Association between cognitive impairment and criteria for frailty syndrome among older adults

Abstract The association between cognitive impairment and physical frailty has been studied in older adults. The criteria degree of frailty may be keys to associated cognitive impairment. Objective: To analyze the association between cognitive impairment and the criteria for frailty. Methods: We cross-sectionally examined data from 667 older adults (≥60 years of age) from a study entitled ‘Variables associated to cognition in elderly caregivers’ involving patients in an urban and rural primary healthcare center. We defined cognitive impairment based on different groups of scores on the Mini Mental State Examination, and defined frailty and prefrailty using the criteria by the Cardiovascular Health Study. We performed multinomial regression models to analyze the association between levels of frailty and cognitive impairment. Results: Similar proportions of women (54.8%) and men (45.2%) participated in the study (mean age: 71 years old). We found cognitive impairment, prefrailty and frailty in 34, 54, and 24% of the participants, respectively. Concomitant cognitive impairment and frailty was found in 13% of them. The chances of cognitive impairment increased up to 330% ( Odds Ratio [OR]: 4.3; 95% confidence interval [95%CI] 2.4‒7.7; p<0.001) among frail individuals, and 70% (OR: 1.7; 95%CI 1.0‒2.8; p=0.033) among prefrail individuals compared to robust/non-frail individuals. After controlling for age, education, place of residence and functional dependence, slowness and fatigue criteria were significantly associated with cognitive impairment. Conclusion: Older adults with frailty have a greater likelihood of concomitant cognitive impairment than prefrail and robust older adults. The prevalence of cognitive impairment and frailty is consistent with data reported in literature. The present findings contribute to the investigation of cognitive frailty.

A slight decrease in cognitive function is expected during the ageing process. However, cognitive impairment can occur when one' s performance regarding memory, judgment, language, and attention is lower than that expected for one' s age and educational level 1,2 . Cognitive impairment can be caused by neurodegeneration, vascular problems and metabolic problems. Nonetheless, chronic stress, depressive symptoms and anxiety can contribute to a poorer mental performance during old age 3 . Lately, the poor physical function, such as frailty, is considered another strong factor linked to cognitive impairment, seen that these conditions share similar pathophysiological mechanisms on the cellular and systemic levels 4,5 .
Different theories and particular (but complementary) evaluations of frailty in older adults have contributed to the health care, to the comprehensive geriatric and gerontological assessment, and supported the interventions planning. The Cardiovascular Health Study (CHS) defined frailty as a geriatric syndrome that could be assessed using the measurement of five clinical criteria: unintentional weight loss, fatigue (exhaustion), muscle weakness, slow gait/slowness and low levels of physical activity 6 . More recently, Morley et al. contributed to the definition of the clinical syndrome as a multiple-cause condition that leads to vulnerability, functional dependence, and death 7 .
Other frailty theories and measures are also useful to predict cognitive impairment 8 . However, a physical examination using the CHS frailty clinical criteria 6 may indicate changes in cognitive functions. Systematic reviews and meta-analyses confirmed the existence of a strong link between physical and cognitive impairment 5,9,10 . Although there is evidence associating frailty with cognitive impairment, a small number of studies on this subject have been conducted considering the five clinical criteria, individually, in low and middle-income countries.
In Brazil, an analysis of the FIBRA study described the criteria of slow gait speed (slowness) and low grip strength (weakness) as the strongest measures associated with cognitive performance among older adults 11 . A further FIBRA study analysis with 384 community-dwelling older adults confirmed that frailty and specific cognitive domains are linked, with a poorer performance as to time orientation and working memory prevalent among frail older adults 12 . Similar findings were observed in 737 participants of a study conducted in Rio de Janeiro City 13 and in a multi-centric Brazilian study 14 . Older adults with frailty had consistently lower Mini Mental State Examination (MMSE) scores compared to prefrail and robust older adults 13,14 . A systematic review with 29,664 participants in 19 studies, which were mostly conducted in Latin America, found that memory is the main function affected in older adults with frailty, and slowness and weakness are the most prevalent frailty clinical criteria in cognitively impaired older adults 15 .
Despite the growing interest in investigating the association between cognitive and physical status, further studies should be conducted with older adults living in low and middle-income countries. Therefore, the aim of the present study was to analyze the association between cognitive impairment and the clinical criteria for frailty syndrome. We hypothesized that some frailty clinical criteria are strongly associated with cognitive impairment in older adults. Additionally, we want to confirm whether frailty presents a close association with cognitive impairment, compared to prefrailty and robust older adults.

