Diverging Frailty Trajectories After Retirement: The Protective Role of Education

Abstract

Retirement is a life-course transition that reshapes daily routines, social engagement, and exposure to stressors that influence later-life health. However, little is known about how frailty, an integrative indicator of biological aging, changes around retirement or whether patterns differ by education. This study examined frailty trajectories among Korean adults using seven waves of the Korean Longitudinal Study of Ageing. Distributed fixed-effects models estimated within-person changes before and after retirement, with frailty measured with a 41-item index across physical, cognitive, and psychosocial domains. Frailty remained stable prior to retirement but increased significantly during the first 2 years afterward, with elevated levels persisting. Increases were larger and sustained among individuals with lower education, whereas higher-educated retirees experienced smaller and temporary changes. Findings suggest that retirement marks a pivotal period in the accumulation of health deficits and that education serves as a protective factor, highlighting the need for policies supporting equitable post-retirement resources.

Keywords

retirement frailty educational attainment moderation longitudinal analysis

What This Paper Adds

• Demonstrates, using within-person distributed fixed-effects models, that frailty remains stable before retirement but rises sharply in the first 2 years afterward, identifying retirement as a key turning point in biological aging.

• Reveals substantial educational disparities in post-retirement frailty trajectories, with lower-educated individuals experiencing larger and more persistent increases, highlighting cumulative life-course disadvantage.

• Advances the literature by integrating a multidimensional frailty index and temporal modeling approach to clarify when frailty changes occur and for whom risks are greatest.

Applications of Study Findings

• Supports gerontological practice by emphasizing the need for early post-retirement screening and prevention programs targeting physical, cognitive, and psychosocial frailty, particularly among lower-educated retirees.

• Informs policy development promoting equitable access to post-retirement resources, such as community-based health services, social engagement opportunities, and flexible retirement pathways to reduce vulnerability.

• Guides future research to explore mechanisms driving socioeconomic differences in frailty progression and to evaluate interventions that mitigate accelerated frailty accumulation following retirement.

Introduction

Retirement is a major life-course transition that reshapes the rhythms of daily life, patterns of social engagement, and exposure to psychosocial and economic stressors—all of which represent key pathways through which health evolves with age (Wang & Shi, 2014). Theories of cumulative advantage (Dannefer, 2003) and the stress process framework (Pearlin et al., 2005) suggest that such transitions can have divergent consequences: for some, withdrawal from demanding work environments may alleviate chronic strain and promote recovery, whereas for others, the loss of income, purpose, and social connectedness may exacerbate vulnerability and accelerate health decline (Park & Kim, 2023b; Wahrendorf et al., 2013). Thesedivergent outcomes likely arise because retirementsimultaneously disrupts multiple dimensions of working life—physical, psychosocial, and functional—suggesting that its health consequences are best assessed using a measure that captures changes across these domains (Iavicoli et al., 2018).

Frailty addresses this need by providing a multidimensional measure of health in the context of retirement transitions. Unlike single-domain health indicators, frailty captures the cumulative nature of age-related decline across multiple biological and functional systems (Cheng & Chang, 2017). Specifically, it encompasses a wide range of deficits, including chronic disease burden, physical functioning, depressive symptoms, and cognitive capacity (Fried et al., 2001; Rockwood & Mitnitski, 2007). Its cumulative nature makes frailty particularly well-suited for examining the health consequences of retirement, as it encodes the health legacy of working life, capturing how occupational exposures and resources shaped physiological resilience prior to retirement (Cesari et al., 2017). The loss of these occupational resources at retirement may then disrupt processes that had been decelerating biological aging, accelerating deficit accumulation across the very systems that frailty is designed to capture (Vigezzi et al., 2025). Despite growing recognition of frailty as an integrative indicator of biological aging, empirical research directly linking retirement timing to frailty trajectories remains limited (Calvo et al., 2013).

A growing body of evidence suggests that retirement may accelerate the development or progression of frailty through a combination of behavioral, psychosocial, and physiological mechanisms. Exiting the workforce often disrupts established routines, leading to declines in physical activity and daily mobility (Stenholm et al., 2015; Wahrendorf et al., 2013), both of which are key predictors of frailty onset (Kojima et al., 2019). The loss of work-related social roles and networks can also reduce opportunities for socialinteraction and cognitive engagement, contributing to loneliness and psychological distress (Nam & Kim, 2026; Van Der Heide et al., 2013). In turn, these psychosocial changes may activate stress responses and inflammatory pathways associated with physiological dysregulation and frailty (Fried et al., 2001; Gale et al., 2015; Kim & Hwang, 2024). Empirical studies provide consistent support for these mechanisms: retirement has been linked to lower physical functioning, greater depressive symptoms, and higher risk of disability and chronic disease accumulation (Calvo et al., 2013; Pinquart & Schindler, 2009; Stenholm et al., 2014; Wahrendorf et al., 2013).

Despite growing interest in the relationship between retirement and physical health and functioning, important limitations persist in the existing literature. Much prior research has focused on single health indicators—such as self-rated health, depressive symptoms, or physical activity—rather than employing multidimensional measures that capture the interconnected physical, psychological, and functional dimensions of health decline (Hoogendijk et al., 2019; Kojima et al., 2019). Because frailty reflects the cumulative process of biological and functional deterioration, short-term or domain-specific assessments may underestimate the enduring health implications of retirement. A particularly salient gap concerns the limited attention to the long-term trajectories of frailty as individuals transition into and adapt to retirement. Most previous studies have relied on cross-sectional or short-term longitudinal designs that document only immediate health changes, overlooking how frailty evolves over extended periods (Calvo et al., 2013; Stenholm et al., 2014). Neglecting the temporal dimension of frailty change may obscure meaningful within-person transitions and underestimate the sustained impact of retirement on health.

Another important consideration is the potential heterogeneity in the association between retirement and frailty by educational attainment. Education can be conceptualized as a multidimensional resource that operates through stratifying exposure to risks accumulated across the work life and shaping individuals’ capacity to adapt to retirement. Education often structures occupational trajectories by shaping access to jobs that differ substantially in their physical demands, stability of employment, and the economic and social resources they confer (Phelan et al., 2010). Education also cultivates health-relevant capacities, including health literacy, self-efficacy, and adaptive coping strategies that sustain physical and mental well-being across biographical transitions (Mirowsky & Ross, 2005). Drawing on the cumulative advantage perspectives (Dannefer, 2003), these advantages can accumulate and be amplified over the life course. As a result, by the time individuals reach retirement, disparities in both exposure to work-related risks and adaptive capacity are already deeply embedded, leading to structured differences in resilience to frailty.

