Longevity has long been a significant pursuit for humanity. In many countries, it is no longer a problem for people to live to the age of 80.¹⁾ According to the World Health Organization (WHO) report, the global average years lived has increased by 4.6 years, from 66.8 years in 2000 to 71.4 years in 2021.²⁾ However, everyone pays attention to the years lived in good health. The difference between the average years lived and the years lived in good health in Taiwan is 7–8 years.^3,4) Aging increases the likelihood of older individuals facing life challenges and difficulties with daily activities due to a decline in function.^5,6)

Frailty is a major concern factor for functional dependence, defined as a clinical syndrome distinguished by diminished physiological capabilities of multiple organ systems, resulting in greater vulnerability to stressors.⁷⁾ Frailty has attracted extensive research attention over the past decade because it is related to a higher likelihood of harmful health effects like falls, hospitalization, and death.⁸⁾ Frailty prevalence varies globally, with estimates ranging from 8% to 14.6% depending on population and measurement approach.^9,10) In Taiwan specifically, approximately 25% community-dwelling older adults is estimated to be frail.¹¹⁾

Healthy aging, defined by the WHO-released World Report on Health and Aging, is the ongoing process of maintaining functional abilities to promote well-being as we age.¹²⁾ WHO issued Integrated Care of Older People (ICOPE) guidelines emphasizing enhancing intrinsic capabilities (IC) in 2017 and 2019,^13,14) which combines an individual’s physical and mental abilities and their interaction with environmental factors. These elements shape a person’s functional abilities throughout life and contribute to healthy aging. ICOPE care pathways are designed to enhance, maintain, or slow the decline of IC^12-15) through assessment and ongoing monitoring of six core domains: movement, energy, vision, hearing, cognition, and depressive symptoms. Unlike other approaches, the IC construct is constructed as a dynamic continuum whose trajectory is monitored throughout the second half of life. This approach provides valuable insights into the effectiveness of clinical interventions, public health measures, and the needs of older age groups.^15,16)

Recent longitudinal studies reveal a strong relationship between IC and frailty.^13,16,19-21) Belloni and Cesari¹⁵⁾ demonstrated that IC and frailty are related yet distinct constructs, with IC representing a broader measure of functional reserves. Jia et al.¹⁹⁾ found that transitions in IC status were significantly associated with changes in frailty status over a 3-year period, suggesting that IC decline may precede frailty onset. Tay et al.²⁰⁾ reported that lower IC scores were associated with increased frailty prevalence in community-dwelling Asian older adults. However, these studies have primarily examined overall IC-frailty relationships, with limited investigation into which specific IC domains are most strongly associated with different stages of frailty (particularly prefrail versus frailty).

The ICOPE approach is designed for integration into primary care for older adults. It comprises five key steps: screening, in-depth assessment, personalized care plan development, ongoing monitoring, community involvement, and caregiver support. This comprehensive framework emphasizes continuous monitoring and personalized care to effectively meet the diverse needs of older adults, ensuring tailored support for their physical, mental, and social well-being.^14,15) Our study focuses on Step 1, screening, which was used to identify potential declines in one or more of the six domains of IC. Since no global consensus exists on a standardized screening tool for assessing ICOPE, each country must adopt a unified tool. In line with this, the Taiwanese government developed ICOPE-TW, a screening tool designed to assess older adults’ IC.¹⁷⁾ The tool encompasses six domains—mental capacity, locomotion, energy (including nutrition and mood), vision, and hearing—based on the framework proposed by the WHO.^14,15)

The main obstacle in addressing IC is detecting declines in associated functional abilities early, enabling interventions to slow or reverse the decline before a person becomes frail.^15,18) Taiwan has initiated pilot programs for the WHO ICOPE framework, utilizing ICOPE-TW as a screening tool. This tool can assist healthcare providers in efficiently evaluating the IC of older Taiwanese adults, particularly those experiencing frailty.¹⁷⁾ However, there is limited evidence on how specific ICOPE-TW domains relate to different frailty stages in the Taiwanese context. Identifying which IC impairments emerge at the prefrail stage versus the frail stage could enable more targeted and timely interventions. Therefore, this study aims to assess IC using the ICOPE-TW tool among community-dwelling older adults in Taiwan, investigate the association between specific IC domains and frailty stages, and identify early markers that distinguish prefrail from robust individuals in our population.

