Did frailty influence older adults’ choices of communication methods during COVID-19? Authors: Jinhan Wang, Bing Niu, Lingling Zhang Published: 2026 ABSTRACT AbstractBackgroundFrailty affects not only the health outcomes of older adults but also their ability to maintain social connections, especially during public health crises such as the coronavirus disease 2019 (COVID-19) pandemic. This study examined whether frailty status influenced older adults’ choice of communication methods during the COVID-19 outbreak.MethodsUsing a cross-sectional dataset, we applied multivariate probit models to analyze the relationship between frailty status and the use of video calls, telephone calls, and face-to-face interactions among older adults. The sample comprised 5,115 community-dwelling older adults living in Sakai City (mean age: 74.66 years). In this study, older adults were defined as individuals aged 65 years and older. The participants’ characteristics were assessed on November 1, 2020. Participants were categorized as frail, pre-frail, or robust based on frailty-related characteristics such as musculoskeletal function, nutritional status, oral function, confinement, and cognitive function. Frailty was assessed using a frailty index based on a 25-item checklist developed from the existing literature.ResultsFirst, older adults with frailty were more likely to report a decrease in their use of telephone calls (0.296, p<.01) compared with their robust counterparts. Second, compared with those aged 80 years and above, younger respondents were more likely to report an increase in video calls (65–69 years: 0.393, p<.01) but less likely to report an increase in face-to-face interactions (65–69 years: -0.498, p<.01; 70–74 years: -0.302, p<.05). Third, a decrease in the report of using telephone calls was positively correlated with a reduction in the report of face-to-face interactions (p<.01).ConclusionsThese findings highlight the complex relationship between frailty and communication patterns. They underscore the need for inclusive interventions that address physical, age, and technological barriers to social connectivity.Keywords: Frailty, Communication methods, Older adults, COVID-19, Multivariate probit Background Frailty recently affected millions of older adults worldwide [1]. In high-income countries, an estimated 10.7% of older adults are frail, and an average of 41.6% are pre-frail. The prevalence of frailty increases significantly with age, rising from 15.7% in individuals aged 80–84 to 26.1% in those over 85 years [2]. In the United Kingdom, the National Health Service reports that 1.8 million people over 60 years live with frailty, with diagnoses predominantly occurring in those older than 85 [3]. An empirical analysis conducted in Japan using data from 2,206 older adults aged 65 and above found that 8.7% were frail, 40.8% pre-frail, and 50.5% robust [4]. Moreover, 29.4% of the Japanese population is aged 65 years and above [5], suggesting that approximately one in four individuals is an older adult. Therefore, even with the reported prevalence (8.7%), the absolute number of older adults affected by frailty (approximately 3.2 million individuals) remains substantial, indicating that frailty continues to be a significant public health issue in Japan. Frailty is characterized by a decline in the functioning of multiple physiological systems, accompanied by increased vulnerability to stressors [1, 6, 7]. It is associated with various adverse health outcomes, including increased risks of falls, mobility decline, hospitalization, admission to long-term care, and mortality, which would increase dependence on others [6]. Individuals living with frailty also face personal burdens, including reduced quality of life and increased loneliness [1, 8, 9]. Moreover, frailty has demonstrated clear patterns of increased healthcare utilization and costs, particularly in cases of higher severity or advanced frailty [10, 11]. Therefore, identifying older individuals who are frail or at risk of becoming frail has become a cornerstone of geriatric care; it is crucial for preventing, reducing, and delaying adverse health outcomes and the associated burden in older adults and their caregivers [12, 13]. The coronavirus disease 2019 (COVID-19) pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), emerged as a major global public health crisis in December 2019, prompting many countries to implement nonpharmacological public health strategies, including isolation, quarantine, social distancing, and community containment [14, 15]. While these preventive strategies helped reduce exposure risk, they also disrupted routine activities and altered the daily behavioral patterns of the public, particularly affecting vulnerable populations such as adults aged 65 and older, especially those living with frailty [16, 17]. Among older adults, social isolation and