Cognitive Frailty as a Risk Factor for Incident Disability During Late Life: A 24-Month Follow-Up Longitudinal Study

Abstract

Objectives

Association between cognitive frailty as identified by a new operational definition and incident disability in the community setting remains unclear. This will be the catalyst for preventive interventions designed to treat adverse health problems among elderlies.

Design

A 24-month follow-up longitudinal study on a community-based cohort.

Setting

Community-setting.

Participants

Participants included a total of 9,936 older adults aged 65 years or older.

Measurements

Frailty was characterized as slow walking speed or/and muscle weakness represented by grip strength. Cognitive function was assessed according to several tests. Cognitive impairment was defined below the age-education reference threshold. Participants were categorized into the four groups: robust, cognitive impairment alone, frailty alone, and cognitive frailty (both frail and cognitive impairment). Incident disability data was extracted from the Japanese Long-Term Care system.

Results

The prevalence of cognitive frailty was 11.2%. The cumulative incidence rates of incident disability in each group were also estimated (robust, 9.6/1,000 person-years (95% CI 7.9 to 11.7); cognitive impairment, 21.3/1,000 person years (95% CI 16.3 to 27.7); frailty, 45.4/1,000 person years (95% CI 39.5 to 52.3); and cognitive frailty, 79.9/1,000 person years (95% CI 68.6 to 93.1)). Adjusted Cox proportional hazard model revealed that the cognitive frailty group had the highest hazard ratio (HR 3.86, 95%CI 2.95–5.05, P < 0.001).

Conclusions

A proper operational definition was developed to determine cognitive frailty among elderlies. Cognitive frailty is more associated with incident disability in community-setting than cognitive impairment or physical frailty alone.

Introduction

Japan implemented mandatory long-term care insurance (LTCI) in 2000. Here, all Japanese residents aged 65 and older are eligible for institutional and community-based services when physical and/or mental care/support is needed. The use of these services has dramatically increased over the 15-year period since the introduction of the LTCI system. Associated costs have thus almost tripled (from 32.2 billion dollars in 2000 to 87.9 billion dollars in 2015) (1). It is essential to reduce disability periods while increasing health among older adults (2); this means reducing LTCI costs to achieve fiscal soundness in Japan. It is therefore crucial to explore the risk factors associated with such disability.

Frailty is a geriatric syndrome associate with dysfunction in multiple organ systems; this reflects a state of vulnerability and the loss of physiological reserve among older adults (3). Frailty is also a strong predictor for various adverse health outcomes (4), including the need for long-term care or support (5). Due to its complex pathophysiological process, frailty has been categorized into different domains (i.e., the physical, cognitive and social) to facilitate further research (6, 7). The international consensus group proposed that “cognitive frailty” was a heterogeneous clinical manifestation characterized by the simultaneous presence of both physical frailty and cognitive impairment (8). Longitudinal population-based studies have also investigated cognitive frailty models associated with increased adverse health problems (i.e., incident dementia (9) and all-cause mortality (10)). Findings suggest that cognitive frailty may impact incident disability, but this association remains unclear.

However, the definition proposed by the consensus group has rendered a low prevalence of cognitive frailty in community-based settings (1.0 to 12.1%) (11). An appropriate operational definition should be used when evaluating high-risk older persons to provide interventions that promote increased health, thus the definition should have higher sensitivity for risk factors to detect a sufficient number of high-risk people in the community setting. Our research group thus developed new operational definitions (9, 12), as follows: 1) Physical frailty is the presence of a slow walking speed or muscle weakness, while 2) cognitive impairment is found when an individual cognitive performance score is lower than 1.5 standard deviations for each cognitive domain in the age-education matched norm among the peer population. This study investigated the association between cognitive frailty according to these new definitions and incident disability using longitudinal cohort data.

Methods

Participants

This prospective cohort study examined data from 10,885 community-dwelling older adults aged 65 years and above who were enrolled in the National Center for Geriatrics and Gerontology-Study of Geriatric Syndromes (NCGG-SGS). NCGG-SGS participants were recruited from both Nagoya and Obu, Japan. All participants resided in Obu (August 2011-February 2012, June 2013) or Nagoya city (July 2013–December 2013) and were aged 65 years or older at the time of examination. Participants were excluded if they required support or care as certified by the Japanese public LTCI system, were dependent on basic activities of daily living, were unable to complete performance-based assessments, had a history of neurodegenerative diseases (e.g., Parkinson's disease or dementia), or scored less than 21 on the Mini-Mental State Examination (MMSE) (an indicator of moderate dementia) (13). Participants with missing data were also excluded. Of the 10,885 initial participants, 949 were excluded from analysis based on these criteria. This resulted in a final sample of 9,936 older adults (mean age 73.5 ± 5.4 years; 4,797 men (48.3%)). Informed consent was obtained from all participants beforehand. This study's protocol was approved by the Ethics Committee of the National Center for Geriatrics and Gerontology.

