Socioeconomic disparities in age at entry into Korea’s Long-Term Care Insurance: a descriptive analysis of 1.07 million adults, 2008–2022

Abstract

Background

This study examined how individual socioeconomic and health characteristics shape the timing of entry into Korea’s Long-Term Care Insurance (LTCI) system by comparing the ages at application and approval.

Methods

Using the National Health Information Database, a nationally representative Korean claims dataset, we analyzed data from 1,068,669 adults aged ≥ 41 years who were followed from 2008 to 2022. This exploratory descriptive study used administrative records to determine ages at LTCI application and approval. Associations with gender, disability, income level, insurance type, and multimorbidity were assessed using descriptive statistics, bivariate comparisons (t-tests, F-tests), and kernel density plots to visualize distributional patterns across subgroups.

Results

The mean age at LTCI application was 79.98 years, and the mean age at approval was 80.95 years, indicating that most beneficiaries enter the system well beyond the initial eligibility age of 65 years. Women, people without disabilities, and individuals with fewer chronic conditions tended to enter and be approved later, whereas individuals with greater disease burden entered earlier and were typically approved earlier. Both the lowest- and highest National Health Insurance contribution groups showed later entry than middle-income groups, while Medical Aid beneficiaries entered earlier than those covered by employment-based or region-based insurance.

Conclusions

Among adults who survive to older ages, LTCI applications and approvals tend to occur well beyond the initial eligibility age of 65 years. Differences in mean ages at entry across subgroups may reflect both variation in help-seeking and differential survival, and our descriptive comparisons cannot disentangle these mechanisms. These patterns provide baseline evidence on ages at entry among survivors and are consistent with possible socioeconomic and informational barriers. Targeted outreach, strengthened health literacy, and more responsive administrative support are therefore best viewed as plausible policy responses and as priorities for future analytic research that explicitly accounts for competing mortality.

Clinical trial number

Not applicable.

Introduction

South Korea is one of the most rapidly aging countries in the world. As of September 2025, the proportion of people in Korea aged ≥ 65 years was about 20.8% [1], with projections suggesting an estimated increase to 50.6% by 2052 [2]. In response to demographic and cultural changes, Korea adopted the Long-Term Care Insurance (LTCI) program in 2008, when people aged ≥ 65 years accounted for about 10% of the population [3, 4]. The LTCI program aimed to socialize care through social insurance, reducing the caregiving burden and improving the quality of life of families caring for frail older adults [4, 5]. Korea’s LTCI program is a mandatory, government-sponsored social insurance scheme that provides in-kind and cash benefits to older adults with substantial limitations in activities of daily living, as well as to some younger adults with severe disabilities. Eligibility is based on a standardized needs assessment, not on income level, and approved beneficiaries receive services such as home-based care, day-care, and institutional care. In parallel, the National Health Insurance (NHI) system covers almost the entire population through employment-based or region-based (generally include the self-employed and others who are not covered through an employer plan) enrollment, while the separate Medical Aid program provides tax-funded coverage for the lowest-income groups who are unable to pay NHI contributions.

The LTCI system has expanded steadily and demonstrated favorable outcomes. In 2013, 380,718 people (6.0% of older adults) were eligible beneficiaries, and 86.9% reported being delighted with LTCI [4]. By July 2025, 1,518,516 individuals had applied for LTCI [6], of whom 1,200,782 were approved, and 135,134 were deemed ineligible. However, previous studies have shown a large gap between the need for care and eligibility for LTCI, which is reflected in low LTCI application rates. One barrier is the requirement that individuals applying for LTCI benefits be well-informed about LTCI, which has disadvantaged older adults who are not well-informed. Another barrier is the uneven distribution of LTC hospitals, nursing homes, and home care agencies [5].

The LTCI application requests information on health status, activity limitations, disability, and insurance coverage. Meanwhile, demand for LTCI and the volume of services received are linked to sex, marital status, activities of daily living, cognitive impairment, and the availability of an informal caregiver [4]. To date, access to LTCI has rarely been examined, and most previous studies have focused on individual factors rather than on family or environmental factors [4]. Furthermore, studies examining accessibility have assessed access at the time of entry, without considering the age at which the LTCI application and approval occurred [4].