Participants
The present cross-sectional study is part of a study entitled "Variables associated with cognition in elderly caregivers" conducted by the Health and Ageing Group of Universidade Federal de São Carlos involving individuals registered in the Family Health Units of São Carlos City, São Paulo State, Brazil. São Carlos is in the Southeastern region of the country and has an estimated population of 221,950 residents, among whom 13% were aged 60 or older, according to the 2010 Brazilian census 16 .
The study was conducted between April and December of 2014. The participant selection process is described elsewhere 3,17,18 , but a brief description follows. All communitydwelling older adults (age≥60 in Brazil, as defined by the World Health Organization) registered at 18 primary healthcare centers (n=1,188) in São Carlos City, Brazil, were contacted in person and invited to participate in the survey. Individuals with auditory, visual or language limitations that could constitute barriers to the data collection instruments were excluded. The response rate was 59.1%. The survey was conducted with 351 community-dwelling older caregivers and 351 communitydwelling older non-caregivers (total: 702 individuals) registered with primary care services in rural and urban regions. For the present study, 667 individuals were included, and the single criterion for entry was having complete data available on demographics, cognitive and frailty status.
The present study was approved by the Human Research Ethics Committee of Universidade Federal de São Carlos (certificate number: 517.182) and all participants signed an informed consent. Household interviews were conducted by trained professionals in the fields of Gerontology and Nursing.

Variables and evaluations
1) Demographic characteristics: gender (male, female), age (continuous and age range), years of education (continuous and education level), retirement (yes, no), place of residence (rural, urban), and ethnicity (black/brown, white and others).
2) Activities of daily living (ADL): Functioning was assessed using the Lawton and Brody Scale to determine the degree of independence on basic activities, such as performing housework, handling money, using the telephone, administering medications, traveling, shopping and preparing full meals. The total score ranges from seven (complete dependence) to 21 (complete independence), with intermediate scores (8 to 20 points) indicative of partial dependence 19,20 .
3) Cognitive impairment: Cognitive screening was performed using the Mini Mental State Examination (MMSE), the score of which ranges from 0 to 30 21 . The cutoff points were adjusted for different levels of formal education: <26 for those with ≥nine years of schooling; <24 for those with five to eight years of schooling, <22 for those with one to four years of schooling, and <17 for illiterate individuals 22 . Addenbrooke's Cognitive Examination -Revised (ACE-R; score: 0-100) was also used to assess global cognition 23,24 .
4) The frailty syndrome and criteria: The five frailty clinical criteria of the Cardiovascular Health Study were considered: unintentional weight loss in the past year, fatigue in the past week, muscle weakness, slowness and decreased physical activity levels when compared to the previous year. Unintentional weight loss in the past year, fatigue in the past week and decreased physical activity level were self-declared. Muscle weakness was assessed using a dynamometer and slowness, with the time required to walk 4.6 meters. Based on Fried' s phenotype, the number of criteria was used to determine the level of frailty: frail ( from three to five criteria), prefrail (one or two criteria) and robust/non-frail (negative for all five criteria) 6 .