Operationalization of cognitive frailty

A consensus group proposed the identification of so-called “cognitive frailty” as a heterogeneous clinical manifestation characterized by the simultaneous presence of both physical frailty and cognitive impairment (8).

First, participants were physically assessed according to walking-speed and grip-strength measurements. Here, walking speed was measured by referring to the seconds frame on a stopwatch. Participants were asked to walk on a flat and straight surface at a comfortable speed. Start and end points were marked to create a 2m stretch on a 2.4m walking path so that participants could achieve a comfortable speed before entering the timed section. Participants were asked to continue walking along a 2m section past the end of the path to ensure a consistent walking pace while on the timed section. Walking-speed deficits were determined according to a previously established cutoff point (<1.0 m/s) (14, 15). Grip strength was then measured in kilograms using a Smedley-type handheld dynamometer (GRIP-D; Takei Ltd., Niigata, Japan). Here, grip-strength deficits were determined based on sex-specific cutoff points (i.e., <26 kg for men and <18 kg for women) (16). Participants with either of these two physical deficits were considered physically frail.

Cognitive screening was conducted according to the National Center for Geriatrics and Gerontology-Functional Assessment Tool (NCGG-FAT) (17). The NCGG-FAT consists of the four following domains: memory (word list memory-I (immediate recognition) and word list memory-II (delayed recall)), attention (an electronic tablet version of the Trail Making Test, TMT-part A), executive function (an electronic tablet version of the TMT-part B), and processing speed (an electronic tablet version of the Symbol Digit Substitution Test). Each participant was given approximately 20 minutes to complete the battery of tests. High test-retest reliability and moderate-to-high validity has been confirmed for the NCGG-FAT (17) among community-dwelling adults. All tests used in this study had established standardized thresholds for the definition of impairment in the corresponding domain (i.e., a score <1.5 SDs below the age and education-specific means) for a population-based cohort of community-dwelling older adults. We determined the presence of cognitive impairment among participants who fell under the standardized threshold in one or more NCGG-FAT test.

Finally, participants were categorized into the four following groups: Robust (non-physically frail older adults without cognitive impairment), cognitive impairment (non-physically frail older adults with cognitive impairment), physical frailty (physically frail older adults without cognitive impairment), and cognitive frailty (physically frail older adults with cognitive impairment) (8).

Acquisition of data for incident disability using the LTCI in Japan

Participants were followed on a monthly basis to determine incident disability according to the LTCI system for a 24-month period after the baseline assessment. Japan implemented a mandatory social LTCI system on April 1, 2000 (18, 19). Japanese citizens aged 65 and older are eligible for benefits (i.e., institutional/community-based services, but not cash) related to their physical and/or mental disability needs. The computer-aided standardized needs-assessment system used by the mandatory social LTCI system categorizes people according to seven levels of need (19). Individual nursing care needs are determined by trained local government officials who visit the homes of prospective recipients to administer questionnaires about current physical and mental statuses. Each questionnaire consists of 73 items among seven dimensions (e.g., paralysis and limitation of joint movement, movement and balance, complex movement, conditions requiring special assistance, activities of daily living/instrumental activities of daily living, communication and cognition, and behavioral problems) and 12 items related to the use of medical procedures. Respondent answers are recorded on a computer to calculate standardized scores based on the seven physical and mental status dimensions. This is also done to estimate the time needed for nine care categories (i.e., grooming/bathing, eating, toileting, transferring, eating, assistance with instrumental activities of daily living, behavioral problems, rehabilitation, and medical services). A needs level is then assigned based on the total estimated time required for care. The Nursing Care Needs Certification Board (comprised of physicians, nurses, and other experts in health and social services) reviews and confirms the care-needs level for each recipient (19). We defined the onset of long-term care/support need as the point when a participant was certified for LTCI.

