Skip main navigation

Temporal Trends in Stroke-Related Memory Change

Results From a US National Cohort 1998–2016
Originally published 2021;52:1702–1711


Background and Purpose:

Findings from the Framingham Heart Study suggest that declines in dementia incidence rates over recent decades are partially due to decreases in stroke incidence and mortality; however, whether trends of declining dementia rates extend to survivors of incident stroke remains unclear. We investigated evidence for temporal trends in memory change related to incident stroke in a nationally representative cohort.


Adults age 50+ in the HRS (Health and Retirement Study) were followed across three successive 6-year epochs (epoch 1: 1998–2004, n=16 781; epoch 2: 2004–2010, n=15 345; and epoch 3: 2010–2016; n=15 949). Participants were included in an epoch if they were stroke-free at the start of that epoch. Annual rates of change in a composite z-standardized memory score were compared using demographic-adjusted linear regression models for stroke-free participants, those who survived after stroke, and those who died after stroke, considering memory change before stroke, at the time of stroke, and for years following stroke.


Crude stroke incidence rates decreased from 8.5 per 1000 person-years in epoch 1 to 6.8 per 1000 person-years in epoch 3. Rates of memory change before and following stroke onset were similar across epochs. Memory decrement immediately after stroke onset attenuated from −0.37 points (95% CI, −0.44 to −0.29) in epoch 1 to −0.26 (95% CI, −0.33 to −0.18) points in epoch 2 and −0.25 (95% CI, −0.33 to −0.17) points in epoch 3 (P value for linear trend=0.02).


Decreases in stroke-related dementia in recent years may be partially attributable to smaller memory decrements immediately after stroke onset. Findings suggest reductions in stroke incidence and improvements in stroke care may also reduce population burden of dementia. Further investigations into whether temporal trends are attributable to improvements in stroke care are needed.


Stroke is a major cause of long-term cognitive impairment and dementia. Nearly, one in 4 stroke survivors is diagnosed with dementia and an additional one in 3 shows signs of cognitive impairment.1–5 Both stroke and dementia incidence rates in the United States and other high-income countries have declined in recent years.3,6,7 Stroke mortality rates have also declined, leading to the hypothesis that reductions in stroke severity or improvements in quality of stroke care may contribute to decreases in poststroke dementia.8 However, whether trends of declining dementia rates extend to survivors of incident stroke remains unclear.6,9

Acute stroke is often preceded by accumulating cerebrovascular disease, causing memory decline years before stroke diagnosis. Additionally, immediate decrements in memory functioning at stroke onset suggest that evaluating links between stroke and memory requires consideration of the entire trajectory of memory performance.10–14 Few studies have longitudinal data to evaluate whether trends of cognitive impairment before and after stroke have changed in recent decades. Results examining temporally nonoverlapping epochs (time periods) in the Framingham Study, corresponding with resampling waves, found that risk of dementia after stroke decreased from a 4.72 (95% CI, 2.17–10.23) times higher risk of dementia in stroke survivors compared with stroke-free participants between 1992 and 1998 to a 1.43 (95% CI, 0.20–10.37) times higher risk between 2004 and 2008.7

We evaluated whether poststroke cognitive outcomes have improved in recent years by defining 3 successive epochs corresponding with study resampling periods in the nationally representative HRS (Health and Retirement Study). Treating each epoch as a distinct cohort for that time period, we estimated cognitive trajectories before stroke, immediate change at stroke onset, and longer-term poststroke cognitive trajectory. We compared epoch-specific memory trajectories in people who experienced stroke to memory trajectories in stroke-free participants to assess temporal trends in stroke-related cognitive change, and to explore when changes occur in the development of stroke-related deficits.


Study Design

HRS is a study of community-dwelling middle-aged and older adults in the United States, initiated in 1992. Enrollment was modified in 1998 to achieve a nationally representative sample of US adults over age 50 and has fielded additional enrollment waves every 6 years to represent subsequent generational birth cohorts. Study recruitment and design have been published elsewhere.15,16 Biennial interviews are ongoing, typically conducted via telephone. Proxies complete exit interviews after death of a participant to provide information on major health events preceding the participant’s death (such as a fatal stroke). HRS is sponsored by the National Institute on Aging, National Institutes of Health and conducted by the University of Michigan, Ann Arbor. All study participants gave informed consent, and HRS was approved by the institutional review board at the University of Michigan. The current study, as a secondary analysis of deidentified publicly available HRS data, was certified as exempt from review by the University of California, San Francisco Institutional Review Board. All data and materials have been made publicly available through the Institute for Social Research at the University of Michigan, Ann Arbor and can be accessed at