Statistical analysis
The Statistical Package for the Social Sciences -SPSS software, version 21.0 program was used for the data analysis. Descriptive statistics were performed to characterize the overall sample and the sample stratified by cognitive status. The values for frequency (n), percentage (%), mean and standard deviation (±) were calculated. The independent t-test was used to compare means and the chi-square test was used to compare categorical variables between groups according to gender ( Table  1). The prevalence of simultaneous cognitive impairment and frailty was estimated with 95% confidence intervals (95%CI). One-way ANOVA was used with Tukey' s post hoc test for comparisons of MMSE scores between frailty levels ( Figure 1).
Multinomial regression models were run to analyze the associations between frailty syndrome/criteria (independent variable) and cognitive impairment (dependent variable). Crude models were run to determine associations between age (continuous), education (continuous), gender (reference: male), place of residence (reference: rural), degree of dependence on ADL (reference: independent), unintentional weight loss, fatigue, weakness, slowness, low physical activity (reference: absence of criteria), prefrailty and frailty (reference: no frailty). Variables with a p-value≤0.20 were selected for the adjusted regression remodel. The first model (Table 2) included all criteria as controlling variables in the same model. Prefrailty (Model B) and frailty (Model C) were incorporated independently in adjusted models. Associations with a p-value≤0.05 were considered statistically significant.

RESULTS
Among the 702 participants, 35 were excluded from the analysis due the missing data on education, cognitive and frailty status. The sample consisted of similar proportions of women (54.8%) and men (45.2%). Women tended to be younger (mean difference: -1.6 years; p=0.012), more independent regarding ADL (w: 33.6% vs. m: 8%) and fewer were retired compared to men (w: 64.5% vs. m: 91.7%). Women also had higher proportions of slowness (w: 27% vs. m: 18.3%) and low physical activity compared to men (w: 54.9% vs. m: 47.2%).
No differences between women and men were found regarding the prevalence of cognitive impairment (Table 1). Cognitively impaired older adults tended to be older (mean difference: 2.8 years; p<0.001) than those with normal cognition. Regarding performance on ADL, 15.5% of cognitively impaired and 25.4% of older adults with normal cognitive were completely independent. Regarding frailty, except for unintentional weight loss, all criteria were more prevalent in older adults with cognitive impairment.
The prevalence of simultaneous condition cognitive impairment and frailty was 13.2% (95%CI 11-16) and the prevalence of concurrent cognitive impairment and prefrailty was . Figure 1 displays the MMSE scores among the levels of frailty. The mean MMSE score was 23.9±3.8 among robust individuals. Prefrail individuals had a poorer MMSE score compared to robust individuals (mean difference: -1.5; p<0.01), and the mean difference in the frail group compared to robust individuals was -5.2 (p<0.01).
As shown in Table 2 and Figure 2, only fatigue/exhaustion and slowness remained associated with cognitive impairment in the model controlled for age, education, place of residence, dependence on ADL and other frailty criteria. Individuals with fatigue were 1.1 times more likely to exhibit cognitive impairment, when compared to those without this criterion. Moreover, individuals with slowness were 2.6 times more likely to exhibit cognitive impairment ( Table 2; Model A). Frailty was more linked to cognitive impairment than prefrailty. The chances of cognitive impairment increased up to 330% in individuals with frailty (Model B) and 70% in individuals with prefrailty, when compared to robust individuals (Model C).

