Transitions across the cognitive frailty spectrum and its association with anxiety in individuals aged 55 years and over: a 7 years longitudinal study

Abstract

This study examined how psychological factors influenced cognitive frailty (CF) transition at 7 years follow-up among Malaysians 55 years at recruitment in 2013. Data from 554 participants, aged (mean ± SD) = 76.03 ± 13.3 years, 56.9% female were included. Frailty was assessed through the Cardiovascular Health Index and Fatigue, Resistance, Ambulation, Illness and Loss of Weight scale at baseline and follow-up respectively. Presence of one or two symptoms were considered prefrail; three or more were categorised as frail. Mild Cognitive Impairment (MCI) was defined as Montreal Cognitive Assessment score of 22 at baseline, and 18 for the blind version administered at follow-up. CF (i.e. frailty and MCI) was present among 6 (1.1%) participants at baseline and 16 (2.8%) at follow-up. 93 (16.8%) experienced improvement and 461 (83.2%) had worsening in CF status over 7 years. The 21-item Depression, Anxiety and Stress Scale scores was associated with CF frailty transition. Improvement in CF status was observed among 92 (100%), 88 (97%) and 91 (100%) of those with normal to mild symptoms of depression, anxiety and stress respectively. Increased anxiety was found to be significantly associated with worsening transition over the 7 years period (odds ratio [95% CI]  3.38 [1.03, 11.10)]. Psychological status represents a potentially modifiable risk factor for progression within the CF continuum.

Introduction

The relentless increase in health and social care burden alongside population ageing highlights an urgent need to identify modifiable risk factors for potentially reversible conditions. Cognitive frailty (CF), defined as the co-occurrence of mild cognitive impairment (MCI) and physical frailty, is associated with increased susceptibility to adverse health outcomes including but not limited to high mortality risk, functional disability, poor quality of life, impaired health functions, disability and hospitalization^1–6. CF is also considered a dementia prodrome and hence a potential target for the prevention of progression to dementia⁴.

Tang et al.⁷ describes CF as a “dynamically changing state”. Individuals with either CF or the presence of MCI or physical frailty in isolation had increased over 6 years. Risk factors for progression to CF highlighted in existing studies include nutrition, pain, comorbidities, age, smoking, poor perceived health status, feeling sad, lack of exercise, social isolation, low income, insomnia, falls^7–11. Conversely, increased sociability, literacy and high levels of education and increased engagement in cognitively stimulating activities were found to be protective against progression to CF^7,12. Published research conducted thus far had primarily addressed predictive models^9,12, while a recent study explored factors which predicted improvement as well as progression related to CF status over a 2 years period¹¹.

A recently published systematic review which included 12 cross-sectional studies, and three longitudinal studies found that 46% of older adults with CF experience depression. Further, CF is associated with a two-fold risk of depression compared to those without CF¹³. The longitudinal relationship between the presence of depression and potential changes in CF status, however, have not previously been investigated. In addition, few published studies have evaluated the relationship between anxiety and stress with CF. This study, therefore, observed psychological status at baseline and its association with CF status transition over a 7 years period among community-dwelling adults aged 55 years and over.

Methodology

Study design

This study utilised a longitudinal study design. Psychological status was determined with the 21-item Depression, Anxiety and Stress Scale at baseline. Cognitive frailty was defined as comorbid frailty based on scores from the Cardiovascular Health Study (CHS) frailty index at baseline or the Fatigue, Resistance, Ambulation, Illness and Loss of Weight (FRAIL) scale at follow-up^14,15, both alongside presence of mild cognitive impairment (MCI) based on the Montreal Cognitive Assessment (MoCA) scores^16,17.

