative post-stroke dementia incidence stratified by different categorical baseline characteristics. Acute phase cognitive impairment was defined as MoCA <26 or MMSE <27. Prediabetes was defined as HbA_1c ≥5·7 and <6·5. Diabetes mellitus was defined as HbA_1c ≥6·5 or treatment with antidiabetic medication. Acute reperfusion therapy indicates intravenous thrombolysis and/or endovascular thrombectomy. Error bars represent the 95% confidence interval for the Kaplan–Meier estimated cumulative incidence. Cumulative incidence rates were compared using Grey's test. CE = cardioembolism. Haemorr. = haemorrhagic stroke. LAA = large artery atherosclerosis. NIHSS = National Institutes of Health Stroke Scale. SVD = cerebral small vessel disease. SVO = small vessel occlusion. TOAST = Trial of Org 10172 in Acute Stroke Treatment. Undet. = stroke of undetermined aetiology.

Fig. 3

Cumulative incidence curves for post-stroke dementia stratified by the presence of metabolic syndrome (top left panel) and individual metabolic syndrome components (top middle to bottom right panel). Metabolic Syndrome was defined as the presence of three or more of the five criteria (Supplementary Methods). BP = Blood pressure. HDL-C = high-density lipoprotein cholesterol. MetS = Metabolic Syndrome. NIHSS = National Institutes of Health Stroke Scale. TG = Triglycerides.

Fig. 4

Cumulative incidence rates for post-stroke dementia stratified by metabolic syndrome and other relevant baseline factors. Error bars represent the 95% confidence interval for the Kaplan–Meier estimated cumulative incidence. Formal interaction tests showed no significant interactions (all p > 0·05). NIHSS = National Institutes of Health Stroke Scale.

Fig. 5

Population attributable fractions (PAF) for different risk factors for early- and delayed-onset post-stroke dementia, defined as dementia that occurred between 3 and 6 or after 6 months, respectively. Acute phase cognitive impairment was defined as MoCA <26 or MMSE <27, and metabolic syndrome as presence of three or more commonly used criteria (Supplementary Methods and Alberti et al.⁹). Error bars represent 95% confidence intervals, derived from 10,000 bootstrap iterations. HDL-C = high-density lipoprotein cholesterol. NIHSS = National Institutes of Health Stroke Scale.