Three epochs with temporally nonoverlapping periods of eligibility, corresponding with new HRS enrollments in 1998, 2004, and 2010 were examined with epoch 1 spanning the period from 1998 to 2004, epoch 2 from 2004 to 2010, and epoch 3 from 2010 to 2016. Full epoch inclusion is shown in Figure 1. Epochs included participants completing the baseline wave of each epoch (1998 for epoch 1 [n=19 819], 2004 for epoch 2 [n=18 523], and 2010 for epoch 3 [n=20 185]) and up to 6 years of follow-up. Respondents were excluded if they had history of a stroke at baseline for that epoch (epoch 1 [1998]: 1461; epoch 2 [2004]: 1356; and epoch 3 [2010]: 1544) or no information on date of stroke experienced during the epoch (epoch 1: 2; epoch 2: 41; and epoch 3: 54). Participants were further excluded if missing memory scores across all waves in an epoch (epoch 1: 1530; epoch 2: 1725; and epoch 3: 2532) or missing covariates (EP1: 45; EP2: 56; and EP3: 106) for a final analytic sample of 16 781 in epoch 1, 15 345 in epoch 2, and 15 949 in epoch 3.

Figure 1.

Figure 1. Epoch eligibility: each epoch included baseline wave and up to 3 waves of biennial follow-up. Participants with prevalent baseline strokes excluded from that epoch to mirror separate cohorts of stroke-free individuals. For example, individuals enrolling in 1998 and reporting first stroke in 2004 would be considered as having a stroke in epoch 1 and would be ineligible for epoch 2. A similar individual enrolling in 1998 but experiencing a first stroke in the 2006 wave would be considered stroke-free in epoch 1 and as having a stroke in epoch 2. HRS indicates Health and Retirement Study.

Dependent Variable

Immediate and delayed (5-minute) recall tests of a verbally given 10-word list of common nouns were used to assess memory. Previous findings have shown that verbal memory may be the strongest predictor of impairment after incident stroke.17 If individuals were too impaired to participate in assessments, proxy informants such as spouses rated participant’s memory on a 5-point Likert scale and the Jorm Informant Questionnaire for Cognitive Decline,18,19 for which validity and reliability have been documented elsewhere.20,21 A previously created composite score combining proxy and direct memory measurements for HRS was used as the measure of memory at each data collection wave.22,23 Memory scores were z-standardized using the population-weighted mean and SD of the full 1998 sample. The final standardized measures ranged from −6.02 to 3.14, with each point corresponding to a one SD change in the 1998 population.

Independent Variable

Stroke status was determined using self-reported doctor’s diagnosis of stroke answering the question, “Has a doctor ever told you that you had a stroke?” If patients were unavailable for direct interview or deceased, proxy informants reported patients’ stroke status. Prior analyses of stroke in HRS have found that self-reported stroke incidence does not differ greatly from physician-verified strokes.24 All memory reports after date of first stroke were considered poststroke cognition; information was not collected on subsequent strokes or stroke subtypes, and transient ischemic attacks were not measured.

Participants were classified into 3 categories for each epoch: (1) stroke survivors: participants experiencing stroke during epoch who survived to participate in a subsequent interview; (2) stroke decedents: participants experiencing stroke who did not survive to participate in subsequent interviews (ie, who died from any cause including but not restricted to stroke); and (3) stroke-free participants: those reporting no stroke during the epoch. Participants were included in an epoch if newly enrolled into HRS and not reporting a previous stroke, or if already-enrolled and stroke-free as of the previous epoch (stroke survivors or decedents were automatically ineligible for inclusion in subsequent epochs as baseline exclusion criteria restricted each epoch to stroke-free participants at the start of each epoch). Trajectories for stroke survivors and stroke decedents were calculated using months before stroke or months since stroke for each memory measurement ascertainment. For stroke-free participants, trajectories of memory were calculated for age-related memory decline.


Models adjusted for stroke-free participants’ interview age or age at stroke for those experiencing stroke, centered at 75 years. The following demographics were collected at HRS entry: race (White, Black, other), gender, birthplace region (Census-defined Northeast, Midwest, South, West, or outside United States), years of attained education (range: 0 to 17+ years), and mother’s education (<8 years, 8+ years, unknown). Height, marital status (married/partnered, separated/divorced/widowed, never married), and wealth were updated at the start of each epoch (1998 for epoch 1, 2004 for epoch 2, and 2010 for epoch 3). Analyses further adjusted for practice effects of repeated memory testing indicating the wave of individuals’ first cognitive assessment.

