Abstract
Background
The association between impairment across cognitive domains and incident stroke has not been thoroughly investigated. This study examined the risk of first‐time stroke among older adults with lower cognitive function in different cognitive domains and how associations vary by age.
Methods
This longitudinal cohort study used baseline data from GÅS (Good Aging in Skåne) and registry data from the Swedish Board of Health and Welfare. Exclusion criteria were previous stroke or transient ischemic attack, baseline dementia, or no cognitive data. The participants (n=4912) were divided into tertiles on the basis of performance on cognitive tests representing multiple cognitive domains. Cox regression was used to estimate the risk of stroke for the low tertile of cognitive function compared with the high tertile. Age‐stratified analysis was performed for participants aged 60 to 69, 70 to 79, and ≥80 years.
Results
The studied population consisted of 2203 men, 3094 aged 60 to 69 years, 494 aged 70 to 79 years, and 1184 aged ≥80 years. During the average follow‐up time of 12.3 years, 572 of the 4912 participants had a stroke. In the fully adjusted models, participants in the low tertile of perceptual speed had higher risk of stroke (hazard ratio [HR], 1.40 [95% CI, 1.13–1.72]). In age‐stratified analysis, higher risk of stroke was observed for participants aged 60 to 69 years in the low tertile of global cognition (HR, 1.86 [95% CI, 1.27–2.69]), episodic memory (HR, 1.68 [95% CI, 1.21–2.33]), and perceptual speed (HR, 1.66 [95% CI, 1.19–2.32]).
Conclusions
Lower cognitive functioning in global cognition, episodic memory and perceptual speed in people aged 60 to 69 years are associated with an increased risk of future stroke.
Keywords: cognitive domains, global cognition, prestroke cognition, stroke
Subject Categories: Cognitive Impairment, Cerebrovascular Disease/Stroke, Aging, Risk Factors
Nonstandard Abbreviation and Acronym
- GÅS
Good Aging in Skåne (Gott Åldrande i Skåne)
Research Perspective.
What Is New?
The findings of this study raise questions about whether low cognitive function in younger older adults can serve as a marker for increased stroke risk.
What Question Should Be Addressed Next?
Subsequent studies in younger cohorts should investigate the potential use of cognitive test performance in combination with biomarkers as indicators of future stroke risk.
Future magnetic resonance (MR) studies should examine associations between lower cognitive performance, preexisting cerebrovascular damage, and subsequent stroke.
In 2021, stroke was the third‐leading cause of death worldwide. 1 The incidence of stroke and cognitive impairment increases with age, and several shared risk factors have been suggested, including hypertension, diabetes, physical inactivity, smoking, and excessive alcohol use. 2 , 3 , 4 , 5 In addition, cognitive impairment has been considered an expression of preexisting cerebrovascular damage, such as silent brain infarcts, leukoaraiosis, and cerebral microbleeds. 6 These conditions are furthermore linked to an increased risk of stroke. 7 , 8 , 9 Cognitive impairment may therefore serve as an important indicator of the need for preventive measures against further cerebrovascular disease.
Two meta‐analyses established a higher relative risk of stroke for those with cognitive impairment. 10 , 11 However, studies considering cognitive impairment across different domains are few. 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 Studies investigating episodic memory have shown various results, where some indicated higher risk of stroke for those with impaired memory, 14 , 15 , 16 while others did not. 12 , 13 , 17 , 18 Similar inconsistent results can be seen for executive function, 16 , 17 , 18 , 19 language, 12 , 13 , 16 , 18 and perceptual speed. 12 , 13 , 16 , 18 , 19 Therefore, more studies are warranted. Although the risk of stroke among individuals with cognitive impairment may vary with age, this stratified risk has not been thoroughly investigated. 20 Furthermore, previous studies have reported sex differences in the associations between major stroke risk factors and incident stroke. 21 Thus, potential sex differences in the relationship between cognitive impairment and incident stroke warrant consideration.
Accordingly, the aim of this study was to investigate the risk of first‐time stroke among older adults with lower cognitive functioning in different cognitive domains. It further aimed to investigate how associations vary across age groups.
Methods
Data Availability
The data that support the findings of this study, as well as analytic models, are available from the corresponding author upon reasonable request.
Study Design and Participants
A prospective cohort study design was used. The majority of the data were extracted from the longitudinal, general population study GÅS (Good Aging in Skåne [Gott Åldrande i Skåne]), which is included in the Swedish National Study on Aging and Care. 22 , 23 The Swedish population register was used to randomly select older adults across 5 municipalities in Skåne County in southern Sweden.
The first 3 waves of participants were included in this study. The first wave (2001) of participants consisted of men and women aged 60 to 99 years. New participants aged 60 and 81 years were invited in 2007 (wave 2) and 2012 (wave 3). The participation rate ranged from 60% to 70% (data on nonparticipation characteristics are not available). For this study, follow‐up lasted until 1 of the following occurred: (1) first‐time stroke, (2) death, or (3) end of registry data (December 31, 2023). Date of death was obtained from the Swedish Cause of Death Register provided by the Swedish Board of Health and Welfare.
The same standardized physiological, psychological, and medical examinations were carried out at all 3 waves. In addition, participants were asked to fill out self‐rate questionnaires. Information was also retrieved from medical records. A total of 5804 participants were examined at baseline. Exclusion criteria for the current study were participants with no cognitive data, dementia, previous stroke, or transient ischemic attack (TIA) at baseline. Participants with dementia or a history of stroke at baseline were excluded, as they would likely be overrepresented in the low cognitive function group compared with higher‐functioning groups. This exclusion was necessary because the study aimed to investigate first‐time stroke and to minimize reverse causation arising from poststroke cognitive impairment. 24 TIA cases before baseline were excluded on the basis that a TIA effects risk of stroke events 25 , 26 as well as the effect of TIA on cognition. 27
Cognitive Assessment
The procedure of cognitive testing has been described in detail elsewhere. 28 Standardized testing methods were implied, reducing issues such as interrater reliability. 29 Episodic memory was measured through immediate recall and recognition of words. The digit cancellation test and the pattern comparison test were used to measure perceptual speed, while word fluency in naming animals and occupations was used for measuring word fluency categories. 28 The tests for episodic memory, perceptual speed, and word fluency categories have been deemed to be reliable and valid to represent the domain in an earlier study. 30 In addition, selected components of executive function were measured by digit span backwards test and a shortened version of the Trail Making Test B, which capture working memory and cognitive flexibility. 28
A Z score for each test score was generated for the participants. A composite score for each cognitive domain, that is, the mean value of the Z scores for representative tests, was calculated. For participants who had completed tests of ≥2 cognitive domains, global cognition was represented by a composite score of the completed tests, that is, the mean value of the Z scores of all available tests. 31 The cognitive domains were then further divided into age‐stratified tertiles, representing low, middle, and high performance. The low tertile represented participants with lower cognitive function.
