A Longitudinal Study on the Association of Interrelated Factors Among Frailty Dimensions,Cognitive Domains,Cognitive Frailty,and All-Cause Mortality

Abstract

Background:

Cognitive frailty integrating impaired cognitive domains and frailty dimensions has not been explored.

Objective:

This study aimed to explore 1) associations among frailty dimensions and cognitive domains over time and 2) the extended definitions of cognitive frailty for predicting all-cause mortality.

Methods:

This four-year cohort study recruited 521 older adults at baseline (2011–2013). We utilized 1) generalized linear mixed models exploring associations of frailty dimensions (physical dimension: modified from Fried et al.; psychosocial dimension: integrating self-rated health, mood, and social relationship and support; global frailty: combining physical and psychosocial frailty) with cognition (global and domain-specific) over time and 2) time-dependent Cox proportional hazard models assessing associations between extended definitions of cognitive frailty (cognitive domains-frailty dimensions) and all-cause mortality.

Results:

At baseline, the prevalence was 3.0% for physical frailty and 37.6% for psychosocial frailty. Greater physical frailty was associated with poor global cognition (adjusted odds ratio = 1.43–3.29, β: –1.07), logical memory (β: –0.14 to –0.10), and executive function (β: –0.51 to –0.12). Greater psychosocial frailty was associated with poor global cognition (β: –0.44) and attention (β: –0.15 to –0.13). Three newly proposed definitions of cognitive frailty, “mild cognitive impairment (MCI)-psychosocial frailty,” “MCI-global frailty,” and “impaired verbal fluency-global frailty,” outperformed traditional cognitive frailty for predicting all-cause mortality (adjusted hazard ratio = 3.49, 6.83, 3.29 versus 4.87; AIC = 224.3, 221.8, 226.1 versus 228.1).

Conclusion:

Notably, extended definitions of cognitive frailty proposed by this study better predict all-cause mortality in older adults than the traditional definition of cognitive frailty, highlighting the importance of psychosocial frailty to reduce mortality in older adults.

Keywords

Cognition cohort study frailty mortality

INTRODUCTION

Cognitive impairment, including dementia and mild cognitive impairment (MCI), has become an important health issue in older populations. In the United States, Alzheimer’s disease (AD) is the most common type of dementia and ranks as the fifth leading cause of death in older adults [1]. In Taiwan, a nationwide survey reported that the age- and sex-adjusted prevalence was 18.8% for MCI and 8.1% for all-cause dementia in people aged 65 and older [2]. Therefore, it is crucial to predict the risk of cognitive impairment and the subsequent mortality at the preclinical phase.

Table 1

Assessment tools and scoring strategies for frailty related variables

Frailty	CHS physical frailty items [31]	Items included in this study	Assessment tools and scoring strategies	Total score
Physical frailty	1) Shrinking (unintentional weight loss or sarcopenia)	1) Shrinking (loss of appetite^a)	Loss of appetite, assessed by CES-D, as a surrogate of unintentional weight loss in the past year 1: Loss of appetite 0: No
	2) Weakness(grip strength)	2) Weakness (poor lower extremity strength)	Transferring or stair climbing in ADL 1: Unable to transfer or climb stairs by oneself or need the assistance/instruction while in motion 0: Could transfer or climb stairs by oneself without the assistance/instruction while in motion
	3) Exhaustion (self-report)	3) Exhaustion (self-report)	Self-report exhaustion was assessed by questions in CES-D: 1) “I felt that everything I did was an effort”, 2) “I could not get going”. 1: Exist 0: Lack of the condition
	4) Slowness(walking speed)	4) Slowness (gait speed)	Time (in seconds) for walking 8 feet, cutoff point by sex (women; men) 1: > 3.5; > 3 s 0:≤3.5; ≤3 s
	5) Low activity(by sex)	5) Low activity (by sex)	Physical activity was assessed by IPAQ (MET-min/week), cutoff point by sex (women; men)
			1: 0–750; 0–1050 (MET-min/week)
			0: > 750–10,000; > 1050–10,000 (MET-min/week)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
			Robust: None of the 5 items of physical frailty	5
			Physically prefrail: 1-2 items of physical frailty
			Physically frail: 3 + items of physical frailty
Psychosocial frailty	–––	Self-rated health	Self-rated health includes 3 questions: 1) Current health status, 2) Current health status compared with his/her status in the previous year, and 3) Current health status compared with peers of the same age. Each question includes 5 levels “good, very good, not too good, poor, and very poor” 1: Three answers were very poor or poor 0.67: Two answers were very poor or poor 0.33: One answer was very poor or poor 0: No answer was very poor or poor
	—	Mood	1) The presence of either one in two questions from CES-D: a) feeling sad, b) feeling depressed, 2) Self-report depressive symptoms, or 3) Taking antidepressant. 1: The presence of any one of the above items 0: None of the above
	—	Social relationship	The frequency involved in social activities. 1: Never 0.67: Seldom (once per week) 0.33: Sometime (2-3 times per week) 0: Usually
	—	Social support	The status of family support 1: Living alone 0: Living with family
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
			Robust: T1 of psychosocial frailty score	4
			Psychosocial prefrailty: T2 of psychosocial frailty score
			Psychosocial frailty: T3 of psychosocialfrailty score
Global frailty = Physical frailty + Psychosocial frailty			Robust: T1 of global frailty scoreGlobal prefrailty: T2 of global frailty scoreGlobal frailty: T3 of global frailty score	5 + 4 = 9

^aThe presence of loss of appetite indicates poor nutritional status, which is a surrogate of shrinkage and has been validated by Cesari’s study [38]. CHS, Cardiovascular Health Study; CES-D, Center for Epidemiology Studies Depression Scale; ADL, activities of daily living; IPAQ, International Physical Activity Questionnaire; MET, metabolomic equivalent of task; T, tertile.

Frailty is a dynamic state, and a multidimensionalapproach to frailty (physical, psychological, and soc-ial frailty) has been used to predict adverse outcomes independent of disease, comorbidity, or disability [3]. Previous studies have focused on physical frailty, which contributes to the wide variation in reported prevalence of frailty among community-dwelling older adults (4% to 59%) [4]. Physical frailty and cognitive impairment have some shared risk factors, e.g., vascular risk factors [5 –7] and chronic inflammation [8, 9]. Cohort studies are limited in number but have found associations of frailty with risk of AD, all-cause dementia [10], and poor performance in cognitive domains (episodic memory, semantic memory, working memory, processing speed, and vis-uospatial ability) [11, 12]. However, only baseline frailty data were available in these studies, with no longitudinal follow-up [10, 13]. In addition, cross-sectional studies have reported inconsistent findings across previous studies [14 –18], which may be attri-butable to different tools for cognitive assessment,heterogeneous operational measures of frailty, and different study designs and populations. Regarding the operational definition of frailty, some studies have mixed physical factors and a limited number ofpsychological factors together. Similarly, the heterogeneous definitions of social frailty have included measures as varied as going out, visiting friends,feeling helpful to friends or families, living alone, tal-king to someone every day, engaging in social activities, and having social support [10 , 19–24].

“Cognitive frailty” was defined as the coexistence of physical frailty and MCI, based on the consensus from the International Academy on Nutrition and Aging and the International Association of Gerontology and Geriatrics (IANA-IAGG) in 2013 [25]. In community settings, the prevalence of cognitive frailty has ranged from 4.9% to 50.3% [13 , 27]. The hazard ratio (HR) of all-cause mortality associated with cognitive frailty has ranged from 1.69 to 6.68 [26 , 29]. A recent meta-analysis indicated that cognitive frailty was associated with the risk of AD, vascular dementia, and all-cause dementia [28]. However, studies relating cognitive frailty to all-cause mortality are limited in number, and there is a lack of studies examining frailty dimensions and cognitive domains. Furthermore, some of these studies combined middle- and old-aged people [30] or included only oldest-old people [27].