The summary of the participants’ characteristics and differences across the frailty groups is described in Table 1. A total of 210 older adults were enrolled in the annual geriatric health examination and outpatient settings for this study. Table 1 displays the prevalence of robust, prefrail, and frail among older adults, which were found to be 68 (32.4%) classified as robust, 82 (39.0%) as prefrail, and 60 (28.6%) as frail. The majority of older adults were between 81 and 85 years old (31.9%), female (52.9%), had attained a college or university education (44.3%), were non-religious (66.2%), and were living with their family (89.2%). For health behaviors, most older adults have no smoking (98.1%), or no drinking history (96.2%), and approximately half of them exercise (54.8%).

Compared to the robust group, older female individuals with elementary school education levels and those who have no religion and no exercise habits were more likely to be classified as frail. Moreover, older individuals who had attended college or university were more likely to be classified as robust (73.5%) and prefrail (42.7%). However, there were no significant differences in the average age, distribution of age groups, gender, living status, smoking, and drinking habits across frailty groups. The difference in the number of comorbidities was significant between groups (p<0.001). The mean±standard deviation, number of comorbidities was 1.74±1.87 for robust, 3.09±2.03 for prefrail, and 4.13±1.69 for frail groups, respectively.

The ICOPE score significantly increased with the level of frailty. The frail group had a higher mean ICOPE score of 2.82, followed by the prefrail group at 1.70, compared to 0.59 for the robust group (p<0.001). Among the six domains of ICOPE, the frail group had the highest prevalence of impairment in all domains, followed by the prefrail. Limited mobility was the most common factor contributing to frailty, particularly in the frail group (87.0%). Furthermore, the prevalence of cognitive decline was significantly higher in prefrail and frail groups, with a prevalence of 37.8% and 50.0%, respectively. The prevalence of depressive symptoms also showed a significant influence in prefrail and frail groups, with a prevalence of 22.0% and 35.0%, respectively. However, no significant differences were observed in visual impairment across frailty groups.

Univariable linear regression revealed a significant positive relationship between the ICOPE score and the Fried phenotype score (Fig. 2). The regression equation was: ICOPE score = 0.56 + 0.67 × Fried phenotype score. Pearson correlation coefficient (r) was 0.600 (p<0.001), indicating a moderate positive linear relationship. The coefficient of determination (R²) was 0.36, indicating that 36% of the variability in the ICOPE score could be attributed to variation in the Fried phenotype score.

The associations between individual ICOPE domains and frailty status, analyzed using multinomial logistic regression, are presented in Table 2. After adjusting for age, sex, education level, and number of comorbidities, several significant associations emerged. Limited mobility and depressive symptoms were significantly associated with both prefrail (aOR=4.44, 95% CI 1.82–10.82; aOR=8.41, 95% CI 1.75–40.37) and frail (aOR=11.57, 95% CI 3.63–36.93; aOR=13.77, 95% CI 2.62–72.49) individuals, respectively. In contrast, malnutrition (aOR=4.01, 95% CI 1.18–13.62) and hearing loss (aOR=4.37, 95% CI=1.09–19.66) were significantly associated with frail older adults, but not with the prefrail group. Cognitive decline showed a positive association but was not significant with both prefrail (aOR=1.98, 95% CI 0.75–5.20) and frail (aOR=1.53, 95% CI=0.52–3.20) groups after adjustment. Visual impairment was not significantly associated with either prefrail (aOR=0.42, 95% CI 0.08–2.27) or frail (aOR=0.40, 95% CI 0.06–2.83) status after adjustment.