Statistical Analysis

Longitudinal trajectories of memory scores were estimated using linear mixed models with random intercepts. Models were estimated separately for each epoch, with the following epoch-specific predictors: stroke status (stroke survivor, stroke decedent, stroke-free), time until stroke for stroke survivors and decedents, time since stroke for stroke survivors, age at stroke or stroke-free age, and covariates noted above. A discontinuity indicator for immediate decrement in memory performance at the time of first stroke was included for participants who survived stroke, representing acute decreases in memory at stroke onset. Changes in memory were thus estimated as a function of time until stroke for participants who experienced stroke and additionally as time following stroke for stroke-survivors, estimating separate prestroke and poststroke slopes, and as age-related memory decline for stroke-free participants. Predicted memory scores were calculated for all stroke categories, with the intercept representing an epoch-specific stroke-free 75-year-old individual. A pooled model including all stroke variables interacted with an epoch indicator was used to assess trends across epochs. Supplementary analyses assessed patterns of stroke-related memory using only direct cognitive assessments (excluding proxy informant memory reports) instead of the previously published imputed memory score. To obtain estimates representative of the community-dwelling US population over age 50, analyses were weighted to be nationally representative of baseline for each epoch. Analyses were conducted in SAS 9.4 (SAS Institute, Cary, NC).


Across all epochs, 2434 individuals experienced a first stroke: 1844 nonfatal and 590 fatal. Average baseline age was 66.3 (SD: 10.0) years in epoch 1, 67.1±10.2 years in epoch 2, and 66.5±10.8 year in epoch 3. Other characteristics are presented in Table 1.

Table 1. Unweighted Baseline Demographic Characteristics

Epoch 1 (1998–2004)Epoch 2 (2004–2010)Epoch 3 (2010–2016)P value for differences across epochs
N=16 781N=15 345N=15 949
Stroke status<0.0001
 Stroke free15 89094.69%14 49394.45%15 25895.67%
 Stroke survivors6724.00%6364.14%5363.36%
 Stroke decedents2191.31%2161.41%1550.97%
Memory score*1.010.560.990.550.990.540.0448
 White13 97983.30%12 62782.29%11 98375.13%
Marital status<0.0001
 Married/partnered11 49968.52%10 27166.93%10 20363.97%
 Never married4902.92%4913.20%8375.25%
Years of education12.323.0112.792.8413.172.62<0.0001
Mother’s education<0.0001
 <8 y788346.98%638741.62%568935.67%
 8+ years889853.02%895858.38%10 26064.33%
Wealth (median, per $1000)74.00IQR (23.33–183.00)102.50IQR (29.81–265.50)86.50IQR (16.56–263.00)<0.0001
Height, m1.690.101.690.101.690.100.0001
Southern birthplace286417.07%247916.16%293018.37%<0.0001

IQR indicates interquartile range. P value for differences across epoch calculated using χ2 statistic for categorical covariates and ANOVA for continuous covariates.

* Memory scores z-standardized to 1998 sample mean and SD; 1 unit change represents 1 SD.

Age-adjusted stroke incidence rates for nonfatal strokes, using 2000 census estimates as the referent population, decreased from 6.6 per 1000 person-years in epoch 1 to 6.1 per 1000 person-years in epoch 2 and 5.3 per 1000 person-years in epoch 3 (Table 2). Age-adjusted incidence rates for fatal strokes similarly decreased, from 1.9 per 1000 person-years in epochs 1 and 2 to 1.4 per 1000 person-years in epoch 3.

Table 2. Age-Standardized* Stroke Incidence Rates (Weighted to be Representative for US Community-Residing Individuals Aged 50 and Older at the Start of Each Epoch)

Total person-time, yEpoch 1 (1998–2003) N=16 781Epoch 2 (2004–2009) N=15 345Epoch 3 (2010–2015) N=15 949
300 758 176370 590 983402 748 397
EventsIncidence rateEventsIncidence rateEventsIncidence rate
All strokes2 777 3498.493 155 6837.972 805 6096.75
 Stroke survivors only2 125 2136.582 387 8256.052 207 4195.33
 Stroke decedents652 1361.91767 8581.91598 1901.42

* Standardized to the US population according to 2000 US census estimates.

† Presented as incidence per 1000 person years.

Estimates for the relationship between stroke and memory trajectories are shown in Table 3 and used to calculate predicted memory scores shown in Figures 2 and 3. Average memory scores in stroke-free participants at age 75 improved from −0.60 (95% CI, −0.65 to −0.55) points per year in epoch 1 to −0.57 (95% CI, 0.61 to −0.52) in epoch 2 and −0.46 (95% CI, −0.50 to −0.42) in epoch 3 (P value for trend: <0.01); memory decline for stroke-free individuals was similar across epochs at −0.07 points per year. For epoch 1 participants who subsequently had stroke, average prestroke memory score was −0.17 (95% CI, −0.24 to −0.09) points worse than stroke-free participants; in epoch 3, average prestroke memory score among stroke survivors was −0.11 (95% CI, −0.18 to −0.03) points worse than stroke-free individuals (P value for trend across epochs: <0.01). Among stroke decedents, average prestroke memory at age 75 in epoch 1 was 0.45 (95% CI, −0.59 to −0.31) points lower than stroke-free participants; −0.33 (95% CI, −0.45 to −0.20) points lower in epoch 2; and −0.38 (95% CI, −0.53 to −0.23) points lower in epoch 3 (P value for trend: <0.01). In sensitivity analyses restricted to direct memory assessments (no proxies), patterns of prestroke memory disadvantage compared to people who did not have stroke attenuated over time, matching primary analyses. The discontinuity at the time of stroke attenuated over time (P=0.025) though the temporal trend was less pronounced than in primary analyses (Table I in the Data Supplement).