Taken together, impairments in different frailty dimensions and cognitive domains may emerge in different stages of a lifetime. However, previous studies have mainly focused on the association of physical and not psychosocial frailty. A limited number of studies have explored cognitive domains, and repeated measures on frailty are lacking. In addition, because definitions of psychological or social frailty have been heterogeneous and have incorporated limited information, accommodating variables from systematic reviews allows us to develop a more comprehensive and acceptable definition. Regarding cognitive frailty, no study has explored the coexistence of frailty dimensions with cognitive domains in relation to mortality. Therefore, this study aimed to explore 1) the association of frailty dimensions (physical dimension: a modified definition from Fried et al. [31]; psychosocial dimension: integrating variables from previous studies; global frailty: combining physical and psychosocial frailty, detailed in Table 1) with global cognition or cognitive domains (memory,attention, executive function, and verbal fluency) over time and 2) the associations between cognitive frailty, defined as the coexistence of MCI/impaired cognitive domains and global frailty/frailty dimensions (detailed in Table 2), with all-cause mortality in community-dwelling older adults. We hope findings from this study will clarify the relationship between different frailty dimensions and cognitive domains and the relationship between extended definitions of cognitive frailty and all-cause mortality in older adults over time. We also hope to add new information and suggest early detection in older adults to reduce the risk of cognitive impairment and lower all-cause mortality in older adults.

Table 2

The traditional and extended definitions of cognitive frailty in this study

Cognition\Frailty	Physical frailty	Psychosocial frailty	Global frailty
MCI: impaired “global” cognition excluding dementia; i.e., MoCA-T score of (22 or 23)	Cognitive frailty (i.e., MCI-physical frailty): traditional definitionbased on the consensusfrom 2013 IANA-IAGG[25]	MCI-psychosocial frailty	MCI-global frailty
	Definition:	Definition:	Definition:
	The presence of MCI and physical frailty	The presence of MCI and psychosocial frailty	The presence of MCIand global frailty
Impaired cognitive domains (memory, attention, executive function, and verbal fluency)	Impaired cognitive domain-physical frailty:– Impaired memory-physical frailty– Impaired attention-physical frailty– Impaired executive function-physical frailty– Impaired verbal fluency-physical frailty	Impaired cognitive domain-psychosocial frailty– Impaired memory-psychosocial frailty– Impaired attention-psychosocial frailty– Impaired executive function-psychosocial frailty– Impaired verbal fluency-psychosocial frailty	Impaired cognitive domain-global frailty– Impaired memory-global frailty– Impaired attention-global frailty– Impaired executive function-global frailty– Impaired verbal fluency- global frailty
	Definition: The presence of impaired cognitive domain and physical frailty	Definition: The presence of impaired cognitive domain and psychosocial frailty	Definition: The presence of impaired cognitive domain and global frailty

Impairment of specific cognitive “domain” was defined as the lowest tertile (T1) of the corresponding z-score; the higher tertiles (T2 and T3) referred to normal cognitive domain. MCI, mild cognitive impairment; MoCA-T, Taiwanese version of Montreal Cognitive Assessment; IANA-IAGG, the International Academy on Nutrition and Aging and the International Association of Gerontology and Geriatrics; T, tertile.

MATERIALS AND METHODS

Study population

This cohort study is part of an ongoing prospective cohort study, which recruited 605 older adults (aged 65+) from the senior health checkup program of a tertiary medical center at baseline (2011–2013) with a 4-year follow-up (2015–2017). Participants with the following conditions were excluded at baseline: prescribed medications for AD; suspected dementia [defined as a score of≤21 on the Taiwanese version of Montreal Cognitive Assessment (MoCA-T)] [32]; and a brain tumor or head injury with a loss of consciousness. Participants lacking data on frailty or cognitive variables at baseline or follow-up were excluded. A total of 521 (86%) participants were included for statistical analyses. We also linked our research data with the National Death Registry between 2011 and 2019. Participants in this study included cognitively impaired and cognitively normal persons. The information collected in the study was ascertained by the family member(s) who lived with the participant and reviewing the medical records. This research was approved by the Research Ethics Committee of the medical center. Written informed consent was obtained from all study participants. The research framework is shown in Supplementary Figure 1. This study complies with the Helsinki Declaration of 1975.

Assessment of cognitive function

Cognition was assessed at baseline and at the 4-year follow-up. Global cognition was assessed by the MoCA-T (≤21: suspected dementia, 22 or 23: MCI, ≥24: normal cognition; these cutoffs have been validated in a Taiwanese population) [32]. The third edition of the Wechsler Memory Scale was used to assess memory (logical memory immediate and delayed theme and free recall) and attention domains (digit span forward and backward) [33]. The time to complete Trail Making Tests A and B was used to assess attention performance and executive function, respectively [34]. Digit span forward and backward measures the number a person can successively distribute his/her attention; the latter was also used to assess working memory and short-term memory [35]. Category/semantic fluency performance was assessed by verbal fluency tests in which fruits, fish, and vegetables were named for one minute each and summed up into a composite score. All cognitive scores were standardized based on the mean and standard deviation at baseline. The lowest tertile (T1) of z-scores indicated impaired cognition in that domain; the higher tertiles (T2 and T3) indicated normal cognition in that domain. Similar quantile approaches were adopted in other studies previously [36, 37].

Assessment of frailty

The assessment of frailty dimensions and global frailty was performed at baseline and the 4-year follow-up. The scoring strategies were shown in Table 1.

Physical frailty was modified from Fried’s [31] definition. First, unintentional weight loss (an indicator of shrinking) was assessed by loss of appetite, which has been validated in another study [38] using a question in the Center for Epidemiology Studies Depression Scale (CES-D) [39]. Second, muscle strength was assessed by two items in activities of daily living (ADL) [40], namely, the “ability to transfer from bed to chair” and “stair-climbing,” which have been previously adopted. A study used “10 steps of stair-climb” to assess resistance (i.e., weakness) with the remaining four phenotypes similar as Fried’s frailty phenotype [41]. They found this modified Frailty Phenotype Questionnaire was a highly accurate screening tool for Fried frailty phenotype in community-dwelling older adults [41]. Third, poor endurance (self-reported exhaustion) was assessed by two questions on the CES-D: “I felt that everything I did was an effort” and “I could not get going.” Fourth, slowness was assessed by the time to walk 8 feet. Finally, physical activity was assessed by using the International Physical Activity Questionnaire [42]. The presence of at least three out of the five criteria above indicated physical frailty; the presence of 1 or 2 criteria indicated physical prefrailty, and the lack of any criteria indicated physical robustness.

Psychosocial frailty was defined based on variables summarized in a systematic review on frailty [43]. Because the definition of psychosocial frailty has been heterogeneous across studies, accommodating variables from systematic reviews allows us to develop a more comprehensive and acceptable definition. In addition, due to the close interrelation between psychological and social frailty, these two dimensions were integrated into “psychosocial” frailty. First, the assessment of psychological variables included three questions about individuals’ self-rated health. Second, the mood was assessed by “feeling sad or depressed” on the CES-D, self-reported depressive symptoms, or the use of antidepressants. Third, social relationships were evaluated by the frequency of social activities. Finally, social support was assessed by the status of family support. Each item was scored with either a 0 or 1 for a total maximal score of 4. T1 indicated robustness, T2 indicated psychosocial prefrailty, and T3 indicated psychosocial frailty.

Global frailty was the combination of physical and psychosocial frailty. Combining these two frailty dimensions is due to the importance of psychosocial frailty and the close relation between physical and psychosocial health. Because this work proposed global frailty for the first time; no study has the same definition that could be cited. Indirectly, we found that a study on patterns of wellbeing explored different domains (physical, emotional, and psychosocial) of function [44]. Their finding supported a largely one-dimensional construct of wellbeing in old age, although some individuals had uneven profiles [44], which in part explained the reason we combined physical and psychosocial frailty into global frailty. Global frailty had a maximum total score of 9, and was subsequently grouped into robust (T1), globally prefrail (T2), and globally frail (T3) categories.

Cognitive frailty

The IANA-IAGG defined cognitive frailty as the coexistence of MCI (MoCA-T score: 22 or 23) and physical frailty (≥3 out of 5 criteria present) [25]. This study, for the first time, further proposed and assessed five categories of extended definitions, which included 1) MCI-psychosocial frailty, 2) MCI-global frailty, 3) impaired cognitive domain-physical frailty, 4) impaired cognitive domain-psychoso-cial frailty, and 5) impaired cognitive domain-global frailty (Table 2).