Dementia and TIA Assessment
Dementia diagnosis was obtained from the examining physician, using the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition 32 for all‐cause dementia, and history of dementia from medical records. Complementary dementia diagnosis and TIA events were obtained from the Swedish National Patient Register of specialized in‐ and outpatient care provided by the Swedish Board of Health and Welfare.
Outcome
Data on first‐time stroke events were extracted from the Swedish National Patient Register of specialized in‐ and outpatient care. Stroke events included brain hemorrhages classified under International Classification of Diseases, Tenth Revision (ICD‐10) I60 to I61, brain infarction classified under ICD‐10 I63.0 to I63.5 and I63.7 to I63.9, as well as unspecified stroke classified under ICD‐10 I64. TIA was not included as an outcome.
Covariates
Age, sex, and years of education, in addition to smoking status, alcohol usage, and physical activity status, were self‐reported at baseline examination. To create the age‐stratified tertiles and to explore age associations in stratified analyses, age was categorized into 3 groups: participants aged 60 to 69, 70 to 79, and ≥80 years. However, in all adjusted analyses, age was included in its continuous form to reduce residual confounding. The total number of years of education was considered a continuous variable. Previous smokers and individuals who had never smoked were considered nonsmokers, while current smokers were considered smokers. Alcohol consumption was defined as either high (twice or more a week) or low to moderate (less than twice a week). Individuals engaging in regular or irregular physical activity were considered active, while individuals not taking part in any physical activity were considered inactive.
Body mass index was calculated as the measured weight/(length/100)2 (kg/m2). Blood pressure was measured 3 times sitting down in the upper left arm by an automatic sphygmomanometer, where the third recording was used here. If a third measurement was not available, the mean value of the first and second measurements was used.
Blood samples were drawn during the visit and non–high‐density lipoprotein cholesterol was calculated as the difference between total cholesterol and high‐density lipoprotein cholesterol. The use of antihypertensives, as well as the use of cholesterol‐lowering treatment, were self‐reported.
Comorbidities included diabetes types 1 and 2, atrial fibrillation including atrial flutter, and ischemic heart disease. These were reported (1) by detection by the physician, ECG, or self‐report; (2) by detection in medical records; and (3) by the in‐ and outpatient registers provided by the Swedish Board of Health and Welfare. Ischemic heart disease was considered if the participant had a diagnosis of angina pectoris, prior infarcts, chronic ischemic heart disease, or heart failure.
Statistical Analysis
Baseline characteristics were compared across the tertiles of global cognition. ANOVA was used for numerical data. While inspection of residual Q‐Q plots indicated some deviation from normality in some covariates, ANOVA was still considered robust due to the large sample size with close to equal group sizes. Furthermore, while post hoc Brown–Forsythe tests for homogeneity of variance showed significant results, inspection of group SDs and variance ratio indicated minor differences. Hence, results of the F‐test were deemed robust and reported. Post hoc Welch ANOVAs were performed as a sensitivity analysis and yielded similar P values. For nominal and ordinal data, the χ2 test was used. Multicollinearity among covariates was analyzed using the variance inflation factor and generalized variance inflation factor. All values were <2.0, which indicated no significant multicollinearity. Multiple imputations were carried through the covariates, where missing data ranged between 0.2% and 3.7%.
The risk of stroke for the tertiles of each cognitive domain was calculated with Cox proportional hazards regression analysis and pooled results were described in hazard ratios (HRs) with 95% CIs. To assess if lower cognitive functioning was associated with incident stroke, the highest tertile was used as reference and compared with the lowest and the middle tertile. In a sensitivity analysis, the middle tertile was used as the reference group. Furthermore, in post hoc sensitivity analysis, subdistribution HRs were estimated using Fine–Gray regression to account for the competing risk of death. For all regression analyses, a P for trend was obtained by treating the tertiles as a continuous variable, reflecting stepwise risk if dropping from a higher tertile to a lower. Minimum detectable HRs were calculated post hoc using the Schoenfield formula.
Three models were run to investigate associations in a stepwise manner, each including the variables of the former model(s): (1) demographic data (age, sex, and years of education), (2) lifestyle factors (physical activity, body mass index, smoking, and alcohol use), and (3) vascular factors (diabetes, systolic blood pressure, non–high‐density lipoprotein cholesterol, antihypertensive treatment, ischemic heart disease, atrial fibrillation or flutter, and cholesterol‐lowering treatment). Covariates were preselected on the basis of clinical rationale, with regards to literature of risk factors of stroke and cognitive impairment, as well as prior studies. 2 , 3 , 4 , 13 , 14 , 15 , 16 , 17 , 18 Testing of the proportional hazard assumption using Schoenfeld residuals showed no violation.
Stratified analyses for age groups (60–69, 70–79, and ≥80 years) were conducted. To address potential sex differences in associated risk of stroke, a sensitivity analysis on the group with the largest number of participants (60–69 years) was performed. To facilitate result presentation and interpretation, only the results from the fully adjusted model (model 3) are presented in the tables. Unadjusted results are found in the Supplemental Material.
A P value of <0.05 was considered statistically significant. Benjamini–Hochberg false discovery rate correction was used to address multiple testing. The correction was applied to the P for trend in the primary and age‐stratified analyses. Sensitivity analyses were considered exploratory. R software version 2026.01.0 Build 392 (R Foundation for Statistical Computing, Vienna, Austria) was used for all analyses. The plot was graphed in R software using the survminer package.
The study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. The checklist is found in the Supplemental Material.
Ethical Considerations
The GÅS study was approved by the Ethics Committee of Lund University (LU 744‐00). Written consent was gathered from all participants of the study. Information about the participants was stored securely and treated confidentially to prevent violation of integrity.
Results
Baseline Characteristics
A total of 5238 participants underwent baseline cognitive testing (Figure 1). Participants with prior TIA or stroke (n=217), dementia (n=93), or both (n=16) at baseline were excluded, leaving 4912 participants for analysis. Out of these, a global cognition score was calculated for 4772 participants.
Figure 1. Flowchart showing participant inclusion and exclusion in the study.

The proportion of participants who died were similar across cognitive domains. TIA indicates transient ischemic attack.