loneliness have been linked with numerous adverse outcomes, including increased risks of depression, social withdrawal, functional decline, and even premature mortality [18–20]. These risks were exacerbated during the pandemic, as quarantine and social distancing measures limited opportunities for interpersonal interaction. Quarantine and social isolation have also been associated with negative consequences for both the physical and mental health of older adults, including elevated risks of cardiovascular, neurocognitive, autoimmune, and depressive conditions and anxiety [21, 22]. In addition, evidence from Japan indicates that physical activity time among older adults significantly decreased after the COVID-19 outbreak, which may lead to a higher incidence of disability in the near future [23]. During the pandemic, primary care systems faced substantial barriers to the provision of preventive care, chronic disease management, and timely responses to common acute conditions [24]. Despite these challenges, the use of information and communication technologies (ICT) positively influences social support, connectedness, and social isolation, particularly among older adults [25, 26]. Older adults who are frail and pre-frail are considered a target group that benefits from ICT use [27]. Previous studies have primarily focused on the use of telephone and video calls among older adults during the pandemic [28]. However, few have examined whether different health conditions, such as frailty, influence their choice of communication tools. Given that self-restraint measures during the pandemic did not entirely prohibit face-to-face interaction, in-person communication as a relevant mode of interaction is also a topic of interest. This study contributes to the literature by examining whether frailty status was associated with the choices of communication methods among older adults living with frailty during the COVID-19 pandemic, addressing a gap in research on the interplay between physiologic vulnerability and social connectivity. By using a large sample size dataset and employing a multivariate probit model, we offer a more comprehensive analytical estimation framework that controls the potential correlations among different communication methods: video calls, telephone calls, and face-to-face interactions. Methods Data This study utilized a cross-sectional dataset from Sakai City, Osaka, collected during the COVID-19 pandemic in 2020, a period characterized by repeated self-restraint measures and restrictions on social activities in Japan [29]. Participant characteristics were assessed on November 1, 2020. A random sample of 9,400 community-dwelling residents aged 65 years and older living in Sakai City was drawn, comprising individuals who had not been certified by long-term care and those who had been certified with the mildest levels of needing support (requiring support 1–2). This secondary dataset can be accessed by applying for information disclosure with the Longevity Support Division of the Longevity Society Department, Health and Welfare Bureau of Sakai City, Osaka Prefecture, Japan. This dataset is representative, as the distributions of population, age, sex, and socioeconomic status closely reflect those of the national population. Initially, 6,982 responses were obtained. Of these, 11 self-respondents with a diagnosis of Alzheimer’s disease and 1,856 participants who did not complete the frailty checklist were excluded. The final sample comprised 5,115 respondents with complete frailty data. Dependent variables The dependent variables in this study were the use of video calls (face visible), telephone calls (audio only), and in-person face-to-face interactions. Appendix Table 4 provides the survey questions. Based on respondents’ answers, we determined whether the use of each communication method increased, decreased, or remained unchanged by calculating the difference in usage frequency before and during the COVID-19 pandemic. Specifically, the frequency of use before the pandemic was subtracted from that during the pandemic. Since higher-frequency responses were coded with smaller numerical values in the questionnaire, a negative value indicated an increase in the use of a given communication method, while a positive value indicated a decrease. For example, if a respondent reported using video calls “several times a year” before COVID-19 (coded as “5”) and “2–3 times a week” during the pandemic (coded as “2”), the change was calculated as 2 − 5 = − 3; a negative value indicated increased usage of video calls. Conversely, if the frequency changed from “once a week” (coded as “3”) to “several times a year” (coded as “5”), the change was calculated as 5 − 3 = + 2; a positive value indicated decreased usage of video calls. Based on the above calculations, we created six binary (dummy) variables identifying changes in the