Empirically, these pathways manifest in systematic differences between educational groups. Those with lower education often enter retirement after years of physically demanding or unstable employment, with limited financial reserves and fewer social or psychological resources to support healthy aging (Hoven & Siegrist, 2013; Song & Kim, 2024). In contrast, higher-educated individuals typically retire from less strenuous and more autonomous jobs, possess greater economic security, and maintain broader social and cognitive engagement (Kim & Kwon, 2024; Mirowsky et al., 2000). These advantages can serve as protective buffers, enabling them to sustain health-promoting routines, access medical care, and remain socially and mentally active after leaving the workforce (Börsch-Supan & Schuth, 2014; Szinovacz & Davey, 2005). Moreover, higher education is linked to stronger health literacy, self-efficacy, and adaptive coping strategies, which facilitate adjustment to lifestyle changes during retirement and help prevent functional decline (Eibich, 2015; Ross & Mirowsky, 2010). As a result, the increase in frailty associated with retirement is likely to be smaller and more transient among the highly educated, whereas those with lower education may experience a steeper and more persistent deterioration—reflecting how cumulative socioeconomic advantage confers resilience against the physiological and psychosocial challenges of retirement.

South Korea provides a particularly important context for examining the relationship between retirement and frailty. The country is experiencing one of the most rapid population aging processes in the world, accompanied by persistently high levels of old-age poverty (Bloom & Finlay, 2009). Despite these pressures, public social protection systems remain relatively limited, and a substantial proportion of older adults continue to depend on labor market participation as a primary source of income (J. Kang et al., 2022). These structural conditions have fueled ongoing policy debates on extending working lives and delaying retirement as potential responses to population aging (Jun, 2020). However, such discussions often overlook the potential health consequences of retirement transitions, particularly for vulnerable populations. In addition, access to stable employment, pension benefits, and post-retirement opportunities is highly stratified by socioeconomic status, suggesting that the health implications of retirement are likely to vary substantially across population subgroups (Hoogendijk et al., 2019). Against this backdrop, examining how retirement shapes frailty—and how these patterns vary by socioeconomic position—can provide important insights into both the timing of health deterioration and the structural inequalities that underlie it.

This study investigates how frailty changes around the time of retirement among older Korean adults, with a particular focus on potential heterogeneity in frailty trajectories by educational attainment. While prior research has examined the association between retirement and general health or functioning, few studies have explored how frailty—a multidimensional indicator of physiological and functional vulnerability—evolves before and afterretirement over an extended period. Using distributed fixed-effects models and nationally representative longitudinal data from the Korean Longitudinal Study of Ageing, this study captures within-person changes in frailty surrounding the retirement transition while controlling for unobserved, time-invariant individual characteristics. In addition, it examines whether the effects of retirement on frailty differ by educational attainment, recognizing that education, as a key dimension of socioeconomic status, may buffer or exacerbate the health consequences of labor force exit.

Data and Methods

Data

This study employed data from the Korean Longitudinal Study of Ageing (KLoSA)—a nationally representative biennial panel survey of community-dwelling adults aged 45 and older residing across all 15 administrative regions of South Korea. Initiated in 2006, KLoSA was designed to monitor temporal changes in key domains of aging, including socioeconomic status, physical and mental health, and psychosocial characteristics. The baseline cohort was drawn using a multistage stratified probability sampling approach. Stratification considered both geographic area (urban vs. rural) and housing type (apartment vs. non-apartment), with enumeration districts from the Population and Housing Census provided by Statistics Korea serving as the sampling frame. This sampling design ensured broad representativeness across diverse residential environments and regional contexts.

The present analysis draws on longitudinal data from Waves 1 through 7 of KLoSA, spanning a 12-year period (2006–2018). Item non-response across analytic variables was minimal. Of the initial 54,131 person-wave observations, 39 observations (<0.1%) were excluded due to missing values on control variables. The final analytic sample comprised 54,092 person-wave observations from 10,241 unique individuals. On average, respondents participated in approximately 5.3 survey waves over the study period. All participants provided informed consent at each wave of data collection. As the dataset is fully anonymized and publicly available, this study was exempt from ethical review by institutional review board (KUIRB-2020-0194-02).

Measures

Dependent Variable

The dependent variable, frailty, was operationalized using a frailty index constructed from 41 variables spanning seven domains, following established approaches (Baek & Min, 2022; Fan et al., 2020; Pérez-Zepeda et al., 2021). Consistent with the deficit accumulation framework developed by Rockwood and Mitnitski (2007), frailty is conceptualized as the proportion of accumulated health deficits across multiple physiological and functional systems. This approach intentionally incorporates heterogeneous indicators—including symptoms, signs, disabilities, diseases, and self-reported health measures—because frailty is understood as a multidimensional and probabilistic state emerging from the cumulative burden of diverse deficits rather than any single condition.

These domains encompassed self-rated health, physical condition, mental health, cognitive function, limitations in activities of daily living (ADLs), limitations in instrumental activities of daily living (IADLs), and chronic diseases. Self-rated health was coded on a 0-1 scale (excellent = 0, very good = 0.25, good = 0.5, fair = 0.75, poor = 1). Physical condition indicators, including grip strength, were coded dichotomously (0 = normal, 1 = impaired) using thresholds of 28.6 kg for men and 16.4 kg for women (Yoo et al., 2017). Body mass index (BMI) values below 18.5 were also coded as 1 (Li et al., 2020). Mental health was measured using the 10-item Center for Epidemiologic Studies Depression Scale (CES-D10), with responses recoded from 0 (“rarely or never”) to 1 (“all of the time”) (Irwin et al., 1999). Cognitive function was assessed using the Korean Mini-Mental State Examination (K-MMSE), with item responses rescaled from 0 to 1 (Baek & Min, 2022). ADL (e.g., dressing, bathing, and hygiene) and IADL (e.g., meal preparation, laundry, and housework) limitations were coded as 0 (“able”), 0.5 (“needs help”), or 1 (“unable”) (Won et al., 2002). Chronic conditions included seven diseases—hypertension, diabetes, chronic lung disease, heart disease, stroke, arthritis, and urinary incontinence—as well as current use of prescribed medications, all coded as binary indicators (0 = no, 1 = yes).

To assess the internal coherence of domain-specific components, we computed Cronbach’s alpha for each subdomain at baseline (Wave 1). Reliability coefficients were 0.57 for the physical domain, 0.89 for the psychological domain, 0.82 for the cognitive domain, 0.96 for the functional (ADL/IADL) domain, and 0.59 for the chronic conditions domain. These values indicate acceptable to high internal consistency for most domains, particularly for psychological, cognitive, and functional components, while the comparatively lower alpha values for physical and chronic domains reflect the intentionally heterogeneous nature of deficit accumulation measures.