Study population

This study utilized baseline data from the Malaysian Elders Longitudinal Research (MELoR), which was later merged as a sub-cohort of the larger Transforming Cognitive Frailty into Later-Life Self-Sufficiency (AGELESS) at point of follow-up. Baseline cognition and frailty data were available for 993 participants. Participants aged above 55 years old were initially recruited in 2013 for the MELoR study via stratified random sampling by age deciles and ethnicity through the electoral rolls of three parliamentary constituencies within the Klang Valley¹⁸. The age cut-off of 55 years selected was in keeping with the retirement age of Malaysia at the time the study commenced. Further, this facilitated longitudinal assessments since those aged 55 years would then cross the 65 years threshold in 10 years. Follow-up data were obtained from 2020 to 2022 through virtual interviews held during the Coronavirus Disease 2019 (COVID-19) pandemic¹⁹. Exclusion criteria at both baseline and follow-up included communication difficulties due to major neurocognitive, speech and/or mental disorders. Figure 1 displays the flowchart of sample selection for longitudinal analyses.

Fig. 1.

Flowchart of sample selection for longitudinal analyses.

Cognitive frailty

Mild cognitive impairment

Presence of MCI was assessed using the MoCA test. The MoCA test was administered in its entirety during baseline assessments. However, the blind version of the MoCA test was administered during virtual interviews. The cut-off score to determine presence of MCI was 22 for the full-MoCA and 18 for MoCA-Blind^16,17.

Frailty

Frailty was assessed at baseline and follow-up using the CHS index and FRAIL scale respectively. The CHS data guided the development of Fried et.al. (2001)’s frailty phenotype which comprises five components—exhaustion, weakness, slowness, low levels of physical activity and unexplained weight loss¹⁴. These components were adapted to fit data from the MELoR survey²⁰. One point is awarded for each criteria met. Participants are characterized pre-frail if they fulfilled one or two criteria and frail if they met three or more criteria, the remaining participants were considered robust.

The FRAIL scale was used to define frailty at follow-up. Individual items were extracted from the AGELESS virtual survey that was administered via telephone/video call by trained enumerators. The five components of the FRAIL scale equated to verbal reports of deficits corresponding to domains the CHS index. The AGELESS questionnaire assessed each item²¹. Scores range from 0 to 5, where 0, 1 to 2, and 3 or more indicate robust, pre-frail and frail respectively¹⁵.

As older adults were shielded throughout the entirely of the COVID-19 pandemic, it was necessary to pivot physical assessment to virtual interviews¹⁹. Nonetheless, the study performed additional Kappa weighted agreement and Spearman’s correlation analyses to assess its interchangeability and suitability for the population (refer to Supplementary Table 1).

Table 1.

Participant characteristics based on cognitive frailty transition over 7 years period.

Characteristics (n = 554)	Improved	Worsened	p-value
Age, M (SD)	73.65 (6.73)	72.99 (6.56)	0.382
Female, n (%)	52 (55.9%)	267 (58.0%)	0.395
Educational category
No formal education	2 (1.8%)	7 (1.3%)	0.001*
Primary education	30 (26.5%)	56 (10.5%)
Secondary education	48 (42.5%)	240 (44.6%)
Tertiary education (Certificate/College/University)	33 (29.2%)	235 (43.7%)
Marital status, n (%)
Single/Divorced/Separated	8 (8.6%)	57 (12.4%)	0.553
Married	72 (77.4%)	335 (73.0%)
Widowed	13 (14.0%)	67 (14.6%)
Mortality, n (%)
2 (2.2%)	12 (2.6%)	0.573
Falls, n (%)
24 (25.8%)	82 (17.8%)	0.054*
Medical history
Ischemic heart disease	7 (7.5%)	37 (8.0%)	0.533
Hypertension	53 (57.0%)	231 (50.2%)	0.140
Cholesterol	61 (65.6%)	264 (57.4%)	0.088
Diabetes mellitus	21 (22.6%)	106 (23.0%)	0.522
Joint problem	13 (14.0%)	82 (17.8%)	0.231
Cataract/Glaucoma	29 (31.2%)	168 (36.5%)	0.195
Asthma	1 (1.1%)	38 (8.3%)	0.005*
Lung disease	1 (1.1%)	14 (3.0%)	0.251
Gout	11 (11.8%)	31 (6.7%)	0.075
Thyroid	6 (6.5%)	24 (5.2%)	0.392
Dyspepsia	7 (7.5%)	45 (9.8%)	0.324
Cancer	3 (3.2%)	40 (8.7%)	0.048*
Problem urinating	20 (21.5%)	102 (22.2%)	0.505
Alcohol consumption	29 (80.6%)	158 (88.3%)	0.162
Smoking	20 (21.5%)	80 (17.9%)	0.248
Living alone	6 (6.5%)	21 (4.6%)	0.291
Lawton’s independent activity of daily living
Moderate functioning	28 (30.1%)	92 (20.0%)	0.025*
Independent/High functioning	65 (69.9%)	367 (80.0%)
Lubben social network scale (LSNS-6)
Total score	16.57 (6.23)	17.63 (5.74)	0.131
Social isolation	22 (23.9%)	84 (18.4%)	0.140
Familial isolation	29 (31.5%)	131 (28.7%)	0.332
Friends isolation	27 (29.0%)	113 (24.7%)	0.255
Quality of life—CASP-12	28.43 (5.07)	28.39 (5.01)	0.949