Baseline characteristics of the global cognition tertiles are shown in Table 1. The tertiles differed significantly in all described characteristics except age group, atrial fibrillation, cholesterol‐lowering treatment, and stroke subtypes. In total, there were 3094 participants aged 60 to 69 years, 494 aged 70 to 79 years, and 1184 aged ≥80 years. The studied population consisted of 2203 (46.2%) men.
Table 1.
Baseline Characteristics of the Global Cognition Tertiles
| Characteristics | Total (n=4772) | Tertiles of composite global cognitive score | P value | ||
|---|---|---|---|---|---|
| High (n=1589) | Middle (n=1591) | Low (n=1592) | |||
| Age, y, n (%) | 1.000 | ||||
| 60–69 | 3094 (64.8) | 1031 (33.3) | 1031 (33.3) | 1032 (33.3) | |
| 70–79 | 494 (10.3) | 164 (33.2) | 165 (33.4) | 165 (33.4) | |
| ≥80 | 1184 (24.8) | 394 (33.3) | 395 (33.4) | 395 (33.4) | |
| Men, n (%) | 2203 (46.2) | 670 (30.4) | 738 (33.5) | 795 (36.1) | <0.001 |
| Education, y, mean±SD | 10.9±3.9 | 12.3±4.0 | 10.8±3.7 | 9.6±3.4 | <0.001 |
| Physical inactivity, n (%) | 498 (10.4) | 107 (21.5) | 155 (31.1) | 236 (47.4) | <0.001 |
| Body mass index, kg/m2, mean±SD | 27.1±3.9 | 26.8±4.2 | 27.0±4.5 | 27.5±4.8 | <0.001 |
| Current smoker, n (%) | 803 (16.8) | 216 (26.9) | 266 (33.1) | 321 (40.0) | <0.001 |
| High alcohol consumption, n (%) | 1129 (23.7) | 521 (46.1) | 381 (33.7) | 227 (20.1) | <0.001 |
| Diabetes, n (%) | 367 (7.7) | 100 (27.2) | 110 (30.0) | 157 (42.8) | <0.001 |
| Systolic blood pressure, mm Hg, mean±SD | 143.7±22.0 | 141.7±21.1 | 143.6±21.7 | 145.8±22.8 | <0.001 |
| Antihypertensive treatment, n (%) | 1434 (30.1) | 441 (30.8) | 469 (32.7) | 524 (36.5) | 0.005 |
| Non–high‐density lipoprotein cholesterol, mmol/L, mean±SD | 4.3±1.1 | 4.2±1.1 | 4.3±1.1 | 4.4±1.1 | 0.001 |
| Cholesterol‐lowering treatment, n (%) | 749 (15.7) | 257 (34.3) | 235 (31.4) | 257 (34.3) | 0.082 |
| Ischemic heart disease, n (%) | 1018 (21.3) | 274 (26.9) | 344 (33.8) | 400 (39.3) | <0.001 |
| Atrial fibrillation, n (%) | 269 (5.6) | 83 (30.9) | 85 (31.6) | 101 (37.5) | 0.321 |
| Global cognition, mean±SD* | 0.46±0.74 | 1.12±0.47 | 0.50±0.43 | −0.23±0.56 | <0.001 |
| Episodic memory, mean±SD* | 0.32±0.92 | 0.94±0.71 | 0.36±0.73 | −0.34±0.84 | <0.001 |
| Perceptual speed, mean±SD* | 0.85±1.07 | 1.54±0.89 | 0.86±0.85 | 0.12±0.97 | <0.001 |
| Selected components of executive function, mean±SD* | 0.22±0.77 | 0.70±0.58 | 0.26±0.56 | −0.32±0.78 | <0.001 |
| Verbal fluency, mean±SD* | 0.50±1.06 | 1.30±0.89 | 0.51±0.79 | −0.34±0.79 | <0.001 |
| Stroke, n (%) | 549 (11.5) | 158 (28.9) | 179 (32.6) | 212 (38.6) | 0.011 |
| Stroke subtypes | 0.163 | ||||
| Ischemic stroke, n (%) | 419 (8.9) | 121 (28.9) | 141 (33.7) | 157 (37.5) | |
| Hemorrhagic stroke, n (%) | 83 (1.7) | 25 (30.1) | 25 (30.1) | 33 (39.6) | |
| Both, n (%) | 2 (<0.1) | 1 (50.0) | 1 (50.0) | 0 (0) | |
| Unspecified stroke, n (%) | 45 (0.9) | 12 (26.7) | 12 (26.7) | 21 (46.7) | |
ANOVA was used for comparing numerical data, while χ2 test was used for categorical and ordinal data. Differences among the global cognition tertiles are noted by the P value. A P value<0.05 was considered statistically significant.
Analysis performed on composites of Z scores of tests included in the cognitive domains.
Primary Analyses
In Table 2, the adjusted HRs of incident stroke for the tertiles across cognitive domains are shown. During a mean follow‐up time of 12.3 years, 572 of the participants included sustained a first‐time stroke (Table S1). Minimum detectable HRs given the number of events are found in Table S2. After adjustments for demographic data, a higher risk of stroke was found for the low compared with the high tertile of global cognition, although with P for trend >0.05 (data not shown). In the fully adjusted models, a higher risk of stroke was found for the low tertile of perceptual speed in comparison with the high tertile (HR, 1.40 [95% CI, 1.13–1.72]). No significant associations were found in the fully adjusted models when comparing the low tertile with the high tertile for the whole sample for global cognition (HR, 1.17 [95% CI, 0.95–1.47]), episodic memory (HR, 1.11 [95% CI, 0.91–1.37]), selected components of executive function (HR, 0.95 [95% CI, 0.77–1.17]) and verbal fluency (HR, 1.18 [95% CI, 0.95–1.46]).
Table 2.
