Latent class analysis of intrinsic capacity and its association with quality of life in hospitalized older people with stroke: a cross-sectional study.

Latent class analysis of intrinsic capacity and its association with quality of life in hospitalized older people with stroke: a cross-sectional study

Objective

This study aimed to identify latent classes of intrinsic capacity among hospitalized older people with stroke and to examine the associations of these classes with stroke-specific quality of life.

Methods

A cross-sectional design was used, and 314 hospitalized older people with stroke were consecutively recruited. Stroke severity was assessed using the National Institutes of Health Stroke Scale (NIHSS), comorbidity burden using the Age-adjusted Charlson Comorbidity Index (ACCI), perceived social support using the Perceived Social Support Scale (PSSS), intrinsic capacity across five domains using the Intrinsic Capacity Assessment Tool, and stroke-specific quality of life using the Stroke-Specific Quality of Life Scale (SS-QOL). Latent class analysis identified intrinsic-capacity profiles, and multivariable logistic regression was used to examine factors associated with latent-class membership, and ANOVA was used to compare SS-QOL scores across the identified classes.

Results

LCA identified three latent classes of intrinsic capacity among hospitalized older people with stroke: the low motor–low vitality group (27.70%), the low cognitive–moderate vitality, psychological, and sensory group (17.20%), and the function maintenance group (55.10%). Multinomial logistic regression showed that age, stroke severity, sleep duration, and social support were associated with membership in the low motor–low vitality group, whereas education level was associated with membership in the low cognitive–moderate vitality, psychological, and sensory group. Significant differences were observed in most stroke-specific quality-of-life domains across the identified groups (p < 0.05).

Conclusion

Three distinct intrinsic-capacity profiles were identified among hospitalized older people with stroke, and these profiles were associated with different levels of stroke-specific quality of life. Both clinical and psychosocial factors were associated with profile membership, highlighting the need for targeted interventions to preserve intrinsic capacity and improve quality of life.

1Introduction

Stroke remains the second leading cause of death and long-term disability worldwide and in China and has been identified as a priority chronic disease for prevention and control (1, 2). The burden of stroke increases substantially with age, making older adults a particularly vulnerable population (3). Beyond mortality and disability, stroke has a profound and lasting impact on quality of life (QOL), including physical function, mood, cognition, and social participation, even beyond the acute stage (4, 5).

In 2015, the World Health Organization (WHO) introduced the concept of intrinsic capacity (IC) in the World Report on Aging and Health (6). IC refers to the composite of all physical and mental capacities that an individual can draw on at a given time and includes five core domains: locomotion, vitality, sensory function, cognition, and psychological wellbeing (7). Decline in IC has been associated with poor quality of life, disability, frailty, dependence, and other adverse health outcomes in older adults (8–10). However, much of the existing literature has focused on single functional domains rather than multidimensional IC patterns (6, 10).

Recently, person-centered analytical approaches have been increasingly used to identify multidomain IC profiles in older populations. For example, latent class analysis (LCA) has been used to classify heterogeneous patterns of IC impairment and to explore factors associated with class membership (11, 12). Nevertheless, evidence remains limited regarding IC profiles specifically among hospitalized older people with stroke. This issue is clinically important because people with stroke often present with marked heterogeneity in motor, cognitive, sensory, nutritional, and psychological status, as well as in stroke severity, comorbidity burden, sleep, and social support (13, 14). In addition, major stroke rehabilitation guidelines emphasize multidisciplinary, patient-centered care and individualized goal setting across physical, cognitive, emotional, and participation domains (15).

LCA is a probability-based statistical approach that can identify unobserved subgroups with similar response patterns and therefore provides a useful framework for characterizing multidimensional IC heterogeneity (16). To our knowledge, few studies have applied LCA to examine IC profiles specifically in hospitalized older people with stroke and further linked these profiles to stroke-specific quality of life. Therefore, the aims of this study were to identify latent classes of IC in hospitalized older people with stroke, examine factors associated with class membership, and explore differences in stroke-specific quality of life across these classes. The findings may provide evidence for more individualized rehabilitation and nursing strategies based on domain-specific risk patterns.

2Methods

This cross-sectional study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement (17).

2.1Study participants

Participants were consecutively recruited from May 2024 to November 2024 from the Department of Neurology, Jiangsu Province Hospital (the First Affiliated Hospital of Nanjing Medical University). Eligible participants met all of the following inclusion criteria: (1) diagnosis of ischemic or hemorrhagic stroke according to the Key Points for Diagnosis of Major Cerebrovascular Diseases in China (2019), confirmed by brain computed tomography (CT) or magnetic resonance imaging (MRI) (18); (2) age ≥60 years; and (3) clinically stable stroke beyond the acute stage, as judged by the treating neurologist on the basis of the medical record and current clinical condition, including patients in the recovery or sequelae stage. To ensure clinical stability, enrollment was performed at least 48 h after admission, when the patient had stable vital signs and no acute deterioration.

Exclusion criteria were as follows: (1) severe cognitive impairment or a history of mental illness; and (2) comorbid cancer or vital organ failure. Severe cognitive impairment was screened at enrollment by trained investigators using the Chinese version of the Mini-Mental State Examination (MMSE), together with an assessment of communication ability and response capacity during the enrollment interview. Participants were excluded if they were unable to understand the study information or interview questions, unable to communicate meaningfully with the investigator, or unable to provide coherent and consistent responses sufficient to complete the questionnaire assessment.

Sample size estimation followed the variable multiple method, yielding a theoretical sample size range of 160–320 (16 variables × 10–20) (18). After allowing for a potential 10% rate of questionnaire loss or invalidity, the minimum required sample size was set at 178.

2.2Investigation tools

The main study outcomes were intrinsic capacity (five-domain status and total score) and stroke-specific quality of life. Additional measured variables included neurological deficit severity, comorbidity burden, and perceived social support. The instruments were selected to align with the WHO intrinsic capacity framework and the study objectives, and ensure feasibility for bedside assessment in hospitalized older people with stroke. All measures are widely used in stroke/geriatric research and were feasible for bedside assessment in hospitalized older people with stroke.

2.2.1General information questionnaire

A self-designed questionnaire was administered to collect demographic data (gender, age, body mass index [BMI], education level, marital status), socioeconomic status (employment situation, medical insurance type, residence), lifestyle factors (smoking, alcohol consumption, sleep duration), and stroke-related characteristics (stroke type, stroke severity, limb muscle strength). Limb muscle strength was assessed using the Lovett grading system (0–5). “Normal” was defined as grade 5 in all limbs; “abnormal” was defined as grade ≤ 4 in any limb on either side. Spasticity was not assessed (e.g., Modified Ashworth Scale was not collected).

2.2.2National Institutes of Health Stroke Scale

Developed in 1989, the National Institutes of Health Stroke Scale (NIHSS) assesses neurological function using 11 items, including level of consciousness, visual fields, limb motor function, sensation, and language (19). Scores range from 0 to 42, with 0 to 1 defined as normal, 2–4 as mild, 5–15 as moderate, and >15 as severe. The NIHSS is a well-validated instrument for quantifying neurological deficit severity in people with stroke (20). The Cronbach's α coefficient for this scale is 0.810.