usage of communication methods during the COVID-19 pandemic: increased usage of video calls, increased usage of telephone calls, increased usage of face-to-face interactions, decreased usage of video calls, decreased usage of telephone calls, and decreased usage of face-to-face interactions. These dummy variables were used as the dependent variables. Independent variables The primary independent variable in this study was the frailty status of older adults. Current debates on the definition of frailty center on two main approaches: defining frailty based strictly in terms of physical phenotype or adopting a broader definition encompassing physical, psychological, and social dimensions [2]. To capture this complexity, we adopted a multidimensional definition of frailty encompassing seven domains: instrumental activities of daily living, musculoskeletal function, nutritional status, oral function, confinement, cognitive function, and depressed mood. Frailty was assessed using a frailty index based on a 25-item checklist (see Appendix Table 5 for a detailed description of the items and scoring procedure) developed from the existing literature [30–32]. Items included “Do you go out alone by bus or train (or use your car)?” “Do you shop for food and daily necessities by yourself?” and “Do you make deposits and withdrawals yourself?” Each item contributed a specified number of points toward the total frailty score; higher scores reflected greater frailty. The total score ranged from 0 to 25 points. As noted in the frailty checklist (Appendix Table 5), each item was assigned 1 point only if the respondent’s answer matched the option specified in the “Scoring” column. Based on the total score, participants were categorized into three frailty levels: robust (scored 0–3 points), pre-frail (4–7 points), and frail (8 points or more) [31, 32]. In addition to frailty status, the analysis controlled for a set of sociodemographic and socioeconomic characteristics shown in previous studies to be associated with older adults’ communication behaviors and use of communication technologies [28, 33]. These included sex (female, male), age group (65–69, 70–74, 75–79, and ≥ 80 years), years of education, self-assessed economic status (experiencing poverty, average, wealthy), living arrangement (living alone or not), and marital status, which are commonly considered key determinants of communication preferences and access among older populations. Model This study analyzed whether frailty and other covariates influenced the choice of communication methods among older adults during the COVID-19 outbreak. Traditional logistic or probit models assume independence across outcome variables and thus cannot account for potential correlations between the use of different communication methods. To address this limitation, we applied a multivariate probit model, allowing for the simultaneous estimation of multiple binary outcomes (video calls, telephone calls, and in-person interactions) while considering their correlations. The multivariate probit model is explained in detail in the Appendix. Statistical analyses were conducted using Stata/SE 17 (StataCorp LLC, College Station, TX, USA). Results Descriptive statistics Table 1 shows that 7.3% of participants reported increased video call usage, 8.2% increased telephone usage, and 2.4% increased face-to-face interactions. Contrarily, 1.1% reported decreased video call usage, 7.8% decreased telephone usage, and 50.2% decreased face-to-face interactions. Table 1Descriptive statistics of dependent, independent, and covariate variablesVariableObsProp.MinMaxDependent Variables Increased usage of video calls39677.26%01 Increased usage of telephone calls45748.22%01 Increased usage of face-to-face interactions42612.39%01 Decreased usage of video calls39671.08%01 Decreased usage of telephone calls45747.80%01 Decreased usage of face-to-face interactions426150.22%01Main Independent Variables Frail511530.22%01 Pre-frail511535.84%01 Robust511533.94%01Covariates Female507954.48%01Age: 65–69 years508822.98%01 70–74 years508830.25%01 75–79 years508825.26%01 80 + years508821.50%01Years of education: Up to 9 years509423.03%01 10–12 years509447.25%01 13 + years509429.07%01 Others50940.65%01Economic status (self-assessed): Experiencing poverty509126.91%01 Average509158.75%01 Wealthy509114.34%01Living alone510319.44%01Married508370.37%01Data source: A Survey on Older Adults of Sakai City [29] Descriptive statistics of dependent, independent, and covariate variables Data source: A Survey on Older Adults of Sakai City [29] Regarding frailty-related characteristics, 30% of participants were classified as frail, 36% as pre-frail, and 34% as robust. Figure 1 presents the normalized distribution of frailty index scores across the sample. The distribution was unimodal and right-skewed, peaking at approximately 3. The number of individuals