There are several important reasons to consider the potential impact of discrepancies between the ages at the time of LTCI application and approval. First, unlike NHI, the LTCI services provide benefits only after a formal application and approval process [7]. As such, there may be a discrepancy between the application and approval ages that can be extensive in some cases. Care deficit, or care poverty, can arise from insufficient, inadequate, or ineffective provision of LTCI services in addressing the diverse needs of older adults [8]. Moreover, such delays can accelerate disease progression and impose considerable financial and systemic burdens [8–10]. Second, accessibility to healthcare services is closely linked to equity. Socioeconomic, demographic, and historical factors can lead to disparities in access to care [11], and similar mechanisms have been shown to shape long-term care wait times and delays in institutional admission in other high-income settings [12]. Identifying these determinants is essential to enhance equity [11]. These findings can help prevent vulnerable populations from experiencing disproportionate delays in LTCI services. Finally, access to LTCI services is closely related to a strong social safety net, which aims to reduce the caregiving burden by socializing care [4]. Clarifying the discrepancy between application and approval, along with the determinants of each, can inform supply–demand planning, resource allocation, and long-term policy design.

In this study, we distinguish between three related but distinct concepts: (1) the statutory eligibility age for LTCI (typically ≥ 65 years or younger with severe disability), (2) the onset of long-term care need, which reflects functional decline and is not directly observed in our data, and (3) the actual ages at which individuals apply for and are approved for LTCI. Our analysis focuses on the latter, describing when people enter the LTCI system, rather than when care needs first arise.

Therefore, this study aimed to examine and compare the ages at which LTCI applications and approvals occur in Korea, using a large, nationally representative claims dataset. By focusing on descriptive timing patterns rather than on individual waiting times or causal effects, we sought to document how the timing of LTCI entry varies across key sociodemographic and health-related characteristics. These descriptive findings provide baseline evidence for monitoring equity in LTCI access and for generating hypotheses about potential bottlenecks in application and approval processes that can be examined in future analytic research. Given that many high-income countries face similar challenges in balancing universal eligibility for long-term care with timely access, our findings may inform policy design and equity monitoring strategies in other rapidly aging societies.

Methods

Data and subjects

This study uses the National Health Information Database (NHID), a large dataset from the National Health Insurance Service (NHIS) of Korea, compiled from reimbursement and claims data related to healthcare and long-term care services, including general health examinations, provided to all Korean adults. Access to the NHID is available to users approved by the NHIS.

We included data from a representative sample of 1,068,669 Korean adults—5% of the population who were ≥ 41 years old on the first day of 2008—and followed their records up to the last day of 2022. We included adults who were ≥ 41 years on January 1, 2008, to ensure that a substantial proportion of the cohort would reach age ≥ 65 years, the standard eligibility threshold for LTCI, during the 15-year follow-up period through 2022. This lower age cutoff allowed us to capture incident LTCI applications and approvals across the typical window of old-age onset while maintaining a large, nationally representative baseline sample. The data were initially recorded annually and, for this study, merged using personal identification key codes to create a wide-form longitudinal dataset for all participants. These data were initially prepared for a study that aimed to understand the association between healthcare use and long-term care use across individuals’ life courses, but we used them to explore factors associated with the age at LTCI application and the age at LTCI approval by the NHIS.

We acknowledge that the youngest cohort members (aged 41 years in 2008) would have reached only their mid-50s by 2022 and thus would have remained below the usual LTCI eligibility age throughout the follow-up period. These individuals contribute to the baseline description of the cohort but are unlikely to appear among LTCI applicants or approved beneficiaries during the study period, which introduces left truncation and immortal-time selection at younger ages.

This study was approved by the Institutional Review Board of Seoul Women’s University (SWU IRB-2023-36), which waived the need for informed consent because the data were secondary and all personally identifiable information was anonymized.

Variables and measurements

Data for most variables were sourced from individuals’ healthcare records collected by the NHIS. Since 2008, when LTC services were institutionalized and combined with the NHIS, every individual aged ≥ 65 years has been asked whether they have applied for LTCI and, if so, when the NHIS approved the application. Using their answers to these questions, we calculated the ages at which the LTCI application and approval occurred, respectively. Because the birth year, without month or date, was the only information included in the NHID, we calculated the ages at LTCI application and approval using only the year data.

