1. Background
The vulnerable groups are special groups with poverty, low quality of life, and vulnerability of endurance, and are as such defined in terms of poor social status and social circumstances. A direct cause of vulnerability is the lack of personal ability (natural or acquired incapacity), and the deep reason is the social structure flaw [
1], namely the unfair social system arrangement. China has the world’s largest elderly population, and faces a rapidly aging process. The proportion of population over 60 years old in China has increased from 10% in 2000 to 18.7% in 2020, and those 65 years old and above in 2020 had reached 190 million [
2]. With aging, the proportion of vulnerable older adults, such as solitary individuals [
3] and oldest old [
4] will continue to rise [
5]. According to preliminary analysis of “The Fourth Sample Survey on the Living Conditions of China’s Urban and Rural Older Persons” by China National Commission on Aging, 59.4% of older adults had economic difficulties, 47.0% had health problems in varying degrees, and more than 40 million disabled older adults need priority assistance. The distribution of vulnerable older adults was unbalanced between urban and rural areas and varies among regions, with the proportion of vulnerable older adults significantly higher in rural areas than in urban areas, and higher in the central and western regions than in the eastern regions [
6]. China’s antipoverty program has eradicated extreme poverty in 2020 [
7] and significantly improved economic conditions of older adults, but they are still at risk of becoming poor due to illness and old age. Therefore, the improvement of public health policies mainly focusing on the medical care system, and the improvement of social support policies related to mental health and active aging, are core issues in the next stage [
8]. Social changes such as population aging and population mobility [
9] have made older adults the high-risk social group. To effectively improve relevant policies, characteristics of older adults and differences between supply and demand of current public policies in different regions should be fully considered. However, the quantitative analysis on particularity, internal differences, and distribution characteristics of vulnerable older adults are rare, which is detrimental to consolidate the achievements of anti-poverty governance and will restrict the foreseeable overall design of public health policies.
Compared with young adults, older adults have a great decline in physiological functions, are less functional in the family, and are more likely to rely on others or society for self-care and livelihood. Vulnerable older adults usually have no economic income, low political influence, high psychological pressure, and a strong sense of social alienation [
10]. There are different types of vulnerability for elderly adults. Previous studies mostly identified vulnerable older adults based on part of their vulnerable dimensions, such as individual/family characteristics, health status, economic level or others, and focused on special groups such as women, the aged, those living alone/empty nest, the disabled, the poor, and older adults who lack community services. Some studies also focused on the care needs and social assistance for older adults with “double difficulties” (difficulties in economy and self-care) from the perspective of economic and health [
11]. With aging and urbanization, the traditional vulnerable older adults, such as women, the aged, living alone/empty nest, the disabled, and the poor older adults, not only continue to expand, but also show new characteristics and superimposing vulnerability, making it more difficult to design a diversified accurate old-age assistance system. Accordingly, it is necessary to identify vulnerable older adults more precisely, which helps to understand their demand from a policy perspective.
Research on social vulnerability have found that social transition in China has created new vulnerable groups [
12]. Among them, older adults, such as left behind women [
13] and childless individuals [
14], have to suffer higher vulnerability. Most of the research focus on conceptual definition and situation analysis of some special older groups in a certain area, such as left behind women in rural China [
13] or urban older adults in poverty [
15], trying to put forward reasonable proposals through sufficient qualitative description. The urban and rural older adults are distinguished by hukou, namely the household registration type, which induces the differences in social security, welfare system, and medical resources between urban and rural residents. Accordingly, it is necessary to conduct a comparative analysis of the older adults with different hukou types. While research on overall elder adults with rural or urban hukou by using quantitative methods to measure social vulnerability are seldom seen, this paper aims to construct a comprehensive identification system for the vulnerable older adults, which could accurately reflect basic characteristics of older vulnerable groups in urban and rural China. It is helpful to reduce the negative impact of health and support deficit on vulnerable older adults, which is the key issue after the eradication of extreme poverty [
7]. It also provides practical evidence and policy recommendations for health and happiness improvement of older adults in the process of population aging.
4. Identification of Vulnerable Older Adults
4.1. Identifying Variables
The measurement of social vulnerability is multi-dimensional and multi-index. Existing research generally incorporate economic status, social support, and social participation. Because of the limitation of specific research purpose, most of the studies only included mental health indicators without considering the factors of physical health. However, as one of important human capital of elder adults, physical health is particularly important in measurement of vulnerability. Therefore, this paper regarded health factors and social factors as two important dimensions to identify vulnerable older adults, and attempted to construct a comprehensive index system for the identification of vulnerable older adults. The index system covers four aspects: health status, socioeconomic status, social support, and social engagement. Considering data accessibility of CLASS 2014, the following 12 variables are selected to classify and identify the vulnerable older adults (
Table 2).