Table 3. Covariate-Adjusted* Associations Between Stroke-Status and Memory Scores in Population-Weighted Mixed Linear Regression Models

Epoch 1 (1998–2004) N=16 781Epoch 2 (2004–2010) N=15 345Epoch 3 (2010–2016) N=15 949P value (linear trend across epochs)
β95% CIβ95% CIβ95% CI
Stroke-free participants
 Age, y−0.07−0.07−0.07−0.07−0.07−0.07−0.07−0.07−0.07<0.0001
Stroke survivors
 Age of stroke, y−0.10−0.10−0.09−0.09−0.10−0.09−0.09−0.10−0.09<0.0001
 Time until stroke, y−0.12−0.14−0.10−0.13−0.15−0.11−0.12−0.14−0.100.6342
 Discontinuity at time of stroke§−0.37−0.44−0.29−0.26−0.33−0.18−0.25−0.33−0.170.0196
 Time since stroke, y−0.08−0.11−0.06−0.13−0.15−0.10−0.10−0.13−0.080.1599
Stroke decedents
 Age of stroke, y−0.09−0.10−0.08−0.10−0.10−0.09−0.11−0.12−0.10<0.0001
 Time until stroke, y−0.21−0.25−0.17−0.18−0.21−0.14−0.20−0.25−0.160.0831

* Adjusted for gender, race, memory practice effects, marital status, wealth, education, height, birthplace, mother’s education.

† Average memory score in stroke-free participants at age 75 who are female, White race, married/partnered, not born in the Southern United States with 12 y of education, mother’s education of <8 y, and median income and height for that epoch.

‡ Indicator coefficients represent difference in average memory for this stroke group immediately compared to otherwise similar stroke-free individuals, assuming stroke occurred at age 75.

§ Discontinuity at time of stroke is a binary indicator for assessments taken after stroke representing short-term poststroke decrement in cognition associated with stroke onset.

Figure 2.

Figure 2. Trends in memory decline by epoch in the Health and Retirement Study, stratified by stroke status; solid curves represent trends in Epoch 1, dashed curves represent Epoch 2, and dotted curves represent Epoch 3. Time 0 indicates stroke onset for participants who experienced incident stroke at 75 y of age (survivors and decedents), or age 75 for (A) stroke-free participants. For (B) stroke survivors and (C) decedents, curves to the left of stroke onset indicate change in memory before stroke onset and curves to the right indicate change after stroke; B and C include stroke-free trends in light gray.

Figure 3.

Figure 3. Predicted memory scores for stroke-free participants and poststroke decline in stroke survivors. Time 0 indicates time of stroke onset for stroke survivors experiencing a stroke at age 75 or age 75 for stroke-free participants. A, Trajectories (dashed curves for stroke-free participants, dotted curves for stroke survivors) of decline at 0, 1, 2, and 3 y poststroke or postage 75. B, Trends in mean absolute difference and 95% CI between average memory scores of stroke survivors and stroke-free participants at 0, 1, 2, and 3 y poststroke or postage 75.

Figure 2 shows prestroke and poststroke decline in stroke survivors and prestroke decline in decedents. Memory decline rates before stroke for survivors was similar across epochs at −0.12 (95% CI, −0.14 to −0.10) points per year before stroke in epoch 1, −0.13 (95% CI, −0.15 to −0.11) in epoch 2, and −0.12 (95% CI, −0.14 to −0.10) in epoch 3 (P value for trend: 0.63). These coefficients indicate that over a 1-year period before stroke, we expect similar memory decline as 1.7 years among stroke-free participants. At the time of stroke, the immediate decrement in memory was −0.37 (95% CI, −0.44 to −0.29) points in epoch 1 but only −0.26 (95% CI, −0.33 to −0.18) points in epoch 2 and −0.25 (95% CI, −0.33 to −0.17) points in epoch 3 (P value for trend: 0.02). Following stroke, average annual memory decline was −0.08 (95% CI, −0.11 to −0.06) points per year in epoch 1, −0.13 (95% CI, −0.15 to −0.10) points in epoch 2, and −0.10 (95% CI, −0.13 to −0.08) points in epoch 3 (P value for trend: 0.16), or a 1-year poststroke decline being approximately equivalent to a 1.1-year decline in stroke-free participants. Before stroke for decedents, average annual memory decline was −0.21 (95% CI, −0.25 to −0.17) points in epoch 1, −0.18 (95% CI, −0.21 to −0.14) in epoch 2, and −0.20 (95% CI, −0.25 to −0.16) in epoch 3 (P value for trend: 0.08), or a 1-year decline before stroke in decedents being approximately equivalent to a 3.3-year decline in stroke-free participants.