increased from frailty index scores of 0 to 3 and then gradually declined as scores increased further, indicating that most individuals had low-to-moderate levels of frailty, while a smaller proportion had higher frailty index scores. Normalized distribution of frailty index (FI) scores in this study sample (n = 5115). Data source: A survey on older adults of Sakai City [29] Female respondents constituted 55% of the sample. The mean age of the participants was 74.66 years, 23% were aged 65–69, 30% aged 70–74, 25% aged 75–79, and 22% aged 80 and above. In terms of educational level, 23% received up to 9 years of education, 47% received 10–12 years, 29% received 13 or more years, and 1% fell into the “other” category. Regarding self-assessed economic status, 27% reported experiencing poverty, 59% had an average economic status, and 14% considered themselves wealthy. In terms of household composition and marital status, 19% lived alone, while 70% were married. We further examined the relationship between frailty status and changes in communication methods using chi-square tests. Table 2 presents the test results. Percentages represent the proportion of individuals within each frailty category and each respective change in communication method. These results indicate that changes in communication methods varied by frailty status, and significant differences across frailty groups were observed in the dependent variable of “decreased usage of telephone calls.” Table 2Changes in communication methods by frailty status (chi-square test results)VariableRobustPre-frailFrail p (n = 1290)(n = 1277)(n = 959)Increased usage of video calls7.91%8.46%5.94%0.072Increased usage of telephone calls8.29%9.40%7.30%0.205Increased usage of face-to-face interactions1.78%2.51%1.56%0.232Decreased usage of video calls0.78%0.94%1.77%0.064Decreased usage of telephone calls5.89%6.42%10.43%0.000Decreased usage of face-to-face interactions52.64%51.37%48.91%0.212Data source: A survey on older adults of Sakai City [29]Values are percentages calculated within each frailty category based on non-missing observations. Frailty status was treated as a three-category variable (robust, pre-frail, and frail) for the chi-square tests Changes in communication methods by frailty status (chi-square test results) Data source: A survey on older adults of Sakai City [29] Values are percentages calculated within each frailty category based on non-missing observations. Frailty status was treated as a three-category variable (robust, pre-frail, and frail) for the chi-square tests Estimation results of the multivariate probit model We estimated the coefficients from the multivariate probit model (Appendix Table 6). Figure 2 summarizes the key results. Several key patterns in the communication choices of older adults during the COVID-19 pandemic are observed (coefficients in parentheses). Coefficient estimates from the multivariate probit model of changes in the use of video calls, telephone calls, and face-to-face interactions. Panel A (a)–(b): Effects on video call usage. Panel B (c)–(d): Effects on telephone call usage. Panel C (e)–(f): Effects on face-to-face interaction. Note: Confidence intervals for “other years of education” in (a), (b), (d), and (e) exceed the plotted range and are truncated for clarity. The figures display the corresponding coefficient estimates, standard errors, and significance levels (p-values). *** p<.01, ** p<.05, and * p<.1 First, older adults with frailty were more likely to report a decrease in their use of telephone calls (0.296, p<.01) compared with their robust counterparts. Second, compared with their male counterparts, the female group reported more variability in video call usage, with a higher likelihood of both increasing (0.265, p<.01) and decreasing (0.399, p<.01) usage. In addition, the female group was significantly more likely to report an increase in their use of telephone calls (0.371, p<.01) and a decrease in face-to-face interactions (0.605, p<.01) during the pandemic. Third, age group differences were observed. Compared with those aged 80 years and above, younger respondents were more likely to report an increase in their use of video calls (65–69 years: 0.393, p<.01) but less likely to report an increase in their use of face-to-face interactions (65–69 years: -0.498, p<.01; 70–74 years: -0.302, p<.05). Fourth, several individual characteristics, including years of education, economic status, household composition, and marital status, were associated with different communication patterns. Compared with older adults with less than 10 years of education, those with higher education levels were more likely to report fewer face-to-face interactions. Wealthier adults were less likely to report an increase in their use of telephone calls (-0.229, p<.05). Additionally, older adults living alone were more likely to report a decrease in video