Introduction
Home healthcare (HHC) refers to healthcare services that provide longitudinal interdisciplinary care to homebound or functionally impaired patients who have difficulty accessing traditional outpatient clinics (Ritchie et al. Reference Ritchie, Leff and Garrigues2018). These services can vary widely geographically and internationally, encompassing diverse programs and models. The scope of services ranges from acute hospital care at home to long-term support and treatment for chronic or terminal illnesses. The primary goal is to deliver integrated, optimal care to patients in their familiar home environments (Liao et al. Reference Liao, Chen and Wu2020).
Patients receiving HHC often exhibit higher levels of frailty, disability, and malnutrition (Kojima et al. Reference Kojima, Iliffe and Walters2018; Huang et al. Reference Huang, Umegaki and Kamitani2019; Proietti and Cesari Reference Proietti and Cesari2020). For instance, a Norwegian study revealed that 76% of older HHC patients had moderate or severe frailty, with 67% of this group dying within a 2-year follow-up period (Krogseth et al. Reference Krogseth, Rostoft and Benth2021). Given the increased vulnerability of individuals with frailty to poor health outcomes and adverse events (Soones et al. Reference Soones, Federman and Leff2017; Ritchie and Leff Reference Ritchie and Leff2018), advance care planning (ACP) is a critical aspect of care for this population (Crooms and Gelfman Reference Crooms and Gelfman2020).
ACP is a process facilitating the discussion and understanding of personal values, life goals, future medical choices, and end-of-life care preferences (Sudore et al. Reference Sudore, Lum and You2017). It is important to distinguish between ACP engagement, which represents the relational and psychological process of readiness and communication, and advance directive (AD) completion, which is the formal, often legal, outcome of that process. While effective communication regarding ACP enhances the ability of older people to participate in shared decision-making, “readiness” to engage in these discussions does not always result in the immediate completion of formal documentation (Houben et al. Reference Houben, Spruit and Groenen2014; Kinley and Flemming Reference Kinley and Flemming2021).
Decision-making self-efficacy, or an individual’s perceived ability to engage in discussions and make one’s own decisions, emerges as a crucial factor for ACP and AD (Farley Reference Farley2020). However, people with frailty frequently manifest diminished self-efficacy, further compounded by age-related multimorbidity and functional dependence (Cybulski et al. Reference Cybulski, Cybulski and Krajewska-Kulak2017; Whitehall et al. Reference Whitehall, Rush and Gorska2021). Furthermore, communication barriers, including cognitive and physical decline related to frailty, may hinder individuals’ ability to actively engage in decision-making processes (Doba et al. Reference Doba, Tokuda and Saiki2016; Maruta et al. Reference Maruta, Makizako and Ikeda2022).
In Asian countries, cultural factors such as collective decision-making and misperceptions of frailty as a total loss of capacity – may influence the decisional self-efficacy among older people (Martina et al. Reference Martina, Lin and Kristanti2021b). Health literacy, patients’ knowledge of their disease, and ACP may also pose challenges in facilitating ACP discussions (de Vries et al. Reference de Vries, Banister and Dening2019; Martina et al. Reference Martina, Geerse and Lin2021a). In HHC settings, family dynamics, social support, and the availability of healthcare resources also play pivotal roles in shaping the effectiveness of ACP among frail patients (Boerner et al. Reference Boerner, Carr and Moorman2013; Martina et al. Reference Martina, Geerse and Lin2021a). However, little is known about the relationship between frailty and ACP in patients receiving HHC. This study aims to investigate the association between levels of frailty, decision-making self-efficacy, ACP engagement, and AD completion among frail older people receiving HHC in a nationwide cohort in Taiwan.