*p-value significant at p < 0.050.

Cognitive frailty spectrum and transition

Within this study, individuals were classified into six groups within the CF spectrum according to frailty and MCI categories: robust (C₀F₀), MCI only (C_mF₀), prefrail only (C₀F_p), MCI and prefrail (C_mF_p), frail (C₀F_f), MCI and frail (C_mF_f). CF classifications were determined separately for baseline and follow-up.

Transition between CF classes from baseline to follow-up were dichotomized into the two groups of “improvement” or “worsening”. Improvement was defined as maintenance or reversal to the robust state (C₀F₀) or improvements in frailty scores without progression in cognitive scores, or improvements in cognitive scores without progression in frailty status. Worsening in CF status included those who progressed across frailty states (F₀ to F_p, F₀ to F_f or F_p to F_f) with or without development of MCI (C₀ or C_m), remained frail (F_f)/prefrail (F_p)/MCI (C_m) or progressed to MCI (C_o to C_m) regardless of frailty status.

Measures of psychological status

Baseline depression, anxiety and stress were assessed using the 21-item Depression, Anxiety and Stress Scale (DASS-21). There were 7-items for each subscale of depression, anxiety and stress. Items were rated based on a 4-point Likert scale ranging from 0 (did not apply to me at all) to 3 (applied to me very much, or most of the time). Scores were multiplied by two to obtain final score. Individuals with at least mild depression, anxiety or stress, which corresponded to the published cut-offs of five, four and eight or greater respectively were considered to have depressive, anxiety or stress symptoms within this study²².

Sociodemographic and baseline measures

Sociodemographic and baseline information such as age, sex, marital status, education, falls, medical history, quality of life, social network, activities of daily living, and physical activity, among others were assessed through the MELoR survey. Smoking, alcohol consumption and medical history were determined through participant self-reports. Mortality data was obtained from the Malaysian National Registry Department upon formal request.

Data analysis

Data were analysed using the IBM Statistical Package for the Social Sciences (SPSS) Version 27. As the FRAIL scale was not yet available in 2013–2016 when MELoR was conceived and was selected as a tool that could be used exclusively virtually for AGELESS, the FRAIL scale was correlated via Spearman’s correlation with the CHS frailty tool to determine its validity. Kappa weighted agreement for presence and absence of frailty was also performed (see Supplementary Table S1).

An alluvial plot was generated through the online platform RAWGraph (DensityDesign Research Lab, Milan, Italy) to visualise the transition within the CF spectrum from baseline to follow-up²³. Descriptive and frequency analyses were then conducted using the Chi-squared test for categorical variables and independent t-test for continuous variables based on of the two categories for CF status change (worsening vs. maintenance and improvement). Chi-squared tests were administered to identify potential differences in depression, anxiety and stress on cognitive frailty transition categories. Binomial logistic regression analyses were then conducted to observe the association between cognitive frailty transition and categories of depression, anxiety and stress. Potential confounders were selected based on p < 0.05 on the comparison table and other variables gathered through literature review and were adjusted for in the binomial logistic regression analyses. Statistical significance of p < 0.05 were enforced for all analyses conducted.