Adjusted HRs for Incident Stroke Across Tertiles of Composite Cognitive Scores
| Cognitive domain | Tertiles of composite cognitive scores | P for trend† | |||||
|---|---|---|---|---|---|---|---|
| High | Middle | Low | |||||
| n/N | Adjusted HR (95% CI)* | n/N | Adjusted HR (95% CI)* | n/N | Adjusted HR (95% CI)* | ||
| Global cognition | |||||||
| Total | 158/549 | 1 (reference) | 179/549 | 1.03 (0.83–1.28) | 212/549 | 1.17 (0.95–1.47) | 0.349 |
| Age 60–69 y | 44/245 | 1 (reference) | 82/245 | 1.54 (1.06–2.24) | 119/245 | 1.86 (1.27–2.69) | 0.007 |
| Age 70–79 y | 36/96 | 1 (reference) | 25/96 | 0.58 (0.33–0.99) | 35/96 | 0.94 (0.56–1.59) | 0.948 |
| Age ≥80 y | 78/208 | 1 (reference) | 72/208 | 0.95 (0.68–1.32) | 58/208 | 0.88 (0.60–1.28) | 0.793 |
| Episodic memory | |||||||
| Total | 180/548 | 1 (reference) | 158/548 | 0.81 (0.65–1.01) | 210/548 | 1.11 (0.91–1.37) | 0.487 |
| Age 60–69 y | 57/242 | 1 (reference) | 74/242 | 1.18 (0.83–1.67) | 111/242 | 1.68 (1.21–2.33) | 0.007 |
| Age 70–79 y | 37/95 | 1 (reference) | 24/95 | 0.58 (0.34–0.99) | 34/95 | 1.09 (0.62–1.64) | 0.978 |
| Age ≥80 y | 86/211 | 1 (reference) | 60/211 | 0.65 (0.46–0.91) | 65/211 | 0.83 (0.59–1.16) | 0.487 |
| Perceptual speed | |||||||
| Total | 158/549 | 1 (reference) | 147/549 | 0.84 (0.67–1.05) | 244/549 | 1.40 (1.13–1.72) | 0.006 |
| Age 60–69 y | 55/248 | 1 (reference) | 56/248 | 0.79 (0.54–1.15) | 137/248 | 1.66 (1.19–2.32) | 0.006 |
| Age 70–79 y | 31/95 | 1 (reference) | 30/95 | 0.89 (0.52–1.52) | 34/95 | 1.15 (0.67–1.97) | 0.793 |
| Age ≥80 y | 72/206 | 1 (reference) | 61/206 | 0.89 (0.62–1.26) | 73/206 | 1.38 (0.95–2.02) | 0.349 |
| Selected components of executive function | |||||||
| Total | 182/557 | 1 (reference) | 189/557 | 1.00 (0.82–1.23) | 186/557 | 0.95 (0.77–1.17) | 0.793 |
| Age 60–69 y | 73/249 | 1 (reference) | 80/249 | 1.01 (0.73–1.39) | 96/249 | 1.05 (0.77–1.44) | 0.887 |
| Age 70–79 y | 31/97 | 1 (reference) | 40/97 | 1.18 (0.72–1.91) | 26/97 | 0.83 (0.48–1.44) | 0.793 |
| Age ≥80 y | 78/211 | 1 (reference) | 69/211 | 0.93 (0.67–1.30) | 64/211 | 0.92 (0.65–1.30) | 0.793 |
| Verbal fluency | |||||||
| Total | 163/555 | 1 (reference) | 193/555 | 1.12 (0.91–1.38) | 199/555 | 1.18 (0.95–1.46) | 0.349 |
| Age 60–69 y | 61/243 | 1 (reference) | 77/243 | 1.08 (0.77–1.52) | 105/243 | 1.31 (0.94–1.82) | 0.349 |
| Age 70–79 y | 34/97 | 1 (reference) | 33/97 | 0.85 (0.51–1.42) | 30/97 | 0.93 (0.55–1.59) | 0.887 |
| Age ≥80 y | 68/215 | 1 (reference) | 83/215 | 1.34 (0.96–1.85) | 64/215 | 1.21 (0.84–1.74) | 0.487 |
HR indicates hazard ratio.
Adjusted for age, sex, years of education, body mass index, physical inactivity, current smoking, alcohol consumption, diabetes, systolic blood pressure, antihypertensive treatment, non–high‐density lipoprotein cholesterol, cholesterol‐lowering therapy, ischemic heart disease, and atrial fibrillation.
A significant change in risk when stepping down 1 tertile is marked by a P for trend <0.05. P for trend was adjusted for false discovery rate by the Benjamini–Hochberg‐method.
The results of the age stratified analyses are shown in Table 2. Participants aged 60 to 69 years had a mean follow‐up time of 14.2 years, during which 252 of the 3134 participants had a stroke (Table S1). The mean follow‐up time was 12.1 years for the participants aged 70 to 79 years, during which 97 of the 507 participants had a stroke. Furthermore, the participants aged ≥80 years had a mean follow‐up time of 7.4 years, in which 223 of the 1271 participants had a stroke. Minimum detectable HRs given the number of events for each age strata are found in Table S2.
For global cognition, the participants aged 60 to 69 years in the low tertile had a higher risk of stroke compared with the high tertile in the fully adjusted model (HR, 1.86 [95% CI, 1.27–2.69]). The unadjusted cumulative incidence of stroke for the global cognition tertiles of participants aged 60 to 69 years is illustrated in Figure 2.
Figure 2. Kaplan–Meier estimates of unadjusted cumulative incidence of stroke for the 3 tertiles of global cognition of participants aged 60 to 69 years.

The study population consisted of 3094 participants. Difference in cumulative incidence between tertiles analyzed by log‐rank test (P<0.0001). The low tertile represents participants with lower cognitive function.
In addition, for those participants aged 60 to 69 years, the risk of stroke was higher for the low tertile of episodic memory (HR, 1.68 [95% CI, 1.21–2.33]) and for the low tertile of perceptual speed (HR, 1.66 [95% CI, 1.19–2.32]) compared with the high tertile in the fully adjusted model. After adjustments for demographic factors, an increased risk of stroke was observed for participants with impairment in verbal fluency, although with nonsignificant P for trend (data not shown). For the low tertile of the selected components of executive function, no significant increased risk of stroke was found when compared with the high tertile (HR, 1.05 [95% CI, 0.77–1.17]).
Higher risk of stroke was not observed across cognitive domains for either the participants aged 70 to 79 or ≥80 years when comparing the low tertile with the high tertile in the fully adjusted models. No associations were found for the low tertile of perceptual speed for participants aged 70 to 79 years across models. However, for participants aged ≥80 years, higher risk of stroke was found for the low tertile of perceptual speed in model 1 (HR, 1.47 [95% CI, 1.02–2.14]) and model 2 (HR, 1.46 [95% CI, 1.00–2.11]), although with P for trends >0.05. Nonetheless, the association was attenuated further in model 3 (HR, 1.38 [95% CI, 0.95–2.02]).
The unadjusted HRs for the tertiles of global cognition for the total population and age subgroups are found in Table S3.