Health status. Health status includes health self-assessment, Barthel index, and chronic and mental health. The health self-assessment was coded from 5 to 1 according to the level of health from high to low. Barthel index was measured by the activities of daily life (15 = no difficulty, 10 = few difficulty, 5 = some difficulty, 0 = cannot do) in 10 performing tasks. The Barthel index ranged from 0 to 100, and the lower the score, the more serious the health problem. Chronic is measured through “What chronic diseases do you have?” as a continuous variable. On the “Did you feel sad in the past week?” deficit, which includes three response categories, scores are 1 if the answer is “No”, 2 if the answer is “Sometimes”, and 3 for “often”.
Socioeconomic status. Socioeconomic status contains income, economic independence, and housing property. Income is measured through “in the past 12 months, what is your personal total income?” as a continuous variable. Economic independence is measured through “What is your main source of income?”. If the main source of older adults’ income is their or their spouse’s pension or labor payment, this variable is coded as “1”; Otherwise coded as “0”. As housing property today represents great wealth of older adults, it can reflect the socioeconomic situation of older adults, which is measured through “How many houses do you (and your spouse) have?” as a continuous variable.
Social support. Social support includes social network support and children’s economic support. Social network support includes three variables: someone that older adults meeting/contacting, discuss personal affairs, or look for help. Children’s economic support is assessed by monetary or material assistance within the past 12 months.
Social engagement. Social engagement contains community security patrols, caring for other older adults, caring for children, environmental protection, dispute mediation, chatting, volunteer service, and others. This paper measures the number of activities older adults had participated in the past 12 months as a continuous variable.
4.2. Identifying Indicators of Vulnerable Older Adults: Factor Analysis
Identifying indicators of the vulnerable older adults is studied by factor analysis method. Some multi-category variables are re-coded quantitatively and standardized, by assigning values of each option according to the degree and the direction of “social deficits”. For example, the answers of “How do you feel about your current health status? (1) Very healthy, (2) Relatively healthy, (3) General, (4) Relatively unhealthy, and (5) Very unhealthy” are assigned 1–5 respectively. In the process of standardization, the MIN-MAX range standardization method is used to unify the variable range between [0,10] intervals, eliminating the dimensional influence of the original variables, and the influence of the variation size and the numerical size.
To test data, the rationality and sampling adequacy of factor analysis, the Kaiser–Meyer–Olkin test and the Bartlett test were used. A KMO value of 0.713 (>0.5) revealed the sufficient sampling, and a significance level from the Bartlett test <0.01 indicated that the data are appropriate and useful to substantially reduce data dimension. Then, three common factors with eigenvalues >1, as the eigenvalue of those three is 2.34, 1.71, and 1.43 respectively, are extracted for the varimax-rotated analysis. The three common factors explained a tremendous 45.74% of the total variance, and it is reasonable to select them as the main factors (VF1, VF2, and VF3).
Table 3 shows that the variables with factor loading up to 0.5 is very suitable and is selected into the Factors. Specifically, VF1 is largely composed of “someone to meeting/contacting”, “someone to discuss personal affairs”, and “someone for help” which represent social support. VF2 has strong positive loading on “health self-assessment”, “Barthel index”, and “chronic”, suggesting the health status. VF3, implying the socioeconomic status, consists of the “income”, “economic independence”, and “housing property”. The social engagement weighted on leisure activities failed to form one VF.
4.3. Typologies of Vulnerable Older Adults: Latent Class Analysis
All the identification indicators of the vulnerable older adults were treated as binary variables, and the responses to each item were assigned a value of “1” if representing a deficit and “0” otherwise.
Table 4 reports distribution of items in different dimensions. Firstly, social support included three variables, persons whom older adults meet/contact, discuss personal affairs, or look for help. It was scored 0 if he/she had potential support from family members, relatives, or friends on the above aspects, and 1 if he/she did not have. Secondly, health status included health self-assessment, Barthel index, and chronic. Health self-assessment was coded as 0 = healthy, 1 = unhealthy. The Barthel index was used to determine whether older adults rely on others for life-care (0 = no, 1 = yes). Chronic was coded as 0 = no chronic disease, 1 = have chronic disease. Thirdly, economic status included income, economic independence, and housing property. Income was coded as 0 = higher than older adults’ average income, 1 = lower than older adults’ average income. Economic independence was coded as 0 = the main income resource is older adults themselves, 1 = the main income resource is not older adults themselves. Housing property was coded as 0 = owning property, 1 = no property.