The frailty index was computed as the mean of all 41 item scores multiplied by 100, yielding a continuous measure ranging from 0 to 100. A detailed description of the individual items within each domain is provided in Table S1 of the online supplemental file. Following standard practice in the deficit accumulation literature, all items were equally weighted and coded on a 0–1 scale. Prior research demonstrates that the predictive validity and construct stability of the frailty index derive primarily from the number (proportion) of accumulated deficits rather than differential weighting of specific components, and that indices constructed from sufficiently large sets of age-associated deficits yield comparable properties across populations (Rockwood & Mitnitski, 2007).

Independent Variable

The key independent variable represents respondents’ temporal proximity to retirement, measured as the number of months between each survey interview and the respondent’s self-reported retirement date. This measure was categorized into eight time intervals to capture both pre- and post-retirement periods: 0–11 months (first year), 12–23 months (second year), 24–35 months (third year), and 36 months or more afterretirement (fourth year or later); as well as −1 to −12 months (final year before retirement), −13 to −24 months (second-to-last year), −25 to −36 months (third-to-last year), and −37 months or earlier (4 or more years before retirement). For individuals who did not retire during the observation period, the retirement proximity variable was set to zero across all waves in the fixed-effects models. Although these respondents do not contribute to within-person changes in retirement status, they were retained to improve the efficiency of estimates for time-varying covariates and to reduce potential selection bias. In cases where respondents reported multiple retirement events (approximately 2% of the sample), the most recent retirement date was used, reflecting their final and stable withdrawal from the labor force.

Control Variable

The empirical models included a set of time-varying covariates to adjust for potential confounding influences. Age was treated as a continuous variable. Marital status was captured with a binary indicator distinguishing currently married respondents from those not married, the latter encompassing individuals who were never married, widowed, divorced, or separated. Household size reflected the total number of household members. Socioeconomic status was measured using two indicators: household income, categorized into quartiles (with a separate category for missing income values) to capture relative economic position, and homeownership, coded as a binary variable (1 = homeowner, 0 = non-homeowner). Residential location was classified by urbanicity into three categories—large city, small city, and rural area.

Statistical Analysis

To examine the association between retirement timing and frailty, we estimated distributed fixed-effects (FE) regression models. The FE approach is well-suited for longitudinal data because it accounts for all time-invariant individual characteristics, both observed and unobserved, by exploiting within-person variation over time. This strategy minimizes bias from stable confounders and yields more robust estimates of the relationship between retirement timing and changes in frailty. A series of indicator variables was created to capture each observation’s position relative to the respondent’s retirement, distinguishing between pre- and post-retirement periods. The reference category was defined as 4 or more years before retirement (≤−37 months), allowing all coefficients to be interpreted relative to this baseline. The general model specification is expressed as:

y_{i t} = α + β_{1} {(- 3 n d y e a r)}_{i t} + β_{2} {(- 2 n d y e a r)}_{i t} + β_{3} {(- 1 s t y e a r)}_{i t} + β_{4} {(+ 1 s t y e a r)}_{i t} + β_{4} {(+ 2 n d y e a r)}_{i t} + β_{5} {(+ 3 r d y e a r s)}_{i t} + β_{6} (+ 4 t h o r m o r e y e a r s) + Z_{i t} δ + v_{i} + ε_{i t}

In this equation, $y_{i t}$ denotes the frailty score of individual i at time t; $Z_{i t}$ is a vector of time-varying covariates; $v_{i}$ represents unobserved individual-specific effects that are constant over time; and $ε_{i t}$ is the idiosyncratic error term. The $β$ coefficients indicate the average within-person change in frailty at each time interval surrounding retirement relative to the pre-retirement reference period. Robust standard errors were used in all models to adjust for potential heteroskedasticity.

KLoSA employs a multistage stratified sampling design and provides sampling weights. Because our analyses use individual fixed-effects models that account for all time-invariant characteristics, including those related to sampling design (e.g., region and cohort), survey weights were not applied in the main models. As FE estimation removes time-invariant factors, the application of survey weights—designed primarily to adjust for baseline sampling probabilities—does not generally improve consistency and may reduce efficiency. Moreover, survey weights are available only for respondents participating in all seven waves, and restricting the sample accordingly would substantially reduce the sample size. As a robustness check, we re-estimated the models using the available cross-sectional weights, and the results were substantively unchanged (Table S2 in the online supplemental file).

To explore educational heterogeneity, we conducted subgroup analyses by educational attainment. Education was dichotomized into lower education (middle school or below) and higher education (high school or above), reflecting meaningful socioeconomic divisions among older Korean cohorts, where few individuals attained tertiary education (Kim et al., 2025; Yang et al., 2025). In addition to stratified analyses, we estimated interaction models between retirement timing and education level to formally test for differences in effects across subgroups.

As in most longitudinal studies of older adults, the KLoSA panel experienced attrition over time due to mortality and non-mortality dropout. The number of observations declined gradually across waves: Wave 1 (n = 10,236), Wave 2 (n = 8,683), Wave 3 (n = 7,917), Wave 4 (n = 7,482), Wave 5 (n = 7,026), Wave 6 (n = 6,615), and Wave 7 (n = 6,133). This pattern is consistent with expected sample reduction in aging cohorts. Because frailty increases mortality risk, selective attrition could potentially attenuate estimated associations if individuals with higher frailty are disproportionately lost to follow-up. Our analyses use individual fixed-effects models estimated on an unbalanced panel, allowing respondents to contribute all available observations prior to attrition. This approach mitigates bias arising from time-invariant unobserved characteristics related to both frailty and dropout. In addition, sensitivity analyses using inverse probability weighting yielded substantively similar results (Table S3 in the online supplemental file), suggesting that attrition does not materially affect our conclusions.

Results

Table 1 presents descriptive statistics for the analytic sample drawn from the 2006 baseline of the KLoSA. The mean frailty index for the entire sample was 0.145, with higher frailty observed among retirees (0.153) compared with non-retirees (0.141). Within the retirement group, individuals with lower educational attainment exhibited substantially higher frailty (0.179) than those with higher education (0.109; p < 0.05). Retirees were, on average, older (64.2 years) and less likely to be female (44%) than non-retirees (60.5 years; 63%). They were also more likely to be married (80% vs. 77%) and homeowners (77% vs. 76%), but tended to live in smaller households (2.84 vs. 3.05 persons) and had lower household income levels. Educational differences were evident within the retired population, with highly educated retirees more often residing in large cities and having higher income and homeownership rates than their lower-educated counterparts. The distribution of time to retirement by educational level is presented in Table S4 of the online supplemental file.

Table 1.

