Abstract
Keywords
INTRODUCTION
As people age, there is a decline in physical capability often accompanied by a decline in cognitive function. Together, the two conditions pose a great challenge to individual quality of life and public health. Physical function and cognition may be interrelated, as both may involve the central nervous system and share some common age-related mechanisms [1–3]. Indeed, individuals who are cognitively impaired have poorer physical capabilities compared to cognitively intact persons [4, 5]. Moreover, motor dysfunction in persons with Alzheimer’s disease (AD) is common [6, 7].
Using tests of handgrip strength and walking speed as surrogates of physical capability and muscle strength has gained considerable attention in recent years. Handgrip strength and gait speed are excellent markers of both current health and future health outcomes [8] including disability [9, 10] and mortality [11]. Importantly, they have the advantages of being quick, simple, and inexpensive, and can be administered with minimal training and equipment in research and clinical settings.
Poor grip strength and slow gait speed emerge as potential novel risk factors for dementia and AD, as recent findings demonstrate an increased risk of future dementia and AD in cognitively intact individuals who perform poorly on those tests [12–14]. Thus, in parallel to the well-characterized changes in brain structure and function, the long prodromal phase of dementia and AD [15] may also be characterized by changes in physical capability and muscle strength. Nevertheless, to our knowledge, the relationship of grip strength and gait speed with subclinical measures of cognition and brain disease in cognitively intact individuals has not been studied.
Data on the relationship between physical function and risk of stroke is scarce [16, 17]. Stroke and dementia share common mechanisms and risk factors, and it is increasingly recognized that stroke may also have a preclinical phase of subclinical vascular brain dysfunction [18]. While changes in cognition [18, 19] and brain structure [18] have been demonstrated years prior to stroke occurrence, it is yet unknown whether changes in measures of grip strength and gait speed are also implicated.
In the current study, we used data from the Framingham study to further explore the role of physical function as independent risk factors for cognitive outcomes. Thus, we assessed the relationships of handgrip strength and fast-paced gait speed with the risk of dementia, AD, and stroke, and tested whether they improve the predictive value of previously validated dementia and stroke risk scores. In addition, the cross-sectional associations with subclinical cognitive and brain MRI measures were assessed.
METHODS
Study sample
The Framingham Offspring Cohort was recruited in 1971 and has been examined nine times, once every 4 years. This cohort consists of 5,124 children (and their spouses) of the Original Framingham cohort and have been previously described [20].
Of the 5,124 enrolled Offspring, 3,539 survivors attended the 7th examination cycle between 1998 and 2002. These participants were invited to attend a call-back visit for detailed brain magnetic resonance imaging (MRI) testing, neuropsychological assessment, and testing of walking speed and of handgrip strength, all on the same day. Call-back visits were performed between 1999 and 2005, and 2,433 Offspring participants aged 35–84 years were able to complete neuropsychological assessment and testing of walking speed and of handgrip strength. For the purposes of the current sub-study, 257 subjects were excluded due to prevalent dementia, stroke, known neurological conditions that would confound cognitive and/or motor testing (e.g., brain tumors, multiple sclerosis, hydrocephalus, sarcoidosis, Lyme disease, or a history of head trauma severe enough to produce loss of consciousness for >30 min), or if there was missing covariate information. This yielded a primary sample of 2,176 subjects on whom cross sectional analyses of cognitive and motor testing were performed. Of these, a subset of 1,936 subjects also underwent detailed brain MRI. Participants in the primary sample were subsequently followed longitudinally for up to 11 years for assessment of stroke, transient ischemic attack (TIA) and dementia, our outcomes of interest. After exclusions for lack of pertinent follow-up data, incident dementia was assessed in 2,046 participants and incident stroke/TIA was assessed in 2,149 (Supplementary Figure 1). All subjects provided informed consent, and the research was approved by the institutional review board at Boston University Medical Campus/ Boston Medical Center.
Motor testing
Motor testing, including walking speed and handgrip strength, was administered by trained examiners who taught subjects how to perform the motor tasks prior to their final evaluation. Persons who required a walker or cane and those who were unable to perform the handgrip task due to arthritis remained in the overall sample if they performed at least one of the two tasks, but were excluded from analyses for the specific test that they had been unable to perform.