2.2.3Age-adjusted Charlson Comorbidity Index

The Age-adjusted Charlson Comorbidity Index (ACCI) is a structured scoring instrument designed to quantify the complexity of patients' clinical conditions and to prospectively estimate mortality risk in those with chronic diseases and cancer (21). This index integrates 19 clinically relevant diseases, each assigned a specific weight according to its potential impact on mortality. The scoring system is based on two primary dimensions. The first dimension, age-stratified scoring, assigns points as follows: 40–49 years, 1 point; 50–59 years, 2 points; 60–69 years, 3 points; and 70 years or older, 4 points. The second dimension, disease burden, accounts for the weighted scores of the 19 specified comorbid conditions. Data on pre-existing comorbidities prior to stroke diagnosis were obtained from medical record review. The cumulative score reflects the overall comorbidity burden, with scores ≥5 indicating multiple comorbidities (high burden) and scores < 5 indicating non-multiple comorbidities. The CCI/ACCI has been extensively evaluated, and evidence supporting its clinimetric properties has been summarized in a critical review (21). The Cronbach's α coefficient for the ACCI is 0.832.

2.2.4Perceived Social Support Scale

The Perceived Social Support Scale (PSSS) measures multidimensional self-perceived social support (22). It consists of 12 items across three domains: family support (items 3, 4, 8, 11), friend support (items 6, 7, 9, 12), and other support (items 1, 2, 5, 10). Items are rated on a 7-point scale (1–7), with total scores ranging from 12 to 84. Scores of 12–36 indicate low social support, 37–60 indicate moderate social support, and 61–84 indicate high social support. The PSSS has been validated and has shown acceptable psychometric properties in Chinese hospitalized populations (22). The Cronbach's α coefficient for this scale is 0.840.

2.2.5Intrinsic capacity assessment tools

Intrinsic capacity was assessed according to the World Health Organization's five-domain framework (7). For latent class analysis, each domain was dichotomized as preserved (1) or impaired (0) based on predefined criteria.

Vitality: assessed using the Mini-Nutritional Assessment Short-Form (MNA-SF) (23). Scores < 11 were classified as impaired vitality (risk of malnutrition or malnutrition), whereas scores ≥11 were classified as preserved vitality. The MNA-SF is a widely used and validated screening tool for nutritional status in older people (23). The Cronbach's α coefficient in this study was 0.816.

Psychological: assessed using the 15-item Geriatric Depression Scale (GDS-15) (24). Scores ≥8 indicated impaired psychological status, whereas scores < 8 indicated preserved psychological status. The Chinese version of the GDS-15 has demonstrated acceptable psychometric properties in older Chinese populations (24). The Cronbach's α coefficient in this study was 0.793.

Cognition: assessed using the Mini-Mental State Examination (MMSE). Education-specific cutoffs were used to define cognitive impairment: ≤ 17 for illiterate participants, ≤ 20 for those with primary school education, and ≤ 24 for those with junior high school education or above (25). Scores above these thresholds were classified as preserved cognition. The Chinese version of the MMSE was used, and measurement invariance has been examined in older Chinese populations (25). The Cronbach's α coefficient in this study was 0.838.

Locomotion (motor): assessed using the Short Physical Performance Battery (SPPB), which includes standing balance, 4-m walking speed, and the five-times sit-to-stand test (26, 27). Scores ≤ 8 indicated impaired locomotion, whereas scores >8 indicated preserved locomotion. The SPPB is a validated measure of lower-extremity physical performance and has been widely used in older-adult populations (27). The Cronbach's α coefficient in this study was 0.812.

Sensory: assessed using visual acuity and whisper tests (28). Sensory function was classified as impaired if the participant had abnormal visual acuity and/or an abnormal whisper test result; otherwise, it was classified as preserved. Normal visual function was defined as uncorrected bilateral visual acuity ≥0.8.

The total intrinsic capacity score ranged from 0 to 5, with higher scores indicating better overall intrinsic capacity.

2.2.6Stroke-specific quality of life scale

The Stroke-Specific Quality of Life Scale (SS-QOL) was employed to assess quality of life in people with stroke (29). It comprises 49 items across 12 domains: vision, language, mood, cognition, upper extremity function, personality, mobility, family roles, self-care, energy, work productivity, and social roles. Each item is rated on a 5-point scale, with higher total scores indicating better quality of life. This instrument has demonstrated acceptable validity and reliability in previous studies (30, 31). The Cronbach's α coefficient is 0.832.

2.3Data collection and quality control

This study was approved by the Ethics Review Committee of the First Affiliated Hospital of Nanjing Medical University (approval number: 2024-SR-100). All participants provided written informed consent before the interview-based assessment. Data collection was conducted by two graduate nursing students who had received standardized training before the study, including training in questionnaire administration, communication procedures, and assessment consistency.

After informed consent was obtained, trained investigators conducted structured, interviewer-administered, one-to-one assessments using standardized paper questionnaires. Before formal data collection, participants' ability to understand the interview content and provide reliable responses was verified during enrollment screening. Demographic and clinical information was primarily obtained from the hospital electronic medical record system, and missing items were supplemented through direct inquiry of participants or family members whenever possible.

Eligibility screening was completed before questionnaire administration. A total of 322 eligible participants were enrolled and underwent the interviewer-administered assessment. Eight questionnaires were subsequently excluded in their entirety because of invalid or incomplete responses, resulting in 314 valid questionnaires and an effective response rate of 97.5%.

2.4Statistical analysis

Data entry followed a double-independent process with cross-verification. Latent class models were constructed using Mplus version 8.0 (Muthén & Muthén, Los Angeles, CA, United States) Model fit was evaluated using the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and sample-size adjusted BIC (aBIC), with lower values indicating better fit. The optimal number of classes was determined using the Lo-Mendell-Rubin adjusted likelihood ratio test (LMR-LRT) and bootstrap likelihood ratio test (BLRT), with p < 0.05 indicating that the k-class model fit significantly better than the (k−1)-class model. Classification accuracy was assessed using entropy, with values >0.80 indicating good classification quality (32).

Statistical analyses were performed using SPSS version 27.0 (IBM Corp., Armonk, NY, United States). Normality of continuous variables was assessed using the Shapiro–Wilk test together with visual inspection of histograms and Q–Q plots. Homogeneity of variances was examined using Levene's test before ANOVA. Normally distributed data are presented as mean ± standard deviation and were compared using one-way analysis of variance (ANOVA), followed by least significant difference (LSD) post-hoc tests where appropriate. Skewed data are presented as median (interquartile range) and were compared using non-parametric tests. Categorical variables are described as frequencies and percentages and were compared using the chi-square test.