Covariates

A self-report questionnaire was administered to collect demographic information and anthropometric data at baseline and at the 4-year follow-up. Moreover, information regarding disease history, health behaviors, and physical function (ADL [40] and instrumental activities of daily living (IADL) [45]) was also recorded. The number of diseases was determined by seven common diseases (hypertension, diabetes, heart diseases, osteoarthritis, stroke, parkinsonism, and chronic respiratory diseases) based on self-reporting or medication use in this olderpopulation. For details of APOE genotyping, please refer to our previous work [46]. In brief, APOE ɛ4 status was determined by two single-nucleotide polymorphisms, rs429358 and rs7412. Participants carrying at least one ɛ4 allele was defined as carriers; the remaining participants were defined as non-carriers.

Statistical analyses

Frailty and cognition. Analysis of variance, Kru-skal-Wallis tests, and chi-square tests were used to compare the differences across three levels of frailty for continuous and categorical variables at baseline. T tests and Mann-Whitney U tests were used to compare continuous variables between cognitively frail and robust participants. Generalized linear mixed models (GLMMs) were utilized to analyze longitudinal data for estimating 1) the adjusted odds ratio (aOR) and 95% confidence intervals (CIs) for the risk of cognitive impairment (global cognition: MoCA-T score < 24; cognitive domain: T1 of a z-score) and 2) the changes in β coefficients for either global/domain-specific cognition when frailty scores increased one unit or when frailty status moving to a more severe level (prefrailty or frailty). Both cognitive- and frailty-related variables were time varying in the GLMMs, which incorporated a random intercept to account for individual variations. Additional covariates adjusted in the GLMMs included age, sex, years of education, APOE ɛ4 status, alcohol consumption (ever drinkers), number of chronic diseases, body mass index, and follow-up time.

Cognitive frailty and all-cause mortality. Time-dependent Cox proportional hazard (PH) regression models were used to estimate the adjusted HR (aHR) for all-cause mortality in participants with or without cognitive frailty (traditional and extended definitions). The PH assumption was assessed by the interaction of the independent variable and the function of survival time [47]. All Cox models were adjusted for age, sex, years of education, alcohol consumption, number of chronic diseases, and body mass index. Covariates chosen for GLMMs and time-dependent Cox models PH included 1) variables with biological importance to the outcome variables, and 2) significant variables identified by the univariable analyses (Table 3) and remained significant in the multivariable analyses. The –2 log likelihood (–2 LogL) and Akaike information criterion (AIC) [48] were used to compare Cox PH models using different definitions of cognitive frailty. All statistical tests were two-sided, and p < 0.05 indicated statistical significance. All analyses were performed by SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

Table 3

Characteristics of the study population by the status of frailty dimensions, global frailty, and cognitive frailty at baseline (2011–2013)

Variables	Frailty dimensions
	Physical frailty^a			Psychosocial frailty^b			Global frailty^b			Cognitive frailty^c
	Robust	Prefrail	Frail	Robust	Prefrail	Frail	Robust	Prefrail	Frail	Yes	No
	(n = 231)	(n = 269)	(n = 21)	(n = 158)	(n = 148)	(n = 215)	(n = 173)	(n = 198)	(n = 150)	(n = 5)	(n = 516)
	N (%)
Women	128 (55.4)	131 (48.7)	14 (66.7)	79 (50.0)	65 (43.9)	129 (60.0)	83 (48.0)	105 (53.0)	85 (56.7)	4 (80.0)	269 (52.1)
APOEɛ4 carriers	38 (16.6)	42 (15.8)	4 (20.0)	21 (13.5)	28 (19.1)	35 (16.5)	30 (17.5)	29 (14.8)	25 (16.9)	1 (20.0)	83 (16.3)
Alcohol drinking	60 (26.0)	62 (23.1)	6 (28.6)	47 (29.8)	42 (28.4)	39 (18.1)	51 (29.5)	46 (23.2)	31 (20.7)	2 (40.0)	126 (24.4)
Cigarette smoking	32 (13.9)	51 (19.0)	5 (23.8)	29 (18.4)	23 (15.5)	36 (16.7)	29 (16.8)	32 (16.2)	27 (18.0)	1 (20.0)	88 (16.9)
	Mean (SD)
MoCA-T score	27.1 (2.0)	26.8 (2.3)	26.2 (2.4)	27.2 (2.0)	27.0 (2.2)	26.6 (2.4)	27.1 (2.0)	26.8 (2.2)	26.8 (2.5)	22.6 (0.5)	26.9 (2.2)
Age (y)^d	71.5 (5.0)	73.3 (5.3)	77.2 (5.6)	71.7 (4.9)	72.9 (5.6)	73.3 (5.4)	71.5 (5.1)	72.1 (5.0)	74.8 (5.5)	78.2 (4.8)	72.6 (5.3)
Education (y)^d	13.8 (3.7)	13.8 (3.5)	13.2 (3.9)	14.1 (3.6)	13.9 (3.6)	13.4 (3.5)	14.1 (3.7)	13.7 (3.6)	13.6 (3.4)	11.2 (3.8)	13.8 (3.6)
BMI (kg/m²)^d,e	23.6 (2.8)	24.1 (2.9)	23.6 (3.9)	23.6 (2.8)	24.4 (3.0)	23.7 (2.8)	23.6 (2.7)	24.1 (2.9)	23.8 (3.0)	24.6 (3.0)	23.8 (2.9)
ADL^d	99.7 (1.3)	99.3 (2.3)	95.0 (4.5)	99.7 (1.4)	99.2 (2.3)	99.1 (2.7)	99.7 (1.4)	99.6 (1.4)	98.4 (3.4)	98.0 (2.7)	99.3 (2.2)
IADL^d	7.9 (0.4)	7.8 (0.6)	7.1 (1.7)	7.9 (0.5)	7.9 (0.4)	7.7 (0.8)	7.9 (0.4)	7.9 (0.4)	7.6 (0.9)	8.0 (0.0)	7.8 (0.6)
Number of chronic	1.4 (1.1)	1.8 (1.1)	2.3 (1.2)	1.5 (1.1)	1.6 (1.2)	1.7 (1.1)	1.3 (1.1)	1.6 (1.1)	1.9 (1.1)	1.8 (1.1)	1.6 (1.1)
diseases^d,f

^aParticipants with≥3 frailty items indicated physical frailty, one or two frailty items indicated prefrailty, and no frailty item was referred to robust. ^bThe scores of psychosocial frailty and global frailty (sum of physical frailty and psychosocial frailty) were tertiled, respectively. The lowest tertile (T1) indicated robust, the second tertile (T2) indicated pre-frail, and the highest tertile (T3) indicated frail. ^cThe traditional “cognitive frailty” (i.e., MCI-physical frailty) was defined based on the consensus from 2013 IANA-IAGG. MCI was defined as a MoCA-T score of 22 or 23; MoCA-T≥24 indicated normal cognition. ^dKruskal-Wallis tests were applied to compare non-normally distributed continuous variable across levels of frailty (robust, prefrail, and frail); Mann-Whitney U test was utilized to compare non-normally distributed continuous variable between two groups of cognitive frailty. ^eBody mass index (BMI) was the ratio of weight (kg) to the square of height (m²). ^fThe number of diseases was determined based on the presence of seven common diseases in older adults, which included hypertension, diabetes, heart diseases, osteoarthritis, stroke, Parkinsonism, and chronic respiratory diseases. Numbers in bold indicated statistically significant differences across groups (p < 0.05). APOE, apolipoprotein E; MoCA-T, Taiwanese version of Montreal Cognitive Assessment; BMI, body mass index; ADL, activity of daily living; IADL, instrumental activity of daily living; SD, standard deviation; T, tertile; MCI, mild-cognitive impairment; IANA-IAGG, the International Academy on Nutrition and Aging and the International Association of Gerontology and Geriatrics.