Methods
This study is part of the HOme-based Longitudinal Investigation of the multidiSciplinary Team Integrated Care (HOLISTIC) study, which is the first nationwide prospective longitudinal study of HHC in Taiwan (Liao et al. Reference Liao, Chen and Wu2020). The HOLISTIC study involved a 2-year observation period starting in November 2019, covering 18 sites nationwide. It included 6 assessments of patients receiving HHC services and their caregivers. Detailed recruitment information is available in our published article (Liao et al. Reference Liao, Chen and Wu2020). For the present study, we have utilized the baseline patient data from the HOLISTIC study, including age, sex, Charlson Comorbidity Index (CCI), education, marital status, living status, religious beliefs, and income.
Participants
In the HOLISTIC study, patients were considered eligible if they met the following inclusion criteria: (1) age over 50 years, (2) receipt of HHC for at least 2 months, (3) Barthel index <60, (4) ability to communicate with the interviewer in a mutually understood language, and (5) cognitive impairment, if present, was compensated for by a cognitively competent caregiver who facilitated communication with the interviewer. In the HOLISTIC protocol, caregiver facilitation was employed primarily to support patients with cognitive impairment and to help interviewers’ understanding of patients’ pre-dementia values and preferences, rather than to act as proxy decision-makers. The interviews were specifically designed to prioritize patients’ perspectives, utilizing a first-person approach to reflect their personal experiences while promoting inclusivity for vulnerable populations. This approach aligns with the concept of supported decision-making, ensuring accurate representation of the patients’ perspectives through those who know them best and can advocate for them.
During the initial recruitment phase of the HOLISTIC study, HHC team members provided an overview of the study protocol. To ensure effective communication, interviewers received training that included standardized explanations and education on ACP, enabling them to effectively address any questions or misconceptions. These trained interviewers then contacted participants by telephone to schedule home visits, where they offered a more detailed explanation of the study’s purpose and procedures. After confirming participants’ willingness and eligibility, written informed consent was obtained, and interviews were conducted. The caregivers or family members, who knew the patients well and could best advocate for them, facilitated the interviews, ensuring accurate representation of the patients’ perspectives.
Measures
Within the HOLISTIC study, a structured questionnaire consisting of 5 domains (health-related outcomes, end-of-life issues, continuity and coordination of care, care resource utilization and costs, and caregiving burden) was used (Liao et al. Reference Liao, Chen and Wu2020). For the purpose of this study, only measures relevant to our research objectives are discussed, including the 9-point Clinical Frailty Scale (CFS), the DEcision-making Participation Self-efficacy scale (DEPS), ACP as measured by the 4 questions in ACP Engagement Survey, and demographic information.
A trained nurse conducted a comprehensive geriatric assessment, evaluating frailty using the 9-point CFS. This scale offers a distinct description for each score, ranging from 1 (very fit) to 9 (terminally ill), and is internationally recognized as a common tool for evaluating frailty among older individuals (Church et al. Reference Church, Rogers and Rockwood2020). The translated CFS has shown strong validity and reliability in the Taiwanese population (Chou et al. Reference Chou, Tsou and Chan2022). In this study, we classified patients with CFS scores of 4–5 as mildly frail, 6 as moderately frail, 7 as severely frail, and 8–9 as very severely frail. We included patients aged 65 and older, excluding those with CFS scores of 1–3 (n = 25), as they were not considered frail.
The DEPS was used to assess patients’ confidence in their participation in shared decision-making (Arora et al. Reference Arora, Weaver and Clayman2009). The scale consists of 5 questions, with patients responding on a 5-point Likert scale ranging from 1 (not at all confident) to 5 (completely confident). Higher scores indicate greater perceived self-efficacy in engaging with treatment-related decisions.
A 4-item ACP Engagement Survey was used to assess patients’ readiness to engage in discussion concerning their personal beliefs and future treatment preferences with both family members and healthcare teams (Sudore et al. Reference Sudore, Heyland and Barnes2017). Patients responded on a 5-point Likert scale, ranging from 1 (never thought about it) to 5 (already completed). Higher scores indicate greater preparedness in understanding and sharing personal values, life goals, and preferences for future medical care. The reliability and validity of the ACP Engagement Survey in Taiwan have been verified, demonstrating good internal consistency (Wei et al. Reference Wei, Hsu and Wu2022).