Sociodemographic variables and covariates

We conducted face-to-face interviews with participants to determine sociodemographic characteristics (age, gender, and educational level), medical history (number of medications and chronic diseases (hypertension, hyperlipidemia, diabetes, stroke, and osteoarthritis)), and health-related behaviors (current smoking habit, current alcohol use, and physical inactivity). Body mass indexes (BMIs) and depressive symptoms were also included as covariates.

Depressive symptoms were measured at baseline using the 15-item Geriatric Depression Scale (GDS) (20), which consists of 15 yes/no questions that are combined to determine a score between 0 and 15. We evaluated physical activity according to the following questions: (1) “Do you engage in moderate levels of physical exercise or sports?” and (2) “Do you engage in low-intensity physical exercise for the purpose of maintaining health?” Participants who answered “no” to both questions were considered physically inactive (14). GDS scores were categorized using cutoffs derived from previous studies (GDS: <5/≥6 (20)).

Statistical analysis

All data-entry processes and analyses were performed using IBM SPSS Statistics 25.0 (SPSS Inc., New York, NY, USA). All statistical significance levels were set at P < 0.05 a priori. An analysis of variance and chi square test were used to investigate the differences between cognitive frailty groups. We calculated cumulative incident disability during follow-ups for each of the four abovementioned groups (i.e., the robust, cognitive impairment, physical frailty, and cognitive frailty groups) according to Kaplan-Meier curves. Intergroup differences were estimated using the log-rank test. Crude and adjusted Cox proportional hazard models were constructed to calculate hazard ratios (HR) with 95% confidence intervals (CI) for risk of incident disability.

Results

The demographic characteristics and scale scores for each group are shown in Table 1. Mean scores varied significantly for the following variables between groups: Age, sex, education, medications, all chronic disease, physical function, cognitive function, GDS, current drinking habit, current smoking habit, physical inactivity, and incident disability. BMI was not significantly related to intergroup differences (P = 0.725).

Table 1.

Participants' characteristics

Variables	All participants	Robust	Cognitive impairment	Physical frailty	Cognitive frailty	P value
	n = 9,936	n= 5,274 (53.1%)	n = 1,303 (13.1%)	n= 2,250 (22.6%)	n = 1,109 (11.2%)
Sociodemographic factor
Age, year	73.5±5.4	72.1±4.7	72.3±4.8	76.0±5.7	76.5±5.6	< 0.001
Men, number	4797 (48.3)	2637 (50.0)	667 (51.2)	999 (44.4)	494 (44.5)	< 0.001
Education, year	11.8±2.6	12.1±2.5	11.6±2.4	11.5±2.7	11.0±2.6	< 0.001
BMI, kg/m2	23.2±3.1	23.2±2.9	23.1±3.1	23.1±3.3	23.1±3.4	0.725
Medical factor
Medication, number	2.7±2.6	2.3±2.3	2.4±2.4	3.5±2.9	3.6±2.9	< 0.001
Hypertension, number	4661 (46.9)	2340 (44.4)	580 (44.5)	1152 (51.2)	589 (53.1)	< 0.001
Hyperlipidemia, number	4074 (41.0)	2209 (41.9)	509 (39.1)	952 (42.3)	404 (36.4)	0.002
Diabetes, number	1285 (12.9)	568 (10.8)	166 (12.7)	359 (16.0)	192 (17.3)	< 0.001
Stroke, number	564 (5.7)	222 (4.2)	84 (6.4)	151 (6.7)	107 (9.6)	< 0.001
Osteoarthritis, number	1797 (18.1)	857 (16.2)	199 (15.3)	515 (22.9)	226 (20.4)	< 0.001
Physical function
Grip strength, kg	26.8±7.8	28.8±7.3	28.4±7.2	22.9±7.2	22.4±7.2	< 0.001
Gait speed, m/sec	1.15±0.22	1.25±0.16	1.22±0.16	0.97±0.20	0.91±0.19	< 0.001
Neuropsychological function
Word list memory, score	11.2±2.9	11.4±2.7	11.1±2.9	10.2±3.1	11.2±2.9	< 0.001
TMT-A, sec	21.4±6.5	19.2±3.9	24.6±8.1	21.1±4.5	28.4±10.1	< 0.001
TMT-B, sec	43.5±17.5	35.7±9.7	58.2±19.9	41.1±11.4	68.4±19.7	< 0.001
SDST; score	37.9±7.9	41.1±6.5	34.3±7.2	37.0±6.7	29.0±7.5	< 0.001
GDS, score	2.8±2.7	2.4±2.4	2.9±2.7	3.3±2.9	3.8±2.9	< 0.001
Life-style factor
Current drinker, number	4414 (44.4)	2523 (47.8)	592 (45.4)	893 (39.7)	406 (36.6)	< 0.001
Current smoker, number	818 (8.2)	412 (7.8)	148 (11.4)	161 (7.2)	97 (8.7)	< 0.001
Physical inactivity, number	2442 (24.6)	1122 (21.3)	316 (24.3)	643 (28.6)	361 (32.6)	< 0.001
Incident disability	513 (5.2)	100 (1.9)	54 (4.1)	195 (8.7)	164 (14.8)	< 0.001

