Development and Validation of a Social Frailty Index Among Older Adults in the Philippines

Abstract

Background

Social frailty has not been studied in the Philippines, a Southeast Asian country with distinct sociocultural characteristics.

Objective

To (i) develop and validate the Social Frailty Index-Philippines (SFI-Phil), using all-cause mortality (up to 4 years) as the outcome and (ii) evaluate performance of SFI-Phil across age, sex, and residence.

Methods

Performing regression analyses on baseline and 4-year follow-up data from 5153 older adults aged 60+ from the nationally representative Longitudinal Study of Ageing and Health in the Philippines (LSAHP), we selected and validated a parsimonious model of social predictors.

Results

The resulting 6-item SFI-Phil demonstrated satisfactory accuracy in predicting mortality up to 4 years (C-statistic 0.640, 95% confidence interval [CI]: 0.608–0.673) in a validation sample and across demographic strata, although better amongst 60–69 year olds, men, and urban residents.

Conclusion

SFI-Phil, developed considering Philippines’s sociocultural context, can be used to assess social frailty among older adults in the Philippines.

Keywords

social frailty gerontology mortality index

Introduction

Frailty is generally defined as a state of vulnerability to adverse health outcomes (Fried et al., 2004). While traditionally conceived as solely or primarily physical frailty, it is increasingly recognized as a multidimensional construct that includes cognitive, psychological, and social components (Yu et al., 2023). Of these components, social frailty remains the least studied (Pek et al., 2020; Yu et al., 2023; Zhang et al., 2023).

Bunt et al. proposed an integrative definition of social frailty as “a lack of resources to fulfill one’s basic social needs” (Bunt et al., 2017). According to this conceptualization, social frailty is a multi-dimensional construct that encompasses general resources (e.g., financial status, education), social resources (e.g., social network, living arrangement) and social activities (e.g., organizational membership, engaging in neighborhood activities). Lack of these resources hinders fulfillment of basic social needs (e.g., sense of belonging, warm and trusting relationships), leading to a state of social frailty (Bunt et al., 2017).

Social frailty is increasingly shown to be associated with a spectrum of adverse health outcomes such as disability (Goto et al., 2024; Teo et al., 2017), motoric cognitive risk syndrome (Zhang et al., 2024), institutionalization (Lee et al., 2022), and mortality (Garre-Olmo et al., 2013; Goto et al., 2024; Lee et al., 2022; Ma et al., 2018) independent of physical frailty. These associations highlight the potential utility of incorporating social frailty into routine geriatric assessment and care. However, a major challenge limiting the application of social frailty in clinical settings is the lack of global consensus on its assessment. Some assessment tools do already exist for social frailty, and have been used across both Western and Eastern contexts (Gobbens et al., 2010; Makizako et al., 2015; Pek et al., 2020; Teo et al., 2017; Yamada & Arai, 2015). A recent systematic review and meta-analysis demonstrated that depending on country, setting, and assessment tool used, the prevalence of social frailty differs substantially, ranging from 3.6% to 66.5% (Zhang et al., 2023). One of the main factors contributing to this wide range of estimates is the difference in the overall focus of measures of social frailty as well as the components included in the measures. Unlike physical frailty or cognitive frailty, which have well defined biological underpinnings, what is considered “socially frail” likely differs across countries depending on social and economic contexts such as insurance and welfare policies, lifestyle, and general perceptions of aging (Yu et al., 2023).

Several studies in Singapore, India, and the U.S. have created and validated social frailty indices based on the four domains of social frailty outlined in Bunt’s conceptual framework: general resources, social resources, social activities, and social need fulfillment (Bunt et al., 2017). A Social Frailty Scale (Pek et al., 2020) created from a study of older adults in Singapore included social predictors from three out of the four social need domains (i.e., social resources, social activities, and social need fulfillment), while another Social Frailty Index (Irshad et al., 2023) derived from a study of Indian older adults included predictors from only two domains (i.e., social resources and social activities). Using 4-year mortality as an outcome, Shah et al. developed a 10-item Social Frailty Index based on older adults in the United States that included social predictors from all four social need domains (Shah et al., 2023). These studies demonstrate that while Bunt’s four social need domains serve as a useful conceptual framework for defining social frailty, the underlying meaning of each domain in each context and the relative importance of the domains, as well as specific social predictors measured within them, differs across cultural contexts.

Social frailty has not been studied in the Philippines, a developing Southeast Asian country with distinct sociocultural characteristics and an aging population. Religion plays a central role in the social life of older adults in the Philippines (Badana & Andel, 2018; Esteban, 2015), with most older adults identifying as Roman Catholic (Cruz et al., 2019). Caring for family members is strongly valued, such that many older adults are involved in the partial or full care of grandchildren even in advanced age (Cruz et al., 2019). Moreover, considering the country’s’ limited social security pension, older adults in the Philippines tend to remain employed longer in order to financially support themselves and their other family members (Badana & Andel, 2018). These relatively unique social factors suggest that what it means to be “socially frail” in the Philippines could differ from other countries. Therefore, in order to maximize the potential of social frailty for risk assessment among older persons in the Philippines, it is important to develop a scale that considers the local sociocultural context.