Fast walking speed
Subjects were asked to walk down a measured 4-meter walkway on a firm and even surface, while wearing comfortable shoes or socks. Subjects were instructed to walk as fast as possible but were not allowed to jog or run. The measured time to walk the entire 4 meters course was inverted and multiplied by 4 to obtain walking speed in meters/second (m/s).
Handgrip strength
Handgrip strength was tested using a Jamar Hydraulic Hand Dynamometer (Lafayette Instrument, Lafayette, IN), and measured in kilograms of force exerted. Subjects were seated comfortably with the forearm resting on a table and bent 90 degrees at the elbow. After instructions and coaching, subjects exerted their maximum hand force to the dynamometer for 5 seconds. This was repeated three times for each hand. The highest measurement out of six was recorded as the final handgrip strength value, irrespective of manual dominance.
End points
Our clinical outcomes of interest were incident dementia, AD, and incident TIA or stroke on longitudinal follow-up. Median follow-up was 6.5 years for dementia and 8 years for stroke/TIA. All participants in the Framingham study are under periodic surveillance for impairment in cognitive function, dementia, stroke, and TIA. Surviving Offspring were followed up with history, physical examination, neurologic examination, and administration of the Folstein Mini-Mental State Examination [21]. Assessment of dementia and stroke/TIA also included self, physician, or family referral; telephone health status updates; or records linkage. For the present analyses, data for incident dementia and AD collected until November 2011 were used, and for incident stroke/TIA data obtained until December 2009 were used.
Dementia
Participants who were suspected to have possible cognitive decline based on the Mini-Mental State Examination score or additional health status information underwent in-depth assessment that included neurological and neuropsychological evaluations. We determined whether each person fulfilled criteria for a diagnosis of dementia. The probable date of onset and type of dementia was determined at a consensus review by a panel composed of at least one behavioral neurologist and one neuropsychologist who reviewed all available records including examinations by the Framingham Heart Study investigators, hospital and nursing home records, data from structured family interviews, imaging, and when available, autopsy data. Participants with dementia met criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition), and were required to have symptoms for at least 6 months [22]. Participants with AD met the criteria for definite, probable, or possible AD from the National Institute of Neurological Diseases and Stroke and the Alzheimer’s Disease and Related Disorders Association [23]. More details on dementia ascertainment in the Framingham study can be found elsewhere [24].
Stroke/TIA
Participants who were suspected to have possible TIA or stroke based on their physical examination or additional health status information as above underwent detailed assessment that typically included neurological evaluations. We determined whether each person fulfilled criteria for a diagnosis of stroke or TIA. The date of onset and stroke/TIA mechanism were determined at a consensus review by a panel composed of at least two clinical neurologists with expertise in cerebrovascular disorders who reviewed all available records including examinations by the Framingham Heart Study investigators, hospital and nursing home records, data from structured family interviews, imaging, and when available, autopsy data. Stroke or TIA was defined by clinical criteria, when the acute onset of a focal neurological deficit of vascular etiology occurred. In the case of stroke, deficit (s) lasted for ≥24 hours, whereas for TIA, deficits lasted <24 hours.
Subclinical endpoints
Brain MRI
Brain MRI imaging methods have been previously described elsewhere [25, 26]. Subjects underwent brain MRI at the Framingham Heart Study facilities using Siemens Magnetom (Munich, Germany) 1-Tesla or 1.5-Tesla field strength machines. Double spin-echo coronal imaging sequences of 4-mm contiguous slices from nasion to occiput were acquired. For post-processing, imaging data were transferred to an MRI reading center at the University of California at Davis and analyzed by trained operators blinded to the subject’s identity, age, sex, exposure to dementia and stroke risk factors, and performance on neuropsychologic and motor testing. Analyses were performed using a custom-designed image analysis package (QUANTA 6.2) operating on a Sun Microsystems Ultra 5 workstation (Sun Microsystems, Santa Clara, CA) [27].