note. The continuous variables were analyzed using analysis of variance (mean value ± standard deviation), and categorical variables were analyzed using chi square test (number (%)). Significance set at P < 0.05; BMI, body mass index; TMT, Trail Making Test; SDST, symbol digit substitution task; GDS, Geriatric Depression Scale

During follow-up (mean 23.3 ± 3.2 months), the incident rate of disability was 26.6/1,000 person-years (95% CI 24.4 to 29.0), and there were 116 censored data (death, n = 78 [0.8%]; moving out from research field, n = 38 [0.4%]). Figure 1 shows the probability of independence according to the Kaplan-Meier analysis; the cognitive frailty group showed the highest onset rate of disability. The log-rank test revealed significant intergroup differences among incident disability rates (Table 2).

Figure 1.

Kaplan-Meier survival curves showing the relationship between incident disability and cognitive frailty status

Table 2.

Difference in incident disability by log-rank test

	Robust		Cognitive impairment		Physical frailty
	X2	P value	X2	P value	X2	P value
Cognitive impairment	23.620	< 0.001
Physical frailty	196.507	< 0.001	25.749	< 0.001
Cognitive frailty	405.716	< 0.001	83.121	< 0.001	29.575	< 0.001

Significance set at P < 0.05.

Finally, Table 3 shows the unadjusted and adjusted survival probabilities for incident disability using the Cox proportional hazard risk analysis. Cumulative incidence rates were estimated for each group, as follows: Robust, 9.6/1,000 person-years (95% CI 7.9 to 11.7); cognitive impairment, 21.3/1,000 person-years (95% CI 16.3 to 27.7); physical frailty, 45.4/1,000 person-years (95% CI 39.5 to 52.3); and cognitive frailty, 79.9/1,000 person-years (95% CI 68.6 to 93.1), respectively. The cognitive frailty group exhibited the highest risk of incident disability in both the unadjusted and adjusted models, followed by the physical frailty and cognitive impairment groups, respectively (Table 3).

Table 3.

Association of cognitive frailty with incident disability by Cox proportional hazard models

	At risk	Events	Rate per 1,000 person-years		Crude model			Adjusted model
Risk pattern	n	n	(95% Confidence Interval)		Hazard Ratio (95% Confidence Interval), P value
Cognitive frailty	1,109	164	79.9 (68.6–93.1)	8.39	(6.54–10.76)	< 0.001	3.86	(2.95–5.05)	< 0.001
Physical frailty	2,250	195	45.4 (39.5–52.3)	4.75	(3.73–6.05)	< 0.001	2.40	(1.86–3.10)	< 0.001
Cognitive impairment	1,303	54	21.3 (16.3–27.7)	2.22	(1.60–3.09)	< 0.001	2.09	(1.50–2.91)	< 0.001
Robust	5,274	100	9.6 (7.9–11.7)	Reference		Reference

Significance set at P < 0.05; Adjusted model: adjusted for age, sex, education, BMI, medication, hypertension, hyperlipidemia, diabetes, stroke, osteoarthrosis, current drinking habit, current smoking habit, physical inactivity; BMI, body mass index.

Discussion

The prevalence of cognitive frailty was identified at 11.4% according to a new operational definition. Incident disability was determined at 79.9/1,000 person years for this study's cognitive frailty group over a 24-month follow-up period. Older people with cognitive frailty exhibited the highest incident disability rate. This new operational definition of cognitive frailty also allowed for a higher probability of identifying incident disability compared to methods using cognitive impairment or physical frailty alone.