Gender (female = 0; male = 1) and disability status (0 = no disability; 1 = disability) were obtained from the NHIS data. Age in years was calculated using the individual’s date-of-birth record, with no month or day provided. Household income was proxied by NHI contribution level, which the NHID provides in 20 quantiles based on monthly contributions. Quantile 0 corresponds to Medical Aid beneficiaries, who do not pay NHI contributions. Quantiles 1–2 represent the lowest-contribution groups, whereas quantiles 19–20 represent the highest-contribution groups. The NHID does not permit the release of exact monetary thresholds for each quantile for confidentiality reasons, so we cannot report precise contribution amounts or income ranges; however, these quantiles correspond to the bottom and top deciles of the contribution distribution in the Korean NHI system. The types of health insurance were categorized as region-based, employment-based, and Medical Aid beneficiary. The number of chronic diseases, including hypertensive diseases, diabetes mellitus, mental and behavioral disorders, respiratory tuberculosis, cerebrovascular diseases, nervous system diseases, neoplasms, disorders of the thyroid gland, diseases of the liver, chronic kidney diseases, and arthropathies, before LTCI approval was coded as 0, 1, 2, or 3 or more. Chronic diseases were identified from NHID claims using ICD-10-equivalent diagnosis codes and were not available as pre-defined variables. We constructed binary indicators for each condition and then summed them to create a count of chronic diseases. For each individual, we considered diagnoses recorded through the year of LTCI approval (or, for those who applied but were not yet approved, through the last observed year), implying that the observation window for chronic disease accumulation varied with the timing of LTCI entry.

Statistical analysis

A set of descriptive analyses was used to characterize the sample. To explore the association between the ages at LTCI application and approval and various relevant factors, a series of bivariate analyses, including t-tests for binary factors and F-tests for multinomial factors, was conducted. Also presented were a series of graphs showing the distribution of LTCI application and approval ages by potentially associated factors. Stata 17 (StataCorp LP, College Station, TX, USA) was used to manipulate and analyze the data. In addition to tabular summaries, we constructed kernel density plots of ages at application and approval, using key characteristics to visualize the distributions’ shapes and spreads. These figures were intended to complement mean comparisons by highlighting shifts in the entire age distribution across sociodemographic and health-related subgroups. Given the very large sample size, even small differences in mean ages were statistically significant. We therefore report p-values but emphasize the magnitude and patterns of group differences in mean ages (in years) and the shape of the distributions shown in kernel density plots, rather than focusing on statistical significance alone.

We inspected histograms and kernel density plots of ages at application and approval, which showed approximately unimodal, bell‑shaped distributions without pronounced skewness. Given these distributional properties and the very large sample size (n ≈ 122,000–145,000), t‑tests and F‑tests are expected to perform well by the central limit theorem, and we therefore retained these parametric tests as our main analytic tools.

Table 1 describes the full baseline cohort of 1,068,669 adults aged ≥ 41 years in 2008. In contrast, Table 2; Figs. 1 and 2 are restricted to individuals who applied for LTCI (n = 145,057) or were approved for LTCI (n = 122,208) at any point during follow-up, because ages at application and approval are only defined for these subgroups. Overall, missing data were rare. For example, the NHI contribution quantile was missing for 15,321 individuals (1.4% of the baseline cohort), and the chronic-disease count was valid for all individuals. Individuals with missing values were excluded from analyses that used these variables (listwise deletion), resulting in slightly smaller denominators for some rows in Table 2. We did not use imputation because the proportion of missingness was small, and these variables were used primarily for descriptive purposes. This study was designed as an exploratory descriptive analysis to characterize patterns in ages at LTCI application and approval across key sociodemographic and health-related subgroups, rather than to estimate causal effects or fully adjusted associations. In this manuscript, ‘earlier application’ and ‘earlier approval’ refer to entering or being approved for LTCI at a younger chronological age, not to shorter processing times between application and approval.

Table 1.