Brain MRI measures
Brain MRI measures analyzed were total cerebral brain volume, hippocampal volume, white matter hyperintensity volume, and lateral ventricular volume. Brain volume was determined by manual outlining of coronal images of the intracranial vault above the tentorium to determine the total cranial volume as a measure of head size. Once the skull and other non-brain tissues were removed from the image, mathematic modeling was performed to determine total parenchymal brain volume above the tentorium (cerebral). Total cerebral brain volume was computed as a ratio of the total brain parenchymal volume to total intracranial volume, thus correcting for differences in head size. Hippocampal volume was estimated using operator-defined, manually traced boundaries in coronal sections to define the regions of interest. White matter hyperintensity volume was measured according to previously published methods [28]. Specifically, for segmentation of white matter hyperintensities from other brain tissues, the first and second echo images from T2 sequences were summed and a log normal distribution was fitted to the summed data (after removal of cerebral spinal fluid and correction of image intensity non-uniformities). A segmentation threshold for white matter hyperintensity was determined as 3.5 standard deviations (SD) in pixel intensity greater than the mean of the fitted distribution of brain parenchyma, and volumes were expressed as a proportion of the total intracranial volume. Lateral ventricular volume was calculated as the sum of manually outlined coronal volumes of the right and left lateral ventricles and expressed as a proportion of the total intracranial volume.
Cognitive testing
A standardized neuropsychological test battery was administered to all study subjects by trained examiners on the same day of brain MRI testing in 98% of the subjects. The cognitive domains tested and tests administered were: a) Verbal and visual memory: Wechsler Memory Scale (Logical Memory, Visual Reproduction, Verbal Paired Associates); b) Abstraction: Wechsler Adult Intelligence Scale Similarities Test; c) Language: Boston Naming Test; d) Executive function: Halstad-Reitan Trailmaking Parts A and B; and d) Visuoperceptual function: Hooper Visual Organization Test. All cognitive tests were audiotaped and samples reviewed randomly by certified neuropsychologists to guarantee the quality of test battery administration. All cognitive variables were age- and education-standardized before analysis.
Statistical analysis
For statistical analyses, walking speed and handgrip strength were treated as continuous variables, and standardized to z-scores within 10-year age categories. This was done given the known decline with age in walking speed and handgrip strength. Additional analyses were performed using cutoffs: for walking speed we used the lowest 5th percentile versus the rest (walking speed ≤1 m/s versus >1 m/s) and for handgrip strength we used ≤10th sex-specific percentile versus >10th percentile (corresponding to cut-offs of 15 kg in women and 30 kg in men).
Cox proportional hazard models were constructed, to examine the relationships of walking speed and handgrip strength with incident dementia and incident stroke/TIA. Given the possibility that associations might be different in the elderly compared to the younger <65 years of age participants, we re-examined the associations between walking speed and handgrip strength with dementia, AD and stroke risk in those aged 65 years or more. Due to small number of events, these associations could not be examined among a younger subsample.
In order to test whether the relationships of walking speed and grip strength with incident dementia and AD are cause or consequence, we ran a post-hoc analysis excluding individuals who developed dementia (n = 8), died, or were lost to follow-up within the first 3 years from baseline. Another post-hoc analysis was performed estimating integrated discrimination improvement (IDI), using continuous Cox proportional hazards based models [29] to estimate the value of adding handgrip strength or gait speed to a previously validated dementia screening indicator [30]. The dementia screening indicator is a brief test for use in primary care settings, which aims to identify individuals at elevated risk for incident dementia [30].
To assess the relationships between walking speed and handgrip strength and subclinical brain disease in subjects free of clinical dementia or stroke, we constructed multivariable linear and logistic regression models looking at the associations between walking speed, handgrip strength, and standardized cognitive function, and between walking speed, handgrip strength, and brain MRI measurements. These analyses were also conducted among a younger subsample aged <65 years.
All models were adjusted for age and sex and for factors previously associated with risk of dementia and of stroke/TIA, and which have also been shown to affect brain volume and cognitive function in the Framingham Original and Offspring cohorts [31]. Covariates included in the models were: age, sex, prevalent diabetes mellitus, prevalent cardiovascular disease (including coronary heart disease, congestive heart failure and intermittent claudication), prevalent atrial fibrillation, current smoking, and homozygosity or heterozygosity for apolipoprotein ɛ4 allele, treated as categorical variables; and systolic blood pressure, waist-to-hip ratio, total cholesterol level, total homocysteine levels, and physical activity index [32], treated as continuous variables. Adjustments for education were additionally conducted when associations with cognitive performance were examined. Statistical analyses were performed using SAS statistical software version 9.2 (SAS Institute Inc, Cary, North Carolina) and all statistical tests were 2-sided. A p value of less than 0.05 was used to indicate statistical significance.