calls (0.933, p<.01). Married individuals were more likely to report a decrease in face-to-face interactions during the pandemic (0.142, p<.05). Additionally, based on the results presented in Table 3, significant correlations were observed among the choices of communication methods. The use of video and telephone calls were positively correlated (p<.01), suggesting that their use tended to either increase or decrease simultaneously. Similarly, a decrease in the use of telephone calls and face-to-face interactions was also positively correlated (p<.01), indicating a parallel decline in these interaction types. Table 3Estimated correlation (Rho) matrix and standard errors among the choices of communication methodsIncreased choice among communication methodsCorr.(Std. err.)Increased usage of telephone calls × increased usage of video calls0.341***(0.039)Increased usage of face-to-face interactions × increased usage of video calls0.060(0.077)Increased usage of face-to-face interactions × increased usage of telephone calls0.050(0.079)Likelihood ratio testchi2(3) = 67.404 Prob > chi2 = 0.0000Decreased choice among communication methodsCorr.(Std. err.)Decreased usage of telephone calls × decreased usage of video calls0.158***(0.056)Decreased usage of face-to-face interactions × decreased usage of video calls0.050(0.043)Decreased usage of face-to-face interactions × decreased usage of telephone calls0.217***(0.033)Likelihood ratio testchi2(3) = 49.672 Prob > chi2 = 0.0000***p<.01 Estimated correlation (Rho) matrix and standard errors among the choices of communication methods ***p<.01 Discussion Several key insights emerged in this study regarding older adults’ communication choices during the COVID-19 pandemic. First, older adults with frailty were more likely to report a decrease in their use of telephone calls compared with their robust counterparts. Second, compared with those aged 80 years and above, younger respondents were more likely to report using video calls but less likely to report using face-to-face interactions. Third, a decrease in telephone calls was positively correlated with a reduction in the report of face-to-face interactions. First, compared with their robust counterparts, older adults with frailty did not report an increase in any communication method and were more likely to report reduced telephone use. With respect to telephone calls, consistent with previous studies, frailty was significantly associated with communication technology use, with higher levels of frailty corresponding to lower usage rates [27]. This decline may increase their risk of social isolation, further distancing them from interpersonal interactions and support networks. Previous studies have also found that older adults with frailty have smaller social networks, diminished social functioning, and increased loneliness, which often worsen over time [9]. This pattern suggests that multiple stressors during the pandemic, including mobility restrictions, fear of infection, and cognitive or emotional fatigue associated with communication efforts, may have increased older adults’ social frailty. Consequently, rather than compensating by switching between communication modes, older adults with frailty reported reducing both telephone and face-to-face communication under compounded physical and psychosocial stress. The observed communication decline reflects not only pandemic-related behavioral shifts but also deeper, preexisting social vulnerability patterns. Addressing this issue requires attention to both frailty and social isolation risks. Therefore, targeted interventions, such as improving access to digital communication tools or offering alternative social support, are needed to mitigate isolation among this high-risk group. Second, younger groups (65–69 years) were more likely than those aged 80 years and above to report an increase in their use of video calls during the pandemic. This age gradient is consistent with previous surveys conducted in Japan showing that internet and social networking service (SNS) use declines with increasing age among older adults [34]. Even though substantial age-related differences in technology utilization exist, prior evidence suggests that internet use increased among older adults during the pandemic [34, 35]. Previous studies have also shown that younger older adults more readily adopted technology-based activities, such as digital communication, health-related interactions, and online food acquisition, during this period [36]. This group generally reported higher acceptance of digital devices and a higher likelihood of engaging in video-based communication. Consequently, age-specific digital proficiency must be considered in the design and dissemination of technology solutions involving older populations. In contrast, compared with those aged 80 years and above, younger groups (65–69 and 70–74 