To examine factors associated with latent-class membership, multinomial logistic regression was performed with the Function Maintenance Group as the reference category. Candidate predictors were selected based on both their univariate association with latent-class membership and their theoretical or clinical relevance to intrinsic capacity in older people with stroke, as informed by previous literature and the study objectives. Variables entered into the model included age group, education level, sleep duration, stroke severity, and perceived social support. For regression analysis, education level was recategorized into three groups (primary school or below, junior high school, and senior high school or above), and sleep duration was dichotomized as >6 h vs. ≤ 6 h. Categorical predictors were dummy coded, and reference categories were specified in Table 1. Multicollinearity among independent variables was assessed using variance inflation factors (VIFs), and no serious multicollinearity was detected. Odds ratios (ORs), 95% confidence intervals (CIs), and p-values are reported for all model estimates. A two-sided p-value < 0.05 was considered statistically significant.

Table 1Multivariate logistic regression analysis of factors associated with latent classes of intrinsic capacity.

Predictor Reference category Low motor–low vitality group vs. function maintenance group OR (95% CI) p Low cognitive–moderate vitality, psychological and sensory group vs. function maintenance group OR (95% CI) p

Age 60–74 years ≥75 years 0.365 (0.185–0.718) 0.004 0.452 (0.204–1.002) 0.051

Mild stroke severity Moderate/severe stroke 0.158 (0.064–0.393) < 0.001 0.965 (0.233–3.987) 0.960

Low social support High social support 3.420 (1.158–10.103) 0.026 0.560 (0.112–2.808) 0.481

Moderate social support High social support 1.994 (1.070–3.717) 0.030 0.639 (0.303–1.346) 0.239

Primary school or below Senior high school or above 0.643 (0.285–1.453) 0.289 0.191 (0.043–0.859) 0.031

Junior high school Senior high school or above 1.595 (0.833–3.054) 0.159 1.712 (0.870–3.370) 0.120

Sleep duration >6 h ≤ 6 h 0.424 (0.231–0.779) 0.006 0.526 (0.271–1.021) 0.058

Multinomial logistic regression was performed with latent class membership as the dependent variable. The Function Maintenance Group was used as the reference outcome category. Reference categories for predictors were as follows: age ≥75 years, moderate/severe stroke, high social support, senior high school or above, and sleep duration ≤ 6 h. OR, odds ratio; CI, confidence interval.

Homogeneity of variances across latent classes was examined using Levene's test before one-way ANOVA. For SS-QOL domains with homogeneous variances, one-way ANOVA followed by least significant difference (LSD) post-hoc comparisons was used. For domains with unequal variances, pairwise comparisons were interpreted using the Games–Howell procedure. Pairwise comparison results are reported as mean differences, 95% confidence intervals (CIs), and p-values. No additional adjustment for multiple testing across SS-QOL domains was applied.

After the optimal latent class model was selected in Mplus, each participant was assigned to the class with the highest posterior probability (most likely class membership). This assigned class variable was then exported to SPSS and used in subsequent analyses, including chi-square tests, one-way ANOVA, and multinomial logistic regression. Classification uncertainty was not directly incorporated into these downstream analyses.

3Results

3.1General information

A total of 314 older people with stroke were included in this study, consisting of 207 males and 107 females. Participant age ranged from 60 to 86 years, with a mean ± standard deviation of 68.97 ± 6.36 years. Additional demographic and clinical information are presented in Table 2.

Table 2Demographic and clinical characteristics of older adults with stroke.

Variable Category Number Percentage (%)

Gender Male 207 65.9

Female 107 34.1

Age 60–74years 249 79.3

≥75years 65 20.7

BMI < 18.5 11 3.5

18.5–24.9 183 58.3

≥25 120 38.2

Educational level Illiterate 18 5.7

Primary school 36 11.5

Junior high school 93 29.6

Senior high school/technical secondary school 95 30.2

College or above 72 22.9

Marital status Married 296 94.2

Single 9 2.9

Divorced/widowed 9 2.9

Employment situation Employed 53 16.9

Unemployed/retired 261 83.1

Medical insurance type Employee medical insurance 120 38.2

Resident medical insurance 10 3.2

Non-local medical insurance 168 53.5

Self-pay 16 5.1

Residence Urban 219 69.7

Rural 95 30.3

Smoking Yes 80 25.5

Never/quit 234 74.5

Alcohol consumption Yes 82 26.1

Never/quit 232 73.9

Sleep duration >6 h 160 51.0

4–6 h 60 19.1

< 4 h 94 29.9

Stroke type Hemorrhagic stroke 60 19.1

Ischemic stroke 254 80.9

Limb muscle strength (Lovett grade)a Normal 235 74.8

Abnormal 79 25.2

Stroke severity (NIHSS) Mild stroke 281 89.5

Moderate/severe stroke 33 10.5

ACCI Multiple comorbidities 109 34.7

Non-multiple comorbidities 205 65.3

PSSS Low social support 22 7.0

Moderate social support 118 37.6

High social support 174 55.4

aLimb muscle strength was assessed using the Lovett grading system (0–5). Normal = grade 5 in all limbs; abnormal = grade ≤ 4 in any limb on either side.

BMI, body mass index; NIHSS, National Institutes of Health Stroke Scale; ACCI, Age-adjusted Charlson Comorbidity Index; PSSS, Perceived Social Support Scale.

3.2LCA of intrinsic capacity in older people with stroke

The median intrinsic capacity score of older people with stroke was 4 (interquartile range: 3–5). A decline in at least one dimension of intrinsic capacity was observed in 72.0% (226 cases) of people with stroke. Among the five dimensions, decline in the vitality dimension was most common (44.9%, 141 cases), followed by motor (32.5%, 102 cases), sensory (29.6%, 93 cases), cognitive (29.0%, 91 cases), and psychological (12.4%, 39 cases) dimensions. LCA was performed using the five dimensions as manifest variables. Although AIC, BIC, and sample-size aBIC values decreased as the number of classes increased, and the BIC reached its lowest value in the two-class model, the three-class model achieved the optimal entropy value, indicating superior classification accuracy. Furthermore, the Lo-Mendell-Rubin (LMR) and Bootstrap Likelihood Ratio Test (BLRT) results were statistically significant (p < 0.05), confirming that the three-class model was significantly better than the two-class model (Table 3). Comprehensive evaluation of multiple fit indices confirmed the three-class model as optimal. The average posterior probabilities for the three classes were 0.879, 0.976, and 1.000, respectively, demonstrating high stability and reliability of classification. Based on the highest posterior probability, participants were assigned to one of the three latent classes for subsequent analyses. Figure 1 presents the conditional probability distribution from the Mplus software analysis, illustrating the characteristics of the three latent classes of intrinsic capacity. Class 1 (C1): Low Motor–Low Vitality Group–87 participants (27.7%) with markedly reduced motor and vitality dimensions; Class 2 (C2): Low Cognitive–Moderate Vitality, Psychological, and Sensory Group–54 participants (17.2%) with significant decline in cognitive function and moderate impairment in vitality, psychological, and sensory dimensions; Class 3 (C3): Function Maintenance Group–173 participants (55.1%) with relatively preserved functional dimensions.

Table 3Fit indices for latent class analysis models of intrinsic capacity.