We employed Mplus to analyze the response patterns generated from the cross-classification of dichotomous indicators of vulnerable older adults. The latent class model with only a single class was used as the baseline model, and then successively added the number of classes in order to determine the optimal model by checking for model fit [
46]. The goodness-of-fit measures used to select the optimal models were the Akaike information criterion (AIC), Bayesian information criterion (BIC), and sample size-adjusted BIC (aBIC). The above fitness was useful for selecting the best-fitting model among reasonable but competing models, with a smaller value of that fitness providing the better model fit [
47]. The entropy index, which ranged from 0 to 1, is often used to evaluate the accuracy of classification in LCA. When entropy was 0.6, it showed that about 20% of individuals had classification errors, while when entropy equaled 0.8, it showed that the accuracy of classification exceeded 90% [
48]. Mplus also provided the Lo–Mendell–Rubin likelihood ratio test (LMR), and the significant
p-value of the LMR indicated that the model with more classes fit significantly better [
47].
Table 5 reports the goodness-of-fit statistics for seven models of vulnerable older adults. For all seven models, the LMR were significant. But in the first six models, the BIC decreased successively with each additional class added, indicating relative improvements in model fit; while when the classification reached seven, the value of BIC increased, indicating that the six-class model was the fittest.
Table 6 shows the conditional probabilities and latent class probabilities of the six-class model of vulnerable older adults, which helps to facilitate the observation and comparison, and find out more intuitively the prominent vulnerable characteristics of different types of vulnerable older adults.
Table 7 shows the characteristics and proportion of different types of vulnerable older adults. Based on these patterns, elderly adults were divided into four levels: high vulnerability (need assist urgently), moderate vulnerability (focus group), slight vulnerability (care group), and potential vulnerability (vulnerability prevention). According to the characteristics, elderly adults were further divided into six subtypes: (1) multi-vulnerable (1.46%), (2) dual-vulnerable (12.76%), (3) support-vulnerable (1.34%), (4) economy-vulnerable (14.21%), (5) health-vulnerable (19.17%), and (6) potential-vulnerable (51.06%). Each subtype shows a specific demand for social assist, which reveals a more explicit policy implication. First, highly vulnerable older adults are poor in both health and socioeconomic status without close relatives or friends; thus, they could neither take good care of themselves nor ask someone for help. As a result, they are in urgent need of policy assistance. Second, compared with highly vulnerable older adults, moderate vulnerable ones could get some help from their family or friends, so their life is tough but still relatively sustainable. Their plight reflects shortcomings of current public policies, so they are the key objects of policy design and improvement. Third, there are three kinds of slightly vulnerable older adults facing only one kind of vulnerability, such as poor, illness, or loneliness. These older adults are not so frail currently, but may become vulnerable in the near future when they get older and if the public service resources are still limited or the social support pressure increases. Finally, potentially vulnerable older adults are relatively good in all aspects up to now, but it does not mean that they are inessential for public policy improvement. Older adults are inherently vulnerable groups compared with the young, and their vulnerability deepens as they grow older. Therefore, to design forward-thinking public service policies, their demands would be the main basis.
4.4. Characteristics of Different Types of Vulnerable Older Adults
Gender differences. Compared with males, vulnerability of female older adults is relatively high. Males, in particular, are more likely to suffer shortage of social support, and females are more likely to suffer shortage of economic support and deteriorating health in older age.
Age distribution. With the aggravation of vulnerability, the proportion of the oldest old is increasing. Vulnerability is basically caused by the lack of social support for the younger older adults.
Marital status. Older adults with no spouse are more likely to face the dual vulnerability in terms of economy and health. While among the support vulnerable older adults, the proportion of older adults with no spouse is much lower. It comes probably because older adults with no spouse are more likely to get social support from others.
Living arrangement. Compared with other types of vulnerable older adults, the proportion of support vulnerable older adults living alone is relatively high, followed by the multi-/dual-disadvantaged older adults.
Social insurance. Participation in social security plays the role of economic support to a certain extent, and helps to prevent older adults from becoming increasingly vulnerable.
Table 8 shows the characteristics difference of household registration among vulnerable older adults. Among the multi/dual vulnerable older adults, compared with NA-older adults, the proportion of the A-older females was lower, and that of older adults without social insurance was higher. However, among the support vulnerable older adults, compared with NA-older adults, the proportion was higher for A-older females and lower for those without social insurance. Among the economic vulnerable older adults, compared with the A-older adults, the proportions of females and non-spouse ones were relatively high in NA-older adults. For the health vulnerable older adults, compared with the A-older adults, the proportions of older adults with older age, no spouse, and no insurance were relatively high.
6. Conclusions and Policy Implication
By using the latent class analysis method, this paper identifies the vulnerable older adults in urban and rural areas of China in terms of health status, socioeconomic status, and social support, and classifies older adults into six vulnerable types, corresponding to four levels of assistance urgency. The group characteristics and geographic distribution of different vulnerable groups were also analyzed, which provides evidence for policy makers to reduce vulnerability of older adults integrally.