The primary aim of this study was to develop and validate a Social Frailty Index based on the older adult population in the Philippines (SFI-Phil), using all-cause mortality up to 4-years as the outcome. Prior studies have shown that social frailty status differs across age (Irshad et al., 2023; Misu et al., 2023), sex (Irshad et al., 2023; Takagi et al., 2020), and urban–rural residence (Park et al., 2019). Considering the diverse pool of older adults in the Philippines, it is important to ensure that the SFI-Phil performs adequately across key demographic subpopulations. Therefore, we also evaluated its performance in subpopulations defined by age, by sex, and by urban-rural residence as determined by population density and available facilities (Philippine Statistics Authority, 2017).

Methods

Participants

We utilized data from two waves of the Longitudinal Study of Ageing and Health in the Philippines (LSAHP), the first nationally representative longitudinal survey of older adults to be conducted in the country. The survey is designed to assess health status and social and economic well-being of older adults in the Philippines (Cruz et al., 2019). Participants are community-dwelling older adults, aged 60 years and over at baseline, recruited from 11 provinces and 167 barangays (smallest geopolitical unit in the Philippines), with a total of 5985 individual respondents in the baseline interview. Data collection for the baseline wave (response rate: 94.0%) took place from October 2018 to February 2019, while 4-year follow-up wave (93.4% of those interviewed at baseline) was conducted from January 2023 to April 2023. Details about the design of the LSAHP and its data collection have been documented elsewhere (Cruz et al., 2019).

For this analysis, a total of 832 participants were excluded (776 who failed a cognitive test administered at the beginning of interview in the baseline wave and required a proxy respondent, as such respondents were not asked questions informing many of the candidate social predictors; 49 with unknown vital status in the follow-up wave as they lacked data on the outcome of interest; and 7 who passed away within a month following the baseline wave as the minimal unit of time required for survival analyses was ≥1 month), leaving a final analysis sample of 5153 participants.

The LSAHP was conducted after approval from the University of the Philippines Manila Research Ethics Board (UPMREB Code 2018-363-01). Informed consent was obtained from all participants prior to their inclusion in the study. Approval for secondary data analysis of de-identified data from the LSAHP was taken from the National University of Singapore Institutional Review Board (LS-19-295E).

Candidate Social Predictors

We identified 89 potential social predictors from the LSAHP baseline questionnaire, including all individual items from full scales. However, some individual items in the scales did not fit well into the Social Frailty in Older Adults framework (Bunt et al., 2017) based on independent review by study investigators. Discrepancies were thereafter resolved via consensus discussion, finally yielding 33 candidate social predictors.

The social predictors included both objective and subjective measures (Goto et al., 2024) and spanned four broad social need domains as suggested by the Social Frailty in Older Adults framework (Bunt et al., 2017): General Resources, Social Resources, Social Activities, and Fulfillment of Basic Social Needs.

(A) General Resources (11 predictors): Wealth index score (Marquez et al., 2022), presence of liabilities (yes/no), health insurance (yes/no), house ownership (yes/no), lived in a city/población/rural area during childhood (yes/no), childhood financial status (poor/average – well-off), childhood health status (unhealthy/healthy), education level (elementary school and below/high school+) (Cruz et al., 2016; Cruz et al., 2016; Domingo et al., 1996), and has experienced hunger in past 3 months (yes/no).

(B) Social Resources (6 predictors): Marital status (not currently married/currently married), number of siblings, living arrangement (living alone/not living alone), has children (yes/no), has grandchildren (yes/no), and Lubben Social Network Scale-R (LSNS-R) score (Lubben et al., 2006). The LSNS-R score was derived by adding the aggregate score of twelve items that assess social engagement among older adults, reflecting perceived social support from family and friends (Table S1).

(C) Social Activities (6 predictors): Organizational membership (yes/no), participates in social activity (never-seldom/frequent), hangs out with friends and neighbors (never-seldom/frequent), takes care of grandchildren (yes/no), currently employed (yes/no), and participates in religious activity (yes/no).

(D) Fulfillment of Basic Social Needs (10 predictors): Feels left out (often-always/never-seldom), lacks companionship (often-always/never-seldom), feels isolated (often-always/never-seldom), satisfied with level of contact with relatives/friends (yes/no), family, has friends or relatives willing to listen (yes/no), has many people come to you for advice (often-always/never-seldom), feels needed by other people (often-always/never-seldom), has a good influence on the lives of other people (often-always/never-seldom), and has made unique contributions to society (often-always/never-seldom).

Detailed coding strategies are documented in Table S1.

Outcome

The primary outcome was vital status at the follow-up wave. For each participant, we obtained their vital status approximately 4 years after the baseline wave (death: yes/no) and how long they had survived (time: months) since the baseline wave. Vital status was verified through an official death certificate (56%), and if unavailable, through verbal confirmation from a member of the deceased participant’s household.