years) were less likely to report an increase in face-to-face interactions during the pandemic. This pattern may reflect differences in baseline mobility and activity levels across age groups. Decreased frequency of face-to-face interaction may be associated with a higher frequency of going outside, suggesting that those who were more active before the pandemic experienced larger relative declines in in-person contact. In addition, those aged 80 years and above may have been less able to avoid essential face-to-face interactions, such as family check-ins or support services even during the pandemic. Third, a decline in telephone calls correlated with reduced face-to-face interactions. Studies have shown that using a mobile phone (including both basic mobile phones and smartphones) for sociability purposes is associated with increased face-to-face social interactions [37]. For many older adults, telephone-based communication may serve not as a substitute but as an important stepping stone toward in-person contact. Nonetheless, this study has several limitations. First, it relied on self-reported data, which may be affected by recall or social desirability bias. In addition, a proportion of respondents were excluded due to incomplete responses on the frailty checklist, which may have introduced some degree of selection bias. Second, despite including various covariates in the multivariate probit models, unobserved confounders, such as digital literacy and unmeasured sensory factors (e.g., mild hearing difficulties), may have influenced frailty and communication behaviors. Third, although communication patterns may influence frailty over time, the potential for reverse causality in this study was difficult to be examined since the study was based on a cross-sectional dataset, which limited our ability to assess long-term effects or identify reverse causality. Moreover, frailty is a multidimensional condition that develops gradually and is unlikely to be substantially influenced by short-term changes in communication behaviors. Future research should consider using long-term panel data to explore potential bidirectional relationships between frailty and communication patterns, and examine heterogeneity among frail individuals, including variation by frailty severity. Conclusions This study used a cross-sectional dataset of older adults in Japan to examine whether frailty status and individual characteristics are associated with communication method choices among individuals with frailty during the COVID-19 pandemic. Telephone usage among older adults with frailty was reported to have decreased. Frailty and advanced age were associated with a reduction or shift in communication methods, revealing underlying structural vulnerabilities in social connectivity. These findings underscore the need for inclusive and age-appropriate strategies to support diverse communication needs among older adults, particularly during periods of social disruption such as pandemics. This study sheds light on factors that may influence social connectivity among older adults and provides a valuable reference for the development of an age-friendly society. Health policy interventions should address both the physical and technological barriers faced by older adults to prevent isolation during public health crises. Acknowledgements The authors thank the Longevity Support Division of the Longevity Society Department, Health and Welfare Bureau of Sakai City, Osaka Prefecture, Japan, for providing the dataset. Informed consent statement Since the data are fully anonymized and the original data collection process adhered to ethical standards set by the local government, informed consent was waived under the Ethical Guidelines for Life Science and Medical Research Involving Human Subjects issued by the Ministry of Health, Labor and Welfare (MHLW), Japan (Link: https://services.google.com/fh/files/misc/appi_062022_update_eng.pdf/). Abbreviations Coronavirus disease 2019 Geweke–Hajivassiliou–Keane Information and communication technologies Appendix 1 The multivariate probit model used in this study is as follows [38, 39]: We constructed two separate multivariate probit models corresponding to changes in the use of video calls, telephone calls, and face-to-face interactions. Let denote the choice for individual i regarding communication method j (where j=1 represents video calls, j=2 represents telephone calls, and j=3 represents face-to-face interactions), and let denote the responses of i on all j variables. We assume the existence of an underlying latent variable , which follows this model: , where is the vector of covariates for individual i and communication method j, is the parameter vector, and is the error term. The observed binary indicator is defined as According to the multivariate probit model, the probability that , conditioned on parameters , and a set of