To ensure clinical relevance and reflect the distribution within our cohort, outcome variables were categorized as tertiles: DEPS scores (≤9, 10–16, and >16) and ACP scores (4, 5–6, and >6). In both measures, higher scores indicate superior outcomes – specifically, greater perceived self-efficacy in decision-making and higher levels of ACP preparedness, respectively. Covariates encompassed demographic variables, including age, sex, comorbidity status as determined by the CCI (Charlson et al. Reference Charlson, Szatrowski and Peterson1994), educational level, marital status, living arrangements, religious affiliation, monthly income, and the presence of indwelling tubes or catheters. CCI was ascertained based on past and current medical records.
Statistical analysis
A multinomial logistic regression was used to analyze the association between CFS scores and outcome variables. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated after adjusting for all covariates. In the sensitivity analysis, only individuals with a CFS score of 4–5 were included in the mildly frail group to confirm the robustness of our findings. A significance threshold of p < 0.05 was utilized to delineate statistical significance across all analyses. Statistical computations were conducted using SAS software, version 9.4 (SAS Institute, Inc., Cary, NC, USA).
Results
Table 1 offers a comprehensive overview of the characteristics of the study’s patients based on their CFS score. A total of 358 patients (206 women) were enrolled. The study cohort encompassed 60 mildly frail patients (CFS score 4–5), 83 moderately frail patients (CFS score 6), 147 severely frail patients (CFS score 7), and 68 very severely frail patients (CFS score 8–9). The mean age exhibited a progressive increase in tandem with rising CFS scores, ranging from 82 years for mildly frail patients to 86 years for very severely frail patients. CFS scores demonstrated significant disparities in the distribution of age across 3 age categories (<75, 75–84, and ≥85) (p < 0.01). Patients with higher CFS scores (8–9) had elevated CCI scores (mean = 3) and were more likely to live with immigrant caregivers and have household incomes exceeding NT$20,001. Additionally, higher CFS scores were significantly associated with the use of indwelling medical devices (p < 0.01). While 51.7% of the mildly frail group were able to complete the assessment independently, this proportion decreased to 30.1% for the moderately frail, 9.5% for the severely frail, and 0% for the very severely frail group (p < 0.001).
Characteristics of the study participants by Clinical Frailty Scale (CFS) score
CCI = Charlson Comorbidity Index, HBMC = home-based medical care.
a Data were presented as means and standard deviations.
b The Kruskal–Wallis test was used because the data were not normally distributed. The results are presented as medians and interquartile ranges (Q1–Q3).
DEPS scores were significantly higher in mildly frail patients compared to moderately frail patients, severely frail patients, or very severely frail patients (p = 0.03) (Table 2). Among mildly frail patients, 25 (41.7%) exhibited DEPS scores exceeding 16, while the corresponding figures for moderately frail, severely frail, and very severely frail patients were 28 (33.7%), 43 (29.3%), and 16 (23.5%), respectively. Notably, CFS scores showed no substantial correlation with responses to the individual 5 DEPS questions. In the multinomial logistic regression analysis, CFS scores exhibited a trend of negative association with DEPS scores. In comparison to mildly frail patients, very severely frail patients (OR = 0.38, 95% CI = 0.14–1.07) exhibited diminished odds of attaining DEPS scores surpassing 16.
Association between CFS and DEcision-making Participation Self-efficacy scale (DEPS) scores
a The Kruskal–Wallis test was used because the data were not normally distributed. The results are presented as medians and interquartile ranges (Q1–Q3).