Ethics statement

Ethics approval was obtained from the University of Malaya Medical Centre Medical Research Ethics Committee (UMMC-MREC) (MED-ID 20191231-8121). Authors would like to affirm that written informed consent to participate were obtained from all participants during virtual interviews. The study was conducted in accordance with the Declaration of Helsinski and regulations stipulated by the UMMC-MREC.

Results

Transition in cognitive frailty state over 7 years period

A total of 101 (10.2%) participants had died and 338 (34.0%) were lost to follow-up at the end of the 7 years period. Data on cognition and frailty for a total of 554 participants were, therefore, available for analysis.

Figure 2 illustrates the transitions across the CF spectrum from baseline to follow-up. Of the 554 individuals who were successfully followed-up, C_mF_f was present among six (1.1%) participants at baseline and 16 (2.8%) at follow-up. Of the 222 who were C₀F₀ at baseline, 24 (10.8%) remained C₀F₀, 3 (1.4%) transitioned to C_mF₀, 155 (69.8%) transitioned to C₀F_p, 22 (9.9%) transitioned to C₀F_f, 15 (6.7%) transitioned to C_mF_p, while 3 (1.4%) progressed to C_mF_f. Of the 324 individuals who were pre-frail and/or MCI at baseline, 22 (6.8%) transitioned to C₀F₀ while 11 (3.4%) had transitioned to C_mF_f. Of the six C_mF_f participants, two (33.3%) retained their C_mF_f status while four had transitioned to C_mF₀, C₀F_p, and/or C_mF_p states, none returned to C₀F₀. In total, 14 (2.5%) developed C_mF_f over the 7 years period. Supplementary Table S2 provides detailed breakdown of cognitive frailty transition that took place over the 7 years period.

Fig. 2.

Alluvial diagram illustrating transitions in cognitive frailty spectrum over a 7-year period. Colours define baseline cognitive frailty status: robust (peach), prefrail only (blue-green), pre-frailty and mild cognitive impairment (lime-green), mild cognitive impairment only (blue), frail only (purple), cognitive frailty (red).

Participant characteristics

Basic characteristics of participants who had died or were lost to follow-up at the end of 7 years were also observed. Supplementary Table S3 compares the characteristics of participants who had died or were lost to follow-up with those who had completed follow-up. Those who had died were older, more likely to be women, had no formal education and were more likely to be single, divorced or separated. Women were more likely to be lost to follow-up, as well as those with no formal education and who were widowed. Age, sex, educational category, marital status and baseline CF status were comparable among the three groups. Most of the participants who were cognitively frail at baseline were lost to follow-up or had died.

Table 1 depicts characteristics of participants who displayed either improved or worsened cognitive frailty transition over the 7 years period. Significant differences were observed in educational category, falls, asthma, cancer and activities of daily living between groups.

Psychological characteristics

The psychological characteristics of participants were evaluated based on the cognitive frailty transition experienced as per Table 2. Significant differences in depression, anxiety and stress were observed between the improved and worsened groups.

Table 2.

Psychological characteristics based on cognitive frailty transition over 7 years period.

Presence of psychological symptoms	Improved	Worsened	p-value
Depression
Normal/mild	92 (100%)	439 (95.9%)	0.029*
Moderate/severe/extremely severe	0 (0.0%)	19 (4.1%)
Anxiety
Normal/mild	88 (96.7%)	408 (89.7%)	0.019*
Moderate/severe/extremely severe	3 (3.3%)	47 (10.3%)
Stress
Normal/mild	91 (100%)	440 (96.3%)	0.043*
Moderate/severe/extremely severe	0 (0%)	17 (3.7%)

*p-value significant at p < 0.05.