Model K AIC BIC aBIC LMR (P) BLRT (P) Entropy Class probability (%)

Model 1 5 1833.209 1851.956 1836.098 – – – 100

Model 2 11 1774.943 1816.187 1781.298 < 0.001 < 0.001 0.514 51.59/48.41

Model 3 17 1770.463 1834.203 1780.284 < 0.001 < 0.001 0.907 27.70/17.20/55.10

Model 4 23 1770.124 1856.36 1783.411 0.2093 0.1304 0.894 1.91/21.98/17.20/58.91

Model 5 29 1774.919 1883.651 1791.671 0.5995 0.2667 0.796 14.33/19.74/6.69/2.23/57.01

3.3Comparison of characteristics among older people with stroke in different latent classes

Comparison of demographic and clinical characteristics among the three latent classes demonstrated no statistically significant differences in gender, BMI, marital status, employment status, residence, smoking, alcohol consumption, or stroke type (all p > 0.05). Variables with significant differences are presented in Table 4.

Table 4Comparison of demographic and clinical variables across latent classes of intrinsic capacity.

Variable Category Low motor-low vitality group (n = 87) Low cognitive-moderate vitality, psychological and sensory group (n = 54) Function maintenance group (n = 173) Statistic P

Age [n (%)] 60–74 years 58 (66.7) 41 (75.9) 150 (86.7) X2 = 14.612 < 0.001

≥75 years 29 (33.3) 13 (24.1) 23 (13.3)

Educational level [n (%)] Illiterate 6 (6.9) 0 (0.0) 12 (6.9) X2 = 17.862 0.022

Primary school 7 (8.0) 2 (3.7) 27 (15.6)

Junior high school 28 (32.2) 22 (40.7) 43 (24.9)

High school/Technical secondary school 28 (32.2) 21 (38.9) 46 (26.6)

College and above 18 (20.7) 9 (16.7) 45 (26.0)

Medical insurance type [n (%)] Employee medical insurance 26 (29.9) 23 (42.6) 71 (41.0) X2 = 14.351 0.026

Resident medical insurance 4 (4.6) 0 (0.0) 6 (3.5)

Non-local medical insurance 47 (54.0) 30 (55.6) 91 (52.6)

Self-pay 10 (11.5) 1 (1.9) 5 (2.9)

Limb muscle strength (Lovett grade)a [n (%)] Normal 47 (54.0) 42 (77.8) 146 (84.4) X2 = 28.665 < 0.001

Abnormal 40 (46.0) 12 (22.2) 27 (15.6)

Sleep duration [n (%)] >6 h 33 (37.9) 22 (40.7) 105 (60.7) X2 = 19.397 < 0.001

4–6 h 17 (19.5) 10 (18.5) 33 (19.1)

< 4 h 37 (42.5) 22 (40.7) 35 (20.2)

Stroke severity (NIHSS) [n (%)] Mild 65 (74.7) 51 (94.4) 165 (95.4) X2 = 27.982 < 0.001

Moderate/severe 22 (25.3) 3 (5.6) 8 (4.6)

ACCI [n (%)] Multiple comorbidities 40 (46.0) 19 (35.2) 50 (28.9) X2 = 7.454 0.024

Non-multiple comorbidities 47 (54.0) 35 (64.8) 123 (71.1)

PSSS [n (%)] Low support level 8 (9.2) 2 (3.7) 12 (6.9) X2 = 12.176 0.016

Medium support level 42 (48.3) 13 (24.1) 63 (36.4)

High support level 37 (42.5) 39 (72.2) 98 (56.6)

aNormal = Lovett grade 5 in all limbs; abnormal = grade ≤ 4 in any limb on either side.

NIHSS, National Institutes of Health Stroke Scale (higher scores indicate more severe stroke); ACCI, Age-adjusted Charlson Comorbidity Index (higher scores indicate greater comorbidity burden); PSSS, Perceived Social Support Scale (higher scores indicate higher perceived social support).

Data are presented as n (%). Group comparisons were conducted using the chi-square test. Significant p-values are shown in bold (P < 0.05).

3.4Association between latent classes of intrinsic capacity and quality of life scores

One-way ANOVA was conducted to compare SS-QOL scores across the three latent classes. Significant between-class differences were observed for all domains except personality (F = 1.487, p = 0.228). Levene's test indicated unequal variances for vision, language, mood, mobility, self-care, upper extremity function, and work productivity; therefore, Games–Howell post-hoc tests were used for these domains. LSD post-hoc tests were used for the remaining domains with homogeneous variances. Compared with the Function Maintenance Group, the Low Motor–Low Vitality Group had significantly lower scores in vision, language, mood, cognition, upper extremity function, mobility, family roles, self-care, energy, work productivity, and social roles, whereas personality did not differ significantly. Compared with the Function Maintenance Group, the Low Cognitive–Moderate Vitality, Psychological, and Sensory Group had significantly lower scores in cognition, family roles, and social roles, whereas the remaining domains did not differ significantly. Detailed results, including Levene's test results, mean differences, 95% confidence intervals, and p-values for pairwise comparisons, are presented in Table 5.

Table 5A quality of life scores across latent classes of intrinsic capacity.

Dimension Score range Low motor–low vitality group (n = 87) Low cognitive–moderate vitality, psychological, and sensory group (n = 54) Function maintenance group (n = 173) Levene's p F P Post-hoc method Low motor vs. function maintenance MD (95% CI), p Low cognitive vs. function maintenance md (95% CI), p Low motor vs. low cognitive MD (95% CI), p

Vision 3–15 13.00 ± 2.040 14.06 ± 1.089 14.35 ± 1.009 < 0.001 27.692 < 0.001 Games–Howell −1.347 (−1.90, −0.80), < 0.001 −0.291 (−0.69, 0.11), 0.195 −1.056 (−1.68, −0.43), < 0.001

Language 5–25 22.30 ± 3.438 23.61 ± 2.791 24.21 ± 1.575 < 0.001 17.690 < 0.001 Games–Howell −1.909 (−2.83, −0.99), < 0.001 −0.597 (−1.55, 0.36), 0.298 −1.312 (−2.57, −0.06), 0.038

Mood 5–25 20.56 ± 4.037 21.28 ± 5.163 22.34 ± 3.319 0.005 6.366 0.002 Games–Howell −1.778 (−2.96, −0.59), 0.002 −1.063 (−2.85, 0.73), 0.334 −0.715 (−2.68, 1.25), 0.663

Cognition 3–15 9.44 ± 3.190 9.15 ± 3.024 11.65 ± 3.128 0.741 21.634 < 0.001 LSD −2.216 (−3.03, −1.41), < 0.001 −2.505 (−3.46, −1.55), < 0.001 0.289 (−0.78, 1.35), 0.595

Upper extremity function 5–25 20.82 ± 4.285 23.83 ± 2.759 24.39 ± 1.437 < 0.001 50.473 < 0.001 Games–Howell −3.577 (−4.70, −2.45), < 0.001 −0.560 (−1.50, 0.38), 0.331 −3.017 (−4.42, −1.61), < 0.001