Predicted memory scores after stroke onset in survivors and age-related decline in stroke-free participants are shown in Figure 3. Average memory scores for the referent 75-year-old stroke-free individual (−0.60 [95% CI, −0.65 to −0.55] points in epoch 1, −0.57 [95% CI, −0.61 to −0.52] points in epoch 2, and −0.46 [95% CI, −0.50 to −0.42] points in epoch 3) were higher than a similar individual experiencing a nonfatal stroke at age 75 across all epochs (Figure 3A). In epoch 1, absolute differences in memory decline between stroke-free participants and stroke survivors were stable across the three years following stroke (following age 75 for stroke-free participants). Differences between stroke-free and stroke survivors were narrower in epochs 2 and 3 than epoch 1; however, this attenuation was more pronounced 1-year poststroke, with an increasing difference between estimates observed 3-years poststroke (Figure 3B).


Consistent with previous research, patients with stroke had worse memory and faster prestroke decline than stroke-free participants in this large, nationally representative cohort. Across successive epochs, the prestroke memory disadvantage of people with stroke compared with people without stroke had attenuated. The short-term decrement in memory scores associated with stroke onset was also attenuated by 0.11 points across epochs, a 35% reduction in memory decrement (approximately equivalent to 2 years of age-related decline in stroke-free participants). Rate of poststroke memory change did not substantially change across epochs but for 1-year poststroke and to a lesser extent, 3 years poststroke, the gap between stroke-survivors and stroke-free individuals was attenuated.

Our findings are consistent with reduced incidence of stroke in recent years and reduced impact of stroke on memory impairment and related conditions. We observed a decrease in fatal and nonfatal stroke incidence rates across epochs, consistent with previous studies demonstrating downward trends in incidence of first stroke.25–28 The observed decreased incidence from epoch 1 to epoch 3 also mirrors reported decreases in stroke deaths from the US National Vital Statistics System, showing a 3.5% annual decline for national age-standardized stroke death rates from 2000 to 2015 among people aged 65 to 74 and age 75 to 84.8 Improved control of smoking and other vascular risk factors that contribute to stroke has been highlighted as potential contributors for concurrent declines in stroke incidence and subsequent mortality, but existing evidence on associations between vascular risk factors and poststroke dementia is conflicting.3

Our findings suggest that prestroke memory functioning for individuals who experience stroke is better in recent years than earlier periods. Explanations for this change are uncertain but may indicate that determinants of stroke are less strongly associated with memory than in the past. Recent decreases in stroke-related dementia are also likely partially attributable to smaller memory decrements in the immediate aftermath of stroke. The attenuation of short-term adverse consequences of stroke on memory observed in epochs 2 and 3 compared with epoch 1 may be due to improvements in acute stroke care across the last few decades as documented in Get With The Guidelines–Stroke initiatives,29 including wider adoption of tPA (tissue-type plasminogen activator), which is reported to have doubled from 2003 to 2011 and reduced door-to-needle times which improve the efficacy of this treatment.30,31 It is also possible that improved risk factor profiles in people at high risk of stroke—such as improved hypertension control, diabetes management, or increases in physical activity—led to decreases in stroke severity. This may have resulted in an increasing proportion of mild strokes or of subtypes with smaller cognitive consequences. Increases in detection of mild strokes would have similar consequences, but is unlikely given the overall decline in incidence rates. Reduced stroke recurrence would not have contributed to our results because we evaluated only first strokes.32,33

This study contributes significantly to our understanding of trends in poststroke dementia. Previous studies used retrospective samples of only stroke survivors or populations that may not be generalizable to the US population. A major strength of this study is that each epoch was aligned with study resampling, allowing us to mimic continuous monitoring over an extended period of observation corresponding with re-enrollments using consistent recruitment and study procedure methods. Considering prestroke and poststroke trajectories is critical in understanding progression of memory decline around stroke, as long-term outcomes of stroke survivors reflects both consequences of stroke and cognitive status of individuals before stroke. By examining how trajectories differ both before and after stroke onset, we may differentiate between improvements in age-related cognitive decline and changes in acute or poststroke care.