Regarding covariate effects (Table 2), age showed minimal association with frailty stages across all domains (aOR range of 1.03–1.07; all non-significant). Gender showed no significant associations with either prefrail or frail status across all domains. College/university education was consistently and strongly protective against frail status compared to elementary education across all six domains (aOR range of 0.07–0.12; all p<0.01), but showed no significant association with prefrail status. Number of comorbidities was the most consistent predictor, showing significant associations with frailty status in five out of six domains (prefrail, aOR range of 1.39–1.46; frail, aOR range of 1.31–1.86).

This study investigated the relationships between specific IC domains, as assessed by the ICOPE Step 1 screening tool, and frailty stages in community-dwelling older adults in northern Taiwan. Our main findings were that limited mobility and depressive symptoms were significantly associated with both prefrail and frail individuals, suggesting these impairments emerge early in the frailty continuum. In contrast, malnutrition and hearing loss were significantly associated only with frail older adults, indicating these may represent later-stage markers. These stage-specific associations have important clinical implications for targeted screening and early intervention.

Our findings align with and extend those of Zhou et al.,²⁸⁾ who demonstrated that IC impairments identified through ICOPE Step 1 screening were significantly associated with frailty in 1,164 Singaporean community-dwelling older adults. Similar to our study, Zhou et al.²⁸⁾ found that reduced mobility and psychological impairment were significant predictors of frailty. However, our study provides additional clinical insight by distinguishing associations at the prefrail versus frail stages. We found that limited mobility and depressive symptoms emerge as early markers at the prefrail stage (aOR 4.44 and 8.41, respectively), while malnutrition and hearing loss demonstrate significant associations only at the frail stage. This stage-specific finding has important implications for targeted screening: our results suggest that primary care providers should prioritize mobility and mood assessment for early detection of prefrail, potentially enabling earlier intervention before progression to frailty.

Zhou et al.²⁸⁾ reported that sensory impairments (vision and hearing combined) conferred a 47%–99% increased risk of frailty, while we found that hearing loss was significant only for frail individuals, and visual impairment showed no significant association. This discrepancy may reflect several factors: differences in sample characteristics (our older mean age 81.5 vs. their 72.1 years), the relatively small proportion of participants with visual issues in our sample (4.8%), or cultural differences in reporting sensory problems between Taiwan and Singapore. The non-significant results regarding visual impairment were similar to another study using data collected in southern Taiwan,¹⁷⁾ suggesting that this may be a consistent pattern in the Taiwanese context that warrants further investigation. Our findings are consistent with the conceptual framework proposed by Belloni and Cesari,¹⁶⁾ who described IC and frailty as related but distinct constructions, with IC representing broader functional reserves. The moderate correlation (r=0.600) we observed between ICOPE and Fried scores supports this relationship while confirming they measure overlapping but non-identical constructs. This is consistent with the WHO’s vision of IC as a dynamic continuum that can be monitored to prevent or delay the onset of frailty. Ruiz et al.²⁷⁾ indicated that the ICOPE program may be particularly useful for older people who are either prefrail or frail, as interventions at these stages may be more effective than waiting until advanced frailty develops. Our finding that limited mobility and depressive symptoms are already significantly elevated at the prefrail stage (4–8 fold increased odds) supports this recommendation. These two domains could serve as priority screening targets in primary care settings, with positive screens triggering more comprehensive assessment and early intervention before progression to frailty. The stepped approach recommended in ICOPE guidelines—screening (Step 1), followed by in-depth assessment (Step 2) and personalized care planning (Step 3)—appears well-suited to this stage-specific risk profile.

The relationship between educational level and ICOPES-TW scores aligns with existing literature, suggesting that older adults with higher educational levels have better health literacy.²⁹⁾ Improved health literacy may protect against unhealthy behaviors³⁰⁾ and help maintain IC across domains. In our sample, 73.5% of robust individuals had a college or university education compared to only 13.3% of frail individuals, suggesting education may be an important protective factor.