Personality 3–15 11.37 ± 3.275 12.06 ± 3.253 12.06 ± 3.091 0.649 1.487 0.228 LSD −0.690 (−1.51, 0.13), 0.099 0.002 (−0.97, 0.97), 0.996 −0.688 (−1.77, 0.39), 0.211

Mobility 6–30 21.86 ± 4.666 28.72 ± 1.676 28.47 ± 2.911 < 0.001 125.084 < 0.001 Games–Howell −6.606 (−7.90, −5.31), < 0.001 0.254 (−0.50, 1.01), 0.704 −6.860 (−8.17, −5.56), < 0.001

Family roles 3–15 10.15 ± 2.751 10.80 ± 3.129 12.11 ± 2.723 0.207 15.381 < 0.001 LSD −1.960 (−2.69, −1.24), < 0.001 −1.314 (−2.17, −0.45), 0.003 −0.647 (−1.60, 0.31), 0.184

Self-care 5–25 21.08 ± 4.669 24.20 ± 2.184 24.58 ± 1.281 < 0.001 47.669 < 0.001 Games–Howell −3.503 (−4.72, −2.29), < 0.001 −0.380 (−1.13, 0.37), 0.448 −3.123 (−4.50, −1.74), < 0.001

Energy 3–15 8.85 ± 3.384 10.11 ± 3.570 10.39 ± 3.624 0.988 5.524 0.004 LSD −1.537 (−2.45, −0.62), 0.001 −0.276 (−1.37, 0.81), 0.618 −1.261 (−2.47, −0.05), 0.041

Work productivity 3–15 9.95 ± 2.905 11.52 ± 2.626 12.40 ± 2.365 0.011 26.347 < 0.001 Games–Howell −2.451 (−3.30, −1.60), < 0.001 −0.886 (−1.84, 0.07), 0.075 −1.564 (−2.69, −0.44), 0.004

Social roles 5–25 12.63 ± 5.383 15.56 ± 5.964 17.33 ± 5.793 0.719 19.597 < 0.001 LSD −4.697 (−6.17, −3.22), < 0.001 −1.774 (−3.53, −0.02), 0.047 −2.923 (−4.87, −0.98), 0.003

SS-QOL, Stroke-Specific Quality of Life; MD, mean difference; CI, confidence interval.

Values are presented as mean ± standard deviation. Homogeneity of variances was assessed using Levene's test. LSD post-hoc comparisons were used for domains with homogeneous variances, whereas Games–Howell post-hoc comparisons were used for domains with unequal variances. No additional adjustment for multiple testing across domains was applied.

3.5Multivariate logistic regression analysis of factors influencing latent classes

The three latent classes identified by LCA were treated as the dependent variable in the multinomial logistic regression model, with the Function Maintenance Group used as the reference category. Predictor coding and reference categories are shown in Table 1. Full model estimates, including non-significant associations, are presented in Table 1.

Compared with the Function Maintenance Group, participants aged 60–74 years (OR = 0.365, 95% CI: 0.185–0.718, p = 0.004), those with mild stroke severity (OR = 0.158, 95% CI: 0.064–0.393, p < 0.001), and those with sleep duration > 6 h (OR = 0.424, 95% CI: 0.231–0.779, p = 0.006) had lower odds of belonging to the Low Motor–Low Vitality Group. In contrast, participants with low social support (OR = 3.420, 95% CI: 1.158–10.103, p = 0.026) and moderate social support (OR = 1.994, 95% CI: 1.070–3.717, p = 0.030), compared with those with high social support, had higher odds of belonging to this group. Education level was not significantly associated with membership in the Low Motor–Low Vitality Group after adjustment.

For the Low Cognitive–Moderate Vitality, Psychological and Sensory Group, participants with primary school education or below had lower odds of belonging to this group than those with senior high school education or above (OR = 0.191, 95% CI: 0.043–0.859, p = 0.031). Age and sleep duration showed borderline associations with membership in this group, whereas stroke severity, social support, and junior high school education were not significantly associated with class membership after adjustment.

4Discussion

4.1Group heterogeneity in intrinsic capacity

This study identified three distinct intrinsic-capacity (IC) profiles among hospitalized older people with stroke, and most participants showed decline in at least one IC domain. These findings highlight substantial heterogeneity in multidomain IC impairment in this population.

Overall, 72.0% of participants experienced decline in at least one intrinsic-capacity domain, which is slightly higher than the 69.1% reported in general older populations (8). Older people with stroke often experience concurrent impairments in motor, cognitive, sensory, nutritional, and psychological function, which may contribute to reduced quality of life and functional decline (33). The three-class solution therefore provides a useful framework for characterizing these multidimensional patterns.

The Function Maintenance Group (55.1%) showed relatively preserved function across IC domains, although some decline remained, particularly in vitality. This finding suggests that even older people with stroke who appear functionally stable may still have early nutritional vulnerability requiring attention (34).

The Low Motor–Low Vitality Group (27.7%) was characterized by marked impairment in locomotion and vitality. These deficits may be mutually reinforcing, as mobility limitation can reduce daily activity and nutritional intake, while malnutrition may aggravate sarcopenia, weakness, and further mobility decline (34, 35).

The Low Cognitive–Moderate Vitality, Psychological, and Sensory Group (17.2%) showed prominent cognitive impairment together with moderate deficits in vitality, psychological status, and sensory function. This pattern is consistent with the high prevalence of post-stroke cognitive impairment and its frequent co-occurrence with nutritional problems, depressive symptoms, and sensory dysfunction (33, 36–40).

Taken together, these findings suggest that IC profiling may provide added value beyond isolated single-domain assessment by identifying co-occurring patterns of decline. In clinical practice, such profiling may help clinicians prioritize rehabilitation goals, tailor multidisciplinary care, and identify patients who require closer follow-up after hospitalization (50). For example, the Low Motor–Low Vitality Group may benefit most from integrated mobility and nutritional support, whereas the Low Cognitive–Moderate Vitality, Psychological, and Sensory Group may require greater emphasis on cognitive rehabilitation, psychological support, and sensory/environmental management. For the Function Maintenance Group, IC profiling may help support early prevention and continued monitoring before further decline emerges.

4.2Influencing factors of intrinsic capacity latent classes

4.2.1Demographic characteristics

The multinomial logistic regression results showed that age, stroke severity, sleep duration, and social support were mainly associated with membership in the Low Motor–Low Vitality Group, whereas education level was associated with membership in the Low Cognitive–Moderate Vitality, Psychological, and Sensory Group. These findings suggest that different intrinsic-capacity profiles may have distinct correlates in hospitalized older people with stroke, which is broadly consistent with previous research (41).

Participants aged 60–74 years were less likely to be classified into the Low Motor–Low Vitality Group than those aged 75 years or older, suggesting that relatively younger stroke survivors tend to preserve better intrinsic capacity. Age-related decline in intrinsic capacity may reflect the combined effects of physiological aging, reduced physiological reserve, and accumulated disease burden (9, 49).