Statistical Methods

Participants were randomly split into a training set (N = 3675) and a validation set (N = 1478) in an optimized 7:3 ratio (Vrigazova, 2021) using the set.seed () function in R Studio. Considering the number and scope of candidate variables, we derived the prediction model from the training set using a two-step procedure to minimize possible collinearity and improve parsimony (Shah et al., 2023). First, we adopted the least absolute shrinkage and selection operator (LASSO) Cox regression method to reduce data dimension and avoid overfitting (R package: glmnet). LASSO Cox regression utilizes an L1 penalty function to filter out predictors that do not contribute substantially to the model (coefficient = 0). Second, Cox multiple regression analysis included variables with coefficients >0 and additional non-significant predictors were removed. Lastly, we obtained the hazard ratio (HR) and 95% confidence intervals of association with variables selected via the two-step method to assess the independent predictors of mortality up to 4-years.

Discrimination (Harrell’s C-statistic) and calibration (Greenwood Nam-D’ Agostino χ² goodness-of-fit test) of the model derived from the training set was evaluated in the validation set. C-statistic values ≥0.6 were considered satisfactory (Mitchell et al., 2014; Zhang et al., 2022) and non-significant p-values ≥0.05 for the goodness-of-fit test indicated acceptable calibration (Demler et al., 2015).

The overall analysis sample was subsequently stratified by age group (60–69/ 70–79/80+), sex (men/women) and residence (urban/rural). Model discrimination (Harrell’s C-statistic) and calibration (Greenwood Nam-D’ Agostino χ² goodness-of-fit test) were then assessed by age, sex, and rural/urban residence.

Finally, to enhance the practical utility of the model as a risk assessment tool, a predictive nomogram was developed from the overall analysis sample based on the β-coefficients of predictors in the validated model. The predictor with the highest β-coefficient is assigned the maximal score of 100 points in the nomogram (R package: nomogramFormula). Based on their total risk scores on the nomogram, participants were categorized into three risk terciles: low-risk (33.3%), intermediate-risk (33.3%–66.7%), and high-risk groups (66.7%–99.9%). Both unadjusted and adjusted (for age, sex, and physical health status (basic activities of daily living [bADL] fully independent/not fully independent)) hazard ratios for mortality were calculated to ascertain whether significant differences in 4-year mortality rates existed between the three risk groups.

All analyses were conducted using SPSS Version 20 (IBM Corp, Armonk, N.Y., USA) and R Studio Version 4.3.1 (RStudio, Boston, MA, USA). The threshold for statistical significance was set at p < .05 (two-sided).

Results

Baseline Characteristics

Participants in the analysis sample had a median age of 71 years, and 63.5% were women. More than half (55.9%) were rural residents. Within 4 years of the baseline wave, 20.4% (751/3675) of study participants in the training set and 21.2% (314/1478) of participants in the validation set had died. Participants in the training data set were similar to those in the validation set in terms of age group, sex, and place of residence as well as the 33 social predictors (p > .05 across demographic and social predictors) (Table 1).

Table 1.

Participant Characteristics in the Overall Analysis Sample, and in the Training and Validation Sets.