RESULTS

Characteristics of the study population

At baseline, mean age of participants was 72.7 (standard deviation: 5.3), and the prevalence was 3.0% for physical frailty, 37.6% for psychological frailty, 26.6% for global frailty, 8% for MCI, 0.5% for traditional cognitive frailty (i.e., MCI-physical frailty), 4.5% for MCI-psychosocial frailty, and 4.0% for MCI-global frailty (Supplementary Figure 2, Supplementary Table 1). At 4-year follow-up, mean age of participants was 76.1 (standard deviation: 5.0), and the prevalence was 2.8% for physical frailty, 37.1% for psychological frailty, 27.5% for global frailty, 8.3% for MCI, 1.3% for traditional cognitive frailty (i.e., MCI-physical frailty), 4.5% for MCI-psychosocial frailty, and 4.3% for MCI-global frailty. The mean age, physical function (ADL and IADL), and number of chronic diseases significantly differed across the levels of physical frailty and global frailty; sex, alcohol consumption, mean age, and number of chronic diseases varied significantly across the levels of psychosocial frailty; MoCA-T score (global cognition) and mean age differed significantly between cognitively frail and robust older adults (Table 3).

Associations among frailty dimensions/global frailty and cognition over four years

Over four years, greater physical frailty was associated with poor global cognition (aOR = 1.43 to 3.29, β: –1.07; Table 4 and Supplementary Figure 3), logical memory (immediate theme and free recall and delayed free recall, β: –0.14 to –0.10), and executive function/attention (Trail Making Tests A and B, β: –0.51 to –0.12), with significant trends for some associations (p_trend = 0.003 to 0.03). Physical frailty was not associated with attention/working memory (digit span tests) or verbal fluency domains in this population. Greater psychosocial frailty was associated with poor global cognition (β: –0.44) and attention (digit span forward, β: –0.15 to –0.13), with significant trends for some associations (p_trend = 0.01 to 0.02). Greater global frailty was associated with poor global cognition (aOR = 1.39 to 2.29), logical memory (immediate free recall and delayed theme and free recall, β: –0.30 to –0.07), executive function/attention (Trail Making Tests A and B, β: –0.17 to –0.06), and attention (digit span forward, β: –0.12), with significant trends for some associations (p_trend = 0.002 to 0.02).

Table 4

Association between frailty dimensions/global frailty and cognition over four years (N = 521)

Frailty\Cognition	Global cognition (MoCA-T)		Logical memory				Trail Making Test		Digit span		Verbal fluency
	Binary (22/23 versus ≥24) aOR (95% CI)	Continuous β (95% CI)	Immediate theme recall β (95% CI)	Immediate free recall β (95% CI)	Delayed theme recall β (95% CI)	Delayed free recall β (95% CI)	A β (95% CI)	B β (95% CI)	Forward β (95% CI)	Backward β (95% CI)	β (95% CI)
Physical frailty
Continuous	1.43	–0.08	–0.10	–0.14	–0.08	–0.10	–0.13	–0.14	0.07	–0.05	–0.06
	(1.07, 1.92)	(–0.28, 0.13)	(–0.19, –0.02)	(–0.22, –0.06)	(–0.15, 0.001)	(–0.17, –0.02)	(–0.19, –0.06)	(–0.21, –0.06)	(–0.003, 0.14)	(–0.12, 0.03)	(–0.12, 0.001)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Robust^a	1.00	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Prefrail^a	1.52	–0.07	–0.05	–0.10	0.01	–0.05	–0.12	–0.12	0.04	–0.02	–0.02
	(0.86, 2.68)	(–0.35, 0.22)	(–0.19, 0.09)	(–0.23, 0.03)	(–0.12, 0.13)	(–0.18, 0.07)	(–0.22, –0.02)	(–0.23, –0.01)	(–0.07, 0.15)	(–0.14, 0.10)	(–0.13, 0.09)
Frail^a	3.29	–1.07	–0.28	–0.32	–0.34	–0.28	–0.40	–0.51	–0.01	–0.23	–0.19
	(1.20, 9.03)	(–2.11, –0.04)	(–0.66, 0.11)	(–0.66, 0.02)	(–0.68, 0.01)	(–0.58, 0.03)	(–0.70, –0.10)	(–0.83, –0.18)	(–0.35, 0.33)	(–0.57, 0.11)	(–0.42, 0.04)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
p _trend	0.03	0.14	0.22	0.05	0.42	0.17	0.003	0.003	0.64	0.39	0.36
Psychosocial frailty
Continuous	1.31	–0.28	0.01	–0.08	–0.05	–0.09	–0.01	0.07	–0.13	0.01	0.01
	(0.88, 1.93)	(–0.57, 0.01)	(–0.10, 0.11)	(–0.18, 0.02)	(–0.16, 0.05)	(–0.19, 0.01)	(–0.10, 0.08)	(–0.02, 0.16)	(–0.22, –0.04)	(–0.09, 0.12)	(–0.07, 0.10)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Robust^a	1.00	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Prefrail^a	1.08	0.02	0.03	–0.04	–0.05	–0.10	0.10	0.08	–0.05	0.08	0.01
	(0.55, 2.09)	(–0.30, 0.34)	(–0.13, 0.20)	(–0.20, 0.12)	(–0.21, 0.10)	(–0.26, 0.06)	(–0.02, 0.22)	(–0.04, 0.20)	(–0.18, 0.09)	(–0.06, 0.22)	(–0.11, 0.13)
Frail^a	1.60	–0.44	–0.01	–0.12	–0.03	–0.13	–0.02	0.06	–0.15	0.03	0.02
	(0.87, 2.95)	(–0.77, –0.11)	(–0.16, 0.15)	(–0.27, 0.03)	(–0.18, 0.11)	(–0.27, 0.01)	(–0.13, 0.09)	(–0.05, 0.18)	(–0.27, –0.02)	(–0.11, 0.18)	(–0.10, 0.13)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
p _trend	0.12	0.01	0.92	0.11	0.65	0.06	0.77	0.29	0.02	0.67	0.76
Global frailty
Continuous	1.39	–0.15	–0.06	–0.12	–0.07	–0.09	–0.08	–0.06	–0.005	–0.02	–0.03
	(1.12, 1.71)	(–0.31, 0.002)	(–0.12, 0.002)	(–0.18, –0.06)	(–0.13, –0.01)	(–0.15, –0.04)	(–0.13, –0.04)	(–0.11, –0.01)	(–0.05, 0.05)	(–0.07, 0.03)	(–0.08, 0.01)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Robust^a	1.00	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Prefrail^a	1.33	–0.14	0.07	–0.05	0.12	0.02	–0.06	–0.04	–0.12	–0.01	0.06
	(0.69, 2.56)	(–0.44, 0.15)	(–0.08, 0.22)	(–0.20, 0.10)	(–0.03, 0.27)	(–0.13, 0.17)	(–0.17, 0.04)	(–0.16, 0.08)	(–0.24, –0.01)	(–0.13, 0.11)	(–0.05, 0.17)
Frail^a	2.29	–0.31	–0.13	–0.30	–0.12	–0.17	–0.17	–0.12	0.04	0.01	–0.02
	(1.13, 4.64)	(–0.76, 0.14)	(–0.32, 0.06)	(–0.48, –0.11)	(–0.30, 0.07)	(–0.35, 0.001)	(–0.30, –0.03)	(–0.27, 0.03)	(–0.11, 0.19)	(–0.14, 0.17)	(–0.15, 0.11)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
p _trend	0.02	0.17	0.22	0.002	0.25	0.06	0.02	0.11	0.65	0.86	0.81

For each frailty dimension or global frailty, GLMMs were used to estimate aORs for binary outcome [MoCA-T: 22 or 23 (MCI) versus≥24 (cognitive normal)] and β coefficients for continuous cognitive variables adjusted for the other frailty dimension (no need to adjust when global frailty was in the model), age, sex, years of education, APOE e4 status, alcohol drinking, number of chronic diseases, body mass index, and follow-up time. p_trend indicated a linear trend across the tertiles of frailty dimensions following adjustment for the previously described covariates. Numbers in bold indicated statistically significant findings (p < 0.05). ^aFrailty dimensions or global frailty was categorized into robust, prefrail, and frail in the GLMMs. FU, 4-year follow-up; MoCA-T, Taiwanese version of Montreal Cognitive Assessment; aOR, adjusted odds ratio; CI, confidence interval; GLMM, generalized linear mixed model; APOE, apolipoprotein E.