Descriptive statistics, Korean Longitudinal Study of Ageing (KLoSA), 2006

	(1)	(2)	(3)	(4)	(5)
	Entire sample	No retirement group	Retirement group	Retirement group (low education)	Retirement group (high education)
	Mean (SD)/%	Mean (SD)/%	Mean (SD)/%	Mean (SD)/%	Mean (SD)/%
Dependent variable
Frailty index^a,b	0.145 (0.120)	0.141 (0.122)	0.153 (0.114)	0.179 (0.114)	0.109 (0.100)
Demographics and covariates
Female^a,b	0.565 (0.496)	0.627 (0.484)	0.440 (0.496)	0.532 (0.499)	0.281 (0.450)
Age^a,b	61.705 (11.123)	60.483 (11.578)	64.153 (9.700)	65.987 (9.249)	60.985 (9.649)
Married^a,b	77.8%	76.6%	80.0%	73.4%	91.5%
Household size^a,b	2.981 (1.358)	3.053 (1.360)	2.835 (1.344)	2.747 (1.369)	2.986 (1.286)
Household income (Q1)^a,b	23.3%	22.0%	26.1%	29.8%	19.7%
Household income(Q2)^a,b	24.7%	23.0%	28.1%	32.3%	20.8%
Household income(Q3)^a,b	26.7%	27.9%	24.3%	22.9%	26.8%
Household income(Q4)^a,b	17.2%	18.8%	14.0%	8.0%	24.3%
Household income (missing)	8.1%	8.3%	7.5%	7.0%	8.4%
Homeownership^b	76.3%	75.7%	77.4%	75.0%	81.7%
Large city^b	44.7%	44.5%	45.2%	42.6%	49.8%
Small city^b	32.6%	32.5%	32.7%	29.5%	38.3%
Rural^b	22.7%	23.0%	22.1%	27.9%	11.9%
N(Individuals)	10,241	6,832	3,409	2159	1,250

Note. Summary statistics are based on 2006 data. Standard deviations in parentheses.

^ap < 0.05 for differences between no-retirement and retirement groups (columns 2 vs. 3).

^bp < 0.05 for differences between low and high education retirement groups (columns 4 vs. 5).

Table 2 presents the results from the fixed-effects regression models estimating changes in frailty relative to the timing of retirement. After adjusting for age (both linear and quadratic terms) and time-varying covariates, frailty remained largely stable in the years leading up to retirement, with no significant differences observed during the first or second pre-retirement years compared with third or more years before retirement. However, frailty increased notably after retirement, beginning in the first year post-retirement (β = 0.016, p < 0.001) and peaking in the second year (β = 0.022, p < 0.001), with elevated levels persisting 3 or more years after retirement (β = 0.012, p < 0.001). To interpret these changes, the increase in frailty at Year 1 post-retirement (β = 0.016) represents approximately 13% of a within-person standard deviation (SD = 0.12; (Pérez-Zepeda et al., 2021), and the peak increase at Year 2 (β = 0.022) represents approximately 18%. Given that Fan et al. (2020) found each 0.1-unit increment in the frailty index to be associated with a 68% higher risk of all-cause mortality, these changes carry meaningful implications for health risk. For further comparison, the Year 1 increase (β = 0.016) was approximately twice the magnitude of the effect of homeownership (β = −0.008) observed in the same model, a covariate known to represent substantial socioeconomic status in Korea (Park & Kim, 2023a).

Table 2.

Effect of time to retirement on frailty

	(1)
	Frailty index
Sample	Total
Fixed effects	Yes
Age controls	Yes
Time-varying controls	Yes
Time to retirement (ref: ≤ − 3rd year)
− 2nd year	−0.005^*
− 2nd year	[−0.009, −0.001]
− 1st year	−0.004
− 1st year	[−0.009, 0.001]
+1st year	0.016^***
+1st year	[0.010, 0.021]
+2nd year	0.022^***
+2nd year	[0.015, 0.028]
≥ + 3rd year	0.012^***
≥ + 3rd year	[0.007, 0.017]
Covariates
Age	0.005^***
Age	[0.005, 0.006]
Age squared	0.000^***
Age squared	[0.000, 0.000]
Married	−0.003
Married	[−0.008,0.002]
Household size	−0.000
Household size	[−0.001, 0.000]
Household income (Q2)	−0.000
Household income (Q2)	[−0.003, 0.002]
Household income (Q3)	−0.004^**
Household income (Q3)	[−0.007, −0.001]
Household income (Q4)	0.000
Household income (Q4)	[−0.003, 0.004]
Household income (missing)	−0.010^**
Household income (missing)	[−0.016, −0.003]
House ownership	−0.008^***
House ownership	[−0.011, −0.005]
Small city	0.003
Small city	[−0.006, 0.012]
Rural	0.010^*
Rural	[0.002, 0.019]
N(Observations)	54,092
N(Individuals)	10,241

Note. The 95% confidence intervals are in brackets. Robust standard errors were used. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 1 depicts the predicted trajectories of frailty in relation to the timing of retirement. Frailty levels remained relatively stable during the pre-retirement period, showing no significant change up to the year immediately preceding retirement. However, a pronounced increase was observed following retirement, with frailty rising sharply in the first year (p < 0.001) and reaching its peak in the second year after retirement (p < 0.001). Although frailty slightly declined thereafter, it remained significantly higher than pre-retirement levels even three or more years after labor force exit (p < 0.001). Overall, the figure highlights a clear inflection point around the transition to retirement, indicating that health vulnerability—as measured by frailty—tends to accelerate in the years immediately following retirement.

To assess whether specific domains drive the observed patterns, we estimated fixed-effects models separately for each frailty domain (Table S5 in the online supplemental file). Physical frailty increased significantly beginning in the first year after retirement and remained elevated thereafter. Psychological frailty showed a similar post-retirement rise, while chronic conditions also increased consistently following retirement. Cognitive frailty exhibited a more gradual pattern, with significant increases emerging 2 years after retirement. In contrast, the functional domain displayed modest declines in the years immediately preceding retirement and limited change afterward. Overall, the results indicate that the retirement-related increase in frailty reflects multidimensional changes rather than deterioration in a single domain.

Table 3 presents the results from the fixed-effects regression models examining the association between retirement timing and frailty by educational attainment. Among individuals with lower education, frailty increased significantly after retirement, rising in the first year (β = 0.018, p < 0.001), second year (β = 0.029, p < 0.001), and remaining elevated 3 or more years post-retirement (β = 0.017, p < 0.001). In contrast, among those with higher education, the increase in frailty was smaller in magnitude, with significant effects observed only in the first (β = 0.012, p < 0.01) and second (β = 0.011, p < 0.05) years following retirement.

Table 3.

