Korea is experiencing the fastest pace of transformation in the population structure among developed countries and facing social and economic challenges from its aging population (1). Accompanied by a rapid shift in social structure, including the decreasing family size and urbanization, Korea is expecting decades of a super-aged society. Meanwhile, even Western countries with better resources and a slow rate of population aging have been struggling to maintain high-quality services for their older populations (1, 2). The long-term care system in Korea is expected to be overwhelmed in the near future by soaring care demands from older adults with disabilities (3). Hence, there is an urgent need to develop sustainable care models to address the spectrum of individual care demands while optimizing available resources.

Growing knowledge on the importance of the frailty spectrum might be one of the most remarkable achievements in geriatric medicine in this century in person-centered risk assessment and care provision for older adults (4–6). Measured by either the phenotype or the index, the spectrum of frailty has shown high prediction ability for adverse health outcomes encompassing healthcare use and incidence of geriatric syndromes, such as fall and delirium, as well as disability and death (7, 8). These features of frailty have led to a growing research interest in incorporating frailty as a risk measure in many specialized fields of medicine outside geriatrics (9). Translation studies have shown content validity of the frailty spectrum as a measure of human biological age (10, 11). It has been reported that frailty was a dynamic and malleable feature, which responded to appropriately designed multicomponent intervention targeting geriatric functional domains (12, 13). Consequently, addressing the frailty spectrum might be a potential cornerstone in designing tailored care models for the older population, encompassing areas of health promotion, disease treatment, prevention of disabilities, and even long-term caregiving.

Despite its importance, assessing and interpreting frailty has long been regarded as a barrier in incorporating this condition into real-world clinical practice (14). However, the Clinical Frailty scale (CFS), a simplified measure including pictographs and brief descriptions on global fitness and functional capabilities, has been suggested as a viable approach (15, 16). After being initially developed to trace the frailty index, validity of the CFS has been shown in diverse spectrums of care situations, including intensive care units, long-term care facilities, and sometimes in a large population (15, 17). In the Korean population, studies have shown validities of the CFS in geriatric outpatients and inpatients in a hospitalist unit and an intensive care unit (18–20). However, to the authors' knowledge, its utility and validity in community-dwelling older people in Korea are yet to be proven, even though the CFS has potential advantages as an efficient measure of the frailty in community-based public health programs. Furthermore, in previous attempts adopting CFS in the present authors' institution, we faced questions on interpreting and classifying capabilities on instrumental activities of daily living (IADL) of male patients.

Recognizing this knowledge gap, we aimed to assess the characteristics and validities of the CFS in community-dwelling older people in Korea. We measured the CFS by adopting the decision tree, which was recently developed and validated by Theou et al. (21), and assessed its characteristics with the generally well-accepted measures of frailty phenotype and frailty index. In this adoption, we compared the original classification and culturally modified classification considering controversial IADL items of older male population in Korea. We then evaluated whether the CFS could predict a composite outcome of institutionalization and death.

In this study, we evaluated the validities of oCFS and mCFS in community-dwelling Korean older adults, using the classification tree of CFS that was previously validated in a population from the United Kingdom. When the frailty index was used as a standard, mCFS was better than oCFS in construct validity. On the other hand, mCFS and oCFS were comparable in predicting the composite outcome. To the authors' knowledge, this is the first study adopting the classification tree of CFS in an Asian population evaluating construct and criterion validity.

The growing consensus support that frailty is a core clinical feature of the complex system of aging physiology (5), and both the phenotype model and the deficit-accumulation model converge with each other (7), even though controversies still exist to date on the biological and clinical construct of frailty. The CFS was conceptualized from a theoretical model of fitness and frailty (32), encompassing the functional spectrum as a measure of global fitness (16). In the original study, the CFS highly correlated with the frailty index and had comparable prediction ability for outcomes of mortality and institutionalization (16). Hence, the validity of the CFS from the classification tree is predictable because of its high agreement with the CFS by the geriatricians (21). Even though the CFS directly measured by the geriatricians were unavailable, the CFSs determined from the CGA records correlated with both the frailty index and CHS frailty phenotype, supporting convergent validity of the classification tree.