A review study using a previous operational definition among community-dwelling elderly participants estimated the prevalence of cognitive frailty at 1.0–12.1% (10). Our previous study using the same definition indicated that the prevalence of cognitive frailty was 1.2% among 8,864 Japanese elderly adult participants (21). However, the new operational definition used in this study revealed a prevalence of cognitive frailty almost 10 times higher (i.e., 1.2% vs 11.2%). An appropriate operational definition should be used when evaluating high-risk older persons to provide interventions that promote increased health. This definition should include the following features: 1) Higher sensitivity for risk factors (detectable for a sufficient number of high-risk people in the community setting), 2) predictive validity for future adverse health problems, and 3) a state with reversibility. This study's new operational definition of cognitive frailty revealed a different incident disability rate and could more easily be used to detect at-risk elderly persons when compared to the abovementioned previous definition. From this point of view, the new definition is thus a better fit.

This study's findings revealed that older adults with cognitive frailty were more likely have incident disability when compared to healthy individuals. Although they used a different operation definition when testing, previous studies have reported that cognitive frailty was associated with adverse health problems. For instance, a Singaporean three-year population-based longitudinal ageing study conducted among 2,375 participants aged 55 years or older (a mean age of 66) showed that cognitive frailty was associated with an increased incidence of functional disability and mortality (22). In addition, and Italian longitudinal study on aging discovered that frail older people with mild cognitive impairment were at higher risk of basic activities of daily living (BADL) disability and all-cause mortality over a 3.5-year follow-up period (23). Our previous study also revealed significant relationships between incident dementia and cognitive frailty, but not for physical frailty without cognitive impairment (24). Our results provided epidemiological evidence to support previous findings about the association between cognitive frailty and adverse health problems among community-dwelling older people.

The pathological mechanisms underlying the association between cognitive frailty and health problems (including incident disability) remain unclear. This may be because cognitive frailty proficiently represents a later stage of life. In fact, both cognitive and physical function have been found to decline as a result of aging (25, 26). Another explanation is that inflammation may trigger cognitive frailty, which then shifts to incident disability. Increasing levels of inflammatory markers (e.g., interleukin 6) are associated with not only lower grip strength (27), slower gait speed (28), and cognitive impairment (29), but also future hospitalization and mortality (30). Previous findings have also suggested that high inflammation levels in cognitively frail older people can be used to predict BADL disability (23). Cognitive frailty may thus represent a state of significantly increased inflammation. The condition therefore impacts future adverse health problems, including incident disability.

One of this study's major strengths was its use of monthly registry data accrued among a large sample from the Japanese LTCI system. However, there are some limitations. First, the 24-month follow-up period may have been too short to sufficiently capture incident disability among older adults. Second, although participants were randomly recruited through a direct-mailing process, only persons that could perform health checkups at home were selected for participation. This may have led to the recruitment of relatively healthy older people and thus resulted in an underestimation of incident disability. Finally, the registry data used contained no information on why participants were in a state of incident disability. It was thus difficult to clarify the physiological mechanisms underlying the association between cognitive frailty and incident disability.

In summary, a new operational definition was used to detect a more adequate number of elderly persons with cognitive frailty when compared to a previous definition. Cognitive frailty is more significantly associated with incident disability among community-dwelling older people when compared to cognitive impairment or physical frailty alone.

Author contributions

Tsutsumimoto planned the study and wrote the first draft of the manuscript, and coordinated the review and editing process leading to the final manuscript. Doi participated in the design of the study and wrote the paper. Nakakubo, Kim, Kurita and Ishii corrected data and contributed to the editorial process and review of the manuscript. Shimada supervised the study and suggested many of the ideas that have been pursued in this research, and participated in the planning, editorial, and review processes that led to the final manuscript.

Funding

This work received financial support via the Health Labour Sciences Research Grant (23–001) from the Japanese Ministry of Health, Labour, and Welfare, and the Research Funding for Longevity Sciences (22–16) from the National Center for Geriatrics and Gerontology (NCGG), Japan. Additional support was provided by a Grant-in-Aid for Scientific Research (B) to H.S. and a Grant-in-Aid for JSPS Fellows from the Japan Society for the Promotion of Science to H. M. The funding source played no role in the design or conduct of the study; collection, management, analysis, or interpretation of the data; or preparation, review, or approval of the manuscript..

Conflicts of interest

None.

Ethical standards

Informed consent was obtained from all participants beforehand. This study's protocol was approved by the Ethics Committee of the National Center for Geriatrics and Gerontology.