Effect of time to retirement on frailty, by education

	(1)	(2)
	Frailty index	Frailty index
Sample	Low education	High education
Fixed effects	Yes	Yes
Age controls	Yes	Yes
Time-varying controls	Yes	Yes
Time to retirement (ref: ≤ − 3rd year)
− 2nd year	−0.003	−0.008^**
− 2nd year	[−0.009, 0.004]	[−0.014, −0.003]
− 1st year	−0.005	−0.003
− 1st year	[−0.011, 0.002]	[−0.010, 0.004]
+1st year	0.018^***	0.012^**
+1st year	[0.010, 0.026]	[0.003, 0.021]
+2nd year	0.029^***	0.011^*
+2nd year	[0.019,0.038]	[0.002,0.019]
≥ + 3rd year	0.017^***	0.005
≥ + 3rd year	[0.009, 0.024]	[-0.002, 0.013]
Covariates
Age	0.005^***	0.006^***
Age	[0.005, 0.005]	[0.005, 0.006]
Age squared	0.000^***	0.000^***
Age squared	[0.000, 0.000]	[0.000, 0.000]
Married	−0.001	−0.011^*
Married	[−0.008,0.005]	[−0.021, −0.001]
Household size	−0.001	0.000
Household size	[−0.002, 0.000]	[−0.001, 0.001]
Household income (Q2)	0.000	−0.002
Household income (Q2)	[−0.003, 0.003]	[−0.007, 0.002]
Household income (Q3)	−0.004	−0.006^*
Household income (Q3)	[−0.007, 0.000]	[−0.010, −0.001]
Household income (Q4)	−0.000	−0.001
Household income (Q4)	[−0.005, 0.004]	[−0.005, 0.004]
Household income (missing)	−0.007	−0.016^**
Household income (missing)	[−0.015, 0.001]	[−0.025, −0.006]
House ownership	−0.009^***	−0.005^*
House ownership	[−0.014, −0.004]	[−0.010, −0.001]
Small city	−0.003	0.009
Small city	[−0.017, 0.011]	[−0.002, 0.020]
Rural	0.008	0.012
Rural	[−0.004, 0.020]	[−0.000, 0.025]
N(Observations)	33,956	20,136
N(Individuals)	6,476	3,765

Note. The 95% confidence intervals are in brackets. Robust standard errors were used. *p < 0.05; **p < 0.01; ***p < 0.001.

In Table 4, the interaction terms further confirm that these post-retirement increases in frailty were significantly attenuated among higher-educated individuals, particularly from the second year onward (β = −0.018, p < 0.01 for +2 years; β = −0.011, p < 0.05 for ≥ +3 years). These findings indicate that the adverse health consequences of retirement are more pronounced among those with lower educational attainment.

Table 4.

Interaction model of time to retirement on frailty and education

	(1)
	Frailty index
Sample	Total
Fixed effects	Yes
Age controls	Yes
Time-varying controls	Yes
Time to retirement (ref: ≤ - 3rd year)
− 2nd year	−0.003
− 2nd year	[−0.009, 0.004]
− 1st year	−0.005
− 1st year	[−0.011, 0.002]
+1st year	0.018^***
+1st year	[0.010, 0.026]
+2nd year	0.029^***
+2nd year	[0.019, 0.038]
≥ + 3rd year	0.017^***
≥ + 3rd year	[0.009, 0.024]
Interaction terms
− 2nd year × high education	−0.006
− 2nd year × high education	[−0.014, 0.003]
− 1st year × high education	0.002
− 1st year × high education	[−0.008, 0.011]
+1st year × high education	−0.005
+1st year × high education	[−0.017, 0.006]
+2nd year × high education	−0.018^**
+2nd year × high education	[−0.030, −0.006]
≥ + 3rd year × high education	−0.011^*
≥ + 3rd year × high education	[−0.022, −0.001]
N(Observations)	54,092
N(Individuals)	10,241

Note. The 95% confidence intervals are in brackets. Robust standard errors were used. Age controls include both linear and quadratic terms for age. Time-varying controls include marital status, household size, household income, homeownership, economic activity, and region of residence. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 2 illustrates the predicted trajectories of frailty surrounding the transition to retirement, stratified by educational attainment. Among individuals with lower education, frailty increased sharply following retirement, with significant rises observed in the first (p < 0.001), second (p < 0.001), and 3 or more years post-retirement (p < 0.001). In contrast, individuals with higher education exhibited consistently lower levels of frailty across all time points and a more modest post-retirement increase, significant only during the first (p < 0.01) and second (p < 0.05) years after retirement. These diverging patterns indicate that the adverse health consequences of retirement are substantially more pronounced among those with lower educational attainment, whereas higher-educated individuals appear to maintain greater health stability during the retirement transition.

Discussion

This study examined how frailty changes around the time of retirement among older Korean adults and whether these changes differ by educational attainment. Using nationally representative longitudinal data and distributed fixed-effects models, this study traced within-person variations in frailty before and after retirement. The findings revealed that frailty remained largely stable prior to retirement but increased significantly in the years immediately following labor force exit, with elevated levels persisting over time. This pattern suggests that retirement may act as a critical transition point that accelerates physiological and functional decline. These results are consistent with prior studies showing deteriorations in physical health and functioning after retirement (Calvo et al., 2013; Stenholm et al., 2014), yet this study extend this literature by demonstrating that the adverse health effects of retirement are not limited to short-term outcomes but unfold dynamically over multiple years. By focusing on frailty as an integrative measure of biological aging, this study provides stronger evidence that the retirement transition can have enduring consequences for later-life vulnerability.

The domain-specific analyses provide additional insight into the mechanisms underlying the observed retirement–frailty association. The pronounced post-retirement increases in the psychological domain may reflect the psychosocial adjustments accompanying retirement, including changes in daily structure, social roles, and perceived purpose, which can influence depressive symptoms and emotional well-being (Wahrendorf et al., 2013). The concurrent rise in the physical domain may partly reflect reductions in work-related physical activity or routine engagement, as well as emerging health vulnerabilities that become more salient after labor force exit (Kojima et al., 2019). The increase observed in the chronic conditions domain likely reflects the cumulative and progressive nature of diagnosed diseases in later life, which may continue to accumulate independent of retirement but become more evident during this transition (Hoogendijk et al., 2019). Prior studies have also suggested that physical, psychosocial, and functional declines may co-occur following retirement (Dave et al., 2008), whereas our findings extend this literature by demonstrating that frailty captures these multidimensional changes while also revealing their domain-specific trajectories (Vigezzi et al., 2025).