Education level was associated with membership in the Low Cognitive–Moderate Vitality, Psychological, and Sensory Group. In our data, participants with primary school education or below had lower odds of belonging to this group than those with senior high school education or above. This finding should be interpreted cautiously. Educational attainment may reflect not only health literacy and self-management capacity, but also cognitive reserve, which can modify how cognitive impairment manifests after stroke. Recent evidence from a pooled cohort study suggested that post-stroke cognitive trajectories vary by both education level and cognitive domain; compared with stroke survivors with lower educational attainment, those with higher education may have greater cognitive reserve, yet may also show steeper decline in specific domains such as executive function (42). Given the heterogeneity of post-stroke cognitive trajectories across educational strata and cognitive domains, future longitudinal studies using domain-specific cognitive measures are needed to clarify how education shapes intrinsic-capacity profiles after stroke.

Longer sleep duration (>6 h) was associated with lower odds of belonging to the Low Motor–Low Vitality Group and showed a borderline association with the Low Cognitive–Moderate Vitality, Psychological, and Sensory Group. Adequate sleep may support neuroplasticity, reduce fatigue, and promote recovery after stroke, whereas insufficient sleep may adversely affect physical, cognitive, and psychological function (43). These findings underscore the importance of identifying and managing sleep problems in older people with stroke during hospitalization and rehabilitation.

4.2.2Stroke severity and social support

Participants with mild stroke had substantially lower odds of belonging to the Low Motor–Low Vitality Group than those with moderate or severe stroke, indicating that worse neurological deficit is linked to poorer intrinsic-capacity status. This finding is consistent with prior studies (44). Neurological deficits affect core functions such as motor ability, cognition, swallowing, and sensory integration, which may in turn compromise daily functioning and quality of life and accelerate intrinsic-capacity decline (45). Routine evaluation of stroke severity and rehabilitation progress is therefore important for individualized care planning and targeted rehabilitation.

Lower levels of perceived social support were associated with higher odds of belonging to the Low Motor–Low Vitality Group. Compared with participants reporting high social support, those with low or moderate social support were more likely to fall into this more impaired profile. Social support may enhance coping ability, treatment adherence, emotional resilience, and engagement in rehabilitation, thereby helping preserve intrinsic capacity (46). These findings support the importance of strengthening family involvement, peer support, and communication between healthcare professionals and patients during stroke recovery (15, 46).

Notably, the factors associated with latent-class membership in this study included both relatively non-modifiable characteristics (such as age and educational attainment) and potentially modifiable factors (such as sleep duration and social support). This distinction is clinically meaningful because non-modifiable factors may help identify vulnerable subgroups, whereas modifiable factors may represent practical intervention targets during hospitalization and early recovery.

4.2.3Quality of life

Significant differences in stroke-specific quality of life were observed among the latent classes in 11 of the 12 SS-QOL domains, with no significant difference in personality. Overall, the Function Maintenance Group showed the most favorable quality-of-life profile, whereas the Low Motor–Low Vitality Group had the lowest scores across multiple domains. These findings further support the close relationship between intrinsic capacity status and quality of life in older people with stroke (47).

Pairwise comparisons showed that, compared with the Function Maintenance Group, the Low Motor–Low Vitality Group had significantly lower scores in energy, family roles, language, mobility, mood, self-care, social roles, cognition, upper extremity function, vision, and work productivity. This pattern suggests that combined decline in locomotion and vitality may have broad adverse effects on daily functioning, participation, and overall wellbeing, which is consistent with previous findings (48).

In contrast, compared with the Function Maintenance Group, the Low Cognitive–Moderate Vitality, Psychological, and Sensory Group showed significant disadvantages in cognition, family roles, and social roles. This pattern suggests that this profile may be associated not only with difficulties in cognitive functioning and family role performance, but also with reduced social participation in daily life. These findings are broadly consistent with previous evidence indicating that post-stroke multidomain deficits may substantially affect everyday functioning and quality of life (8).

4.3Implications

Intrinsic-capacity (IC) profiling may have important practical and policy implications for the care of people with stroke. In clinical practice, IC profiling may help clinicians rapidly identify heterogeneous care needs and prioritize domain-focused rehabilitation and nursing strategies across subgroups of people with stroke. It may also support more structured multidisciplinary communication by helping physicians, nurses, therapists, dietitians, and family caregivers focus on the domains most likely to limit recovery and quality of life. At the policy level, incorporating routine IC screening and profiling into inpatient stroke care and discharge planning may support risk stratification and facilitate targeted referral pathways, including rehabilitation, nutrition, sleep management, and social-support services. In the longer term, such an approach may contribute to more efficient allocation of rehabilitation resources and more individualized continuity-of-care models for older people with stroke.

4.4Limitations

This study has several limitations. First, the cross-sectional design precludes any inference of causality regarding the associations among intrinsic-capacity profiles, influencing factors, and stroke-specific quality of life. Longitudinal studies are needed to examine how intrinsic-capacity profiles evolve over time and how they relate to subsequent recovery and quality-of-life outcomes.

Second, this was a single-center study conducted in a tertiary hospital, which may limit the generalizability of the findings to other settings, such as community-based populations, primary care institutions, or hospitals in other regions. In addition, although participants were identified as being beyond the acute stage on the basis of medical-record documentation and clinical stability, exact time since stroke onset was not consistently available for all participants. Therefore, stroke stage was determined pragmatically rather than by a standardized onset-to-assessment time threshold, which may have introduced some imprecision in stage classification.

Third, after latent class analysis, each participant was assigned to the class with the highest posterior probability for subsequent analyses. Although the entropy and posterior probabilities indicated good classification quality, classification uncertainty was not directly incorporated into the regression and ANOVA analyses, which may have led to some degree of misclassification bias.

Fourth, several measures relied on questionnaire-based or interviewer-administered assessments and may therefore have been subject to reporting bias or information bias. In particular, some psychosocial and lifestyle variables, such as perceived social support and sleep duration, may be influenced by subjective perception.

Finally, although we adjusted for several demographic, clinical, and psychosocial variables, residual confounding cannot be ruled out. Additional multicenter longitudinal studies with more detailed stroke-related information and domain-specific cognitive assessment are warranted to validate and extend these findings.

5Conclusion

This study identified three distinct intrinsic-capacity profiles among hospitalized older people with stroke, and these profiles were associated with different levels of stroke-specific quality of life. Age, educational level, sleep duration, stroke severity, and social support were important factors associated with profile membership. These findings suggest that intrinsic-capacity profiling may help identify heterogeneous functional needs in this population and support more individualized rehabilitation and nursing strategies. In clinical practice, targeted, domain-focused interventions may be particularly important for people with stroke who are older, have lower educational attainment, shorter sleep duration, more severe neurological deficits, or lower social support. From a broader practice and policy perspective, integrating routine IC screening and profiling into inpatient stroke care and discharge planning may improve risk stratification, continuity of care, and allocation of supportive rehabilitation resources for high-risk older people with stroke.