Sample characteristics recorded in 2008

Variable(Years in 2008)	Total
	(N = 1,068,669)
	n	%
Gender
Male	511,653	47.88
Female	557,016	52.12
Disability status
Not disability	974,654	91.20
Disability	94,015	8.80
NHI contribution (quantile) (n = 1,053,348)
0	52,266	4.96
1	40,523	3.85
2	35,558	3.38
3	43,299	4.11
4	28,130	2.67
5	32,752	3.11
6	36,182	3.43
7	37,038	3.52
8	41,120	3.90
9	38,923	3.70
10	42,849	4.07
11	44,751	4.25
12	46,364	4.40
13	49,360	4.69
14	52,134	4.95
15	56,030	5.32
16	61,553	5.84
17	69,088	6.56
18	77,432	7.35
19	82,514	7.83
20	85,482	8.12
Health insurance type
Region-based insured	424,717	39.74
Employment-based insured	591,686	55.37
Medical Aid	52,266	4.89
Chronic diseases
0	510,650	47.78
1	277,162	25.94
2	176,928	16.56
≥ 3	103,929	9.73
	Mean	SD
Age	56.14	11.62

NHI, National Health Insurance; SD, standard deviation

Table 2.

The ages at long-term care insurance application and approval stratified by their characteristics

Variable	LTCI application			LTCI approval
	(n = 145,057)			(n = 122,208)
	Mean	SD	F (T)	Mean	SD	F (T)
Total	79.98	7.14		80.95	7.14
Gender			−47.60***			−53.44***
Male	78.74	7.26		79.41	7.29
Female	80.61	6.99		81.71	6.94
Disability status			7910.56***	(n = 122,146)		7134.64***
Not disability	81.04	6.82		82.08	6.78
Disability	77.48	7.25		78.46	7.28
NHI contribution (quantile)	(n = 142,343)		140.58***	(n = 119,928)		67.63***
0	78.04	7.46		79.75	7.60
1	81.46	7.20		82.54	7.12
2	80.26	7.34		81.32	7.20
3	79.93	7.47		80.99	7.47
4	79.94	7.32		80.99	7.41
5	80.34	7.40		81.21	7.33
6	79.96	7.31		80.94	7.24
7	79.97	7.31		81.10	7.21
8	80.05	7.23		80.84	7.26
9	80.04	7.22		81.11	7.36
10	80.06	7.20		80.76	7.25
11	79.79	7.34		80.55	7.27
12	79.95	7.33		80.82	7.30
13	79.83	7.38		80.67	7.39
14	79.80	7.46		80.60	7.46
15	79.60	7.28		80.44	7.28
16	79.69	7.23		80.46	7.34
17	79.75	6.95		80.44	7.05
18	80.36	6.61		81.03	6.68
19	80.98	6.39		81.69	6.46
20	81.15	6.34		81.78	6.37
Health insurance type	(n = 145,037)		943.73***	(n = 122,178)		270.06***
Region-based insured	80.25	7.21		81.16	7.23
Employment-based insured	80.35	6.94		81.13	6.97
Medical Aid	78.04	7.46		79.75	7.60
Chronic diseases			284.22***			245.36***
0	82.44	8.26		83.80	8.35
1	80.57	7.96		81.92	8.17
2	79.99	7.52		81.07	7.82
≥ 3	79.79	6.88		80.77	6.89

LTCI, long-term care insurance; NHI, National Health Insurance; SD, standard deviation

*P < 0.05, **P < 0.01, ***P < 0.001

Fig. 1.

Distribution of long-term care insurance application ages by sociodemographic and health-related characteristics. Kernel density plots show age distributions for key subgroups. Left-shifted curves (earlier ages of entry) are observed among individuals with disabilities, multimorbidity, and Medical Aid coverage. Right-shifted curves (later entry) are observed among women, individuals without disabilities, and those in the lowest and highest income quantiles. Data from the National Health Information Database, Korea, 2008–2022 (N = 145,057)

Fig. 2.

Distribution of long-term care insurance approval ages by sociodemographic and health-related characteristics. Kernel density plots show age distributions for key subgroups. Left-shifted curves (earlier ages of entry) are observed among individuals with disabilities, multimorbidity, and Medical Aid coverage. Right-shifted curves (later entry) are observed among women, individuals without disabilities, and those in the lowest and highest income quantiles. Data from the National Health Information Database, Korea, 2008–2022 (N = 122,208)

All analyses of age at LTCI application and approval are conditioned on survival to the time of LTCI entry; individuals who died before applying for or being approved for LTCI are not included in these denominators. Consequently, mean ages at entry reflect both differences in the timing of help-seeking among survivors and differential survival across sociodemographic and health-related subgroups. Because access to the analytic NHID dataset expired at the end of 2025, we were unable to perform a competing-risk survival analysis treating death as a competing event. Our mean-age comparisons should therefore be interpreted as patterns in age at entry among survivors, rather than as conclusive evidence of differential application delay.