The results further revealed that the effects of retirement on frailty varied substantially by educational attainment. The increase in frailty following retirement was more pronounced and persistent among individuals with lower education, whereas those with higher education experienced only modest and short-lived changes. This divergence underscores the protective role of education in shaping health adaptation during the retirement transition. Higher-educated individuals are more likely to retire voluntarily, possess greater financial and psychosocial resources, and engage in health-promoting behaviors that help maintain physical and cognitive functioning after leaving the workforce (Börsch-Supan & Schuth, 2014; Eibich, 2015; Ross & Mirowsky, 2010). In contrast, lower-educated individuals often retire from physically demanding or stressful jobs with fewer financial reserves and weaker social networks, making them more susceptible to post-retirement frailty. These findings align with the theory of cumulative advantage (Dannefer, 2003), which posits that socioeconomic resources accumulated over the life course buffer against age-related vulnerability, thereby widening health disparities in later life.

These findings are particularly meaningful in the South Korean context, where retirement experiences are closely shaped by structural inequalities in access to economic and social resources. In Korea, higher educational attainment is strongly linked to stale employment trajectories, occupational benefits, and favorable post-retirement opportunities, including severance pay, employer-based insurance, and access to bridge employment (J. Kang et al., 2022), providing a continued buffer of material and social resources well beyond the point of retirement. By contrast, formal social protections for retirees remain underdeveloped in Korea, leaving lower-educated individuals with fewer institutional safeguards and limited capacity for individual preparation against the economic and health shocks that retirement can bring (Kang et al., 2022b). Taken together, the observed divergence in frailty trajectories by education can be understood as reflecting unequal access to buffering resources that shape adaptation to retirement, highlighting the role of structural inequality in later-life health outcomes.

This study makes several theoretical and methodological contributions to the literature on retirement and aging. Theoretically, it extends existing research by conceptualizing retirement as a dynamic life-course process rather than a single event, demonstrating that timing of retirement marks a critical inflection point at which cumulative socioeconomic inequalities become more visible. By applying frailty as a multidimensional outcome that consolidates physical, psychosocial, and functional health into a unified framework, it provides an integrative basis for detecting how these disparities extend across multiple domains of later-life health, particularly among low-educated groups (Jokela et al., 2010; Mendes de Leon et al., 2018). The findings further underscore the role of structural inequality such as differences in work conditions, economic security, and social resources as a central contextual factor through which retirement influences frailty trajectories (Wahrendorf et al., 2013), which prove difficult to reverse even after retirement. Methodologically, the distributed fixed-effects approach estimates within-person average effects of retirement on frailty, while controlling for all time-invariant individual heterogeneity. The distributed lag structure further examines contemporaneous associations at each relative time point, capturing how the relationship between retirement and frailty evolves dynamically. By explicitly establishing temporal ordering and ensuring that the retirement precedes the frailty at each lag, this approach also mitigates concerns about reverse causation and thereby overcomes key limitations of cross-sectional or short-term designs by providing stronger evidence on health changes surrounding the retirement transition (Calvo et al., 2013; Stenholm et al., 2014).

This study has several limitations that should be acknowledged. First, although the distributed fixed-effects approach controls for time-invariant individual heterogeneity, unmeasured time-varying factors may still confound the observed associations. For instance, acute health shocks could simultaneously precipitate early labor force exit and accelerate frailty accumulation (König et al., 2019). Second, frailty was operationalized using available indicators from the Korean Longitudinal Study of Ageing, which, while comprehensive, may not fully capture the biological complexity or clinical heterogeneity of frailty, particularly in the absence of biomarker-based measures. Third, retirement status and timing were derived from self-reported information, which may be influenced by recall error or ambiguity in how individuals interpret partial or gradual retirement, especially among older respondents with irregular work histories. Relatedly, our measure of retirement captures only the final retirement state and does not distinguish between different retirement pathways such as voluntary versus involuntary retirement, partial retirement, or bridge employment, which may differently shape health trajectories. Finally, the analytical focus on older Korean adults limits external validity, as the health and social implications of retirement may differ across institutional settings characterized by varying pension schemes, labor market dynamics, and cultural norms surrounding work and aging. Future research should therefore adopt designs that strengthen causal inference, incorporate more detailed measures of retirement pathways and pre-retirement job characteristics, and extend analyses across country settings to assess generalizability, particularly given cross-national differences in retirement systems and social protection.

The findings of this study have important policy implications for promoting healthy aging in the context of population aging and extended working lives. The observed increase in frailty following retirement suggests that the transition out of the labor force represents a critical window for targeted prevention and health promotion (Sims-Gould et al., 2020). Policymakers should consider implementing post-retirement health screening programs, community-based physical activity initiatives, and social engagement opportunities that help older adults maintain functional capacity after leaving the workforce (Arsenijevic & Groot, 2022). Given the pronounced educational disparities in frailty trajectories, policies should also address structural inequalities by enhancing access to health education, financial counseling, and lifelong learning programs that foster resilience among lower-educated retirees (Wilkie et al., 2024). In particular, expanding subsidized community health services, improving outreach and enrollment in public programs, and strengthening locally accessible social participation initiatives such as senior centers or structured group activities may help mitigate these disparities (Jang et al., 2021). Importantly, these interventions should be complemented by efforts to address underlying structural inequalities. Strengthening universal social protection mechanisms, including public pension coverage and employment support for older workers, may reduce the structural disadvantages that compound frailty risk (Aguila et al., 2018). In addition, labor market policies that promote gradual or flexible retirement options may help individuals adjust more smoothly, reducing abrupt declines in physical and psychosocial well-being. Together, these efforts highlight the importance of aligning targeted post-retirement interventions with broader structural reforms to mitigate health deterioration and promote more equitable and sustainable aging trajectories.

Supplemental Material

Supplemental Material - Diverging Frailty Trajectories After Retirement: The Protective Role of Education

Supplemental Material for Diverging Frailty Trajectories After Retirement: The Protective Role of Education by NaKyung Nam, Hyunseo Rim, Jinho Kim in Journal of Applied Gerontology

Footnotes

ORCID iD

Jinho Kim

Author Contributions

All authors contributed equally to this research.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

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

Data Availability Statement

The KLoSA data are available at with the permission of the Korea employment Information Service.

Supplemental Material

Supplemental material for this article is available online.