Our study aimed to evaluate temporal trends by emulating continuous monitoring in a nationally representative sample. Consistent with recent trends observed in the Framingham Study, we observed improvements in cognitive functioning after stroke in recent years.7 However, we found that improvements in poststroke memory functioning were driven primarily by lessening of immediate memory deficits at the time of stroke onset, not differences in memory decline around stroke onset. Furthermore, Framingham Study results compared time periods from 1977 to 2008; our results extend findings through 2016.

Important limitations of this study include self-report of stroke status and inability to differentiate between stroke subtypes or consider transient ischemic attacks. Our results reflect a weighted average effect of different stroke types and likely represent more common ischemic strokes. As with many self-report measures, self-reported strokes are subject to recall bias and overreporting such as possible misclassification of TIAs or other related cardiovascular conditions, the frequency of which may also differ across epochs.24,34 However previous investigations have suggested that misreporting of strokes in HRS is nonsystematic with a sensitivity of 74% and specificity of 93% to detect strokes recorded through Centers for Medicare and Medicaid Services.35 Though imperfect, self-reported strokes in HRS allowed us to estimate trends across the US in the absence of a national infrastructure for systematic stroke reporting.24 Second, we had insufficient data to explore potential mediation of secular trends by vascular risk factors and other comorbidities such as depression, known contributors to poststroke dementia, although previous reports have shown that vascular risk factors do not significantly predict cognitive change after stroke and, therefore, were unlikely to affect our findings.17,36 Third, our results are limited to the cognitive domain of verbal memory after stroke. Though our measure of immediate and delayed recall have been previously shown to be associated with diagnoses of poststroke dementia and can more easily be used to assess large population cohorts than full cognitive batteries, we were unable to determine whether temporal trends in other domains or overall cognition were present. Finally, follow-up time within epochs, available only in 2-year intervals with a maximum of 6 years, may have been insufficient to capture the full effects of stroke on subsequent memory decline. Through sensitivity analyses, we concluded that inclusion of follow-up outside of each epoch biased our results towards healthier participants, as those with lesser decline and better memory performance were more likely to remain eligible across epochs.

This work adds to current knowledge of stroke-related decline by considering potential trends in the relationship between stroke and cognition across almost 2 decades. In the HRS, stroke onset was associated with a smaller decrement in memory for stroke survivors in recent epochs (2004–2010 and 2010–2016) compared with stroke survivors in the first period we observed (1998–2004). Although improvements in quality of care for stroke survivors may explain trends in poststroke memory outcomes, further research with direct measures of quality of care and other possible mediators are needed. Future studies may consider the impact of specific changes in stroke care during this study period to further address possible underlying mechanisms behind these observed trends.

Nonstandard Abbreviations and Acronyms


Health and Retirement Study


tissue-type plasminogen activator

Supplemental Materials

Online Table I

Disclosures Dr Kim received funding from SanBio (TBI03). The other authors report no conflicts.


The Data Supplement is available with this article at

For Sources of Funding and Disclosures, see page 1710.

Correspondence to: Chloe W. Eng, MSPH, Department of Epidemiology and Biostatistics, University of California San Francisco, 550 16th St, San Francisco, CA 94158. Email