RESULTS
Subjects in our study sample were relatively young (mean age 62±9 years, range 35–84 years) with a slight predominance of women (54%). Compared to Offspring cohort participants who attended exam 7 but did not have cognitive data, individuals included in the current sample were on average significantly younger (p < 0.001), more educated (p = 0.003), and were less likely to smoke (p < 0.001) or to have prevalent diabetes (p < 0.001), atrial fibrillation (p = 0.001), and cardiovascular disease (p < 0.001). In addition, the current study sample had on average smaller waist-to-hip ratio (p = 0.003), lower total homocysteine (p < 0.001), and systolic blood pressure (p < 0.001), but higher total cholesterol (p = 0.022) compared to the rest with no cognitive data (Table 1). During a follow-up period of up to 11 years (median 6.5 years), 40 participants developed incident dementia (34 of whom had AD) and 79 developed incident stroke/TIA (eight had hemorrhagic strokes). Mean handgrip strength was 33±13 Kg, and mean walking speed was 1.7±0.4 m/s.
Association of fast walking speed and handgrip strength with incident dementia and AD
Table 2 demonstrates the associations between walking speed, handgrip strength, and incident all-cause dementia and AD. For every one SD decrease in walking speed we noted a significant 76% increase (p = 0.004) in the risk for dementia in the entire study sample, and a 91% increase (p = 0.003) in the risk for dementia in participants aged ≥65 years. Similarly significant effects of walking speed on the risk for incident AD were noted. For every one SD decrease in walking speed there was a 68% increase in the risk of AD in the entire study sample (p = 0.014), and a 78% increased risk of AD in those restricted to age ≥65 years (p = 0.012). Subjects with the slowest walking speed (≤1 m/s) had a nearly three-fold increased risk of dementia (HR 2.53 [95% CI 1.11–5.74, p = 0.027] for the entire study sample, and HR 2.72 [95% CI 1.15–6.41, p = 0.023] for those restricted to age ≥65 years) and of AD (HR 2.92 [95% CI 1.19–7.14, p = 0.019] for the entire study sample, and HR 2.98 [95% CI 1.19–7.47, p = 0.020] for those restricted to age ≥65 years).
Decreased handgrip strength did not show a continuous, graded association with risk of incident dementia or AD but as seen in Table 2, handgrip strength in the lowest 10% (≤15 kg in women and ≤30 kg in men) was significantly associated with the risk of incident dementia (HR 2.17 [95% CI 1.00–4.69, p = 0.05]) and AD HR 2.75 [95% CI 1.18–6.39, p = 0.019)] in the entire study sample, and HR 2.38 [95% CI 1.05–5.39, p = 0.037] for dementia and 3.22 [95% CI 1.31–7.90, p = 0.011] for AD in those aged ≥65 years.
After excluding participants who developed dementia, died, or were lost during the first 3 years of follow-up, the associations of slow walking speed with high risk of dementia and AD attenuated and were no longer statistically significant. However, weak grip strength remained strongly associated with these outcomes (HR 2.69 [95% CI 1.17–6.18, p = 0.02]) and HR 4.10 [95% CI 1.65–10.18, p = 0.002)] for incident dementia and AD, respectively (Supplementary Table 1).
Adding information on walking speed (≤1 m/s) significantly improved the risk prediction assessed by the dementia screening indicator, as indicated by IDI [95% CI] of 0.062 [0.040, 0.106] for dementia and 0.065 [0.041, 0.114] for AD. Similarly, adding handgrip strength (below or above the sex-specific 10th percentile) to the previously validated screening indicator significantly improved the risk prediction (IDI [95% CI] = 0.037 [0.023, 0.064] and 0.042 [0.026, 0.080] for dementia and AD, respectively (data not tabulated).
Association of fast walking speed and handgrip strength with incident stroke/TIA
Table 3 demonstrates the associations between walking speed, handgrip strength, and stroke/TIA. For every 1 standard-deviation (SD) decrease in handgrip strength we noted a significant increase of 61% (p = 0.036) in the risk for stroke/TIA in participants aged ≥65 years. However, a similar significant relationship was not noted in the overall group including all ages. Slower walking speed was not significantly associated with the risk of incident stroke/TIA.