Extended definitions of cognitive frailty and all-cause mortality

From 2011 to 2019, a total of 23 participants has died. We found that increased all-cause mortality was associated with the traditional definition of cognitive frailty (i.e., MCI-physical frailty: aHR = 4.87, Table 5) and four extended definitions, namely, “MCI-psychosocial frailty” (aHR = 3.49), “MCI-global frailty” (aHR = 6.83), “impaired logical memory delayed theme recall-global frailty” (aHR = 3.06), and “impaired verbal fluency-global frailty” (aHR = 3.29) compared with robust older adults. Other extended definitions of cognitive frailty were not associated with all-cause mortality. Among these significant associations, we found that “MCI-psychosocial frailty,” “MCI-global frailty,” and “impaired verbal fluency-global frailty” outperformed traditional cognitive frailty based on –2 logL and AIC (–2 log-likelihood = 208.2, 207.8, 212.05 versus 212.09; AIC = 224.3, 221.8, 226.1 versus 228.1, respectively; Table 5). Supplementary Figure 4 demonstrated survival curves for significant associations of cognitive frailty and its extended definitions with all-cause mortality between 2011 and 2019.

Table 5

Association of cognitive frailty (traditional and extended definitions) with all-cause mortality between 2011 and 2019

Types of cognitive frailty	aHR	(95% CI)	–2 LogL^e (Ranking)	AIC ^e (Ranking)
Cognitive impairment^a &physical frailty^b
MCI^c-physical frailty (traditional “cognitive frailty”)	4.87	(1.06, 22.40)	212.09 (4)	228.1 (5)
Impaired logical memory immediate theme recall^a-physical frailty	3.18	(0.62, 16.43)	–	–
Impaired logical memory immediate free recall^a-physical frailty	1.75	(0.35, 8.78)	–	–
Impaired logical memory delayed theme recall^a-physical frailty	3.06	(0.61, 15.24)	–	–
Impaired logical memory delayed free recall^a-physical frailty	2.38	(0.49, 11.47)	–	–
Impaired Trail Making Test A^a-physical frailty	2.49	(0.64, 9.70)	–	–
Impaired Trail Making Test B^a-physical frailty	3.17	(0.80, 12.60)	–	–
Impaired digit span forward^a-physical frailty	2.04	(0.43, 9.78)	–	–
Impaired digit span backward^a-physical frailty	1.14	(0.14, 9.18)	–	–
Impaired verbal fluency^a-physical frailty	1.76	(0.36, 8.67)	–	–
Cognitive impairment^a &psychosocial frailty^d
MCI^c-psychosocial frailty	3.49	(1.26, 9.73)	208.2 (2)	224.3 (2)
Impaired logical memory immediate theme recall^a-psychosocial frailty	1.17	(0.32, 4.26)	–	–
Impaired logical memory immediate free recall^a-psychosocial frailty	1.01	(0.30, 3.41)	–	–
Impaired logical memory delayed theme recall^a-psychosocial frailty	2.39	(0.80, 7.12)	–	–
Impaired logical memory delayed free recall^a-psychosocial frailty	1.95	(0.65, 5.89)	–	–
Impaired Trail Making Test A^a -psychosocial frailty	1.64	(0.53, 5.11)	–	–
Impaired Trail Making Test B^a -psychosocial frailty	1.08	(0.36, 3.28)	–	–
Impaired digit span forward^a-psychosocial frailty	0.82	(0.26, 2.63)	–	–
Impaired digit span backward^a-psychosocial frailty	1.01	(0.27, 3.79)	–	–
Impaired verbal fluency^a-psychosocial frailty	1.43	(0.47, 4.39)	–	–
Cognitive impairment^a &global frailty^d
MCI^c-global frailty	6.83	(2.42, 19.31)	207.8 (1)	221.8 (1)
Impaired logical memory immediate theme recall^a-global frailty	1.89	(0.61, 5.87)	–	–
Impaired logical memory immediate free recall^a-global frailty	1.83	(0.62, 5.44)	–	–
Impaired logical memory delayed theme recall^a-global frailty	3.06	(1.05, 8.96)	213.7 (5)	227.7 (4)
Impaired logical memory delayed free recall^a-global frailty	2.58	(0.88, 7.56)	–	–
Impaired Trail Making Test A^a-global frailty	2.40	(0.86, 6.68)	–	–
Impaired Trail Making Test B^a-global frailty	1.51	(0.52, 4.35)	–	–
Impaired digit span forward^a-global frailty	1.35	(0.43, 4.27)	–	–
Impaired digit span backward^a-global frailty	1.87	(0.60, 5.81)	–	–
Impaired verbal fluency^a-global frailty	3.29	(1.24, 8.77)	212.05 (3)	226.1 (3)

^aImpairment of specific cognitive test referred to the lowest tertile (T1) of z-score; cognitive normal referred to higher tertiles (T2 and T3) of z-score. ^bParticipants with≥3 frailty items indicated physical frailty, one or two frailty items indicated prefrailty, and no frailty item referred to robust. ^cMCI was defined as a MoCA-T score of 22 or 23; MoCA-T≥24 indicated cognitive normal. ^dThe scores of psychosocial frailty and global frailty (integration of physical frailty and psychosocial frailty) were tertiled, respectively. The lowest tertile (T1) indicated robust, the second tertile (T2) indicated prefrail, and the highest tertile (T3) indicated frail. ^e–2 LogL and AIC were used to compare five Cox PH models with significant findings. The smaller the statistic, the better the model (i.e., higher ranking). The PH assumption, which was assessed by the interactions of each cognitive frailty variable and survival time, was not violated (p_interaction > 0.05) for all models. All time-dependent Cox PH models were adjusted for age, sex, years of education, alcohol drinking, number of chronic diseases, and body mass index with a median follow-up time of 6.2 years. Numbers in bold indicated statistically significant findings (p < 0.05). aHR, adjusted hazard ratio; CI, confidence interval; –2 LogL, –2 Log-likelihood; AIC, Akaike information criterion; MCI, mild cognitive impairment; T, tertile; MoCA-T, Taiwanese version of Montreal Cognitive Assessment; PH, proportional hazard.

DISCUSSION

This study was the first to use longitudinal frailty data to explore the relationship between frailty dimensions and cognitive domains over time. We found that physical frailty was associated with MCI and cognitive impairments in multiple domains (memory and executive function), which tended to progress to AD or vascular dementia or were part of the normal cognitive aging process [49]. In contrast, psychosocial frailty was associated with single- and non-memory domains (attention), which tended to progress to non-AD dementia [49]. However, another study did not support using the single- and multi-domain MCI to differentiate subtypes of dementia [50]. The association of physical frailty with MCI was consistent with some previous studies [10, 11] but was not observed in a UK cohort study [12]. At baseline, only 3% of participants had physical frailty, but 37.6% of participants had developed psychosocial frailty (Supplementary Figure 2). Over 4 years, physical prefrailty significantly increased (approximately 2.5% increment per year), while psychosocial prefrailty slightly decreased (approximately 1.6% decrement per year). In Taiwan, the culture of filial piety [51], sufficient activities and resources, and high socioeconomic status in the Taipei metropolitan area [52, 53] may explain the slightly decline of psychosocial frailty. In addition, part (11.6%) of participants were lost to follow-up over four years due to poor health status, which may also explain this phenomenon. Our findings suggested that the differentfrailty dimensions affected cognitive domains via heterogeneous pathways. In addition, psychosocial frailty was more prevalent than physical frailty in younger older adults (age 65–74), and the effects of frailty dimensions on cognition changed over time (Supplementary Table 3).

Second, this study found that cognitive frailty, defined by the 2013 IANA-IAGG, was associated with all-cause mortality in older adults, which was consistent with previous studies [26–30 , 54]. Importantly, we found that four extended definitions of cognitive frailty were significantly associated with all-cause mortality. Three of the four, specifically “MCI-psychosocial frailty”, “MCI-global frailty”, and “impaired verbal fluency-global frailty,” outperformed the traditional “cognitive frailty” (i.e., MCI-physical frailty) in predicting all-cause mortality, associations that have not been previously explored and identified. Because traditional cognitive frailty only includes physical frailty and the effects of cognitive domains have remained unclear, our findings suggest the importance of incorporating psychosocial frailty assessments and information about impairments in the cognitive domain(s) in the prediction of all-caused mortality. These extended definitions (global frailty-impaired cognition) and longitudinal data allowed us to predict all-cause mortality over time and across a wider age range of older adults.