  • 1. Pendlebury ST, Rothwell PM. Prevalence, incidence, and factors associated with pre-stroke and post-stroke dementia: a systematic review and meta-analysis.Lancet Neurol. 2009; 8:1006–1018. doi: 10.1016/S1474-4422(09)70236-4CrossrefMedlineGoogle Scholar
  • 2. Merino JG, Hachinski V. Stroke-related dementia.Curr Atheroscler Rep. 2002; 4:285–290. doi: 10.1007/s11883-002-0007-5CrossrefGoogle Scholar
  • 3. Mijajlović MD, Pavlović A, Brainin M, Heiss WD, Quinn TJ, Ihle-Hansen HB, Hermann DM, Ben Assayag E, Richard E, Thiel A, et al. Post-stroke dementia - a comprehensive review.BMC Med. 2017; 15:11. doi: 10.1186/s12916-017-0779-7CrossrefMedlineGoogle Scholar
  • 4. Ivan CS, Seshadri S, Beiser A, Au R, Kase CS, Kelly-Hayes M, Wolf PA. Dementia after stroke: the Framingham Study.Stroke. 2004; 35:1264–1268. doi: 10.1161/01.STR.0000127810.92616.78LinkGoogle Scholar
  • 5. Elias MF, Sullivan LM, D’Agostino RB, Elias PK, Beiser A, Au R, Seshadri S, DeCarli C, Wolf PA. Framingham stroke risk profile and lowered cognitive performance.Stroke. 2004; 35:404–409. doi: 10.1161/01.STR.0000103141.82869.77LinkGoogle Scholar
  • 6. Feigin VL, Lawes CM, Bennett DA, Barker-Collo SL, Parag V. Worldwide stroke incidence and early case fatality reported in 56 population-based studies: a systematic review.Lancet Neurol. 2009; 8:355–369. doi: 10.1016/S1474-4422(09)70025-0CrossrefMedlineGoogle Scholar
  • 7. Satizabal CL, Beiser AS, Chouraki V, Chêne G, Dufouil C, Seshadri S. Incidence of dementia over three decades in the Framingham Heart Study.N Engl J Med. 2016; 374:523–532. doi: 10.1056/NEJMoa1504327CrossrefMedlineGoogle Scholar
  • 8. Yang Q, Tong X, Schieb L, Vaughan A, Gillespie C, Wiltz JL, King SC, Odom E, Merritt R, Hong Y, et al. Vital signs: recent trends in stroke death rates — United States, 2000–2015.MMWR Morb Mortal Wkly Rep. 2017; 63:933. doi: 10.15585/mmwr.mm6635e1CrossrefGoogle Scholar
  • 9. Brainin M, Tuomilehto J, Heiss WD, Bornstein NM, Bath PM, Teuschl Y, Richard E, Guekht A, Quinn T; Post Stroke Cognition Study Group. Post-stroke cognitive decline: an update and perspectives for clinical research.Eur J Neurol. 2015; 22:229–38, e13. doi: 10.1111/ene.12626CrossrefMedlineGoogle Scholar
  • 10. Wang Q, Capistrant BD, Ehntholt A, Glymour MM. Long-term rate of change in memory functioning before and after stroke onset.Stroke. 2012; 43:2561–2566. doi: 10.1161/STROKEAHA.112.661587LinkGoogle Scholar
  • 11. Wang Q, Mejía-Guevara I, Rist PM, Walter S, Capistrant BD, Glymour MM. Changes in memory before and after stroke differ by age and sex, but not by race.Cerebrovasc Dis. 2014; 37:235–243. doi: 10.1159/000357557CrossrefGoogle Scholar
  • 12. Levine DA, Galecki AT, Langa KM, Unverzagt FW, Kabeto MU, Giordani B, Wadley VG. Trajectory of cognitive decline after incident stroke.JAMA. 2015; 314:41–51. doi: 10.1001/jama.2015.6968CrossrefMedlineGoogle Scholar
  • 13. Elkins JS, Knopman DS, Yaffe K, Johnston SC. Cognitive function predicts first-time stroke and heart disease.Neurology. 2005; 64:1750–1755. doi: 10.1212/01.WNL.0000161850.01792.77CrossrefMedlineGoogle 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.xCrossrefMedlineGoogle Scholar
  • 15. Juster FT, Suzman R. An overview of the health and retirement study.J Hum Resour1995; 30:S7–S56.CrossrefGoogle Scholar
  • 16. Heeringa SG, Connor JH. Technical description of the Health and Retirement Survey sample design.Ann Arbor, MI; 1995. Scholar
  • 17. Sachdev PS, Lipnicki DM, Crawford JD, Wen W, Brodaty H. Progression of cognitive impairment in stroke/TIA patients over 3 years.J Neurol Neurosurg Psychiatry. 2014; 85:1324–1330. doi: 10.1136/jnnp-2013-306776CrossrefMedlineGoogle Scholar
  • 18. Jorm AF, Christensen H, Korten AE, Jacomb PA, Henderson AS. Informant ratings of cognitive decline in old age: validation against change on cognitive tests over 7 to 8 years.Psychol Med. 2000; 30:981–985. doi: 10.1017/s0033291799002299CrossrefMedlineGoogle Scholar
  • 19. Jorm AF. A short form of the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE): development and cross-validation.Psychol Med. 1994; 24:145–153. doi: 10.1017/s003329170002691xCrossrefMedlineGoogle Scholar
  • 20. Ofstedal M, Fisher G, Herzog A. Documentation of Cognitive Functioning Measures in the Health and Retirement Study.HRS Doc. Rep. DR-006 [online]. 2005. Scholar