Sensitivity Analyses With the Middle Tertile as Reference
Sensitivity analysis on the whole sample, with the middle tertile as reference, revealed higher risk of stroke for the lower tertile of global cognition compared with the middle tertile after adjustments for demographic data (data not shown). This risk was attenuated in further models of adjustment. Furthermore, in the fully adjusted models, an increased risk of stroke for the low tertile of episodic memory (HR, 1.37 [95% CI, 1.12–1.69]) and perceptual speed (HR, 1.66 [95% CI, 1.35–2.05]) were found, although the P for trend for episodic memory was >0.05 (Table S4). No increase in stroke risk was found for the low tertile compared with the middle tertile across models for verbal fluency (HR, 1.05 [95% CI, 0.86–1.28]) and the selected components of executive function (HR, 0.94 [95% CI, 0.77–1.16]).
In the sensitivity analysis using the middle tertile as a reference for participants aged 60 to 69 years, the low tertile of episodic memory (HR, 1.42 [95% CI, 1.06–1.91]) and perceptual speed (HR, 2.10 [95% CI, 1.53–2.89]) had a higher risk of stroke in the fully adjusted models (Table S4). No other associations were found when comparing the middle tertile with the low tertile for the age strata.
Sex Differences
In Table 3, the results of the sensitivity analysis on women and men aged 60 to 69 years are shown, and unadjusted results are found in Table S5. Similar to the results for the whole cohort aged 60 to 69 years, there were no significant associations between the low compared with the high tertile and incident stroke in the fully adjusted models for either men or women in measures of verbal fluency or executive function.
Table 3.
Adjusted HRs for Incident Stroke Across Tertiles of Composite Cognitive Scores, Stratified by Age Group (60–69 Years) and Sex
| Cognitive domain | Tertiles of composite cognitive scores | P for trend† | |||||
|---|---|---|---|---|---|---|---|
| High | Middle | Low | |||||
| n/N | Adjusted HR (95% CI)* | n/N | Adjusted HR (95% CI)* | n/N | Adjusted HR (95% CI)* | ||
| Global cognition | |||||||
| Women | 21/106 | 1 (reference) | 39/106 | 1.60 (0.93–2.75) | 46/106 | 1.59 (0.91–2.75) | 0.136 |
| Men | 23/139 | 1 (reference) | 43/139 | 1.49 (0.89–2.50) | 73/139 | 2.11 (1.28–3.47) | 0.003 |
| Episodic memory | |||||||
| Women | 37/105 | 1 (reference) | 32/105 | 0.98 (0.60–1.59) | 36/105 | 1.27 (0.79–2.03) | 0.325 |
| Men | 20/137 | 1 (reference) | 42/137 | 1.51 (0.88–2.59) | 75/137 | 2.30 (1.39–3.81) | <0.001 |
| Perceptual speed | |||||||
| Women | 26/107 | 1 (reference) | 20/107 | 0.62 (0.34–1.12) | 61/107 | 1.60 (0.98–2.60) | 0.015 |
| Men | 29/141 | 1 (reference) | 36/141 | 0.92 (0.56–1.51) | 76/141 | 1.77 (1.12–2.80) | 0.006 |
| Selected components of executive function | |||||||
| Women | 23/107 | 1 (reference) | 38/107 | 1.44 (0.86–2.43) | 46/107 | 1.34 (0.80–2.25) | 0.333 |
| Men | 50/142 | 1 (reference) | 42/142 | 0.82 (0.54–1.25) | 50/142 | 0.94 (0.62–1.40) | 0.738 |
| Verbal fluency | |||||||
| Women | 25/107 | 1 (reference) | 34/107 | 1.11 (0.65–1.88) | 48/107 | 1.34 (0.80–2.26) | 0.253 |
| Men | 36/136 | 1 (reference) | 43/136 | 1.05 (0.67–1.64) | 57/136 | 1.25 (0.81–1.94) | 0.316 |
HR indicates hazard ratio.
Adjusted for age, sex, years of education, body mass index, physical inactivity, current smoking, alcohol consumption, diabetes, systolic blood pressure, antihypertensive treatment, non–high‐density lipoprotein cholesterol, cholesterol‐lowering therapy, ischemic heart disease, and atrial fibrillation.
A significant change in risk when stepping down 1 tertile is marked by a P for trend <0.05.
Men in the low tertile of global cognition (HR, 2.11 [95% CI, 1.28–3.47]), episodic memory (HR, 2.30 [95% CI, 1.39–3.81]) and perceptual speed (HR, 1.77 [95% CI, 1.12–2.80]) had a higher risk of stroke compared with the high tertile in the fully adjusted models. Furthermore, higher risk of stroke for men in the low compared with the high tertile of verbal fluency was observed until adjustments for lifestyle factors. No significant associations were found for women in the low tertile of global cognition (HR, 1.59 [95% CI, 0.91–2.75]), episodic memory (HR, 1.27 [95% CI, 0.79–2.03]) or perceptual speed (HR, 1.60 [95% CI, 0.98–2.60]) in the fully adjusted models. However, higher risk of stroke for the low tertile of perceptual speed remained until the fully adjusted models, similarly for episodic memory until adjustments of lifestyle factors.
Post Hoc Competing Risk Analyses
Compared to primary analyses where an association for perceptual speed was observed, no association was found for the whole sample in the fully adjusted competing risk models (HR, 1.22 [95% CI, 0.99–1.42]; Table S6), although P for trend remained significant. Results for participants aged 60 to 69 years were confirmed in competing risk models. The direction of the results of the primary analyses remained in competing risk models, with the exception of the higher risk of stroke for participants aged ≥80 years in the low tertile of perceptual speed in model 2 that was not confirmed in this sensitivity analysis.
Discussion
This study, like others, showed that older adults with low cognitive function have a higher risk of first‐time stroke in comparison with those with high cognitive function. 10 , 11 , 16 , 20 These findings were primarily seen for the youngest old, and in men, where the associated risk of those with lower cognitive functioning was 1.66 to 2.30 times the risk of those with higher cognitive functioning depending on the measured cognitive domain. Our results were independent of behavioral, psychological, and physiological factors and adds new insights on age‐ and sex‐specific associations of stroke among those with lower cognitive functioning.
Similar to the results of this study, both meta‐analyses and other recent studies have shown that impairments in global cognition are associated with higher risk of stroke. 10 , 11 , 16 , 20 However, studies that focus on impairment across several cognitive domains are few and inconsistent. 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 Prior studies showed no association between risk of all‐type stroke, 16 , 18 ischemic stroke, 13 , 16 or hemorrhagic stroke 16 for individuals who perform worse on verbal fluency tests, which is consistent with the findings in the present study. However, studies have shown increased risk of fatal 12 and unspecified stroke 16 for these individuals, which suggests that associations for verbal fluency should be further investigated.