Characteristics	Overall (%)(N = 5153)	Training set (%)(N = 3675)	Validation set (%)(N = 1478)	p-value^a
Demographics
Age group (years)				0.74
60–69	2061 (40.0)	1467 (39.9)	594 (40.2)
70–79	2103 (40.8)	1493 (40.6)	610 (41.3)
80+	989 (19.2)	715 (19.5)	274 (18.5)
Sex				0.11
Female	3271 (63.5)	2307 (62.8)	964 (65.2)
Male	1882 (36.5)	1368 (37.2)	514 (34.8)
Residence				0.65
Rural	2883 (55.9)	2064 (56.2)	819 (55.4)
Urban	2270 (44.1)	1611 (43.8)	659 (44.6)
Candidate social predictors
General resources
Wealth index score (mean [SD])^b	0.03 (1.0)	0.0 (1.0)	0.0 (1.0)	0.46
Presence of liabilities (yes)	967 (18.8)	679 (18.5)	288 (19.5)	0.42
Health insurance (no)	1192 (23.1)	837 (22.8)	355 (24.0)	0.84
House ownership (yes)	3781 (73.4)	2708 (73.7)	1073 (72.6)	0.44
Lived in a city during childhood	498 (9.7)	369 (10.0)	129 (8.7)	0.16
Lived in a población during childhood	814 (15.8)	568 (15.5)	246 (16.6)	0.31
Lived in a rural area during childhood	3841 (74.5)	2378 (74.5)	1103 (74.6)	0.95
Childhood financial status (poor)	2311 (44.8)	1621 (44.1)	690 (46.7)	0.10
Childhood health status (unhealthy)	59 (1.1)	39 (1.1)	20 (1.4)	0.46
Education level (elementary and below)	3368 (65.4)	2396 (65.2)	972 (65.8)	0.72
Experienced hunger in the past 3 months (yes)	633 (12.3)	437 (11.9)	196 (13.3)	0.19
Social resources
Marital status (currently married)	1994 (38.7)	1435 (39.0)	559 (37.8)	0.43
Number of siblings (mean [SD])	7.6 (3.0)	7.5 (3.0)	7.7 (3.0)	0.05
Living arrangement (living alone)	664 (12.9)	472 (12.8)	192 (13.0)	0.92
Has children (yes)	4893 (94.9)	3486 (94.9)	1406 (95.1)	0.74
Has grandchildren (yes)	4693 (91.1)	3343 (91.0)	1350 (91.3)	0.71
LSNS-R^c (mean [SD])	24.9 (9.7)	24.9 (9.7)	24.8 (9.7)	0.67
Social activities
Organizational membership (yes)	1573 (30.5)	1122 (30.5)	451 (30.5)	1.00
Participates in social activity (frequent)	349 (6.8)	249 (6.8)	100 (6.8)	1.00
Hangs out with friends and neighbors (frequent)	2977 (57.8)	2129 (57.9)	848 (57.4)	0.74
Takes care of grandchildren (yes)	1184 (23.0)	1307 (22.6)	352 (23.8)	0.35
Currently employed (yes)	1837 (35.6)	1307 (35.6)	530 (35.9)	0.87
Participates in religious activity^d (yes)	4211 (81.7)	2987 (81.3)	1224 (82.8)	0.21
Fulfillment of basic social needs
Feels left out^e (often-always)	1080 (21.0)	761 (20.7)	319 (21.6)	0.51
Lacks companionship^e (often-always)	1354 (26.3)	969 (26.4)	385 (26.0)	0.84
Feels isolated^e (often-always)	991 (19.2)	690 (18.8)	301 (20.4)	0.20
Satisfied with level of contact with relatives (yes)	4602 (89.3)	3293 (89.6)	1309 (88.6)	0.30
Satisfied with level of contact with friends (yes)	4663 (90.5)	3325 (90.5)	1338 (90.5)	1.00
Family, friends or relatives are willing to listen (yes)	4360 (84.6)	3117 (84.8)	1243 (84.1)	0.55
Many people come to you for advice (often-always)^f	838 (16.3)	602 (16.4)	236 (16.0)	0.75
Feels needed by other people (often-always)^f	1056 (20.5)	766 (20.8)	290 (19.6)	0.35
Has a good influence on lives of other people (often-always)^f	1023 (19.9)	737 (20.1)	286 (19.4)	0.59
Has made unique contributions to society (often- always)^f	724 (14.1)	532 (14.5)	192 (13.0)	0.18

^aThe p-value refers to statistical differences between the training and validation sets.

^bThe Wealth Index Score is a composite measure of the cumulative living standard of the household where the older adult resides. It is computed using the household’s ownership of appliances and vehicles, and the characteristics of the dwelling where the older person lives such as type of water source, sanitation facilitates, and materials used for housing construction.

^cLubben Social Network Scale-Revised (Cronbach’s alpha = 0.70) (Lubben et al., 2006).

^dRefers to participation in at least one of the following activities: attending religious services outside your home, attending religious activities outside the home, perform religious activities at home with other family members.

^eItems from the Three-Item Loneliness Scale (Cronbach’s alpha = 0.72) (Hughes et al., 2004).

^fItems from the abbreviated Loyola Generativity Scale (Cronbach’s alpha = 0.78) (Casado et al., 2024; Grossman & Gruenewald, 2017).

SFI-Phil Development and Validation

LASSO Cox Regression analysis yielded 12 out of 33 social predictors with non-zero coefficients based on optimal lambda value. Cox multiple regression analysis yielded a six-predictor model consisting of social predictors from three out of four domains in the Social Frailty in Older Adults Framework (Table S2). There was only a marginal decrease in Harrell’s C-statistic from the 12-predictor model (0.668, 95% CI: 0.650–0.686) to the 6-predictor model (0.660, 95% CI: 0.644–0.677). The six predictors included: education level, participation in religious activity, frequency of hanging out with friends and neighbors, current employment status, taking care of grandchildren, and frequency of feeling left out. The strongest predictor of 4-year mortality was non-participation in religious activity (univariate HR = 2.28, 95% CI = 2.00–2.60) (Table 2). As sensitivity analyses, we generated three different training-validation sets from the overall sample, which demonstrated that the six predictors were consistently selected for in initial LASSO Cox regression analysis (Table S3). In addition, we ran weighted LASSO Cox regression analysis using nationally representative individual-level weights and found that results remained largely unchanged (Table S4).

Table 2.

Social Predictors in SFI-Phil (Cox Regression Analysis).