Association of fast walking speed and handgrip strength with performance on cognitive testing
As seen in Table 4, in people free of clinical dementia and stroke, for each SD decrease in walking speed there was a significant decrease in the performance on tests of visual and verbal memory, language, executive function, and visuoperceptual function. Similarly, for each SD decrease in handgrip strength there was a significant decrease in the performance on tests of visual memory, abstraction, language, executive function, and visuoperceptual function. Among individuals at the ages of 65 years or younger, each SD decrease in walking speed was associated with poorer performance on an attention test (Trail making A), and decreased handgrip strength was related to poorer logical and visual memory, language, and visuospatial perception.
Association of fast walking speed and handgrip strength with brain MRI measures
As seen in Table 5, in people free of clinical dementia and stroke, for each age-adjusted SD decrease in walking speed and in handgrip strength there was a significant decrease in total cerebral brain volume. No significant associations were noted between either walking speed or handgrip strength and hippocampal volumes, ventricular volumes or white matter hyperintensity volumes. Even among individuals who were 65 years of age or younger, those with poorer handgrip strength had on average smaller total cerebral brain volume.
DISCUSSION
In our study of a large middle-aged community sample, we have shown that slower walking speed was associated with an increased risk of incident dementia and AD over a median follow-up period of 6.5 years. Furthermore, in disease-free subjects slower walking speed was associated with decreased total brain volume and poorer performance on cognitive testing, thus proving to be a surrogate marker of subclinical brain disease. Similarly, weaker handgrip strength predicted an increased risk of incident stroke/TIA in subjects aged ≥65 years of age and for persons in the lowest 10 percentile of handgrip strength, increased risk of incident dementia and AD in the entire study sample and in subjects aged ≥65 years. Weaker handgrip strength was also associated with decreased total cerebral brain volume and poorer performance on cognitive testing, even among a subsample of individuals below the age of 65 years. Our data support those of other studies which have implicated walking speed and handgrip strength as markers of functional decline in older age.
Age-related gait dysfunction can be due to multiple conditions such as orthopedic, neuromuscular, vestibular, and visual disorders. Of importance, age-related gait changes may be a manifestation of compromised motor executive function, such as is seen in patients with normal pressure hydrocephalus, extensive leukoaraiosis, or multi-infarct dementia [33]. Indeed, it has been shown that white matter hyperintensity volume and decreased executive function are risk factors for age-related gait abnormalities in the elderly [34]. Cerebrovascular mediation of gait has also been implicated, as common carotid intima media thickness (a marker of prolonged exposure to vascular risk factors), and a larger number of carotid plaques have been shown to be associated with gait impairment in the elderly [35]. An additional pathophysiologic contributor to a slowed walking speed may be chronic inflammation [36–38] which is directly linked to impaired mobility and chronic disease. External factors may also play a role in determination of impaired mobility and health status, such as lower socioeconomic status, decreased access to social services, social isolation, sedentary behavior, and residence in neighborhoods in which there is little green space, often where impoverished population groups reside [39].
Decreased walking speed has been associated with increased risk of stroke in postmenopausal women from the Women Health Initiative [17] and with fatal and non-fatal strokes in subjects >65 years of age in the Cardiovascular Health Study [16]. Additionally, decline in mobility over time in the elderly, as determined by progressive decrease in walking speed, is also correlated with an increase in risk for all-cause mortality [40]. Slower walking speed also predicted functional dependence in an elderly Japanese rural community [41]. In the Einstein Aging Study, slower walking speed in the presence of cognitive complaints was predictive of dementia, especially vascular dementia, in elderly non-demented individuals [42]. These results have biologic plausibility, since optimal walking speed implies the neurological ability to coordinate complex motor tasks, which is mediated through preserved executive functioning and cortical and subcortical integrity. We also expected an association between white matter hyperintensity volume and slower walking speed, which was not apparent in our population. The lack of an association is likely due to the fact that our study sample is relatively healthy with very small amounts of white matter hyperintensity on brain MRI and low sample variability, which may have rendered our study underpowered to detect an association [43].