Individual and total scores of ACP were not associated with CFS scores (Table 3, panel a). The rates of formal completion for ADs were consistently low across all 4 frailty groups, ranging from 4.8% to 10.0%. Notably, frailty status was not significantly associated with the likelihood of having an AD in place (p = 0.44) (Table 3, panel b). Multinomial logistic regression analysis further showed that moderately frail patients (OR = 3.38, 95% CI = 1.4–8.17) and severely frail patients (OR = 2.52, 95% CI = 1.08–5.89) had a higher likelihood of achieving ACP scores exceeding 6 compared to the mildly frail group (Table 4). However, this increase did not follow a linear progression into the very severely frail group (OR = 2.60, 95% CI = 0.96–7.03), suggesting a plateau in engagement as frailty reaches its most advanced stages.
Association between CFS and (a) UCSF-Advance Care Planning (ACP) scores, and (b) advance directives
a The Kruskal–Wallis test was used because the data were not normally distributed. The results are presented as medians and interquartile ranges (Q1–Q3).
Multinomial logistic regression of DEPS scores and ACP scores
All models adjusted for age, sex, CCI, education, marital status, living status, religious beliefs, income (NTD/month), and indwelling tube or catheter.
a DEPS scores: excluding missing (n = 13).
b ACP scores: excluding missing (n = 1).
Discussion
This study provides new insight into how frailty shapes ACP processes among older adults receiving HHC. As frailty increased, decisional self-efficacy declined, yet readiness to engage in ACP was higher among those with moderate-to-severe frailty compared to mildly frail individuals. Importantly, greater engagement did not consistently translate into formal AD completion. Rather than viewing these findings as contradictory, they highlight a clinically meaningful divergence between perceived decision-making confidence and motivation to plan ahead. Recognizing this divergence is essential for identifying when and how to initiate supported ACP discussions in the context of progressive vulnerability.
Consistent with prior reports, self-efficacy for participation in medical decision-making declined as frailty increased (Whitehall et al. Reference Whitehall, Rush and Gorska2021). This pattern aligns with theoretical expectations that frailty, often compounded by multimorbidity, symptom burden, and functional dependence, can erode individuals’ confidence in navigating complex healthcare decisions. At the same time, accumulating experiences of clinical deterioration and repeated healthcare encounters may heighten awareness of future health uncertainties, thereby increasing openness to ACP discussions (Kinley and Flemming Reference Kinley and Flemming2021). In contrast, mildly frail or relatively stable older adults may perceive ACP as less urgent or less relevant to their immediate circumstances (Wan et al. Reference Wan, Chan and Chiu2022).
The co-occurrence of declining decisional self-efficacy and heightened ACP readiness likely reflects 2 related but distinct constructs. Self-efficacy captures an individual’s confidence in actively leading medical discussions, whereas ACP engagement reflects recognition of the importance of articulating future care preferences. As physiological vulnerability progresses, these constructs may move in different directions: patients may feel less confident in managing complex medical conversations while simultaneously becoming more aware of the need for planning. This distinction underscores both the importance of tailoring ACP approaches to the patient’s stage of illness and decision-making capacity, and the proactive role of healthcare professionals in facilitating and supporting ACP initiatives among their clients by all means (Combes et al. Reference Combes, Forbes and Gillett2021). Programs such as Singapore’s tiered national ACP framework, ranging from General ACP to Disease-Specific ACP and Preferred Plan of Care, illustrate how structured, stage-sensitive models may better align ACP engagement with evolving clinical trajectories (Ng et al. Reference Ng, Lip Hoe and Lim2023).
However, the absence of increased AD completion should not be interpreted as a failure of ACP. Rather, it reflects a clinically important adaptation between psychological engagement and formal documentation. As frailty progresses, patients’ ability to engage in the complex process of ACP consultation, including completing documentation, might also decline. In this context, readiness may precede action, and engagement may not immediately translate into signed documents. Prior work has shown that self-efficacy in older adults is influenced by relational dynamics and prior patterns of health communication (Korpershoek et al. Reference Korpershoek, van der Bijl and Hafsteinsdottir2011; Remm et al. Reference Remm, Halcomb and Hatcher2023), suggesting that reduced confidence does not necessarily imply disengagement. Therefore, ACP processes should prioritize structured, iterative, and values-based conversations that support participation before decisional confidence further diminishes. Approaches that emphasize relational dialogue over administrative completion, such as collaborative self-management models, may help bridge the gap between intention and formalization (Lorig et al. Reference Lorig, Sobel and Stewart1999; Thomas et al. Reference Thomas, Liu and Umberson2017). Evaluating ACP success through alignment with patient values and quality of end-of-life engagement, rather than document acquisition alone, may provide a more meaningful measure of impact across the frailty spectrum (Malhotra and Ramakrishnan Reference Malhotra and Ramakrishnan2022; Malhotra Reference Malhotra2023).