The main results, which were confirmed by sensitivity analyses, showed that older adults with lower scores on perceptual speed tests had a higher risk of stroke, which is consistent with the results of some previous studies 16 , 17 , 19 but not all. 12 , 13 , 18 Inconsistencies might be partially attributed to methodological differences, including the use of composite scores derived from multiple cognitive assessments rather than a single test, as well as variations in the specific test used for measurement.
Methodological differences may also account for the inconsistent findings on executive function, as the present study did not observe an increased risk of stroke among individuals with the lowest scores on selected executive function measures. 16 , 17 , 18 , 19 For instance, this study primarily measured working memory and cognitive flexibility, whereas others focus on other aspects of executive functioning such as planning, reasoning, and stronger measurements of inhibition. 16 , 17 Moreover, in the present study, there were several missing values for Trail Making Test B for participants who did not complete the test, as well as those who refused to participate, which might result in a nonresponse bias.
Some studies report an increased risk of stroke for individuals with impairments in episodic memory. 14 , 15 , 16 These studies all included participants aged <70 years. However, in studies with samples aged >70 years were examined, these associations for episodic memory were not found. 12 , 17 Together with our results, this indicates that associated risk for stroke among those with impairments in memory is apparent for the younger older adults. Nonetheless, even though participants aged <70 years were included, Weinstein et al 18 showed no association between impairment in episodic memory and stroke risk. However, age stratified analysis was not performed in that study. 18 Adding on, de Moraes et al investigated a study population aged 48 to 67 years and failed to find an association, although power issues may have existed. 13 , 16
Furthermore, the differences in risk between age groups in this study may also reflect attrition bias. The older participants had shorter follow‐up periods due to a higher number of deaths from other causes. Results from competing risk models had little impact on the results for participants aged 60 to 69 years, and the direction of the results for the other groups remained largely unchanged. However, the higher risk of stroke in the main analyses for participants aged ≥80 years in the low tertile of perceptual speed until adjustments for vascular factors did not remain in competing risk analysis.
Sensitivity analysis also revealed sex‐specific associations, where heightened stroke risk was found for men aged 60 to 69 years in the low tertile of global cognition, episodic memory, and perceptual speed. Prior research has suggested that sex differences influence the relationship between vascular risk factors and stroke. 21 It is therefore plausible that similar sex differences exist in the link between cognitive impairment and stroke risk. Proposed explanations include genetic factors as well as differences in the management of vascular conditions. 21
Nonetheless, in contradiction to our results, Qiao et al presented a higher HR for stroke for women with cognitive impairment. 20 However, the HRs for women in the low tertiles of global cognition and perceptual speed indicate that there might be a higher HR for stroke, even though results were nonsignificant in our study. Furthermore, Qiao et al were not able to show significant sex differences when adding an interaction term between sex and cognitive impairment. 20 Hence, further studies exploring sex differences and possible underlying mechanisms are warranted.
In the present study, the low tertile of each domain represented those with low cognitive functioning. 17 Notably, dividing cognitive test scores into tertiles is a robust measure of level of cognitive functioning, with some degree of misclassification. It is likely that cases of mild cognitive impairment and prodromal dementia will be present in the low tertile. While some studies adjusted for apolipoprotein E4, these data were not available for this study. Hence, it was not taken into consideration that some participants have an increased risk of amyloid angiopathy and Alzheimer dementia, which in turn affects both memory and hemorrhagic stroke risk. 33 , 34 Brief post hoc analysis showed more incident dementia cases in the low tertile of global cognition than the high tertile (6.2% versus 1.6%) and the stroke group compared with the nonstroke group (12.7% versus 3.0%) for participants aged 60 to 69 years, which had the longest follow‐up.
Shared risk factors between stroke and cognitive impairment may explain some of the outcomes, as associations for some domains were attenuated during the stepwise adjustments. However, the risk of stroke across impairments in domains remained even after full adjustments among the youngest participants, suggesting the influence of unmeasured factors or alternative mechanisms. One such unmeasured factor could be poor compliance with medical treatment, where those with low cognitive functioning may have poorer compliance than those in the higher cognitive functioning groups.
Given differences in risk profiles, incidence, and pathogenesis, future research could benefit from stratifying analyses by stroke subtypes, as done in former studies. 16 , 19 On inspection of stroke subtypes across the whole sample and age strata, the risk associations were confirmed for those with ischemic stroke, but not for hemorrhagic stroke (data not shown). Notably, as in other studies, 16 , 19 hemorrhagic stroke events were few, especially in each age strata. Therefore, future studies with more hemorrhagic stroke cases are needed to confirm these results.
Others have discussed that the heightened risk of stroke for those with lower cognitive functioning could be due to previous silent cerebrovascular events, 12 , 14 , 16 , 17 , 18 as this is a suggested risk factor of both stroke and cognitive impairment. 6 , 7 , 8 , 9 Hence, the participants in the lowest tertiles might have sustained more of these events than those in the high tertiles. Since a recent magnetic resonance imaging study on GÅS data showed that >70% of those aged ≥70 years have had cerebrovascular small‐vessel disease, 35 it might be that silent cerebrovascular events serve as a better predictor of future stroke for a younger population. For the younger population, the events are rarer, while the older population also have an increasing accumulation of risk factors, diseases, and death unrelated to stroke that might overshadow associations. This might also partly explain why the middle tertile of episodic memory for individuals aged 80 years appears to have a reduced associated risk of stroke in Cox regression, while the P for trend suggests no pattern in the association.
To the authors' knowledge, no study has described magnetic resonance imaging changes among those with lower cognitive function in relation to the risk of future stroke, although Weinstein et al 18 showed that silent cerebral infarcts could not explain why those with worse performance on Trail Making Test B had a higher risk of stroke. Future studies should explore if the site of possible previous silent cerebrovascular events or the later stroke is linked to the brain area mostly associated with a particular cognitive domain. For instance, it would be valuable to investigate whether individuals with poorer performance on perceptual speed tasks already exhibit structural or functional abnormalities in domain‐relevant brain regions, such as the white matter in the parietal lobes. 36
Strengths of this study are the use of a population‐based design, a large cohort with long follow‐up, and an extensive cognitive battery. Although no information on the characteristics of nonparticipants was available, it is likely that individuals attending GÅS are generally healthier, which may limit the generalizability of the findings. To mitigate this potential bias, home visits were offered to more fragile individuals. Moreover, the baseline cohorts had relatively high participation rates (60%–70%). Other strengths are the adjustments for multiple health covariates and the use of registry data, which prevent loss to follow‐up and recall bias.