References

1.Feigin VL Brainin M Norrving B Martins SO Pandian J Lindsay P . World stroke organization: global stroke fact sheet 2025 . Int J Stroke . (2025 ) 20 :132 –44 . doi: 10.1177/17474930241308142 39635884 PMC11786524

2.Hilkens NA Casolla B Leung TW de Leeuw FE . Stroke . Lancet. (2024 ) 403 :2820 –36 . doi: 10.1016/S0140-6736(24)00642-1 38759664

3.GBD 2019 Stroke Collaborators . Global, regional, and national burden of stroke and its risk factors, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019 . Lancet Neurol . (2021 ) 20 :795 –820 . doi: 10.1016/S1474-4422(21)00252-0 34487721 PMC8443449

4.Youkee D Deen GF Sackley C Lisk DR Marshall I Soley-Bori M . Quality of life and quality-adjusted life years after stroke in Sierra Leone . Int J Stroke. (2024 ) 19 :981 –8 . doi: 10.1177/17474930241249589 38651761 PMC11528921

5.Cheon S Li CY Jeng JS Wang JD Ku LE . The lifetime burden following stroke: long term impact of stroke on survival and quality of life . Int J Stroke. (2023 ) 18 :795 –803 . doi: 10.1177/17474930231165607 36912208

6.Beard JR Officer A de Carvalho IA Sadana R Pot AM Michel JP . The World report on ageing and health: a policy framework for healthy ageing . Lancet. (2016 ) 387 :2145 –54 . doi: 10.1016/S0140-6736(15)00516-4 26520231 PMC4848186

7.Integrated Care for Older People: Guidelines on Community-Level Interventions to Manage Declines in Intrinsic Capacity [EB/OL] (2017 ). Available online at: https://pubmed.ncbi.nlm.nih.gov/29608259/ (Accessed February 18, 2025).29608259

8.Yang Y Ma G Wei S Wei X Yan B Yuan Y . Adverse outcomes of intrinsic capacity in older adults: a scoping review . Arch Gerontol. Geriatr. (2024 ) 120 :105335 . doi: 10.1016/j.archger.2024.105335 38422884

9.Meng LC Chuang HM Lu WH Lee WJ Liang CK Loh CH . Multi-trajectories of intrinsic capacity decline and their impact on age-related outcomes: a 20-year national longitudinal cohort study . Aging Dis. (2023 ) 15 :2697 –709 . doi: 10.14336/AD.2023.1115-1 38029399 PMC11567246

10.Wei X Chen Y Qin J Yang Y Yang T Yan F . Factors associated with the intrinsic capacity in older adults: a scoping review . J Clin Nurs. (2024 ) 33 :1739 –50 . doi: 10.1111/jocn.17017 38345142

11.Zhu L Shen X Shi X Ouyang X . Factors associated with intrinsic capacity impairment in hospitalized older adults: a latent class analysis . BMC Geriatr. (2024 ) 24 :494 . doi: 10.1186/s12877-024-05093-z 38840051 PMC11151595

12.Cheng Y Li W Xiao S Chen Y Qi X . Intrinsic capacity and its dimensions in relation to functional ability in older adults: a systematic review and meta-analysis . Arch Gerontol Geriatr. (2025 ) 135 :105860 . doi: 10.1016/j.archger.2025.105860 40344941

13.Chun M Qin H Turnbull I Sansome S Gilbert S Hacker A . Heterogeneity in the diagnosis and prognosis of ischemic stroke subtypes: 9-year follow-up of 22,000 cases in Chinese adults . Int J Stroke. (2023 ) 18 :847 –55 . doi: 10.1177/17474930231162265 36847304 PMC10374992

14.Buvarp D Viktorisson A Axelsson F Lehto E Lindgren L Lundström E . Physical activity trajectories and functional recovery after acute stroke among adults in Sweden . JAMA Netw Open. (2023 ) 6 :e2310919 . doi: 10.1001/jamanetworkopen.2023.10919 37126346 PMC10152305

15.Winstein CJ Stein J Arena R Bates B Cherney LR Cramer SC . Guidelines for Adult Stroke Rehabilitation and Recovery: a Guideline for Healthcare Professionals From the American Heart Association/American Stroke Association . Stroke. (2016 ) 47 :e98 –169 . doi: 10.1161/STR.0000000000000098 27145936

16.Mori M Krumholz HM Allore HG . Using latent class analysis to identify hidden clinical phenotypes . JAMA. (2020 ) 324 :700 –1 . doi: 10.1001/jama.2020.2278 32808993

17.von Elm E Altman DG Egger M Pocock SJ Gøtzsche PC Vandenbroucke JP . The The strengthening the reporting of observational studies in epidemiology (strobe) statement: guidelines for reporting observational studies . PLoS Med . (2007 ) 4 :e296 . doi: 10.1371/journal.pmed.0040296 17941714 PMC2020495

18.Chinese Guidelines for Diagnosis of Various Cerebrovascular Diseases. Chinese Journal of Neurology (2019 ). Available online at: https://rs.yiigle.com/cmaid/1161330 (Accessed March 16, 2025).

19.Feigin VL Norrving B Mensah GA . Global burden of stroke . Circ Res. (2017 ) 120 :439 –48 . doi: 10.1161/CIRCRESAHA.116.308413 28154096

20.Brott T Adams HP Olinger CP Marler JR Barsan WG Biller J . Measurements of acute cerebral infarction: a clinical examination scale . Stroke. (1989 ) 20 :864 –70 . doi: 10.1161/01.STR.20.7.864 2749846

21.Charlson ME Carrozzino D Guidi J Patierno C . Charlson comorbidity index: a critical review of clinimetric properties . Psychother Psychosom. (2022 ) 91 :8 –35 . doi: 10.1159/000521288 34991091

22.Zhang F Zhu SZ Deng PJ . Evaluation of the Perceived Social Support Scale's application in social support research for hospitalized patients in China . Chin Nurs Res. (2018 ) 32 :2048 –52 .

23.Rubenstein LZ Harker JO Salvà A Guigoz Y Vellas B . Screening for undernutrition in geriatric practice: developing the short-form mini-nutritional assessment (MNA-SF) . J Gerontol Ser A, Biol Sci Med Sci. (2001 ) 56 :M366 –372 . doi: 10.1093/gerona/56.6.M366 11382797

24.Zhang C Zhang H Zhao M Chen C Li Z Liu D . Psychometric properties and modification of the 15-item geriatric depression scale among Chinese oldest-old and centenarians: a mixed-methods study . BMC Geriatr. (2022 ) 22 :144 . doi: 10.1186/s12877-022-02833-x 35183102 PMC8858470

25.Jiang L Zeng H Wu D . Measurement invariance of the Mini-Mental State Examination among older adults with different sex and educational levels . Chin J Clin Psychol. (2022 ) 30 :940 –3 .