covariates , is given by , where is the density of a j-variate normal distribution with mean vector 0 and correlation matrix , is the interval. is an unknown parameter vector and . We denote the p=J(J-1)/2 free parameters of by . The binary indicator takes a value of 1 if individual i reported a change, either an increase or a decrease, in the use of the communication method j and 0 if there was no change. is a vector of explanatory variables for individual i, including frailty status, sex, age, years of education, economic status, living alone, and marital status. is the coefficient associated with the communication method j, and is the error term for individual i and communication method j. The error terms , , and are assumed to be jointly normally distributed: , where is the covariance matrix that captures the correlation between the errors of different communication methods. The model parameters were estimated using simulated maximum likelihood. Given the multivariate nature of the model and the presence of high-dimensional integrals that lack closed-form solutions, we employed the Geweke–Hajivassiliou–Keane (GHK) simulator to approximate the likelihood function, following Cappellari and Jenkins [40]. The simulated log-likelihood function is given by , where denotes the simulated probability of observing the outcome , approximated using the GHK simulator. Table 4Survey Questions Used to Construct Communication-related Dependent VariablesQuestionsCOVID-19 outbreakFrequency of use4 or more times a week2–3 times a weekOnce a week1–3 times a monthSeveral times a yearNeverHow often did you have video calls (face visible) with your family members or friends/acquaintances before the COVID-19 outbreak? How about now?Before (January 2020)123456During (November 2020)123456How often did you have audio-only telephone calls (by telephone, flip phone, smartphone, LINE, etc.) with your family members or friends/acquaintances before the COVID-19 outbreak? How about now?Before (January 2020)123456During (November 2020)123456How often did you meet others (friends/acquaintances) before the COVID-19 outbreak and during the pandemic period?Before (January 2020)123456During (April–May 2020)123456Questions were adapted from A Survey on Older Adults of Sakai City conducted by the Health and Welfare Bureau of Sakai City, Japan [29] Survey Questions Used to Construct Communication-related Dependent Variables Questions were adapted from A Survey on Older Adults of Sakai City conducted by the Health and Welfare Bureau of Sakai City, Japan [29] Table 5Frailty index checklist based on seven domainsQuestionAnswerScoringIADLQ1Do you go out alone by bus or train (or use your car)?(1) Yes, I can and I do. (2) I can, but I don’t. (3) No, I can’t.2 or 3Q2Do you shop for food and daily necessities by yourself?(1) Yes, I can and I do. (2) I can, but I don’t. (3) No, I can’t.2 or 3Q3Do you make deposits and withdrawals yourself?(1) Yes, I can and I do. (2) I can, but I don’t. (3) No, I can’t.2 or 3Q4Do you ever visit a friend’s house?(1) Yes. (2) No.2Q5Do you consult with your family or friends?(1) Yes. (2) No.2Musculoskeletal functionQ6Can you climb stairs without using handrails or walls?(1) Yes, I can and I do. (2) I can, but I don’t. (3) No, I can’t.2 or 3Q7Can you stand up from a chair without holding on to anything?(1) Yes, I can and I do. (2) I can, but I don’t. (3) No, I can’t.2 or 3Q8Can you walk for about 15 min continuously?(1) Yes, I can and I do. (2) I can, but I don’t. (3) No, I can’t.2 or 3Q9Have you fallen in the past year?(1) Many times. (2) Once. (3) No.1 or 2Q10Are you worried about falling?(1) Very (2) A little. (3) Not very much. (4) No.1 or 2Nutritional statusQ11Have you lost more than 2 to 3 kg in weight in the past six months?(1) Yes. (2) No.1Q12BMI< 18.5Oral functionQ13Is it more difficult to eat hard food than it was six months ago?(1) Yes. (2) No.1Q14Have you ever choked on tea, soup, etc.?(1) Yes. (2) No.1Q15Are you worried about your dry mouth?(1) Yes. (2) No.1ConfinementQ16How often do you go out (including visiting local farms or a nearby neighborhood, shopping, going to the doctor)? *Going out at least once a week.(1) 5 or more times a week. (2) 4 times a week. (3) 2 to 3 times a week. (4) Once a week. (5) 1 to 3 times a month. (6) Several times a year. (7) Don’t do that.5 or 6 or 7Q17Do you go out less than you did last year?(1) Much less. (2) Less than. (3) Not too less. (4) No less.1 or 2Cognitive functionQ18Have people around you told you that you are forgetful, such as, “You always ask the same thing”?(1) Yes. (2) No.1Q19Do you look up phone numbers and make calls yourself?(1) Yes. (2) No.2Q20Have you ever had trouble remembering what date it is today?(1) Yes. (2) No.1Depressed moodQ21(Over the past two weeks) There is no sense of fulfillment in your daily life.(1) Yes. (2) No.1Q22(Over the past two weeks) You can no longer enjoy the things you used to enjoy.