References

Aguila

López-Ortega

Gutiérrez Robledo

L. M.

(2018). Non-contributory pension programs and frailty of older adults: Evidence from Mexico. PLOS ONE, 13(11), Article e0206792. https://doi.org/10.1371/journal.pone.0206792

Arsenijevic

Groot

(2022). Health promotion policies for elderly—Some comparisons across Germany, Italy, the Netherlands and Poland. Health Policy, 126(1), 69–73. https://doi.org/10.1016/j.healthpol.2020.01.013

Baek

Min

(2022). Frailty index and gender‐specific mortality in Korean adults: Findings from the Korean longitudinal study of aging (2006–2018). Journal of Advanced Nursing, 78(8), 1–11. https://doi.org/10.1111/jan.15168

Bloom

D. E.

Finlay

J. E.

(2009). Demographic change and economic growth in Asia. Asian Economic Policy Review, 4(1), 45–64. https://doi.org/10.1111/j.1748-3131.2009.01106.x

Börsch-Supan

Schuth

(2014). Early retirement, mental health, and social networks. In Wise

D. A.

(Ed.), Discoveries in the economics of aging (pp. 225–254). University of Chicago Press. https://doi.org/10.7208/chicago/9780226146126.003.0007

Calvo

Sarkisian

Tamborini

C. R.

(2013). Causal effects of retirement timing on subjective physical and emotional health. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 68(1), 73–84. https://doi.org/10.1093/geronb/gbs097

Cesari

Calvani

Marzetti

(2017). Frailty in older persons. Clinics in Geriatric Medicine, 33(3), 293–303. https://doi.org/10.1016/j.cger.2017.02.002

Cheng

M. H.

Chang

S. F.

(2017). Frailty as a risk factor for falls among community dwelling people: Evidence from a meta-analysis. Journal of Nursing Scholarship, 49(5), 529–536. https://doi.org/10.1111/jnu.12322

Dannefer

(2003). Cumulative advantage/disadvantage and the life course: Cross-fertilizing age and social science theory. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 58(6), S327–S337. https://doi.org/10.1093/geronb/58.6.S327

10.

Dave

Rashad

Spasojevic

(2008). The effects of retirement on physical and mental health outcomes. Southern Economic Journal, 75(2), 497–523. https://doi.org/10.1002/j.2325-8012.2008.tb00916.x

11.

Eibich

(2015). Understanding the effect of retirement on health: Mechanisms and heterogeneity. Journal of Health Economics, 43, 1–12. https://doi.org/10.1016/j.jhealeco.2015.05.001

12.

Fan

Guo

Bian

Sun

Yang

Chen

Lei

Sun

Clarke

Chen

China Kadoorie Biobank Collaborative Group . (2020). Frailty index and all-cause and cause-specific mortality in Chinese adults: A prospective cohort study. The Lancet Public Health, 5(12), e650–e660. https://doi.org/10.1016/S2468-2667(20)30113-4

13.

Fried

L. P.

Tangen

C. M.

Walston

Newman

A. B.

Hirsch

Gottdiener

Seeman

Tracy

Kop

W. J.

Burke

McBurnie

M. A.

(2001). Frailty in older adults: Evidence for a phenotype. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 56(3), M146–M157. https://doi.org/10.1093/gerona/56.3.M146

14.

Gale

C. R.

Cooper

Aihie Sayer

(2015). Prevalence of frailty and disability: Findings from the English longitudinal study of ageing. Age and Ageing, 44(1), 162–165. https://doi.org/10.1093/ageing/afu148

15.

Hoogendijk

E. O.

Afilalo

Ensrud

K. E.

Kowal

Onder

Fried

L. P.

(2019). Frailty: Implications for clinical practice and public health. The Lancet, 394(10206), 1365–1375. https://doi.org/10.1016/S0140-6736(19)31786-6

16.

Hoven

Siegrist

(2013). Work characteristics, socioeconomic position and health: A systematic review of mediation and moderation effects in prospective studies. Occupational and Environmental Medicine, 70(9), 663–669. https://doi.org/10.1136/oemed-2012-101331

17.

Iavicoli

Leso

Cesari

(2018). The contribution of occupational factors on frailty. Archives of Gerontology and Geriatrics, 75(June 2017), 51–58. https://doi.org/10.1016/j.archger.2017.11.010

18.

Irwin

Artin

K. H.

Oxman

M. N.

(1999). Screening for depression in the older adult. Archives of Internal Medicine, 159(15), 1701–1704. https://doi.org/10.1001/archinte.159.15.1701

19.

Jang

A. R.

Won

C. W.

Sagong

Bae

Park

Yoon

J. Y.

(2021). Social factors predicting improvement of frailty in community‐dwelling older adults: Korean frailty and aging cohort study. Geriatrics & Gerontology International, 21(6), 465–471. https://doi.org/10.1111/ggi.14160

20.

Jokela

Ferrie

J. E.

Gimeno

Chandola

Shipley

M. J.

Head

Vahtera

Westerlund

Marmot

M. G.

Kivimäki

(2010). From midlife to early old age: Health trajectories associated with retirement. Epidemiology, 21(3), 284–290. https://doi.org/10.1097/EDE.0b013e3181d61f53

21.

Jun

(2020). Social security and retirement in fast-aging middle-income countries: Evidence from Korea. The Journal of the Economics of Ageing, 17, Article 100284. https://doi.org/10.1016/J.JEOA.2020.100284

22.

Kang

J. Y.

Park

Ahn

(2022b). The effect of social pension on consumption among older adults in Korea. The Journal of the Economics of Ageing, 22, Article 100364. https://doi.org/10.1016/J.JEOA.2021.100364

23.

Kang

Park

Cho

(2022a). Inclusive aging in Korea: Eradicating senior poverty. International Journal of Environmental Research and Public Health, 19(4), Article 2121. https://doi.org/10.3390/ijerph19042121

24.

Kim

Hwang

(2024). Separating the effects of transitions into and out of social isolation and loneliness on cognitive function in later life. The Journals of Gerontology: Series B, 79(7), 1996–2018. https://doi.org/10.1093/geronb/gbae082

25.

Kim

Kwon

K. Y.

(2024). Marital dissatisfaction and functional performance in older adults: Heterogeneity in the association by educational level. Social Science & Medicine, 340, Article 116460. https://doi.org/10.1016/j.socscimed.2023.116460

26.

Kim

Park

Kwon

K. Y.

(2025). Spousal loss and physical vulnerability: Diverging frailty trajectories by educational attainment in older adults. Social Science & Medicine, 382, Article 118408. https://doi.org/10.1016/j.socscimed.2025.118408

27.

Kojima

Taniguchi

Iliffe

Jivraj

Walters

(2019). Transitions between frailty states among community-dwelling older people: A systematic review and meta-analysis. Ageing Research Reviews, 50, 81–88. https://doi.org/10.1016/j.arr.2019.01.010

28.

König

Lindwall

Johansson

(2019). Involuntary and delayed retirement as a possible health risk for lower educated retirees. Journal of Population Ageing, 12(4), 475–489. https://doi.org/10.1007/s12062-018-9234-6

29.

Xue

Odden

M. C.

Chen

(2020). Linking early life risk factors to frailty in old age: Evidence from the China health and retirement longitudinal study. Age and Ageing, 49(2), 1–10. https://doi.org/10.1093/ageing/afz160

30.