Results

The basic characteristics of the 2008 sample are summarized in Table 1. Men (47.9%) and women (52.1%) were almost evenly represented, although women accounted for a larger proportion, reflecting gender differences in longevity. About 8.8% of the sample reported any disability. Regarding the type of health insurance, almost one-twentieth (4.9%) were Medical Aid beneficiaries, 55.4% were insured through their employment, and 39.7% were insured through their region. Nearly half of the sample (47.8%) reported no chronic diseases, while 25.9% had one, 16.6% had two, and 9.7% had at least three (Table 1).

Table 2 presents unadjusted differences in age at LTCI application and approval across sociodemographic and health-related characteristics. The average age at the time of LTCI application was 79.98 years (standard deviation = 7.14), and the average age at the time of LTCI approval was 80.95 years (standard deviation = 7.14). In terms of LTCI application age, women (t = 47.6, P < 0.001) and people without disabilities (t = 7,910.6, P < 0.001) applied later than their counterparts. Women applied at an average age of 80.61 years, compared with 78.74 years among men, a difference of about 1.9 years. Approval ages showed a similar pattern (81.71 vs. 79.41 years, difference ≈ 2.3 years). Individuals without disabilities applied at a mean age of 81.04 years, whereas those with disabilities applied at 77.48 years, a gap of approximately 3.6 years, and approval ages differed by a similar margin. Among NHI contributors (quantiles 1–20), individuals at both the lowest and highest contribution quantiles tended to apply for and be approved for LTCI at older ages than those in the middle of the contribution distribution (F(20, 142,322) = 140.6, P < 0.001). By contrast, Medical Aid beneficiaries (contribution group 0) applied for and were approved for LTCI at younger ages than most contribution-paying groups, suggesting earlier entry among this means-tested, low-income assistance group. Region-based insureds and employment-based insureds applied later for LTCI than Medical Aid beneficiaries did (F(2, 145,034) = 943.7, P < 0.001). Finally, the number of chronic diseases appeared to be associated with the age at LTCI application (F(3, 145,053) = 284.2, P < 0.001), with the application time trending earlier as the number of chronic diseases increased.

Figure 1 illustrates the distribution of age at LTCI application across key sociodemographic and health-related subgroups. Compared with their counterparts, individuals with disabilities, multiple chronic diseases, or Medical Aid coverage show application-age curves that are shifted toward younger ages and have heavier lower tails, indicating earlier entry among groups with greater functional and disease burdens. In contrast, women (compared with men), those without disabilities (compared with those with disabilities), and those at both the lowest and highest ends of the NHI contribution quantiles (excluding Medical Aid beneficiaries; compared with middle-contribution groups) exhibit right-shifted, more concentrated curves, suggesting that these groups tend to enter LTCI at older ages (Fig. 1).

In additional exploratory analyses, we stratified these application-age distributions by gender within disability and chronic disease strata and observed qualitatively similar left-shifts among individuals with disabilities and higher multimorbidity for both men and women, suggesting that these patterns are robust across gender.

The age of LTCI approval showed similar patterns of association with the exact age of LTCI application. Gender (t = 53.4, P < 0.001), disability status (t = 7,134.6, P < 0.001), NHI contribution distribution (F(20, 119,907) = 67.63, P < 0.001), type of health insurance (F(2, 122,175) = 270.06, P < 0.001), and the number of chronic diseases (F(3, 122,204) = 245.4, P < 0.001) were all associated with the age at LTCI approval (Table 2).

Figure 2 shows similar patterns for the age at LTCI approval. Approval-age distributions for individuals with disabilities, multimorbidity, and Medical Aid coverage are left-shifted relative to other groups, reflecting earlier approval among those with higher care needs and fewer financial barriers. In this manuscript, ‘earlier approval’ refers to approval at a younger age, rather than a shorter interval between application and approval. By contrast, women (compared with men), those without disabilities (compared with those with disabilities), and people at the extremes of the contribution distribution (excluding Medical Aid beneficiaries; compared with middle-contribution groups) have approval-age distributions that are skewed toward older ages, consistent with later entry into the LTCI system despite formal eligibility (Fig. 2).