This study is not without limitations. First, results could be underestimated due to missing data in cognitive testing, and some loss of information is expected since cognitive results were categorized into tertiles to facilitate interpretation. Second, in sex and age strata 70 to 79 years, analyses of specific cognitive domains were based on a limited number of events, which reduced statistical power. Moreover, the fully adjusted models included a large number of covariates, introducing a potential risk of overfitting in some strata. Therefore, results based on small numbers of events should be interpreted with caution. Nevertheless, there were few instances in which adjustment for covariates altered the significance levels, suggesting that the findings are relatively robust. However, future studies could possibly benefit from using more flexible methods to model age‐specific associations, as some loss of information is expected with categorization. Finally, by excluding TIA cases from the outcome, there is a risk that some misclassified cases of minor stroke were missed as computed tomographic scan was the usual imaging modality for stroke in Sweden during a significant part of the study period. Consequences could be both lower power and risk of missing effects and noneffects.
Conclusions
The results of this study show an association between lower cognitive function in global cognition, episodic memory, and perceptual speed in younger older adults and incident stroke. Furthermore, it suggests that there are sex differences in these associations. Cognitive testing might be an important screening tool for future cerebrovascular events for this age group. More studies in the area, including magnetic resonance imaging data, analysis of stroke subtypes, and analyses on age and sex differences are needed.
Sources of Funding
The project Good Aging in Skåne, part of the Swedish National Study on Aging and Care (www.snac.org), was supported by the Swedish Ministry of Health and Social Affairs, the County of Region Skåne, and the Swedish Research Council under Grant Number 2017‐01613; 521‐2013‐8604, 2021‐00178, and 2021‐01437. This research received no specific grant from any other funding agency in the public, commercial, or not‐for‐profit sectors.
Disclosures
None.
Supporting information
Tables S1–S6
STROBE Checklist
Acknowledgments
The authors thank all participants on the GÅS project, as well as the teams working with data collection, for making this project possible.
This manuscript was sent to Fadar O. Otite, MD, SM, Associate Editor, for review by expert referees, editorial decision, and final disposition.
Supplemental Material is available at https://www.ahajournals.org/doi/suppl/10.1161/JAHA.125.048423
For Sources of Funding and Disclosures, see page 11.
References
- 1. Global, regional, and national burden of stroke and its risk factors, 1990‐2021: a systematic analysis for the global burden of disease study 2021. Lancet Neurol. 2024;23:973–1003. doi: 10.1016/s1474-4422(24)00369-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Mahinrad S, Sorond F, Gorelick PB. The role of vascular risk factors in cognitive impairment and dementia and prospects for prevention. Clin Geriatr Med. 2023;39:123–134. doi: 10.1016/j.cger.2022.07.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Boehme AK, Esenwa C, Elkind MS. Stroke risk factors, genetics, and prevention. Circ Res. 2017;120:472–495. doi: 10.1161/circresaha.116.308398 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. O'Donnell MJ, Xavier D, Liu L, Zhang H, Chin SL, Rao‐Melacini P, Rangarajan S, Islam S, Pais P, McQueen MJ, et al. Risk factors for ischaemic and intracerebral haemorrhagic stroke in 22 countries (the INTERSTROKE study): a case‐control study. Lancet. 2010;376:112–123. doi: 10.1016/s0140-6736(10)60834-3 [DOI] [PubMed] [Google Scholar]
- 5. Plassman BL, Langa KM, Fisher GG, Heeringa SG, Weir DR, Ofstedal MB, Burke JR, Hurd MD, Potter GG, Rodgers WL, et al. Prevalence of cognitive impairment without dementia in the United States. Ann Intern Med. 2008;148:427–434. doi: 10.7326/0003-4819-148-6-200803180-00005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Jansma A, de Bresser J, Schoones JW, van Heemst D, Akintola AA. Sporadic cerebral small vessel disease and cognitive decline in healthy older adults: a systematic review and meta‐analysis. J Cereb Blood Flow Metab. 2024;44:660–679. doi: 10.1177/0271678x241235494 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Gupta A, Giambrone AE, Gialdini G, Finn C, Delgado D, Gutierrez J, Wright C, Beiser AS, Seshadri S, Pandya A, et al. Silent brain infarction and risk of future stroke: a systematic review and meta‐analysis. Stroke. 2016;47:719–725. doi: 10.1161/strokeaha.115.011889 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Debette S, Markus HS. The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: systematic review and meta‐analysis. BMJ. 2010;341:c3666. doi: 10.1136/bmj.c3666 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Akoudad S, Portegies ML, Koudstaal PJ, Hofman A, van der Lugt A, Ikram MA, Vernooij MW. Cerebral microbleeds are associated with an increased risk of stroke: the Rotterdam study. Circulation. 2015;132:509–516. doi: 10.1161/circulationaha.115.016261 [DOI] [PubMed] [Google Scholar]
- 10. Lee M, Saver JL, Hong KS, Wu YL, Liu HC, Rao NM, Ovbiagele B. Cognitive impairment and risk of future stroke: a systematic review and meta‐analysis. Cmaj. 2014;186:E536–E546. doi: 10.1503/cmaj.140147 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Rostamian S, Mahinrad S, Stijnen T, Sabayan B, de Craen AJ. Cognitive impairment and risk of stroke: a systematic review and meta‐analysis of prospective cohort studies. Stroke. 2014;45:1342–1348. doi: 10.1161/strokeaha.114.004658 [DOI] [PubMed] [Google Scholar]
- 12. Batterham PJ, Mackinnon AJ, Christensen H. The association between change in cognitive ability and cause‐specific mortality in a community sample of older adults. Psychol Aging. 2012;27:229–236. doi: 10.1037/a0024517 [DOI] [PubMed] [Google Scholar]
- 13. de Moraes SA, Szklo M, Tilling K, Sato R, Knopman D. Cognitive functioning as a predictor of ischemic stroke incidence. Epidemiology. 2003;14:673–679. doi: 10.1097/01.ede.0000083262.58396.a3 [DOI] [PubMed] [Google Scholar]