Mendes de Leon

Henkens

Bültmann

(2018). Work, retirement, and healthy aging: Life-course issues and challenges. Innovation in Aging, 2(suppl_1), 426–427. https://doi.org/10.1093/geroni/igy023.1597

31.

Mirowsky

Ross

C. E.

(2005). Education, cumulative advantage, and health. Ageing International, 30(1), 27–62. https://doi.org/10.1007/BF02681006

32.

Mirowsky

Ross

C. E.

Reynolds

(2000). Links between social status and health status. In Bird

C. E.

Conrad

Fremont

A. M.

(Eds.), Handbook of medical sociology (5th Edition, p. 4). Prentice Hall.

33.

Nam

Kim

(2026). Diverging trajectories of depressive symptoms after retirement: The moderating role of social engagement. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 81(4), gbag010. https://doi.org/10.1093/geronb/gbag010

34.

Park

G. R.

Kim

(2023a). Trajectories of life satisfaction before and after homeownership: The role of housing affordability stress. Journal of Happiness Studies, 24(1), 397–408. https://doi.org/10.1007/s10902-022-00601-7

35.

Park

Kim

(2023b). The use of assistive devices and social engagement among older adults: Heterogeneity by type of social engagement and gender. GeroScience, 46(1), 1385–1394. https://doi.org/10.1007/s11357-023-00910-6

36.

Pearlin

L. I.

Schieman

Fazio

E. M.

Meersman

S. C.

(2005). Stress, health, and the life course: Some conceptual perspectives. Journal of Health and Social Behavior, 46(2), 205–219. https://doi.org/10.1177/002214650504600206

37.

Pérez-Zepeda

M. U.

Godin

Armstrong

J. J.

Andrew

M. K.

Mitnitski

Kirkland

Rockwood

Theou

(2021). Frailty among middle-aged and older Canadians: Population norms for the frailty index using the Canadian longitudinal study on aging. Age and Ageing, 50(2), 447–456. https://doi.org/10.1093/ageing/afaa144

38.

Phelan

J. C.

Link

B. G.

Tehranifar

(2010). Social conditions as fundamental causes of health inequalities: Theory, evidence, and policy implications. Journal of Health and Social Behavior, 51(1_suppl), S28–S40. https://doi.org/10.1177/0022146510383498

39.

Pinquart

Schindler

(2009). Change of leisure satisfaction in the transition to retirement: A latent-class analysis. Leisure Sciences, 31(4), 311–329. https://doi.org/10.1080/01490400902988275

40.

Rockwood

Mitnitski

(2007). Frailty in relation to the accumulation of deficits. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 62(7), 722–727. https://doi.org/10.1093/gerona/62.7.722

41.

Ross

C. E.

Mirowsky

(2010). Why education is the key to socioeconomic differentials in health. In Bird

C. E.

Conrad

Fremont

A. M.

Timmermans

(Eds.), Handbook of medical sociology (6th ed., p. 3). Vanderbilt University Press.

42.

Sims‐Gould

Franke

Lusina‐Furst

McKay

H. A.

(2020). Community health promotion programs for older adults: What helps and hinders implementation. Health Science Reports, 3(1), Article e144. https://doi.org/10.1002/hsr2.144

43.

Song

Kim

(2024). Prolonged exposure to intimate partner violence and depressive symptoms among Korean married couples: The intersection of gender and education. Society and Mental Health, 14(2), 129–144. https://doi.org/10.1177/21568693231185853

44.

Stenholm

Westerlund

Head

Hyde

Kawachi

Pentti

Kivimäki

Vahtera

(2015). Comorbidity and functional trajectories from midlife to old age: The health and retirement study. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 70(3), 332–338. https://doi.org/10.1093/gerona/glu113

45.

Stenholm

Westerlund

Salo

Hyde

Pentti

Head

Kivimäki

Vahtera

(2014). Age-related trajectories of physical functioning in work and retirement: The role of sociodemographic factors, lifestyle and disease. Journal of Epidemiology and Community Health, 68(6), 503–509. https://doi.org/10.1136/jech-2013-203555

46.

Szinovacz

M. E.

Davey

(2005). Predictors of perceptions of involuntary retirement. The Gerontologist, 45(1), 36–47. https://doi.org/10.1093/geront/45.1.36

47.

Van Der Heide

Van Rijn

R. M.

Robroek

S. J. W.

Burdorf

Proper

K. I.

(2013). Is retirement good for your health? A systematic review of longitudinal studies. BMC Public Health, 13(1), 1–11. https://doi.org/10.1186/1471-2458-13-1180

48.

Vigezzi

G. P.

Barbati

Maggioni

Stenholm

Odone

Amerio

Ardito

Bertuccio

Costa

d’Errico

Gentile

Odone

Vigezzi

G. P.

(2025). Impact of retirement transition on health, well-being and health behaviours: Critical insights from an overview of reviews. Social Science & Medicine, 375, Article 118049. https://doi.org/10.1016/j.socscimed.2025.118049

49.

Wahrendorf

Dragano

Siegrist

(2013). Social position, work stress, and retirement intentions: A study with older employees from 11 European countries. European Sociological Review, 29(4), 792–802. https://doi.org/10.1093/esr/jcs058

50.

Wang

Shi

(2014). Psychological research on retirement. Annual Review of Psychology, 65(1), 209–233. https://doi.org/10.1146/annurev-psych-010213-115131

51.

Wilkie

R. Z.

Choi

E. Y.

Farina

M. P.

Lee

Ailshire

J. A.

(2024). Cross national patterns in educational inequalities in functional limitations among middle aged and older adults at two time points. SSM - Population Health, 28, Article 101725. https://doi.org/10.1016/j.ssmph.2024.101725

52.

Won

C. W.

Yang

K. Y.

Rho

Y. G.

Kim

S. Y.

Lee

E. J.

Yoon

J. L.

Cho

K. H.

Shin

H. C.

Cho

B. R.

J. R.

Yoon

D. K.

Lee

H. S.

Lee

Y. S.

(2002). The development of Korean activities of daily Living (K-ADL) and Korean instrumental activities of daily Living (K-IADL) scale. Journal of the Korean Geriatrics Society, 6(2), 107–120.

53.

Yang

Lee

Lynch

Wilkie

van Tongeren

Kang

M.-Y.

(2025). Educational and gender disparities in healthy working life expectancy among middle-aged and older adults in South Korea. Occupational and Environmental Medicine, 82(6), 293–299. https://doi.org/10.1136/oemed-2025-110239

54.

Yoo

J.-I.

Choi

Y.-C.

(2017). Mean hand grip strength and cut-off value for sarcopenia in Korean adults using KNHANES VI. Journal of Korean Medical Science, 32(5), 868–872. https://doi.org/10.3346/jkms.2017.32.5.868

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.

0.27 MB

0.00 MB