Discussion

This descriptive study investigated how the ages at LTCI application and approval differ across sociodemographic and health-related groups in Korea, using a large, nationally representative sample of adults followed from 2008 to 2022. The mean age at LTCI application was 79.98 years, and that at approval was 80.95 years. The beneficiaries of the LTCI include older adults aged ≥ 65 years and younger adults (< 65 years) with chronic illnesses or disabilities [7]. Nevertheless, the average age at which people apply for and receive approval for LTCI is substantially higher. This suggests that long-term care needs tend to emerge well beyond the initial eligibility age of 65 years. Consistent with this finding, the 2022 Long-Term Care Status Survey in Korea reported that 80.4% of LTCI users are aged ≥ 75 years, and 81.4% of non-users are also in this age group [13]. Previous studies have also shown that, despite being eligible for LTCI, older adults have not entered the LTCI system [4]. Park (2021) reported that only 16.4% of individuals in need of long-term care applied for LTCI. Furthermore, a prior study showed that older adults tend to seek LTCI after deterioration in functional status. They also indicate that the administrative capacity of residential areas, potentially linked to a greater ability to provide information and support, was associated with LTCI application [4]. Similar patterns of unmet or delayed long-term care use have been documented in European welfare states, where ‘care poverty’ persists despite formal coverage [8, 10]. Together, the delayed LTCI application and approval may reflect the need to improve awareness and outreach to encourage earlier engagement with the LTCI system.

It is important to emphasize that our analysis focuses on the ages at which individuals apply for and are approved for LTCI, rather than on the interval between these events or the age at which long-term care needs first arise. Because the NHID does not contain exact application and approval dates or direct measures of the onset of functional need, we cannot construct individual waiting times or determine when care needs begin at the person level. Our findings should therefore be interpreted as descriptive timing patterns along the LTCI pathway, not as precise estimates of administrative delay or the onset of unmet need.

The approximately 1-year difference between mean application and approval ages indicates that some time elapses between these stages of the LTCI process. However, our data do not include exact application and approval dates, so we cannot directly measure individual waiting times or quantify group-specific delays. This average gap may reflect administrative processing and assessment steps required for LTCI approval, but we cannot determine from these data whether specific structural barriers are responsible for longer waiting times or for differences between groups [12, 14]. Older adults become eligible through an LTCI approval process [15]. They have to pass a standardized needs assessment that evaluates physical, cognitive, behavioral, nursing-care, and rehabilitative characteristics [16]. Before the revision of the eligibility criteria, individuals with dementia who did not have severe physical limitations were ineligible for LTCI. Their families suffered from the absence of formal care [16]. Similarly, delays in the eligibility process may cause substantial distress for families who provide care without formal support during the waiting period.

An important interpretive implication of this design is that the ages at LTCI application and approval we observe are conditional on survival to entry. Longer-living groups, such as women, individuals without disabilities, and higher-income groups, have more opportunities to enter LTCI at older ages, whereas individuals in higher-mortality groups may die before they can apply or be approved. As a result, later observed entry among longer-living subgroups and earlier observed entry among higher-mortality subgroups may both be shaped, in part, by differential survival, rather than solely by differences in help-seeking behavior, awareness, or access barriers. Our descriptive comparisons of mean ages at entry cannot disentangle these mechanisms, and the gradients we document should be interpreted as patterns in ages at entry among survivors rather than as unbiased measures of when all individuals in each subgroup would enter LTCI in the absence of competing mortality.

The factors associated with the ages at LTCI application and approval showed similar patterns. On average, women applied for LTCI about 1.9 years later than men and were approved about 2.3 years later, reflecting older ages at entry rather than longer processing times between application and approval. Gender differences in various social contexts may help explain part of this gap. Compared to women, men experienced care deficits more often, especially older men without a spouse. Everyday activities such as cooking, meal preparation, and laundry appeared challenging for men [17]. The study further noted that older men often struggle to find caregivers beyond their spouses, both within and outside the family. However, older women tend to rely more on family members and delay using formal care services to avoid an economic burden on their family [17]. It must also be considered that women’s life expectancy in Korea is significantly longer than men’s (approximately four years, based on age 65 as of 2024) [18]. This is a factor that mechanically increases the likelihood that women will survive to the age at which LTCI enrollment begins. Therefore, it appears that both differences in survival rates and gender-based care norms influence women’s late participation, and our descriptive data do not allow us to determine the relative importance of these mechanisms.