First, in the context of aging and urbanization in China, the characteristics of disadvantaged older adults became more complex and diverse. This paper revealed six types of vulnerable older adults, i.e., the most vulnerable older adults facing high deficit accumulation (multi-vulnerable), the most common vulnerable older adults with health and economic disadvantages (dual-vulnerable), and older adults with a certain advantage (potentially vulnerable), as well as the support, economy, or health vulnerable older adults based on one disadvantage. The results showed that 1.46% of older adults had high vulnerability, 12.76% had moderate vulnerability, 34.72% had slight vulnerability, and about one half of older adults (51.06%) were in the potentially vulnerable group, which is highly consistent with the results of data from “The Fourth Sample Survey on The Living Conditions of China’s Urban and Rural Older Persons” [
6], which confirms the reliability of the typology analysis method and the type identification of the vulnerable older adults in China is robust. From the prospective of public policy, the identification of vulnerability based on single dimension was not convincing since older adults who are at higher risk of vulnerability usually suffer from multi-dimensional vulnerability, interacting with health, socioeconomic, or social support factors. Therefore, in the context of limited medical resources, unbalanced regional development and incomplete health policies, a multi-dimensional and multi-standard identification index system for the vulnerable older adults is necessary for policy makers, which helps to more accurately evaluate vulnerability and design more specific policies to improve welfare of older adults with high efficiency and low cost.
Second, identifying characteristics of different types of vulnerable older adults helps to improve effectiveness of policies. Among females, aged, widowed and nest older adults, the proportion of multiple vulnerable older adults was higher, as was the degree of vulnerability. Specifically, female and widowed older adults were more vulnerable in terms of health and economy, while male and young older adults were more likely to lack social support. From the prospective of social insurance, the health condition was closely related to socioeconomic status, as social insurance could prevent older adults from economic plight and multi-/dual-vulnerable to a certain degree. Speeding up and perfecting the social insurance system is of great practical significance for improving the living welfare of older adults. To improve further relevant policies, especially public health policies, characteristics of different vulnerable older adults should be fully considered: (1) taking advantages of social insurance security to meet basic demands of older adults with high/moderate vulnerability; (2) formulating precise support policies to promote diversified old-age care model for different types of vulnerable older adults; (3) referring to the framework of Active Aging and Healthy China policy, the development needs of the potentially vulnerable older adults can be met, and the overall improvement of older adults’ welfare can be realized.
Third, due to the population mobility and unbalanced allocation of medical resources, urban and rural differences were significant among vulnerable older adults in China. The distribution of different types of vulnerable older adults varied greatly. Among the support vulnerable older adults, male and agricultural household registration older adults accounted for a higher proportion. This may be because, on the one hand, female older adults are more likely to have social relations (such as brothers and sisters, friends, groups, etc.) [
66], while males are more likely to face the lack of social relations. On the other hand, with more and more rural labors migrating to cities for higher-paid job, social support from young adults to rural older adults has been greatly weakened since out-migration of adult children has destroyed traditional extended family in rural China.
In addition, health vulnerable older adults with urban hukou are significantly different from those with rural hukou, which is caused by the imbalance of medical resources allocation between urban and rural areas. The urban older adults have better social security support and more abundant medical resources, which guarantees their health and welfare in their later years. However, the medical conditions in rural areas are relatively backward and the accessibility of medical services is poor, which leads to the rural older adults usually suffer from a higher health dilemma and higher death risk. It is urgent to ensure health status of rural older adults by vigorously developing rural economy, continuously promoting rural revitalization and promoting rural medical care.
Fourth, local policies should be suitable for their own reality in terms of economy, degree of aging and stock of vulnerable older adults. According to the indices of economic development and population aging, China’s provinces can be roughly divided into five types: super high–high area, high–multistage area, medium–high area, low–medium area, and low–low area. The distribution of vulnerable older adults in different areas varies greatly. On the whole, with regional economic level increasing, the proportion of potential vulnerable older adults gradually increased, while the multiple-/dual-vulnerable older adults gradually decreases, and the scale of multiple vulnerable older adults gradually reduces. Government departments should comprehensively assess the scale of vulnerable older adults in various areas, and ensure that they can support and deal with the plight of aging accurately.
Specifically, super high–high areas should prevent the deterioration of the disadvantage for potentially vulnerable older adults, and provide one-to-one assistance to the highly vulnerability older adults. In high–multistage and medium–high areas, it is necessary to formulate targeted support policies for older adults with high vulnerability, strengthen the construction of social insurance system, and provide full play to the most critical role of social security. In low–medium areas, the government needs to account the specific situation of various types of vulnerable older adults, and different assistance strategies should be formulated according to the local conditions. For low–low areas, although the aging degree is still low and the scale of vulnerable older adults is small, the governments should be aware of local degree of aging, and make a sound social support policy in advance.