Taken together, physical frailty and cognitive im-pairment share some common biological pathways. First, some vascular risk factors (e.g., atherosclerotic vascular diseases or embolic events) limit blood flow [5, 6] and affect endocrine regulation (e.g., high insulin and low growth hormone) [61, 62], which then decreases muscle mass/strength [7] and cognition. These vascular factors may increase the risk of stroke, and if the affected area is related to language, the ability of verbal fluency is then affected [55, 56]. Second, chronic inflammation may be a compensatory response to or a marker of a causal mechanism of physical frailty [8]. Regarding the relationship between chronic inflammation and cognition, peripheral inflammatory markers (e.g., interleukin-6) may pass the blood-brain barrier (BBB), and chronic inflammation of intracranial vasculature leads to neuronal lesions and the deposition of amyloid-β [9]. Although C-reactive protein is a large molecule that could not pass BBB freely, it can reach the central nervous system and activate local inflammatory responses via passing leakage region of BBB, active intake, or triggering the inflammatory response in the brain [63 –66]. Third, older adults with higher stress and not participating in the aerobic exercise had significantly higher cortisol:dehydroepiandrosterone (DHEA) ratio and flatter DHEA diurnal rhythm than those who regularly participated in aerobic exercise [67]. In addition, a more dynamic cortisol secretion pattern across the day (from morning to evening) was associated with better cognitive function and physical performance [68]. In contrast, high-level cortisol [57] and IL-6 release due to inflammation [58] may affected verbal fluency. Fourth, some nutrients(e.g., vitamin B12, folic acid, and omega-3 polyunsaturated fatty acids) are known to protect against cognitive impairment [69]. Among them, some studies reported that consuming food containing folate may be beneficial to verbal fluency [59, 60]. In addition, deficiencies in some nutrients (e.g., vitamin D) or energy intake decreasing protein synthesis and muscle mass, which leads to energy imbalance, decreased antioxidative ability, and physical frailty [70]. Up to date, insufficient intervention randomized clinical trials showed the effect of whole diets or specific dietary components on cognitive outcomes in people with MCI [71].

Last, social isolation or loneliness, e.g., living alone or has no friend to confide in, are closely related to the psychological health of older adults [72]. Another study found that psychological well-being variables (including perceived stress, anxiety level, feelings of fear, sleep disturbances, or psychosomatic conditions assessed by the General Health Questionnaire) significantly predicted cognitive frailty [73]. However, other studies incorporate limited psychological variables with physical frailty [10 , 74], which is not easy to differentiate the effect of physical and psychological frailty on cognition. In addition, psychosocial stress increases serum cortisol levels and inflammatory responses affecting the central nervous system (e.g., hippocampal volume) and contributes to cognitive impairment development [57]. Participants in this population appeared relatively healthy at baseline (2011–2013) with a higher prevalence of psychosocial frailty (37.6%; Supplementary Table 1) than physical frailty (3%) as a result of tertile the population for defining psychosocial frailty. However, physical prefrailty significantly increased over four years, while many instances of psychosocial prefrailty converted to psychosocial frailty.

This cohort study has the following strengths. First, it included repeated measures (baseline and 4-year follow-up) on both frailty and cognition (i.e., both variables were time varying), which enabled us to explore their associations over time. Second, our data showed that impairments in cognitive domains tended to start at different periods of time in life, and the assessment of four cognitive domains allowed us to identify which cognitive domain(s) and cognitive domain-frailty measures were associated with frailty and all-cause mortality, respectively. Third, most previous studies assessed a single frailty dimension (i.e., physical frailty) or combined it with psychosocial frailty defined by a limited number of factors. This study included four categories of variables to define psychosocial frailty, and its combination with physical frailty (i.e., “global frailty”) captured a full spectrum of frailty states among older adults. The findings showed that global frailty serves as a better predictor for all-cause mortality than physical frailty, which is the only dimension used to define traditional cognitive frailty.

This study has some limitations. First, our study participants were recruited from the senior health checkup program and thus tended to have good physical status at baseline, which may have introduced participation bias. However, this bias diminished as follow-up time increased, as the participant’s health status decreased over time and became increasingly similar to that of the general population [75]. Second, this study used lower extremity strength to access weakness in physical frailty due to a lack of handgrip strength data. However, a high correlation was found between handgrip strength and lower limb moments (Pearson’s correlation coefficient: 0.56–0.78) previously [76]. In addition, several studies [77 –79] used handgrip strength to predict the risk of reduction of lower extremity muscle, assuming that the reduction of handgrip strength started earlier than the reduction of lower extremity strength. Therefore, using two ADL items (transferring and stair climbing) to measure lower extremity strength may underestimate the prevalence of weakness in this community-dwelling older population, i.e., our findings should be more significant if handgrip strength data is available.

CONCLUSIONS AND IMPLICATIONS

Taken together, we found that physical prefrailty increased faster than psychosocial prefrailty over time, which had not been previously reported. In addition, physical and psychosocial frailty were associated with different cognitive domains, which suggests the involvement of different biological mechanisms. Last, this study, for the first time, explored the extended definitions of cognitive frailty, of which “MCI-psychosocial frailty,” “MCI-global frailty,” and “impaired verbal fluency-global frailty” outperformed traditional cognitive frailty for predicting all-cause mortality. In summary, our findings suggested that the different frailty dimensions were associated with the impairment of different cognitive domains, so did their combination with all-cause mortality over time. Prevention toward multiple frailty dimensions (e.g., physical exercise, psychological, and social support) may effectively reduce cognitive impairment and all-cause mortality in older adults. Cohort studies with longer follow-up periods and repeated measures of frailty dimensions and cognitive domains are warranted to confirm ourfindings.

Footnotes

ACKNOWLEDGMENTS

This study was supported by the Ministry of Science and Technology in Taiwan (100-2314-B-002-103, 101-2314-B-002-126-MY3, 103-2314-B-002-033-MY3, 104-2314-B-002-038-MY3, 107-2314-B-002-186-MY3, 107-2314-B-002-230, and 108-2314-B-002-128-MY2). We thank Prof. Marion Lee and Prof. Wen-Chung Lee for scientific suggestions, and the staff of the Sequencing Core, Department of Medical Research, National Taiwan University Hospital for technical support.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Kochanek

, Murphy

, Xu

, Tejada-Vera

(2016) Deaths: Final data for 2014. Natl Vital Stat Rep 65, 1–122.

Sun

, Lee

, Yang

, Chen

, Lin

, Wang

, Tang

, Chiu

(2014) A nationwide survey of mild cognitive impairment and dementia, including very mild dementia, in Taiwan. PLoS One 9, e100303.

Gobbens

, Luijkx

, Wijnen-Sponselee

, Schols

(2010) Toward a conceptual definition of frail community dwelling older people. Nurs Outlook 58, 76–86.

Collard

, Boter

, Schoevers

, Oude Voshaar

(2012) Prevalence of frailty in community-dwelling older persons: A systematic review. J Am Geriatr Soc 60, 1487–1492.

Halil

, Kizilarslanoglu

, Kuyumcu

, Yesil

, Jentoft

(2015) Cognitive aspects of frailty: Mechanisms behind the link between frailty and cognitive impairment. J Nutr Health Aging 19, 276–283.

Moorhouse

, Rockwood

(2008) Vascular cognitive impairment: Current concepts and clinical developments. Lancet Neurol 7, 246–255.

Morley

, Malmstrom

(2013) Frailty, sarcopenia, and hormones. Endocrinol Metab Clin North Am 42, 391–405.

Hubbard

, O’Mahony

, Savva

, Calver

, Woodhouse

(2009) Inflammation and frailty measures in older people. J Cell Mol Med 13, 3103–3109.