Logistic regression analyses

Binomial logistic regression analyses was conducted to observe the association between psychological status and cognitive frailty transition (see Table 3). From the analyses, regression analyses could not be conducted for stress and depression as no participants from the improved transition group reported moderate/severe/extremely severe levels of depression and stress. The unadjusted model displayed significant association between anxiety and worsened CF transition (OR (95% CI))  3.38 (1.03, 11.10)). The association remained significant after adjusting for potential confounders of educational category, falls and independent activity of daily living, age, sex, quality of life and social network (OR (95% CI)  4.053 (1.168, 14.061)). Presence of cancer or asthma did, however, attenuate the relationship between anxiety and CF transition.

Table 3.

Association between anxiety symptoms and worsened cognitive frailty transition over 7 years period.

Model	Anxiety
	B (95% CI)	p-value
Unadjusted	3.38 (1.03, 11.10)	0.045*
Age + Sex (Adjustment 1)	3.44 (1.04, 11.35)	0.043*
Adjustment 1 + Education category (Adjustment 2)	3.62 (1.09, 12.03)	0.036*
Adjustment 2 + Falls + QoL + Social Network + InstrumentalActivities of Daily Living (Adjustment 3)	3.98 (1.15, 13.78)	0.029*
Adjustment 2 + Cancer or Asthma	3.15 (0.94, 10.54)	0.063
Adjustment 3 + Cancer or Asthma	3.45 (0.98, 12.20)	0.054

Depression and stress could not be computed to non-presence of moderate to severe level symptoms in the improved transition group.

*p-value significant at p < 0.050.

Discussion

Overall improvement or sustainment and worsening of CF transition over the 7 years period were observed among 93(16.79%) and 461(83.21%) participants respectively. Improvement in cognitive frailty were observed among those with normal to mild depression, anxiety and stress. Notably, while all the participants with depression and stress at baseline progressed within the CF spectrum, none of those with moderate to severe depression and stress fell into the improved transition group. A trend of increased anxiety was also associated with worsening of cognitive frailty over the 7 years period.

Improvement in CF status has been described in 6.6% to 46.7% with varying classifications^7,11,12. Lu et al.⁸ describes CF progression from normal to frail then cognitively frail, advocating for the evolution of CF from physical frailty, while a retrospective study found that those with cognitive impairment at baseline were more likely to transition to CF^8,12. This study, on the contrary, revealed a great deal more fluidity in the CF state, whereby those who were C₀F₀, C_mF₀, C₀F_p and C_mF_p could transition to C_mF_f, while those who were C_mF_f could return to C_mF₀ , C₀F_p or C_mF_p. However, those within the C_mF_f group at baseline, despite the ability for improvement to some extent, could not return to C₀F₀. This suggests that once an individual develops CF, it is no longer possible to revert to a robust state, while all the other states could be considered reversible. The presence of MCI, prefrail with or without the presence of MCI, or frail alone, should be considered, before the onset of CF to ensure the possibility of complete reversibility. Further the likelihood of reversibility is better in those with MCI or prefrail in isolation that those with both MCI and prefrail in combination or frailty in isolation. A 2-year multidomain lifestyle intervention has been shown to be effective in improving or maintaining cognitive function in cognitively impaired individuals, with a subsequent study suggesting that multidomain intervention is also effective in improving physical frailty and depression^24,25 for frailty interventions, resistance or muscle strength training and protein supplementation have been suggested by two published systematic reviews^26,27.

There is a general paucity of research relevant to anxiety and stress towards CF. Depression has, however, been attributed as a risk factor for CF progression^7,28,29, while chronic stress has been shown to exacerbate underlying conditions, leading to development of certain pathologies and cognitive decline^30–32. Anxiety in older adults have been found to manifest differently in comparison to their younger counterparts, viewed as a pre-dementia syndrome presaging neurodegeneration and progressive cognitive decline³³, possibly explaining the link between increased anxiety and worsening of cognitive frailty. Alternatively, the presence of psychological symptoms, particularly depression, are known to affect cognitive performance³⁴. In addition, the presence of psychological symptoms may also reduce the likelihood of observing favourable lifestyle behaviours such as exercise and healthy diet to avoid physical frailty³⁵.