  • 21. Sonnega A, Faul JD, Ofstedal MB, Langa KM, Phillips JW, Weir DR. Cohort Profile: the Health and Retirement Study (HRS).Int J Epidemiol. 2014; 43:576–585. doi: 10.1093/ije/dyu067CrossrefMedlineGoogle Scholar
  • 22. Wu Q, Tchetgen Tchetgen EJ, Osypuk TL, White K, Mujahid M, Maria Glymour M. Combining direct and proxy assessments to reduce attrition bias in a longitudinal study.Alzheimer Dis Assoc Disord. 2013; 27:207–212. doi: 10.1097/WAD.0b013e31826cfe90CrossrefMedlineGoogle Scholar
  • 23. Langa KM, Plassman BL, Wallace RB, Herzog AR, Heeringa SG, Ofstedal MB, Burke JR, Fisher GG, Fultz NH, Hurd MD, et al. The aging, demographics, and memory study: study design and methods.Neuroepidemiology. 2005; 25:181–191. doi: 10.1159/000087448CrossrefMedlineGoogle Scholar
  • 24. Glymour MM, Avendano M. Can self-reported strokes be used to study stroke incidence and risk factors?: evidence from the health and retirement study.Stroke. 2009; 40:873–879. doi: 10.1161/STROKEAHA.108.529479LinkGoogle Scholar
  • 25. Koton S, Schneider AL, Rosamond WD, Shahar E, Sang Y, Gottesman RF, Coresh J. Stroke incidence and mortality trends in US communities, 1987 to 2011.JAMA. 2014; 312:259–268. doi: 10.1001/jama.2014.7692CrossrefMedlineGoogle Scholar
  • 26. Koton S, Schneider AL, Rosamond WD, Gottesman RF, Coresh J. Twenty-five years trends in stroke incidence in the atherosclerosis risk in communities (ARIC) study.Circulation. 2014; 129:AP214–AP214.LinkGoogle Scholar
  • 27. Sacco RL, Dong C. Declining stroke incidence and improving survival in US communities: evidence for success and future challenges.JAMA. 2014; 312:237–238. doi: 10.1001/jama.2014.7693CrossrefMedlineGoogle Scholar
  • 28. Carandang R, Seshadri S, Beiser A, Kelly-Hayes M, Kase CS, Kannel WB, Wolf PA. Trends in incidence, lifetime risk, severity, and 30-day mortality of stroke over the past 50 years.JAMA. 2006; 296:2939–2946. doi: 10.1001/jama.296.24.2939CrossrefMedlineGoogle Scholar
  • 29. Smaha LA; American Heart Association. The American Heart Association Get with the guidelines program.Am Heart J. 2004; 148(5 Suppl):S46–S48. doi: 10.1016/j.ahj.2004.09.015CrossrefMedlineGoogle Scholar
  • 30. Schwamm LH, Ali SF, Reeves MJ, Smith EE, Saver JL, Messe S, Bhatt DL, Grau-Sepulveda MV, Peterson ED, Fonarow GC. Temporal trends in patient characteristics and treatment with intravenous thrombolysis among acute ischemic stroke patients at Get With The Guidelines-Stroke hospitals.Circ Cardiovasc Qual Outcomes. 2013; 6:543–549. doi: 10.1161/CIRCOUTCOMES.111.000303LinkGoogle Scholar
  • 31. Virani SS, Alonso A, Benjamin EJ, Bittencourt MS, Callaway CW, Carson AP, Chamberlain AM, Chang AR, Cheng S, Delling FN, et al; American Heart Association Council on Epidemiology and Prevention Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics-2020 update: a report from the American Heart Association.Circulation. 2020; 141:e139–e596. doi: 10.1161/CIR.0000000000000757LinkGoogle Scholar
  • 32. Marshall IJ, Wang Y, McKevitt C, Rudd AG, Wolfe CD. Trends in risk factor prevalence and management before first stroke: data from the South London Stroke Register 1995-2011.Stroke. 2013; 44:1809–1816. doi: 10.1161/STROKEAHA.111.000655LinkGoogle Scholar
  • 33. Thiele I, Linseisen J, Heier M, Holle R, Kirchberger I, Peters A, Thorand B, Meisinger C. Time trends in stroke incidence and in prevalence of risk factors in Southern Germany, 1989 to 2008/09.Sci Rep. 2018; 8:11981. doi: 10.1038/s41598-018-30350-8CrossrefMedlineGoogle Scholar
  • 34. Woodfield R, Sudlow CL; UK Biobank Stroke Outcomes Group; UK Biobank Follow-up and Outcomes Working Group. Accuracy of patient self-report of stroke: a systematic review from the UK biobank stroke outcomes group.PLoS One. 2015; 10:e0137538. doi: 10.1371/journal.pone.0137538CrossrefGoogle Scholar
  • 35. Osypuk TL, Ehntholt A, Moon JR, Gilsanz P, Glymour MM. Neighborhood differences in post-stroke mortality.Circ Cardiovasc Qual Outcomes. 2017; 10:e002547.LinkGoogle Scholar
  • 36. del Ser T, Barba R, Morin MM, Domingo J, Cemillan C, Pondal M, Vivancos J. Evolution of cognitive impairment after stroke and risk factors for delayed progression.Stroke. 2005; 36:2670–2675. doi: 10.1161/01.STR.0000189626.71033.35LinkGoogle Scholar


eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.