(1) Yes. (2) No.1Q23(Over the past two weeks) Things that used to be easy for you now feel like a chore.(1) Yes. (2) No.1Q24(Over the past two weeks) You don’t regard yourself as useful.(1) Yes. (2) No.1Q25(Over the past two weeks) You feel tired for no reason.(1) Yes. (2) No.1Each item was assigned 1 point only if the respondent’s answer matched the option specified in the “Scoring” column. The total score ranged from 0 to 25. IADL: instrumental activities of daily living Frailty index checklist based on seven domains Each item was assigned 1 point only if the respondent’s answer matched the option specified in the “Scoring” column. The total score ranged from 0 to 25. IADL: instrumental activities of daily living Table 6Coefficient estimates from the multivariate probit modelIncreased choice among communication methodsDecreased choice among communication methodsIncreased usage of video callsIncreased usage of telephone callsIncreased usage of face-to-face interactionsDecreased usage of video callsDecreased usage of telephone callsDecreased usage of face-to-face interactionsCoef.Coef.Coef.Coef.Coef.Coef. (Std. err.) (Std. err.) (Std. err.) (Std. err.) (Std. err.) (Std. err.) Health status (ref: robust)Frail-0.043-0.120-0.266*0.328*0.296***-0.040(0.090)(0.086)(0.147)(0.177)(0.086)(0.059)Pre-frail0.0750.0570.0590.0830.036-0.010(0.074)(0.072)(0.118)(0.172)(0.080)(0.051)Sex (ref: male)Female0.265***0.371***0.0910.399***0.0080.605***(0.068)(0.066)(0.110)(0.149)(0.068)(0.046)Age (ref: 80 + years)65–69 years0.393***0.007-0.498***0.0220.0420.098(0.112)(0.100)(0.168)(0.199)(0.105)(0.070)70–74 years0.211*-0.084-0.302**-0.149-0.0110.103(0.111)(0.098)(0.147)(0.205)(0.101)(0.068)75–79 years0.145-0.079-0.075-0.0470.0710.039(0.118)(0.103)(0.145)(0.210)(0.104)(0.071)Years of education (ref: up to 9 years)10–12 years-0.037-0.018-0.0110.3760.0910.239***(0.096)(0.087)(0.136)(0.240)(0.091)(0.062)13 + years0.1410.013-0.0390.454*-0.0880.390***(0.102)(0.096)(0.154)(0.253)(0.102)(0.068)Other-3.7590.034-3.3341.389***-4.1070.152(112.305)(0.391)(167.805)(0.469)(143.884)(0.282)Economic status (self-assessed) (ref: average)Experiencing poverty-0.1190.126*0.177-0.0760.056-0.011(0.083)(0.074)(0.120)(0.169)(0.078)(0.054)Wealthy-0.021-0.229**0.0150.0680.1010.087(0.089)(0.096)(0.151)(0.177)(0.091)(0.062)Living alone0.1500.121-0.0900.933***0.1190.064(0.124)(0.112)(0.159)(0.263)(0.121)(0.080)Married0.1750.142-0.344**0.526**0.1290.142**(0.112)(0.101)(0.144)(0.266)(0.109)(0.071)Constant-1.981***-1.659***-1.633***-3.630***-1.752***-0.703***(0.169)(0.151)(0.215)(0.410)(0.157)(0.104)***p < .01, **p < .05, *p < .1 Coefficient estimates from the multivariate probit model ***p < .01, **p < .05, *p < .1 Authors’ contributions JW: writing: reviewing, editing, and original draft preparation; visualization; validation; software; methodology; funding acquisition; formal analysis; and conceptualization. BN: writing: reviewing and editing; supervision; resources; conceptualization, and project administration. LZ: writing: reviewing and editing.All authors read and approved the final manuscript. Funding This work was supported by JST SPRING (grant no. JPMJSP2139). Data availability The data can be accessed by applying for information disclosure with the Longevity Support Division of the Longevity Society Department, Health and Welfare Bureau of Sakai City, Osaka Prefecture. Declarations Ethics approval and consent to participate This study used anonymized secondary data obtained from the Longevity Support Division of the Longevity Society Department, Health and Welfare Bureau of Sakai City, Osaka Prefecture. This study was conducted in accordance with the Declaration of Helsinki (Link: https://www.wma.net/policies-post/wma-declaration-of-helsinki/), particularly regarding the protection of participant confidentiality and the responsible use of research data. In accordance with Japanese regulations, specifically the Act on the Protection of Personal Information (APPI) and the Ethical Guidelines for Life Science and Medical Research Involving Human Subjects issued by the Ministry of Health, Labor and Welfare (MHLW), the secondary use of anonymized administrative data does not require prior approval from an Institutional Review Board (IRB) or ethics committee. Therefore, this study did not require submission to or approval from an ethics committee. Informed consent was not required in accordance with the relevant guidelines and regulations. For further reference, please see Japan’s Act on the Protection of Personal Information (Link: https://services.google.com/fh/files/misc/appi_062022_update_eng.pdf). Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The data can be accessed by applying for information disclosure with the Longevity Support Division of the Longevity Society Department, Health and Welfare Bureau of Sakai City, Osaka Prefecture.