Rosano

, Marsland

, Gianaros

(2012) Maintaining brain health by monitoring inflammatory processes: A mechanism to promote successful aging. Aging Dis 3, 16–33.

10.

Wang

, Ji

, Wu

, Tang

, Zhang

, Guan

, Liu

, Fang

(2017) Frailty in relation to the risk of Alzheimer’s disease, dementia, and death in older Chinese adults: A seven-year prospective study. J Nutr Health Aging 21, 648–654.

11.

Boyle

, Buchman

, Wilson

, Leurgans

, Bennett

(2010) Physical frailty is associated with incident mild cognitive impairment in community-based older persons. J Am Geriatr Soc 58, 248–255.

12.

Bunce

, Batterham

, Mackinnon

(2019) Long-term associations between physical frailty and performance in specific cognitive domains. J Gerontol B Psychol Sci Soc Sci 74, 919–926.

13.

Sathasivam

, Kamaruzzaman

, Hairi

, Ng

, Chinna

(2015) Frail elders in an urban district setting in Malaysia: Multidimensional frailty and its correlates. Asia Pac J Public Health 27, 52S–61S.

14.

Tsutsumimoto

, Doi

, Makizako

, Hotta

, Nakakubo

, Makino

, Suzuki

, Shimada

(2017) Association of social frailty with both cognitive and physical deficits among older people. J Am Med Dir Assoc 18, 603–607.

15.

Nishiguchi

, Yamada

, Fukutani

, Adachi

, Tashiro

, Hotta

, Morino

, Shirooka

, Nozaki

, Hirata

(2015) Differential association of frailty with cognitive decline and sarcopenia in community-dwelling older adults. J Am Med Dir Assoc 16, 120–124.

16.

Rosado-Artalejo

, Carnicero

, Losa-Reyna

, Guadalupe-Grau

, Castillo-Gallego

, Gutierrez-Avila

, Alfaro-Acha

, Rodriguez-Artalejo

, Rodriguez-Manas

, Garcia-Garcia

(2017) Cognitive performance across 3 frailty phenotypes: Toledo Study for Healthy Aging. J Am Med Dir Assoc 18, 785–790.

17.

, Liu

, Chen

, Lee

, Peng

, Wang

, Chen

(2015) Cognitive function in individuals with physical frailty but without dementia or cognitive complaints: Results from the I-Lan Longitudinal Aging Study. J Am Med Dir Assoc 16, 899.e899–816.

18.

Brigola

, Ottaviani

, Carvalho

DHT

, Oliveira

, Souza

É N

, Pavarini

SCI

(2020) Association between cognitive impairment and criteria for frailty syndrome among older adults. Arq Neuropsiquiatr 78, 2–8.

19.

Wang

, Karp

, Winblad

, Fratiglioni

(2002) Late-life engagement in social and leisure activities is associated with a decreased risk of dementia: A longitudinal study from the Kungsholmen project. Am J Epidemiol 155, 1081–1087.

20.

Fratiglioni

, Wang

, Ericsson

, Maytan

, Winblad

(2000) Influence of social network on occurrence of dementia: A community-based longitudinal study. Lancet 355, 1315–1319.

21.

Scarmeas

, Levy

, Tang

, Manly

, Stern

(2001) Influence of leisure activity on the incidence of Alzheimer’s disease. Neurology 57, 2236–2242.

22.

Wilson

, Mendes De Leon

, Barnes

, Schneider

, Bienias

, Evans

, Bennett

(2002) Participation in cognitively stimulating activities and risk of incident Alzheimer disease. JAMA 287, 742–748.

23.

Seeman

, Lusignolo

, Albert

, Berkman

(2001) Social relationships, social support, and patterns of cognitive aging in healthy, high-functioning older adults: MacArthur studies of successful aging. Health Psychol 20, 243–255.

24.

Yeh

, Liu

(2003) Influence of social support on cognitive function in the elderly. BMC Health Serv Res 3, 9.

25.

Kelaiditi

, Cesari

, Canevelli

, Van Kan

, Ousset

P-J

, Gillette-Guyonnet

, Ritz

, Duveau

, Soto

, Provencher

(2013) Cognitive frailty: Rational and definition from an (IANA/IAGG) international consensus group. J Nutr Health Aging 17, 726–734.

26.

Esteban-Cornejo

, Cabanas-Sanchez

, Higueras-Fresnillo

, Ortega

, Kramer

, Rodriguez-Artalejo

, Martinez-Gomez

(2019) Cognitive frailty and mortality in a national cohort of older adults: The role of physical activity. Mayo Clin Proc 94, 1180–1189.

27.

Hao

, Dong

, Yang

, Dong

, Wei

(2018) Frailty and cognitive impairment in predicting mortality among oldest-old People. Front Aging Neurosci 10, 295.

28.

, Huang

, Xue

, Li

, Bai

, Xu

(2021) Cognitive frailty as a predictor of adverse outcomes among older adults: A systematic review and meta-analysis. Brain Behav 11, e01926.

29.

Liu

, Chen

, Lee

, Wu

, Hwang

, Lin

, Shimada

, Peng

, Loh

, Arai

, Chen

(2018) Cognitive frailty and its association with all-cause mortality among community-dwelling older adults in Taiwan: Results from I-Lan Longitudinal Aging Study. Rejuvenation Res 21, 510–517.

30.

Lee

, Peng

, Liang

, Loh

, Chen

(2018) Cognitive frailty predicting all-cause mortality among community-living older adults in Taiwan: A 4-year nationwide population-based cohort study. PLoS One 13, e0200447.

31.

Fried

, Tangen

, Walston

, Newman

, Hirsch

, Gottdiener

, Seeman

, Tracy

, Kop

, Burke

, McBurnie

(2001) Frailty in older adults: Evidence for a phenotype. J Gerontol A Biol Sci Med Sci 56, 146–157.

32.

Tsai

, Lee

, Wang

, Shia

, Nasreddine

, Fuh

(2012) Psychometrics of the Montreal Cognitive Assessment (MoCA) and its subscales: Validation of the Taiwanese version of the MoCA and an item response theory analysis. Int Psychogeriatr 24, 651–658.

33.

Wechsler

(1997) Wechsler Memory Scale for Adults, 3rd Ed, Psychological Corporation, San Antonio, TX.

34.

Arnett

, Labovitz

(1995) Effect of physical layout in performance of the Trail Making Test. Psychol Assess 7, 220–221.

35.

Blankenship

(1938) Memory span: A review of the literature. Psychol Bull 35, 1–25.

36.

Finlay

, Khan

, Gronlund

, Sol

, Jang

, Melendez

, Judd

, Clarke

(2020) Weather woes? Exploring potential links between precipitation and age-related cognitive decline. Int J Environ Res Public Health 17, 9011.

37.

Benito-León

, Romero

, Louis

, Bermejo-Pareja

(2014) Faster cognitive decline in elders without dementia and decreased risk of cancer mortality. Neurology 82, 1441–1448.

38.

Cesari

, Vellas

, Hsu

, Newman

, Doss

, King

, Manini

, Church

, Gill

, Miller

, Pahor

, LIFE Study Group (2015) A physical activity intervention to treat the frailty syndrome in older persons-results from the LIFE-P study. J Gerontol A Biol Sci Med Sci 70, 216–222.

39.

Radloff

(1977) The CES-D scale: A self-report depression scale for research in the general population. Appl Psychol Meas 1, 385–401.

40.

Mahoney

, Barthel

(1965) Functional Evaluation: The Barthel Index. Md State Med J 14, 61–65.

41.

Kim

, Kim

, Jung

, Won

(2020) Development of a frailty phenotype questionnaire for use in screening community-dwelling older adults. J Am Med Dir Assoc 21, 660–664.

42.

Hagströmer

, Oja

, Sjöström

(2006) The International Physical Activity Questionnaire (IPAQ): A study of concurrent and construct validity. Public Health Nutr 9, 755–762.

43.

de Vries

, Staal

, van Ravensberg

, Hobbelen

, Olde Rikkert

, Nijhuis-van der Sanden

(2011) Outcome instruments to measure frailty: A systematic review. Ageing Res Rev 10, 104–114.

44.