Individuals without disabilities showed a later application and approval of LTCI than those with disabilities, with approximately a 3-year age gap. In addition, individuals with multiple chronic diseases tended to be younger at application and approval of LTCI. These results can be explained by the LTCI eligibility assessment process, which prioritizes functional decline, disease burden, and care needs [7]. During the selection process, applicants undergo comprehensive multidimensional evaluations that include functional limitations, cognitive impairment, behavioral symptoms, nursing and rehabilitation needs, sensory deficits, environmental risks, and comorbid illnesses [7]. Additionally, previous studies have shown that multimorbidity is strongly associated with functional decline, defined as a progressive deterioration in self-care abilities and independence [19]. These procedures and functional decline may enable earlier eligibility for individuals with greater disease burdens or disability.

Individuals at the lowest and highest ends of the NHI contribution distribution had later ages at both LTCI application and approval than those in the middle of the distribution. These findings suggest that middle-income groups may access LTC services more promptly, whereas very low-income and very high-income contributors enter at older ages. Low socioeconomic status is associated with limited access to healthcare and formal information sources, which can delay timely care [20–22], but differential survival may also play a role. Among low-income contributors with substantial health burdens, selective survival of comparatively robust individuals into older ages could shift the observed entry ages upward, while high-income groups may have lower need for LTC services and support. The asymmetric pattern we observe at the extremes of the contribution distribution is therefore likely to reflect a combination of informational barriers, differences in need, and survival selection, which our descriptive design cannot disentangle.

Medical Aid beneficiaries were more likely to apply for and be approved for LTCI than those covered by National Health Insurance, potentially because of the absence of copayment requirements. The Medical Aid program is a component of Korea’s national social security system, designed to support the lowest-income group, which primarily depends on government financial assistance [20]. According to Article 40 of the LTCI Act in Korea, Medical Aid beneficiaries are fully exempt from copayments, thereby reducing financial barriers to accessing LTC services. However, previous studies have suggested that Medical Aid beneficiaries may have higher levels of healthcare use, possibly reflecting lower financial barriers to service use in the context of minimal or zero co-payments [20].

The graphical distributions in Figs. 1 and 2, together with the mean differences in Table 2, suggest that earlier entry into LTCI is concentrated among individuals with disabilities, those with greater multimorbidity burdens, and those with Medical Aid coverage. In contrast, later entry is more common among women, those without disabilities, and both the lowest- and highest-income contribution groups.

These gradients may also be influenced by left truncation and survival selection. Because our cohort excludes individuals who died or developed care needs before age 41 in 2008, and because higher-mortality subgroups are more likely to be removed from the risk set before reaching older ages, younger observed ages at LTCI entry in groups such as Medical Aid beneficiaries and individuals with disabilities may partly reflect differential survival rather than purely earlier help-seeking. Similarly, later observed entry among longer-living groups may reflect their greater opportunity to survive to ages at which LTCI entry is observed.

Although our analyses are intentionally descriptive and unadjusted, documenting these systematic patterns in application and approval ages provides essential baseline evidence on when older adults actually enter the LTCI system and which groups experience earlier or later entry. Such descriptive mapping is a necessary first step for monitoring equity in LTCI access and for motivating more targeted, hypothesis-driven analytic studies.

The patterns documented in this study suggest several plausible policy priorities. First, targeted outreach and information campaigns could focus on older adults who appear to enter LTCI at later ages—such as women and individuals without disabilities—as well as low‑income groups, recognizing that our data cannot distinguish whether their later observed entry reflects delayed application, differential survival, or both. Community‑based health literacy programs, delivered through primary care clinics and senior centers, may help improve awareness of LTCI benefits and application procedures, particularly among groups that previous studies identify as facing informational and financial barriers to timely care [11, 20–22]. Second, proactive case‑finding mechanisms, such as alerts for older adults with multimorbidity or functional decline during routine health check‑ups, could encourage earlier consideration of LTCI before the onset of severe disability. Third, reducing administrative burdens in the application and approval process—such as simplifying documentation requirements and expanding mobile assessment services in rural areas—may help shorten the interval between application and approval and lessen the strain on family caregivers during this period. Although our study does not directly measure awareness, information barriers, or health literacy, these recommendations are motivated by the descriptive timing disparities we document and by prior evidence on barriers to access, and they should be viewed as hypotheses for future analytic research rather than as interventions definitively mandated by our data. Similar descriptive mapping of ages at entry could inform equity monitoring in other long‑term care systems, but system‑specific analyses will be needed to identify the most appropriate policy responses.