It must be highlighted, however, that comorbidities asthma and cancer had a confounding effect towards this association. The influence of asthma towards the relationship between progression across the CF spectrum or persistence of a non-robust state and increased anxiety could be due to symptom perception and management of asthma, which is common to anxiety and poor asthma control^36,37. While Curran, Sharpe and Butow’s³⁸ integrated model describes the possible influence of cancer towards the association, individuals who have lived with cancer may experience distress due to stigma, but equally the adverse effects of cancer treatment in terms of chemotherapy, extensive curative resection, and radiotherapy, may all lead to an increased risk of CF³⁹. The mechanistical link between cancer and aging is also an emerging theme⁴⁰.

A potential limitation to the study is that follow-up data was collected during the COVID-19 pandemic period, which may have had an impact on the psychological symptoms reported by participants. The use of different tools to assess frailty at baseline and follow-up was also unavoidable given the resultant circumstances from the COVID-19 pandemic and may pose a limitation in interpretation of the study’s findings. But the moderate levels of agreement and correlation implies valid use of and interchangeability between both tools. The study was still able to clearly highlight the role of psychological status in CF trajectory despite the use of different frailty measures at baseline and follow-up. The MoCA has also been found to educationally biased as a cognitive assessment tool, presenting as a limitation. However, within this study interindividual comparisons of actual MoCA scores were not conducted., Instead comparisons were made between observed longitudinal changes in MoCA scores. Further, the small number of individuals who fulfilled the definition of CF or moderate to severe psychological impairment limited the possibility of in-depth exposition of the effect of psychological issues on CF transition. A sizeable portion of the participants were lost to follow-up, especially those with MCI regardless of frailty status at baseline. As the pandemic hampered our capacity to reach out to these participants physically, future studies will benefit from face-to-face outreach and implementing measures in maintaining adherence among participants.

The findings suggests that CF should not be considered phenotypically nor as a progressive state. Instead, it is a dynamic state with transitions occurring with a dose–response relationship. This study is novel in exploring the association between psychological correlates and cognitive frailty transition, with evidence of a relationship between moderate to severe anxiety, depression and stress with worsening of CF status. Future studies should also now evaluate the possibility of modification of psychological risk factors as a preventive measure for worsening of CF status.

The categories of robust, prefrail and frail with and without the presence of MCI exists as a dynamic state. Findings also seeks to inform clinicians on the potential for improvement for frailty and MCI states, not to exists as an unmodifiable diagnosis. Transitions occur in all directions between the six cognition-frailty status categories examined in this study, with the presence of moderate to severe anxiety, depression and stress appearing to obliterate the likelihood of recovery in terms of improvement in CF status. Future studies should now evaluate the role of psychological interventions among those with moderate to severe anxiety, depression and stress as a potentially modifiable risk factor for progression in CF status.

Author contributions

Kiirtaara Aravindhan: Formal analysis, investigation, conceptualization, writing—original draft. Sumaiyah Mat: Supervision, investigation, data curation, formal analysis, writing-review and editing. Maw Pin Tan: Supervision, Funding acquisition, project administration, formal analysis, writing—review and editing. Xu Xiang Jiang: writing—review and editing. Nurul Nabilah Akmal Hashim: writing—review and editing. Moses Raj Muraly: writing—review and editing. Nor I’zzati Saedon: writing – review and editing. Kejal Hasmukharay: writing—review and editing. Azlina Ahmad Annuar: writing—review and editing. Hazlina Mahadzir: writing—review and editing. Maw Pin Tan: Supervision, Funding acquisition, project administration, formal analysis, writing—review and editing.

Data availability

Supporting data are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors have no potential conflict of interest to declare.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-91650-4.