Social predictor	Categories	Unadjusted HR^a (95% CI)	Adjusted HR^a (95% CI)
Education level	Elementary school and below	1.37 (1.20–1.57)**	1.29 (1.13–1.48)**
Education level	High school +	Ref	Ref
Participates in religious activity	Does not participate	2.28 (2.00–2.60)**	1.86 (1.63–2.12)**
Participates in religious activity	Participates	Ref	Ref
Currently employed	No	2.08 (1.80–2.41)**	1.74 (1.50–2.01)**
Currently employed	Yes	Ref	Ref
Takes care of grandchildren	No	1.89 (1.59–2.24)**	1.67 (1.40–1.98)**
Takes care of grandchildren	Yes	Ref	Ref
Hangs out with friends and neighbors	Never-seldom	1.38 (1.22–1.55)**	1.20 (1.07–1.36)*
Hangs out with friends and neighbors	Frequent	Ref	Ref
Feels left out	Often-always	1.34 (1.17–1.54)**	1.21 (1.05–1.39)*
Feels left out	Never-seldom	Ref	Ref

*p < .05; **p < .001.

^aHazard Ratio. Adjusted HR is from models that included the listed social predictors as well as age, sex, and physical health status (b-ADL status).

To validate the final six-predictor model derived in the training dataset, its discrimination and calibration was evaluated in the validation set. Discrimination remained acceptable (C-statistic 0.640, 95% CI: 0.608–0.673) and the model (henceforth referred to as Social Frailty Index-Philippines, SFI-Phil) demonstrated good calibration with Greenwood Nam-D’ Agostino χ² goodness-of-fit test (χ² = 5.83, p = .76) (see Figure S1 for a visual representation).

SFI Performance by Age, Sex, and Urban–Rural Residence

Discrimination and calibration of SFI-Phil was evaluated in the overall sample and across age, sex, and urban–rural residence strata. Discrimination (C-statistic 0.651, 95% CI: 0.636–0.637) and calibration (χ² = 4.13, p = .90) of SFI-Phil in the overall sample was acceptable.

Based on point estimates of the C-statistic, although there was an overlap in the 95% CIs, the SFI-Phil had lower discriminative ability among participants aged 70–79 years (C-statistic 0.623, 95% CI: 0.602–0.644) and 80+ years (0.594, 95% CI: 0.568–0.624) compared to participants aged 60–69 years (0.642, 95% CI: 0.606–0.673). It also had lower discriminative ability amongst women (C-statistic 0.649, 95% CI: 0.623–0.672) compared to men (0.670, 95% CI: 0.650–0.694) and rural residents (0.649, 95% CI: 0.628–0.666) compared to urban residents (0.660, 95% CI: 0.637–0.684). Calibration was acceptable across all strata (p > .05) (Table S5).

A predictive nomogram generated from SFI-Phil demonstrated good discriminative ability in the overall sample (Figure 1). Compared to participants in the low-risk group, participants in the intermediate-risk (HR = 1.77, 95% CI: 1.49–2.10) and high-risk groups (HR = 3.19, 95% CI: 2.73–3.73) had significantly higher 4-year mortality hazard rates (Figure 2). Adjusting for age, sex, and physical health status (bADL) did not change our conclusions (intermediate-risk group: HR = 1.42, 95% CI: 1.19–1.70; high-risk group: HR = 2.03, 95% CI: 1.72–2.40). Figure S2 illustrates a flowchart depicting all analytic steps.

Discussion

Using data from the first longitudinal survey of older adults in the Philippines and guided by the Social Frailty in Older Adults framework (Bunt et al., 2017), we developed and validated a local Social Frailty Index (SFI-Phil) using mortality up to 4-years as the outcome. SFI-Phil performed adequately in the overall sample, as well as across demographic subpopulations.

Compared to SFIs created in other countries, SFI-Phil was different in several ways. First, SFI-Phil included predictors from only three of the four previously specified social need domains, with no predictors under the “social resources” domain. This contrasts with SFIs created from studies based in United States (Shah et al., 2023), India (Irshad et al., 2023) and Singapore (Pek et al., 2020), which included predictors related to social resources. Considering that most older adults in the Philippines report high levels of satisfaction with their social networks including both family and friends (Cruz et al., 2019), predictors related to social resources may be less predictive of mortality outcomes. On the other hand, most predictors in SFI-Phil fall under the “social activities” domain, suggesting that despite having a large social network, a lack of meaningful engagement with family members and with the wider community could be central to what it means to be “socially frail” in the Philippines context.

Second, lack of participation in religious activity emerged as the strongest predictor of mortality up to 4 years, highlighting the central role of religion in the social life of older adults in our sample. Within the Philippines, partaking in religious activities confers at least two psychosocial benefits. Psychologically, having religious beliefs may provide resilience against potential hardships that accompany the aging process (Badana & Andel, 2018). Socially, taking part in religious activities facilitates a sense of belonging among co-religionists and is perceived as equally if not more important than the religious belief itself (Esteban, 2015). Overall, these findings suggest that older adults in the Philippines who do not partake in or withdraw from religious activities are more likely to be socially frail. As such, additional measures should be taken to identify and provide support to at-risk older adults via engagements with both family members and faith communities.