26.Li W Rao Z Fu Y Schwebel DC Li L Ning P . Value of the short physical performance battery (SPPB) in predicting fall and fall-induced injury among old Chinese adults . BMC Geriatr. (2023 ) 23 :574 . doi: 10.1186/s12877-023-04290-6 37723438 PMC10507826

27.Guralnik JM Simonsick EM Ferrucci L Glynn RJ Berkman LF Blazer DG . short physical performance battery assessing lower extremity function: association with self-reported disability and prediction of mortality and nursing home admission . J Gerontol. (1994 ) 49 :M85 –94 . doi: 10.1093/geronj/49.2.M85 8126356

28.Integrated Care for Older People (ICOPE) Implementation Framework: Guidance for Systems and Services . Available online at: https://www.who.int/publications/i/item/9789241515993 (Accessed February 19, 2025).

29.Sakr F Dabbous M Akel M Salameh P Hosseini H . Construction and validation of the 17-item stroke-specific quality of life scale (SS-QOL-17): a comprehensive short scale to assess the functional, psychosocial, and therapeutic factors of QOL among stroke survivors . Int J Environ Res Public Health. (2022 ) 19 :15668 . doi: 10.3390/ijerph192315668 36497746 PMC9741031

30.Cao WH Li J Guo CH . Development and evaluation of a quality of life scale for patients with stroke . Chin J Geriatr Heart Brain Vessel Dis . (2003 ) 5 :252 –4 .

31.Williams LS Weinberger M Harris LE Clark DO Biller J . Development of a stroke-specific quality of life scale . Stroke. (1999 ) 30 :1362 –9 . doi: 10.1161/01.STR.30.7.1362 10390308

32.Wang MC Bi XY . Latent Variable Modeling and Mplus Application: Advanced Level . 1st ed. Chongqing University Press (2018 ). Available online at: https://thinker.cnki.net/bookstore/book/bookdetail?bookcode=9787568908160000&type=book (Accessed February 20, 2025). In Chinese.

33.Report On Stroke Prevention And Treatment In China Writing Group . Summary of the report on stroke prevention and treatment in China 2021 . Chin J Cerebrovasc Dis. (2023 ) 20 :783 –93 .

34.Di Vincenzo O Luisi MLE Alicante P Ballarin G Biffi B Gheri CF . The assessment of the risk of malnutrition (undernutrition) in stroke patients . Nutrients. (2023 ) 15 :683 . doi: 10.3390/nu15030683 36771390 PMC9921740

35.Papadopoulou SK Papadimitriou K Voulgaridou G Georgaki E Tsotidou E Zantidou O . Exercise and nutrition impact on osteoporosis and sarcopenia—the incidence of osteosarcopenia: a narrative review . Nutrients. (2021 ) 13 :4499 . doi: 10.3390/nu13124499 34960050 PMC8705961

36.Yuan L Shen J Ye S Zhou N . Assessing care dependence status and associated influencing factors among middle-aged hemiplegic stroke patients during the post-acute rehabilitation phase: a correlational study . J Clin Nurs. (2024 ) 33 :2249 –58 . doi: 10.1111/jocn.17124 38509780

37.Sun M Wang L Hu Y Wang X Yan S Guo Y . Cognitive impairment mediates the association between dietary inflammation and depressive symptoms in the elderly . Nutrients. (2022 ) 14 :5118 . doi: 10.3390/nu14235118 36501149 PMC9737219

38.Xu Y Chen A Chen R Zheng W . Association between depressive symptoms and cognitive function in the older population, and the mediating role of neurofilament light chain: evidence from NHANES 2013–2014 . J Affect Disord. (2024 ) 360 :221 –8 . doi: 10.1016/j.jad.2024.05.165 38823588

39.Lin FR Pike JR Albert MS Arnold M Burgard S Chisolm T . Hearing intervention versus health education control to reduce cognitive decline in older adults with hearing loss in the USA (ACHIEVE): a multicentre, randomised controlled trial . Lancet. (2023 ) 402 :786 –97 . doi: 10.1016/S0140-6736(23)01406-X 37478886 PMC10529382

40.Kwan RYC Kwan CW Kor PPK Chi I . Cognitive decline, sensory impairment, and the use of audio-visual aids by long-term care facility residents . BMC Geriatr. (2022 ) 22 :216 . doi: 10.1186/s12877-022-02895-x 35296238 PMC8928635

41.Beard JR Si Y Liu Z Chenoweth L Hanewald K . Intrinsic capacity: validation of a new WHO concept for healthy aging in a longitudinal Chinese study . J Gerontol A Biol Sci Med Sci . (2022 ) 77 :94 –100 . doi: 10.1093/gerona/glab226 34343305

42.Springer MV Whitney RT Ye W Briceño EM Gross AL Aparicio HJ . Education levels and poststroke cognitive trajectories . JAMA Netw Open. (2025 ) 8 :e252002 . doi: 10.1001/jamanetworkopen.2025.2002 40136300 PMC11947833

43.Leng Y Knutson K Carnethon MR Yaffe K . Association between sleep quantity and quality in early adulthood with cognitive function in midlife . Neurology. (2024 ) 102 :e208056 . doi: 10.1212/WNL.0000000000208056 38170947 PMC10870739

44.Tang SH Lin JS Wang YY Qiao XX Li YX . Path analysis of influencing factors on quality of life in stroke patients based on structural equation modeling . Chin J Nurs . (2023 ) 58 :2747 –53 . In Chinese.

45.Kim H Teghipco A Rorden C Fridriksson J Chaves M Hillis AE . Hypoperfusion regions linked to National Institutes of Health Stroke Scale scores in acute stroke . Neuroimage Clin. (2025 ) 45 :103761 . doi: 10.1016/j.nicl.2025.103761 40015104 PMC11909741

46.Huang ZT Lai ETC Luo Y Woo J . Social determinants of intrinsic capacity: a systematic review of observational studies . Ageing Res Rev. (2024 ) 95 :102239 . doi: 10.1016/j.arr.2024.102239 38382677

47.Salinas-Rodríguez A González-Bautista E Rivera-Almaraz A Manrique-Espinoza B . Longitudinal trajectories of intrinsic capacity and their association with quality of life and disability . Maturitas. (2022 ) 161 :49 –54 . doi: 10.1016/j.maturitas.2022.02.005 35688495

48.Ahmadi S Afshar PF Malakouti K Azadbakht M . The relationship between intrinsic capacity and functional ability in older adults . BMC Geriatr. (2025 ) 25 :57 . doi: 10.1186/s12877-025-05709-y 39863832 PMC11763112

49.Wang S Chen Y Xiong L Jin N Zhao P Liang Z . Multimorbidity measures associated with cognitive function among community-dwelling older Chinese adults . Alzheimers Dement. (2024 ) 20 :6221 –31 . doi: 10.1002/alz.14117 39072982 PMC11497665

50.Liu-Ambrose T Falck RS Dao E Best JR Davis JC Bennett K . Effect of exercise training or complex mental and social activities on cognitive function in adults with chronic stroke: a randomized clinical trial . JAMA Netw Open. (2022 ) 5 :e2236510 . doi: 10.1001/jamanetworkopen.2022.36510 36227593 PMC9561961