- 14. DeFries T, Avendaño M, Glymour MM. Level and change in cognitive test scores predict risk of first stroke. J Am Geriatr Soc. 2009;57:499–505. doi: 10.1111/j.1532-5415.2008.02132.x [DOI] [PubMed] [Google Scholar]
- 15. Glymour MM, Maselko J, Gilman SE, Patton KK, Avendaño M. Depressive symptoms predict incident stroke independently of memory impairments. Neurology. 2010;75:2063–2070. doi: 10.1212/WNL.0b013e318200d70e [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Heshmatollah A, Mutlu U, Koudstaal PJ, Ikram MA, Ikram MK. Cognitive and physical impairment and the risk of stroke ‐ a prospective cohort study. Sci Rep. 2020;10:6274. doi: 10.1038/s41598-020-63295-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Rostamian S, van Buchem MA, Westendorp RG, Jukema JW, Mooijaart SP, Sabayan B, de Craen AJ. Executive function, but not memory, associates with incident coronary heart disease and stroke. Neurology. 2015;85:783–789. doi: 10.1212/wnl.0000000000001895 [DOI] [PubMed] [Google Scholar]
- 18. Weinstein G, Beiser AS, Decarli C, Au R, Wolf PA, Seshadri S. Brain imaging and cognitive predictors of stroke and Alzheimer disease in the Framingham heart study. Stroke. 2013;44:2787–2794. doi: 10.1161/strokeaha.113.000947 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Wiberg B, Lind L, Kilander L, Zethelius B, Sundelöf JE, Sundström J. Cognitive function and risk of stroke in elderly men. Neurology. 2010;74:379–385. doi: 10.1212/WNL.0b013e3181ccc516 [DOI] [PubMed] [Google Scholar]
- 20. Qiao T, Lu H, Wang X, Wang X. Cognitive impairment is associated with new‐onset stroke: evidence from a nationwide prospective cohort study. Cerebrovasc Dis. 2024;53:288–297. doi: 10.1159/000533503 [DOI] [PubMed] [Google Scholar]
- 21. Peters SAE, Carcel C, Millett ERC, Woodward M. Sex differences in the association between major risk factors and the risk of stroke in the UK biobank cohort study. Neurology. 2020;95:e2715–e2726. doi: 10.1212/wnl.0000000000010982 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Ekström H, Elmståhl S. Pain and fractures are independently related to lower walking speed and grip strength: results from the population study “good ageing in Skåne”. Acta Orthop. 2006;77:902–911. doi: 10.1080/17453670610013204 [DOI] [PubMed] [Google Scholar]
- 23. Lagergren M, Fratiglioni L, Hallberg IR, Berglund J, Elmståhl S, Hagberg B, Holst G, Rennemark M, Sjölund BM, Thorslund M, et al. A longitudinal study integrating population, care and social services data. The Swedish National Study on Aging and Care (SNAC). Aging Clin Exp Res. 2004;16:158–168. doi: 10.1007/bf03324546 [DOI] [PubMed] [Google Scholar]
- 24. El Husseini N, Katzan IL, Rost NS, Blake ML, Byun E, Pendlebury ST, Aparicio HJ, Marquine MJ, Gottesman RF, Smith EE. Cognitive impairment after ischemic and hemorrhagic stroke: a scientific statement from the American Heart Association/American Stroke Association. Stroke. 2023;54:e272–e291. doi: 10.1161/str.0000000000000430 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Kauw F, Takx RAP, de Jong H, Velthuis BK, Kappelle LJ, Dankbaar JW. Clinical and imaging predictors of recurrent ischemic stroke: a systematic review and meta‐analysis. Cerebrovasc Dis. 2018;45:279–287. doi: 10.1159/000490422 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Lioutas VA, Ivan CS, Himali JJ, Aparicio HJ, Leveille T, Romero JR, Beiser AS, Seshadri S. Incidence of transient ischemic attack and association with long‐term risk of stroke. JAMA. 2021;325:373–381. doi: 10.1001/jama.2020.25071 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Ganesh A, Barber PA. The cognitive sequelae of transient ischemic attacks‐recent insights and future directions. J Clin Med. 2022;11:11. doi: 10.3390/jcm11092637 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Overton M, Pihlsgård M, Elmståhl S. Up to speed: birth cohort effects observed for speed of processing in older adults: data from the good ageing in Skåne population study. Intelligence. 2018;67:33–43. doi: 10.1016/j.intell.2018.01.002 [DOI] [Google Scholar]
- 29. Overton M, Pihlsgård M, Elmståhl S. Test administrator effects on cognitive performance in a longitudinal study of ageing. Cogent Psychol. 2016;3:1260237. doi: 10.1080/23311908.2016.1260237 [DOI] [Google Scholar]
- 30. Laukka EJ, Lövdén M, Herlitz A, Karlsson S, Ferencz B, Pantzar A, Keller L, Graff C, Fratiglioni L, Bäckman L. Genetic effects on old‐age cognitive functioning: a population‐based study. Psychol Aging. 2013;28:262–274. doi: 10.1037/a0030829 [DOI] [PubMed] [Google Scholar]
- 31. Overton M, Skoog J, Laukka EJ, Bodin TH, Mattsson AD, Sjöberg L, Hofer SM, Johansson L, Kulmala J, Kivipelto M, et al. Sleep disturbances and change in multiple cognitive domains among older adults: a multicenter study of five Nordic cohorts. Sleep. 2024;47:47. doi: 10.1093/sleep/zsad244 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. American Psychiatric Association . Diagnostic and statistical manual of mental disorders: DSM‐IV: prepared by the task force on DSM‐IV. Washington, D.C.: American Psychiatric Association; 1994. [Google Scholar]
- 33. Waziry R, Chibnik LB, Bos D, Ikram MK, Hofman A. Risk of hemorrhagic and ischemic stroke in patients with Alzheimer disease: a synthesis of the literature. Neurology. 2020;94:265–272. doi: 10.1212/wnl.0000000000008924 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Morley JE, Farr SA. The role of amyloid‐beta in the regulation of memory. Biochem Pharmacol. 2014;88:479–485. doi: 10.1016/j.bcp.2013.12.018 [DOI] [PubMed] [Google Scholar]
- 35. Elmståhl S, Ellström K, Siennicki‐Lantz A, Lätt J, Månsson S, Månsson T, Abul‐Kasim K. Incidence of cerebral small vessel disease‐related MR markers in the Swedish general population ‘good aging in Skåne’ (GÅS) study. J Neurol. 2024;271:5997–6003. doi: 10.1007/s00415-024-12562-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Turken A, Whitfield‐Gabrieli S, Bammer R, Baldo JV, Dronkers NF, Gabrieli JD. Cognitive processing speed and the structure of white matter pathways: convergent evidence from normal variation and lesion studies. Neuroimage. 2008;42:1032–1044. doi: 10.1016/j.neuroimage.2008.03.057 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Tables S1–S6
STROBE Checklist
Data Availability Statement
The data that support the findings of this study, as well as analytic models, are available from the corresponding author upon reasonable request.