Third, our study evaluated SFI performance across demographic subpopulations. We found that while SFI-Phil had acceptable discrimination across demographic subpopulations, it performs better with older adults aged 60–69 years (compared to those 70–79 and 80+ years), men (compared to women), and urban residents (compared to rural residents). Prior studies outside of the Philippines have demonstrated that social frailty often precedes physical frailty in young-old adults (Misu et al., 2023), supported by findings suggesting loneliness tends to peak between ages 50 and 60 (Hawkley et al., 2022; Luhmann & Hawkley, 2016). In addition, the experience of loneliness, a correlate of social frailty (Li et al., 2024), has been found to differ between sexes (Takagi et al., 2020, 2022). Women tend to experience loneliness despite the presence of a robust social network, suggesting a relatively greater emphasis on the fulfillment of psychosocial needs (Takagi et al., 2020). Conversely, men tend to rely primarily on social relationships to cope with loneliness (Takagi et al., 2020). Finally, compared to urban-dwelling older adults, rural-dwelling older adults in the Philippines have a higher proportion of self-reported loneliness, possibly due to smaller social networks (Takagi et al., 2022). These findings highlight the need for further studies elucidating differences in social vulnerabilities between key demographic subpopulations within the Philippines.

Fourth, SFI-Phil has been operationalized with and evaluated using a predictive nomogram. Considering the potential of incorporating social frailty measures into clinical settings (Shah et al., 2023), the nomogram is an example of a simple yet effective tool that can be readily used to complement existing clinical risk assessments, such as a screening tool in primary and community healthcare settings to enhance existing social services.

We acknowledge several limitations of this study. Our study sample may not be entirely representative of the older adult population in the Philippines. Due to the self-report nature of many items in the baseline survey, we had to exclude older adults for whom a proxy respondent was interviewed in the baseline wave. Future studies can add to our findings by assessing solely objective measures of social frailty amongst older adults who require additional assistance. In addition, since the aim of our current study was to create a parsimonious index intended for practical use, we did not distinguish between the intergenerational caregiving subgroups in our study (e.g., between caretakers of grandchildren from biological vs. extended kin; between caretakers with no children but taking care of grandchildren and vice versa) even though these subgroups may be qualitatively and meaningfully different. Future studies can build on our study by defining these subgroups and assessing their effect on social engagement estimates and predictive validity. Finally, we only used a single outcome, namely mortality up to 4-years, to validate SFI-Phil. We did so as mortality is a clearly defined objective outcome, associated with a specific date of death, thus permitting time-to-event analyses. Future studies can support our findings by assessing rates of morbidities such as disability, institutionalization, and functional decline among high-risk adults as identified by SFI-Phil.

Conclusions and Implications

The Social Frailty Index (SFI-Phil), developed and validated considering the local sociocultural context, can be used to assess social frailty among older adults in the Philippines. SFI-Phil performed best among 60–69-year-olds, males, and urban residents. Our study is the first to examine social frailty in the Philippines and can be used as both a theoretical foundation to inform policy and future research, and the resulting SFI-Phil can be used as a practical tool for population health survey and to complement existing clinical risk assessments.

Supplemental Material

Supplemental Material - Development and Validation of a Social Frailty Index Among Older Adults in the Philippines

Supplemental Material for Development and Validation of a Social Frailty Index Among Older Adults in the Philippines by Edina Yi-Qin Tan, BA, Hons, MD, Hanzhang Xu, RN, PhD, Rahul Malhotra, MBBS, MD, MPH, Mark Ryan B. Paguirigan, MA, Grace Trinidad Cruz, PhD, Yasuhiko Saito, PhD, Truls Østbye, MD, PhD in Journal of Aging and Health

Footnotes

Acknowledgements

Thanks to the LSAHP project members at the University of the Philippines Population Institute.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Longitudinal Study of Ageing and Health in the Philippines (LSAHP) is funded by the Economic Research Institute for ASEAN and East Asia (ERIA), with the Demographic Research and Development Foundation, Inc. (DRDF) as the implementing agency in the Philippines.

ORCID iDs

Edina Yi-Qin Tan

Hanzhang Xu

Rahul Malhotra

Mark Ryan B. Paguirigan

Supplemental Material

Supplemental material for this article is available online.

References

Badana

A. N. S.

Andel

(2018). Aging in the Philippines. The Gerontologist, 58(2), 212–218. https://doi.org/10.1093/geront/gnx203

Bunt

Steverink

Olthof

Van Der Schans

C. P.

Hobbelen

J. S. M.

(2017). Social frailty in older adults: A scoping review. European Journal of Ageing, 14(3), 323–334. https://doi.org/10.1007/s10433-017-0414-7

Casado

Sousa

Guerra

Couto

P. S. .

Tavares

J. P. A.

(2024). Loyola generativity scale and generative behavior checklist: A psychometric study with Spanish LGBT+ older adults. Journal of Nursing Referência, 6(3), 1–9. https://doi.org/10.12707/RVI23.64.31274

Cruz

C. J. P.

Cruz

G. T.

Saito

(2019). Ageing and health in the Philippines. Demographic Research and Development Foundation, Inc, & Economic Research Institute for ASEAN and East Asia.