Several contextual factors may further explain why heightened readiness did not translate into document completion. Taiwan’s national health insurance does not routinely reimburse ACP consultations, and the requirement for out-of-pocket payment may deter formalization (He et al. Reference He, Lin and Hsu2021; Wu et al. Reference Wu, Yang and Lin2023). Emotional avoidance of mortality, information overload during discussions, and decisional paralysis may also delay action despite expressed willingness (Lund et al. Reference Lund, Richardson and May2015). Sociocultural norms that frame end-of-life conversations as taboo (Lin et al. Reference Lin, Cheng and Chen2018), limited institutional training or standardized protocols (Lund et al. Reference Lund, Richardson and May2015; Blackwood et al. Reference Blackwood, Walker and Mythen2019), and family disagreement may further complicate the transition from intention to documentation. In addition, common misperceptions of ACP in Asia – that it is primarily financial planning, a legal document, or a conversation solely about death – may obscure its broader relational and values-based purpose (Martina et al. Reference Martina, Geerse and Lin2021a).
An additional nuance emerges at the highest levels of frailty. The odds of ACP engagement did not continue to rise linearly among individuals classified as very severely frail (CFS 8–9). In advanced frailty, profound multimorbidity, and cognitive impairment may constrain patients’ ability to further deepen ACP engagement, even if awareness of vulnerability remains high. At this stage, communication complexity, decisional fatigue, and reliance on surrogate participation may limit measurable increases in readiness. Moreover, the generally low rate of formal document completion in Taiwan may attenuate observable differences in this subgroup. This plateau effect suggests that ACP engagement may follow a curvilinear rather than strictly linear trajectory across the frailty spectrum.
Surrogate involvement in ACP discussions is common among individuals with advanced frailty or cognitive impairment (Golden et al. Reference Golden, Rubim and Thammana2023). Although proxy-reported outcomes may not fully capture patients’ internal perspectives, evidence suggests reasonable alignment between patient and caregiver reports in domains such as quality of life (Takura et al. Reference Takura, Koike and Matsuo2022), palliative care outcomes (Murtagh et al. Reference Murtagh, Ramsenthaler and Firth2019), and life-sustaining treatment decisions (Lin et al. Reference Lin, Peng and Chen2020), with moderate concordance observed in cognitively impaired populations(Clapham et al. Reference Clapham, Daveson and Allingham2021; Kroenke et al. Reference Kroenke, Stump and Monahan2022). The ACP Engagement Survey has demonstrated strong internal consistency when completed by both individuals and surrogate decision-makers (Cronbach’s alpha: 0.90–0.91) (Van Scoy et al. Reference Van Scoy, Day and Howard2019).
Within this context, it is ethically important to distinguish between substituted and supported decision-making. Substituted decision-making entails a surrogate making choices on behalf of an incapacitated individual based on presumed preferences, whereas supported decision-making provides assistance that enables individuals to express their own values and choices to the greatest extent possible. In progressive frailty, this distinction becomes particularly critical. The recent Consensus Definition of Advance Care Planning in Dementia from the European Association for Palliative Care emphasizes inclusivity in developing care plans for individuals with cognitive impairment (van der Steen et al. Reference van der Steen, Nakanishi and den Block L2024). In our cohort, caregivers facilitated rather than dictated responses, reflecting clinical realities in many Asian settings where ACP is embedded within family-centered decision-making traditions (Lin et al. Reference Lin, Cheng and Chen2018; Chiang et al. Reference Chiang, Wang and Hsieh2021; Mori et al. Reference Mori, Chan and Lin2025). These contextual considerations should inform the interpretation of the observed divergence between self-efficacy and ACP engagement.