Handgrip strength is a validated and easy to use bedside tool used to assess muscle strength of the upper extremities [44]. Consistent evidence suggests that handgrip strength is a strong predictor of health-related prognosis [45]. Low handgrip strength has been shown to be associated with post-operative outcomes [46], and mid-life handgrip strength is a strong predictor of functional limitation and disability [45, 48] even 25 years later [45, 47]. In addition, handgrip strength is associated with all-cause mortality [45, 49–51] as well with mortality from heart disease and stroke [49]. Only limited data are available on the predictive value of handgrip strength in relation to cognitive outcomes. In two studies [52, 53], the Mini-Mental State Examination was used, and it was not clear whether a decline in muscle strength preceded the cognitive outcome or vice versa. In the Religious Orders study, handgrip strength [12] as well as a more comprehensive measure of muscle strength [13] were associated with incident AD in a cohort of ∼900 elderly persons free of dementia at baseline. In the Honolulu Heart Program, higher handgrip strength in Japanese middle-aged men was associated with exceptional survival, free of major morbid diseases and cognitive impairment after 40 years of follow-up [54]. In our sample we also found an association between the lowest 10% of handgrip strength and risk of incident AD, although we did not find a continuous, graded association with risk of AD as we did for the risk of stroke. Additionally, our findings suggest that middle-aged people with low handgrip strength may demonstrate subsequent poorer cognitive function. These people may have a higher risk of developing clinical dementia and AD as they age [55, 56].
While walking speed is a measure of strength as well as balance and motor control, handgrip strength is a more specific measure of physical capability and is a good indicator of muscle mass. In turn, reduction in muscle mass may result from aging, co-morbidities, and decrease in hormone balance and regulation [57], and thus may indicate poorer cardiovascular profile which in our results is reflected by an association with vascular rather than neurodegenerative outcomes and by the fact that the association with stroke was restricted to people ≥65 years. Our findings of an association between low handgrip strength and decreased total cerebral brain volume may suggest a possible metabolic dysfunction that co-occurs in brain and muscle.
According to our study, adding either walking speed or handgrip strength to a previously validated screening indicator improves the risk prediction significantly. Hence, these measures may serve as promising tools for detecting high-risk individuals. Nevertheless, unlike walking speed, the association of handgrip strength with incident dementia and AD remained robust after exclusion of participants who developed dementia or died during the first years of follow-up. Furthermore, as opposed to walking speed, weak handgrip strength was associated with poor performance on many cognitive domains and with decreased total brain volume even among participants aged 65 years or younger. These findings suggest that measures of handgrip strength compared to walking speed are less influenced by pre-clinical stages of dementia, and can be used as risk prediction markers in a wider population age-range.
This study has several limitations. First, because participants in this study were relatively young, the number of events, especially dementia and AD is small. Regardless, significant associations with these outcomes were detected, denoting the strength of our results. Second, as an observational study, causality cannot be assumed. In addition, we cannot exclude the possibility that poor cognitive function and brain atrophy preceded slow walking speed or low handgrip strength. Although risks of dementia and stroke were assessed prospectively in participants free of these conditions at baseline, it is still possible that sub-clinical changes occurring before the clinical manifestation of dementia and stroke influenced walking speed and handgrip strength. Third, residual confounding may occur and underlying factors that may influence both the dependent and the outcome variables such as early life nutrition could not be accounted for. Lastly, the overwhelmingly European origin of the study sample limits the generalizability of our results. Strengths of this study include its population-based design with large sample size and long follow-up periods, a comprehensive cognitive battery and volumetric MRI measures performed on the same day, careful surveillance for clinical outcomes and detailed covariate information.
CONCLUSIONS
In conclusion, our findings suggest that middle-aged people who demonstrate slow walking speed on a test of maximal pace may be prone to higher risk of dementia while those with low handgrip strength may be at higher risk of stroke and in those with the lowest handgrip strength, also an increased risk of AD. In addition, both slow walking speed and low handgrip strength may be associated with poorer cognitive performance and with smaller total cerebral brain volume. These findings emphasize the need for interventions targeted to improve physical health in mid-life. Walking speed and handgrip strength are also simple tools for use in clinic: they are fast, inexpensive, and non-invasive. If our findings are confirmed, these measures may serve as additional tools in screening for people with high risk of stroke or dementia even in middle-age. Indeed, our findings suggest that each of those measures can improve the predictive validity of previously developed screening tools for primary care settings. Hence, it may be important to consider these measures when designing a risk score for dementia or stroke and for the development of interventions geared towards improved physical and cognitive health.
Footnotes
ACKNOWLEDGMENTS
This work was supported by the Framingham Heart Study’s National Heart, Lung, and Blood Institute contract (N01-HC-25195) and by grants from the National Institutes of Health, National Institute of Neurologic Disorders and Stroke (R01-NS-17950) and from the National Institute on Aging (R01-AG-008122; AG-016495; AG-033193; AG-031287, P30-AG-013846, U01 AG-049505).
The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