Our findings on the correlation between individual IC loss and frailty were consistent with previous research linking malnutrition,^31-33) sensory loss,^19,28) depressive symptoms,^28,34) and reduced mobility^28,35) to frailty. The specific finding that prefrail is significantly associated only with limited mobility and depressive symptoms were also consistent with previous research.^36-38) Serra-Prat et al.³⁶⁾ found that prefrail patients showed early changes in mobility and mood but not necessarily in nutritional status. Sousa-Santos et al.³⁷⁾ reported that weakness (closely related to mobility) was the most frequent criterion among prefrail Portuguese older adults. Wu et al.³⁸⁾ demonstrated that depressive symptoms were a common risk factor for both prefrail and frailty in patients with Alzheimer's disease. Our study confirms these patterns in the Taiwanese context and suggests that these two domains—mobility and depressive symptoms—might serve as sensitive early indicators of the transition from robustness to prefrail.

Using the Fried frailty phenotype, we found that among 210 participants living in northern Taiwan, 39.0% were prefrail and 28.6% were frail. A previous study using the Taiwan Longitudinal Study on Aging (1989–2011) showed frailty occurrence rates of 33.1% in urban Taiwan and 38.1% in rural Taiwan,¹¹⁾ which are comparable to our findings. These prevalence estimates underscore the substantial burden of frailty in Taiwan, highlighting the need for effective screening and intervention tools. The ICOPE rationale for older adults emphasizes that early interventions aimed at reversing prefrail or preventing older adults from becoming frail are more successful than interventions initiated after frailty is established. Tools like ICOPES-TW, which can identify at-risk individuals, have been shown to help decrease the need for more intensive, step-by-step approaches often better suited for already frail older adults.²⁷⁾

Our study has several strengths. We evaluated the relationship between frailty and IC using validated instruments appropriate for the Taiwanese population. The availability of comparisons between measurements of IC and frailty domains relevant to healthy aging enables the IC to identify older adults at risk of advanced, harmful health effects. The focus on differentiating prefrail from frail stages provides actionable clinical insights for targeted screening. However, several limitations should be acknowledged. First, the cross-sectional design limits our ability to make causal inferences and determine whether IC decline precedes frailty development or occurs concurrently. Longitudinal studies are needed to establish temporal relationships and determine whether interventions targeting mobility and mood in prefrail individuals can prevent progression to frailty. The relatively modest sample size (n=210) limited the statistical power for detecting small effects and precluded subgroup analyses by age categories or gender. In addition, the sampling method represents only the northern region of Taiwan through a single medical center, which limits generalizability to the broader Taiwanese older adult population and other regions or countries. Selection bias may also be present, as participants were recruited through health screenings and outpatient settings, potentially excluding the most frail individuals who cannot access these services. Some questions in this study were derived from a self-report nature, particularly for sensory function and psychological symptoms. Therefore, recall bias and social desirability bias may occur when participants respond to specific self-report measures. Furthermore, the ICOPE domains used are screening instruments rather than gold-standard diagnostic measures. For example, the three-item cognitive screening is less comprehensive than full neuropsychological testing, and the two-question psychological screening is not equivalent to a clinical depression diagnosis. This may affect the precision of our estimates and could overstate the reliability of these associations. However, this limitation must be balanced against the clinical utility of ICOPE as a first-step screening tool, designed for primary care settings where rapid and feasible assessment is prioritized over comprehensive diagnostic evaluation. The ICOPES-TW has demonstrated good validity in validation studies,¹⁷⁾ supporting its utility for the intended purpose despite these inherent limitations of screening tools. Finally, IC measurement approaches differ somewhat across studies conducted in different countries, as there is no universally accepted gold standard for assessing IC. A validated and standardized assessment method would be valuable to ensure consistency when comparing findings across studies and populations.