The strength of our study resides in its utilization of a national cohort, employing stratified sampling across urban, suburban, and rural areas to examine the relationship between frailty, self-efficacy, and ACP in older adults. By recruiting patients from predefined regions, we aimed to minimize sampling bias, as the availability and utilization of healthcare services may exert an influence on self-efficacy and ACP. However, there are still several limitations. First, our findings are based on a national cohort in Taiwan, which limits the generalizability of our results to individuals from diverse ethnic backgrounds, cultural contexts, religious beliefs, and ethical ideals worldwide. Second, our results may not apply to older adults who are fully independent in their daily care. Third, the cross-sectional nature of this baseline analysis does not capture the dynamic, prospective fluctuations in self-efficacy and ACP engagement as frailty progresses or as health crises occur. Fourth, we did not assess our subjects’ baseline understanding of ACP. Given that greater knowledge of ACP is associated with an increased willingness to initiate ACP discussions (Ng et al. Reference Ng, Kuah and Loo2017), this omission may have led us to overlook factors influencing their engagement and responses. Finally, surrogate respondents may not accurately reflect the true decision-making self-efficacy and ACP readiness of patients. We recognize that patient-reported outcome measures enhance identification of unmet need, enabling healthcare professionals to address patients’ concerns more effectively, thereby reducing psychological distress and improving quality of life (Etkind et al. Reference Etkind, Daveson and Kwok2015). However, in real-world settings where cognitive impairment is prevalent, incorporating surrogate perspectives is essential to help healthcare professionals navigate the complexities of shared decision-making in ACP and respond appropriately. To address these challenges, we designed our study protocol based on the concept of supported decision-making (Davidson et al. Reference Davidson, Kelly and Macdonald2015). This approach aligns with the clinical recommendations outlined in the National Guidebook for Dementia Palliative Care in Taiwan, developed by an expert panel (Chen Reference Chen2024). To gain a more comprehensive understanding of the interplay between frailty, self-efficacy, and ACP during the transition, longitudinal follow-up studies and further analyses involving various cultural and geographical backgrounds are required.
Conclusions
Frailty was linked to declining decisional self-efficacy yet greater readiness for ACP. This divergence indicates that growing vulnerability may increase motivation to plan ahead without ensuring formal documentation. Supporting meaningful patient engagement beyond document completion should be a priority in caring for frail older adults.
Data availability statement
The datasets generated and/or analyzed during the current study are not publicly available due to concerns regarding patient privacy and confidentiality but are available from the corresponding author on reasonable request.
Author contributions
C.H.H., C-P.L., K.Y.-C.C., and P.-J.C. originally conceived and designed the study. C.H.H., C.-P.L., K.Y.-C.C., J.-Y.L., W.-Z.T., and P.-J.C. acquired the data and screened records. C.H.H., C.-P.L., K.Y.-C.C., J.-Y.L., W.-Z.T., and P.-J.C. extracted, analyzed, and interpreted the data and were major contributors in writing the manuscript. All authors read and approved the final manuscript.
Funding
This work was supported by the National Health Research Institutes of Taiwan (PH-112-GP-06), National Science and Technology Council of Taiwan (112-2410-H-037-005-MY3), E-Da hospital (EDAHP114016 and EDAHP114019), and a grant from the Kaohsiung Medical University Hospital (KMUH114-4R83). The funders had no role in study design, methods, data collection, analysis, and preparation of the manuscript.
Competing interests
The authors declare that they have no competing interests.
Ethical approval
The study has received approval from the Research Ethics Committee of National Health Research Institutes in Taiwan (EC1080203, EC1080203-R1) and has been registered on ClinicalTrials.gov (NCT04250103).
Open access