For mCFS, we did not consider the presence of dependency in IADL items of household chores and preparing meals as sufficient to classify as CFS 5 in men. Controversies existed in interpreting care needs in IADL items in the older Asian populations in geriatrics and gerontology research (26, 33), as traditionally, household work was largely performed by women in the region. In a Korean study on gender differences of ADL and IADL items, men were more likely to rate themselves as dependent in household activities, such as preparing meals and doing laundry (26). A similar gender bias was noted in a Singaporean study, reporting high dependency in preparing meals, doing laundry, and taking medication in men (34). Despite the relevance of these items in the clinical construct of IADL disability being difficult to conclude, we observed that the convergent validity of the CFS was better when dependencies in household items were interpreted leniently (mCFS) than strictly (oCFS) in men. Furthermore, the frailty index of men reclassified from CFS 5 to CFS 4 by the 'cultural modification' was lower than the population who remained at CFS 5. These findings support the potential cultural impact has on defining the construct of disability and frailty, calling for research on this topic as many Asian countries are experiencing rapid population aging and the need to establish social support systems for their aging population (2).

There was no significant difference between mCFS and oCFA in predicting the composite outcome of death and institutionalization. As HRs of CFS 4 and CFS 5 for both classifications were relatively similar, the effect of the reclassification would be minimal in prediction ability for the outcome. As we merged the two outcomes of death and institutionalization into a composite outcome in this analysis, given limitations in sample size and observation period, we could not dissect the impact of IADL interpretation on mortality and functional decline. The minimal outcome differences between CFS 4 and CFS 5 might be due to availability of long-term care insurance that provides at-home assistance for community-dwelling older adults with disabilities in IADL items. Yet, the impact of this service on preventing institutionalization of people living with IADL impairment cannot be proven under the current study design (3). However, with survival analysis showing a discrete and decisive increment in outcome risk from CFS 4, the CFS could be used to screen community-dwelling older adults at risk of adverse health outcomes.

Evidence supports well-designed multicomponent intervention programs that effectively prevent adverse health outcomes in community-dwelling older adults (12, 35, 36). A recent paper published from Korea, analyzing the long-term outcomes of a 6-month program with a propensity-score-matched control group, showed that both the frailty status and physical performance was better in the intervention group, and the benefit in physical performance remained significant for 2 years after the end of the program (12). Furthermore, the 30-month mean institutionalization-free survival time was longer by 5.2 months in the intervention group than in the comparison group. To scale up these programs with multiple domains in the public health scale for a more extended period, selecting the target population that might benefit maximally from such interventions is necessary. In this regard, early programs may start by focusing on the population of CFS 4–5, considering its prediction ability for institutionalization-free survival. Given the brief nature of the CFS, the burden of finding mass-scale cases might be minimal.

There are several limitations to this study. We used recorded data of the CGA to estimate oCFS and mCFS, rather than prospectively measuring CFS by geriatricians. Nevertheless, as we constructed the CFS from individual items of the CGA, we were able to tag individuals from IADL items; however, recording all relevant IADL items and frailty index parameters might be less feasible in real-world CFS examinations by geriatricians. Generalizability is limited, as we were only able to assess cultural impact in determining the CFS in a community-dwelling population in a rural area in Korea. As noted in the literature, cultural characteristics are rapidly changing with alterations in family and social structures in Asian countries, including China (37). Our data were acquired from a rural area in the mid-2010s, and the results might differ if performed using individuals in an urban area. Also, baseline characteristics of participants with or without IPAQ differed considerably in terms of age, frailty, and other geriatric parameters (Supplementary Table 3), which might act as a limitation for generalizability. As the current analysis was from the observational cohort, we could not assess the potential effect of interventions on the CFS, as studies showing the frailty spectrum could be altered by interventions (12). As the ASPRA primarily comprised of community-dwelling, ambulatory participants, we had no CFS 8–9 individuals.

In conclusion, the classification tree of CFS could be culturally adopted in a community-dwelling population in Korea and considered valid in detecting vulnerable population. As such, further studies are warranted to assess the feasibilities and benefits of CFS when performed in the public health scale to screen vulnerable population who might benefit from community-based programs.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional Review Board of Asan Medical Center. The patients/participants provided their written informed consent to participate in this study.

H-WJ, JB, I-YJ, and EL conceptualized the study and reviewed and edited the manuscript. H-WJ did the formal data analysis. H-WJ, JB, and I-YJ did the investigation. H-WJ and JB wrote the first draft of the manuscript. H-WJ and I-YJ supervised the study and acquired study funding.

This Aging Study of Pyeongchang Rural Area was funded by the Pyeongchang Health Center, Pyeongchang County, Gangwon Province, South Korea. This study was also supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health and Welfare, Republic of Korea (Grant No. HI18C2383), and the Asan Multidisciplinary Committee for Seniors.

H-WJ cofounded Dyphi Inc., a startup company based on sensor technology. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.