This study has several limitations. First, because the NHID was constructed for administrative and reimbursement purposes, it lacks detailed information on family caregiving resources, regional service availability, and other contextual factors that may shape both care needs and application decisions. Second, our analyses were intentionally descriptive and relied on bivariate comparisons without multivariable adjustment for potential confounders, so the observed associations should not be interpreted as causal. Third, although there is an average difference of about one year between mean ages at application and approval, we could not construct a separate waiting-time variable or examine its distribution across subgroups. Our inferences about the interval between these stages are therefore indirect and based on aggregate age differences rather than direct measures of delay. Fourth, the ages at LTCI application and approval that we analyze are observed only for individuals who survive long enough to enter the LTCI system, and our cohort is left-truncated at age 41 in 2008. Individuals who developed care needs or died at younger ages are structurally excluded from the analytic sample, and these exclusions are likely to be more pronounced in higher-mortality subgroups such as those with disabilities, multiple chronic conditions, or very low income. Together, survival conditioning and left truncation may play an important role in shaping the observed gradients in entry age, including the asymmetric pattern at the extremes of the contribution distribution. Our findings should therefore be interpreted as describing ages at entry among survivors in each subgroup, rather than as unbiased estimates of when all individuals in each subgroup would enter LTCI in the absence of competing mortality and left truncation.

Despite these limitations, the use of nationally representative data and the consistency of patterns across application and approval ages highlight systematic differences in ages at LTCI entry along socioeconomic and health gradients, rather than random variation. The descriptive patterns identified here delineate the principal axes along which LTCI entry is stratified and can inform the selection of key variables, interactions, and policy-relevant subgroups for future multivariable or causal analyses. Future research employing longitudinal designs with richer contextual data, more granular timestamps, and multivariable or causal inference methods is needed to disentangle the joint effects of socioeconomic position, health status, local service supply, and explicit delays between application and approval on the timing of LTCI entry and subsequent health and caregiving outcomes.

Conclusion

This study showed that, among adults who survive to older ages, observed entry into LTCI in Korea tends to occur well beyond the initial eligibility age of 65 years. The similarity in the determinants of application and approval ages is consistent with a pattern in which individuals often enter LTCI after substantial functional decline; however, we cannot distinguish this explanation from the alternative that differential survival across groups shifts the ages at which entry is observed. Structural features of eligibility and financing processes may contribute to delayed access and shift burdens to families, but our descriptive data do not allow us to identify specific barriers or to quantify their effects. Targeted outreach, improved health literacy, and responsive administrative support should therefore be viewed as plausible policy responses to the timing disparities documented here, and as priorities for future analytic research that explicitly accounts for competing mortality and left truncation.

Abbreviations

LTCI

Long-term care insurance

NHI

National Health Insurance

NHID

National Health Information Database

NHIS

National Health Insurance Service

Author contributions

Conceptualization: JWN, JK, and YDK. Methods: JWN, JK, and YDK. Formal Analysis: JK. Writing – Original Draft: JWN, JK, and MHH. Writing – Review & Editing: YDK. Project Administration: MHH. Funding Acquisition: JK. All authors read and approved the final manuscript.

Funding

This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2025S1A5A2A03016626).

Data availability

The data supporting this study are available from the National Health Insurance Service (NHIS) of Korea through the National Health Information Database (NHID), but access is restricted. These data were used under license for the current study and are not publicly available. Researchers can request access to the NHID by submitting a research proposal to the NHIS (https://nhiss.nhis.or.kr). Detailed documentation on access procedures is available from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of Seoul Women’s University (SWU IRB-2023-36) and was conducted in accordance with the Declaration of Helsinki and the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. The IRB explicitly waived the requirement to obtain informed consent because this study used de-identified secondary data from an administrative database, in accordance with Korean regulations governing research using routinely collected health data. The authors obtained all necessary administrative permissions from the National Health Insurance Service to access the data before conducting this study. All data handling and analysis procedures adhered to the principles outlined in the WMA Declaration of Taipei on Ethical Considerations Regarding Health Databases and Biobanks.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.