DISCUSSION
One third of the participants presented cognitive impairment, one quarter was frail, and one half was prefrail. The analyses confirmed that frailty was strongly associated with cognitive impairment and fatigue, and that slowness seemed to be the clinical criteria associated with cognitive impairment.
The prevalence of cognitive impairment in the population-based SABE study in Brazil was 7.9% 25 . In another study, the proportion of elderly people with some degree of cognitive impairment was 13.6% 26 . Similar prevalence rates of frailty were found in other middle-income and low-income countries. In studies conducted in Colombia, the prevalence of frailty was 12.2% 27,28 . In Taiwan, the prevalence of frailty and prefrailty was 4.9% and around 40%, respectively 29 . A systematic review analyzing 19 studies held in Latin America found that the prevalence of cognitive impairment ranged from 16 to 25%, and frailty was present in 10% of the population 15 . The proportion of cognitive impairment in studies may vary due to the profile of older adults in the sample, as well as the measures and cut-off points employed. In the present study, most participants had less than five years of schooling and the full version of MMSE was used. Moreover, clinical frailty criterion of low physical activity was more prevalent, which can be explained by the demographics, characterized as female and older, which are conditions associated with physical inactivity 30 .
In a study involving 2,375 Chinese older adults, the estimated prevalence of frailty with cognitive impairment was 1.8% and the estimated prevalence of prefrailty with cognitive impairment was 8.9% 31 . Half of the population had completed high school and scored significantly higher on the MMSE. Moreover, 61 participants were categorized with frailty, using the frailty phenotype criteria, and the prevalence of cognitive frailty increased fivefold among individuals aged 75 and older 32 . A Japanese study involving 4,207 participants found a 2.7% combined prevalence of MCI and frailty (3% in women and 2.4% in men). This combination increased to 4.4% among individuals with a low level of schooling. The regression analysis adjusted by gender, age and education level showed that older adults with frailty had a 100% increased chance of presenting MCI 33 . In a study involving Chilean older adults, individuals with frailty had 3.93 times more chance of presenting MCI 34 . A study conducted in Brazil, with 51 prefrail and frail older adults used a similar MMSE cut-off. Frailty was treated as the dependent variable and global cognition explained up to 19% of the variation in the syndrome 35 . Furthermore, a longitudinal study demonstrated that 27.8% of non-frail individuals will not experience cognitive decline, whereas only 2% of frail older adults improve or stabilize their cognitive status 8 .
Frailty criteria also seem to be associated with cognitive impairment. A longitudinal survey involving 2,817 Japanese men showed that individual frailty factors were associated with a 16 to 18% reduction in their global cognitive status 36 . In another study, slowness and physical exhaustion ( fatigue) were associated with a reduction in global cognition 37 . Slowness is the strongest frailty criterion associated to cognitive impairment, and this association has been frequently seen in literature. Additionally, in this study with 4,649 participants aged ≥50, prefrail individuals (n=1,444) had lower MMSE scores than robust individuals (n=3,155), and frail individuals (n=90) had lower MMSE scores compared to the other two groups 37 . A study involving 395 American older adults found than an increase in walking speed was associated with a subsequent improvement in cognitive performance, especially recalling 38 . This finding underscores the   importance of measuring gait speed and other components of frailty to identify older adults at risk of dysfunctional cognition and its determinants 39,40 .
In Brazil, the FIBRA study conducted in a low-income community used the same MMSE cut-off as that used in the present study and found that weakness was associated with global cognitive impairment, whereas slowness was specifically associated with a poorer performance regarding verbal fluency and the clock drawing test 11 .
Some studies suggest biological pathways that may occur in both cognitive impairment and frailty. These mechanisms involve markers, such as sociodemographic clinical, inflammatory/immunity, and laboratorial characteristics, as well as proteins, metabolism/oxidative stress and genetics. Sociodemographic factors include advanced age, female gender, widowhood, low formal education and financial income 5,41 . Clinical factors, besides others, include cardiovascular conditions (diabetes, dyslipidemias and hypertension), nutritional deficiencies (malnutrition and vitamin D deficiency), functional dependence, hormonal dysregulation (reduction in testosterone and insulin resistance), inflammation and neurotoxic accumulation of the protein betaamyloid in the brain, loss of neurons of the substantia nigra, symptoms of depression, use of medications and other drugs, lifestyle and worse perception of health 5,41 .
The investigation of shared mechanisms in physiological conditions is a new field of study, which limits hypothesizing the pathways of clinical frailty criteria and the decline in cognitive functioning, despite the fact that the outcomes are known. Both frailty and cognitive impairment are risk factors for future adverse outcomes, such as dementia, disability, hospitalization and death. These outcomes have been confirmed in Brazilian and non-Brazilian older adults 42, 43,44,45 .
The major limitation of the present study was the nonevaluation of dementia, which may affect the interpretation of results. The cross-sectional study design also limited us from knowing causal effects. On the other hand, one of the strengths of the study was the use of the MMSE with different cutoff points based on education level, which lends credibility to the assessment of cognition among the participants.
In conclusion, older adults with frailty scored lower on the MMSE than those individuals with prefrailty or robustness. Moreover, the prevalence of cognitive impairment, frailty and prefrailty in the present sample is consistent with data reported in literature. The frailty clinical criteria fatigue and slowness were associated to cognitive impairment; slowness seems to be the strongest criteria associated with this condition. The present findings contribute to the investigation of cognitive frailty in Brazil.