Cruz

G. T.

Natividad

J. N.

Gonzales

M. L.

Saito

(2016). Aging in the Philippines: Findings from the 2007 philippine study on aging. University of the Philippines Population Institute and Demographic Research and Development Foundation.

Demler

O. V.

Paynter

N. P.

Cook

N. R.

(2015). Tests of calibration and goodness-of-fit in the survival setting. Statistics in Medicine, 34(10), 1659–1680. https://doi.org/10.1002/sim.6428

Domingo

L. J.

Perez

A. E.

Cabigon

J. V.

(1996). The Filipino elderly: Findings from the 1996 philippine elderly survey. University of the Philippines Population Institute and Demographic Research and Development Foundation.

Esteban

R. C.

(2015). Thinking about aging: Experience, identity and meaning among an elderly population in the Philippines. Advances in Aging Research, 04(05), 133–153. https://doi.org/10.4236/aar.2015.45015

Fried

L. P.

Ferrucci

Darer

Williamson

J. D.

Anderson

(2004). Untangling the concepts of disability, frailty, and comorbidity: Implications for improved targeting and care. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 59(3), M255–M263. https://doi.org/10.1093/gerona/59.3.M255

10.

Garre-Olmo

Calvó-Perxas

López-Pousa

De Gracia Blanco

Vilalta-Franch

(2013). Prevalence of frailty phenotypes and risk of mortality in a community-dwelling elderly cohort. Age and Ageing, 42(1), 46–51. https://doi.org/10.1093/ageing/afs047

11.

Gobbens

R. J. J.

Van Assen

M. A. L. M.

Luijkx

K. G.

Wijnen-Sponselee

M.Th.

Schols

J. M. G. A.

(2010). The tilburg frailty indicator: Psychometric properties. Journal of the American Medical Directors Association, 11(5), 344–355. https://doi.org/10.1016/j.jamda.2009.11.003

12.

Goto

Kishimoto

Fujiwara

Shirayama

Ichikawa

(2024). Social frailty as a predictor of all-cause mortality and functional disability: A systematic review and meta-analysis. Scientific Reports, 14(1), 3410. https://doi.org/10.1038/s41598-024-53984-3

13.

Grossman

M. R.

Gruenewald

T. L.

(2017). Caregiving and perceived generativity: A positive and protective aspect of providing care? Clinical Gerontologist, 40(5), 435–447. https://doi.org/10.1080/07317115.2017.1317686

14.

Hawkley

L. C.

Buecker

Kaiser

Luhmann

(2022). Loneliness from young adulthood to old age: Explaining age differences in loneliness. International Journal of Behavioral Development, 46(1), 39–49. https://doi.org/10.1177/0165025420971048

15.

Hughes

M. E.

Waite

L. J.

Hawkley

L. C.

Cacioppo

J. T.

(2004). A short scale for measuring loneliness in large surveys: Results from two population-based studies. Research on Aging, 26(6), 655–672. https://doi.org/10.1177/0164027504268574

16.

Irshad

Govil

Sahoo

(2023). Social frailty among older adults in India: Findings from the longitudinal ageing study in India (LASI) – wave 1. Experimental Aging Research, 50(3), 1–17. https://doi.org/10.1080/0361073X.2023.2195291

17.

Lee

Kim

Yun

Chuck

K. W.

(2022). The influence of multiple frailty profiles on institutionalization and all‐cause mortality in community‐living older adults. Journal of Cachexia, Sarcopenia and Muscle, 13(5), 2322–2330. https://doi.org/10.1002/jcsm.13033

18.

Wang

Zhou

Yun

Shi

Wang

(2024). The relationship between social frailty and loneliness in community-dwelling older adults: A cross-sectional study. BMC Geriatrics, 24(1), 73. https://doi.org/10.1186/s12877-024-04666-2

19.

Lubben

Blozik

Gillmann

Iliffe

Von Renteln Kruse

Beck

J. C.

Stuck

A. E.

(2006). Performance of an abbreviated version of the lubben social network scale among three European community-dwelling older adult populations. The Gerontologist, 46(4), 503–513. https://doi.org/10.1093/geront/46.4.503

20.

Luhmann

Hawkley

L. C.

(2016). Age differences in loneliness from late adolescence to oldest old age. Developmental Psychology, 52(6), 943–959. https://doi.org/10.1037/dev0000117

21.

Sun

Tang

(2018). Social frailty is associated with physical functioning, cognition, and depression, and predicts mortality. The Journal of Nutrition, Health & Aging, 22(8), 989–995. https://doi.org/10.1007/s12603-018-1054-0

22.