Zammit

, Starr

, Johnson

, Deary

(2012) Profiles of physical, emotional and psychosocial wellbeing in the Lothian birth cohort 1936. BMC Geriatrics 12, 64.

45.

Lawton

, Brody

(1969) Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist 9, 179–186.

46.

Chen

, Jung

, Chen

, Chiou

, Chen

, Tang

, Yeh

, Lee

(2017) Association of dietary patterns with global and domain-specific cognitive decline in Chinese elderly. J Am Geriatr Soc 65, 1159–1167.

47.

Fisher

, Lin

(1999) Time-dependent covariates in the Cox proportional-hazards regression model. Annu Rev Public Health 20, 145–157.

48.

Goerdten

, Carriere

, Muniz-Terrera

(2020) Comparison of Cox proportional hazards regression and generalized Cox regression models applied in dementia risk prediction. Alzheimers Dement (N Y) 6, e12041.

49.

Portet

, Ousset

, Visser

, Frisoni

, Nobili

, Scheltens

, Vellas

, Touchon

, MCI Working Group of the European Consortium on Alzheimer’s Disease (2006) Mild cognitive impairment (MCI) in medical practice: A critical review of the concept and new diagnostic procedure. Report of the MCI Working Group of the European Consortium on Alzheimer’s Disease. J Neurol Neurosurg Psychiatry 77, 714–718.

50.

Glynn

, O’Callaghan

, Hannigan

, Bruce

, Gibb

, Coen

, Green

, B

, Robinson

(2021) Clinical utility of mild cognitive impairment subtypes and number of impaired cognitive domains at predicting progression to dementia: A 20-year retrospective study. Int J Geriatr Psychiatry 36, 31–37.

51.

Lee

, Tan

(2012) Filial ethics and judgments of filial behaviour in Taiwan and the United States. Int J Psychol 47, 161–168.

52.

Socioeconomic Status Database Platform, Taiwan, Accessed September 10, 2021. https://segis.moi.gov.tw/STAT/Web/Portal/STAT_PortalHome.aspx#

53.

National Statistics, Taiwan, Accessed September 10, 2021. https://eng.stat.gov.tw/ct.asp?xItem=37336&ctNode=2191&mp=5

54.

Solfrizzi

, Scafato

, Seripa

, Lozupone

, Imbimbo

, D’Amato

, Tortelli

, Schilardi

, Galluzzo

, Gandin

, Baldereschi

, Di Carlo

, Inzitari

, Daniele

, Sabba

, Logroscino

, Panza

, Italian Longitudinal Study on Aging Working G (2017) Reversible cognitive frailty, dementia, and all-cause mortality. The Italian Longitudinal Study on Aging. J Am Med Dir Assoc 18, e1–89.e88.

55.

Sinanović

, Mrkonjić

, Zukić

, Vidović

, Imamović

(2011) Post-stroke language disorders. Acta Clin Croat 50, 79–94.

56.

Rolfson

, Majumdar

, Tsuyuki

, Tahir

, Rockwood

(2006) Prevalence, duration and indications for prescribing of antidepressants in primary care. Age Ageing 35, 523–526.

57.

Lee

, Glass

, McAtee

, Wand

, Bandeen-Roche

, Bolla

, Schwartz

(2007) Associations of salivary cortisol with cognitive function in the Baltimore memory study. Arch Gen Psychiatry 64, 810–818.

58.

Alley

, Crimmins

, Karlamangla

, Hu

, Seeman

(2008) Inflammation and rate of cognitive change in high-functioning older adults. J Gerontol A Biol Sci Med Sci 63, 50–55.

59.

Tucker

, Qiao

, Scott

, Rosenberg

, Spiro III

(2005) High homocysteine and low B vitamins predict cognitive decline in aging men: The Veterans Affairs Normative Aging Study. Am J Clin Nutr 82, 627–635.

60.

Bryan

, Calvaresi

, Hughes

(2002) Short-term folate, vitamin B-12 or vitamin B-6 supplementation slightly affects memory performance but not mood in women of various ages. J Nutr 132, 1345–1356.

61.

Neumann

, Rojo

, Navarrete

, Farías

, Reyes

, Maccioni

(2008) Insulin resistance and Alzheimer’s disease: Molecular links & clinical implications. Curr Alzheimer Res 5, 438–447.

62.

Zhong

, Ya

, Jia

, Hong

, Wang

, Jun

(2012) Hyperinsulinemia, insulin resistance and cognitive decline in older cohort. Biomed Environ Sci 25, 8–14.

63.

Felger

, Haroon

, Patel

, Goldsmith

, Wommack

, Woolwine

, Le

, Feinberg

, Tansey

, Miller

(2020) What does plasma CRP tell us about peripheral and central inflammation in depression? Mol Psychiatry 25, 1301–1311.

64.

Quan

, Banks

(2007) Brain-immune communication pathways. Brain Behav Immun 21, 727–735.

65.

Goehler

, Gaykema

, Hansen

, Anderson

, Maier

, Watkins

(2000) Vagal immune-to-brain communication: A visceral chemosensory pathway. Auton Neurosci 85, 49–59.

66.

D’Mello

, Le

, Swain

(2009) Cerebral microglia recruit monocytes into the brain in response to tumor necrosis factoralpha signaling during peripheral organ inflammation. J Neurosci 29, 2089–2102.

67.

Heaney

, Carroll

, Phillips

(2014) Physical activity, life events stress, cortisol, and DHEA: Preliminary findings that physical activity may buffer against the negative effects of stress. J Aging Phys Act 22, 465–473.

68.

Ponce

, Del Arco

, Loprinzi

(2019) Physical activity versus psychological stress: Effects on salivary cortisol and working memory performance. Medicina (Kaunas) 55.

69.

Morris

(2016) Nutrition and risk of dementia: Overview and methodological issues. Ann N Y Acad Sci 1367, 31–37.

70.

Yannakoulia

, Ntanasi

, Anastasiou

, Scarmeas

(2017) Frailty and nutrition: From epidemiological and clinical evidence to potential mechanisms. Metab Clin Exp 68, 64–76.

71.

McGrattan

, McEvoy

, McGuinness

, McKinley

, Woodside

(2018) Effect of dietary interventions in mild cognitive impairment: A systematic review. Br J Nutr 120, 1388–1405.

72.

Greco

, D’Onofrio

(2020) The complexity of frailty: Psychological mechanism and therapeutic interventions in old people – A narrative review. In Frailty in the Elderly - Understanding and Managing Complexity.

73.

Navarro-Pardo

, Facal

, Campos-Magdaleno

, Pereiro

, Juncos-Rabadan

(2020) Prevalence of cognitive frailty, do psychosocial-related factors matter? Brain Sci 10, 968.

74.

Langlois

, Vu

TTM

, Kergoat

, Chassé

, Dupuis

, Bherer

(2012) The multiple dimensions of frailty: Physical capacity, cognition, and quality of life. Int Psychogeriatr 24, 1429–1436.

75.

Strotmeyer

, Arnold

, Boudreau

, Ives

, Cushman

, Robbins

, Harris

, Newman

(2010) Long-term retention of older adults in the Cardiovascular Health Study: Implications for studies of the oldest old. J Am Geriatr Soc 58, 696–701.

76.

Samuel

, Rowe

(2012) An investigation of the association between grip strength and hip and knee joint moments in older adults. Arch Gerontol Geriatr 54, 357–360.

77.

Al Snih

, Markides

, Ottenbacher

, Raji

(2004) Hand grip strength and incident ADL disability in elderly Mexican Americans over a seven-year period. Aging Clin Exp Res 16, 481–486.

78.

Hashimoto

, Hatayama

, Terauchi

, Saito

, Higuchi

, Chikuda

(2020) Preoperative hand-grip strength can be a predictor of stair ascent and descent ability after total knee arthroplasty in female patients. J Orthop Sci 25, 167–172.

79.

Strandkvist

, Larsson

, Pauelsen

, Nyberg

, Vikman

, Lindberg

, Gustafsson

, Röijezon

(2021) Hand grip strength is strongly associated with lower limb strength but only weakly with postural control in community-dwelling older adults. Arch Gerontol Geriatr 94, 104345.