Makizako

Shimada

Tsutsumimoto

Lee

Doi

Nakakubo

Hotta

Suzuki

(2015). Social frailty in community-dwelling older adults as a risk factor for disability. Journal of the American Medical Directors Association, 16(11), 1003.e7. https://doi.org/10.1016/j.jamda.2015.08.023

23.

Marquez

M. P. N.

Tanalgo

K. P.

Paguirigan

M. R. B.

Cabaraban

M. K. S.

(2022). Technical note on the creation of the 2018 LSAHP wealth index: Demographic Research and Development Foundation.

24.

Misu

Katayama

Lee

Makino

Harada

Tomida

Morikawa

Yamaguchi

Nishijima

Fujii

Shimada

(2023). Reciprocal relationship between physical and social frailty among community-dwelling older adults. Archives of Gerontology and Geriatrics, 114, Article 105066. https://doi.org/10.1016/j.archger.2023.105066

25.

Mitchell

L. B.

Southern

D. A.

Galbraith

Ghali

W. A.

Knudtson

Wilton

S. B.

APPROACH investigators . (2014). Prediction of stroke or TIA in patients without atrial fibrillation using CHADS ₂ and CHA ₂ DS ₂ -VASc scores. Heart, 100(19), 1524–1530. https://doi.org/10.1136/heartjnl-2013-305303

26.

Park

Jang

I. Y.

Lee

H. Y.

Jung

H.-W.

Lee

Kim

D. H.

(2019). Screening value of social frailty and its association with physical frailty and disability in community-dwelling older Koreans: Aging study of PyeongChang rural area. International Journal of Environmental Research and Public Health, 16(16), 2809. https://doi.org/10.3390/ijerph16162809

27.

Pek

Chew

Lim

J. P.

Yew

Tan

C. N.

Yeo

Ding

Y. Y.

Lim

W. S.

(2020). Social frailty is independently associated with mood, nutrition, physical performance, and physical activity: Insights from a theory-guided approach. International Journal of Environmental Research and Public Health, 17(12), 4239. https://doi.org/10.3390/ijerph17124239

28.

Philippine Statistics Authority . (2017). PSA Board Resolution No. 01, Series of 2017–098: Approving and adopting the official concept and definition of “Filipino children”. [Superseding NSCB Resolution No. 09, Series of 2003].

29.

Shah

S. J.

Oreper

Jeon

S. Y.

Boscardin

W. J.

Fang

M. C.

Covinsky

K. E.

(2023). Social Frailty Index: Development and validation of an index of social attributes predictive of mortality in older adults. Proceedings of the National Academy of Sciences of the United States of America, 120(7), Article e2209414120. https://doi.org/10.1073/pnas.2209414120

30.

Takagi

Cruz

G. T.

Saito

(2022). Older adults’ loneliness and social relationships in the Philippines: Comparison between rural and urban contexts. Asia-Pacific Journal of Public Health, 34(6–7), 662–666. https://doi.org/10.1177/10105395221106861

31.

Takagi

Saito

Chan

(2020). Gender differences in the association between social relationships and loneliness among older adults in Singapore. Journal of Population Research, 37(3), 243–263. https://doi.org/10.1007/s12546-020-09242-0

32.

Teo

Gao

Nyunt

M. S. Z.

Wee

S. L.

T.-P.

(2017). Social frailty and functional disability: Findings from the Singapore longitudinal ageing studies. Journal of the American Medical Directors Association, 18(7), 637.e13–637.e19. https://doi.org/10.1016/j.jamda.2017.04.015

33.

Vrigazova

(2021). The proportion for splitting data into training and test set for the bootstrap in classification problems. Business Systems Research Journal, 12(1), 228–242. https://doi.org/10.2478/bsrj-2021-0015

34.

Yamada

Arai

(2015). Predictive value of frailty scores for healthy life expectancy in community-dwelling older Japanese adults. Journal of the American Medical Directors Association, 16(11), 1002.e7. https://doi.org/10.1016/j.jamda.2015.08.001

35.

Wang

Zeng

Yang

Tang

(2023). The prevalence of social frailty among older adults: A systematic review and meta-analysis. Geriatric Nursing, 49, 101–108. https://doi.org/10.1016/j.gerinurse.2022.11.009

36.

Zhang

Jiang

Hao

Liu

Jiang

Jin

Wang

(2024). Social frailty and the incidence of motoric cognitive risk syndrome in older adults. The Journal of the Alzheimer’s Association, 20(4), 2329–2339. https://doi.org/10.1002/alz.13696

37.

Zhang

X.-M.

Cao

Gao

Xiao

Xie

(2023). The prevalence of social frailty among older adults: A systematic review and meta-analysis. Journal of the American Medical Directors Association, 24(1), 29–37.e9. https://doi.org/10.1016/j.jamda.2022.10.007

38.

Zhang

Wong

Mann

Muller

Yang

J. Y.

(2022). SurvBenchmark: comprehensive benchmarking study of survival analysis methods using both omics data and clinical data. Gigascience, 11, giac071.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.50 MB