How Does Multidimensional Poverty Affect Sustainable Well-Being Associated with Elderly Cognitive Function? Evidence from the 2018 CLHLS Survey in China

Abstract

This study examines the impact of family multidimensional poverty on cognitive function among older adults in China using the 2018 Chinese Longitudinal Healthy Longevity Survey (CLHLS). Filling a critical gap in the existing literature, we construct a multidimensional poverty index (MPI) based on the Alkire-Foster methodology to evaluate cognitive decline within the context of China’s post-poverty-eradication landscape. Utilizing quantile regression analysis, our findings demonstrate that multidimensional poverty exerts a significant, negative effect on cognitive function, which is more pronounced among individuals at lower cognitive quantiles, consistent with the cumulative disadvantage theory. Furthermore, we identify substantial urban–rural and regional disparities, revealing unique socio-economic inequalities. By linking multidimensional poverty to elderly cognitive health through psychosocial pathways, this study provides empirical evidence that reducing multidimensional deprivation among older adults is integral to achieving both SDG1 and SDG3 in China’s post-eradication context, demonstrating that income-based metrics alone are insufficient to capture the full burden of poverty on elderly cognitive health.

1. Introduction

Sustainable development emphasizes inclusive social equity and population health, with the United Nations Sustainable Development Goals (SDGs) taking SDG1 (No Poverty) and SDG3 (Good Health and Well-being) as foundational pillars for achieving sustainable well-being [1,2]. For aging countries, safeguarding the cognitive health of older adults is not only a concrete practice of SDG3 but also an essential part of achieving sustainable well-being—an important dimension of social sustainability, as cognitive impairment can undermine elderly independence, increase social care burdens, and hinder the long-term sustainability of social systems [3,4]. China, as a middle-income country that eradicated absolute poverty, now faces an accelerating aging process, where the prevalence of cognitive impairment among older adults has risen sharply, becoming a prominent challenge affecting the sustainability of healthy aging [5,6].

China’s declaration of absolute poverty eradication in 2020 marks a critical policy turning point that fundamentally reframes the nature of poverty-related research. Traditional poverty alleviation strategies, largely centered on income-based thresholds, have achieved their primary objective—yet this very success reveals a new and more complex challenge. Older adults who have nominally escaped income poverty may remain deeply deprived across non-monetary dimensions such as healthcare access, social security, educational attainment, and living conditions. In the post-poverty-eradication era, it is precisely this residual multidimensional deprivation that poses the most pressing threat to elderly cognitive health and sustainable well-being. Yet whether this residual non-monetary deprivation translates into measurable cognitive risks among China’s elderly population remains largely untested.

Against this backdrop, existing studies have indicated that family multidimensional poverty restricts the elderly’s access to cognitive protection resources, timely medical services for cognitive decline, and emotional support, while also exacerbating mental health problems such as anxiety and loneliness, all of which directly erode their cognitive function [7,8]. However, the mechanism through which family multidimensional poverty impacts elderly cognitive function, and its implications for sustainable healthy aging, has not been fully explored in the context of post-absolute poverty China.

To fill this research gap, this study aims to: (1) empirically examine the impact of family multidimensional poverty on the cognitive function of older adults in China; (2) explore the mediating mechanisms underlying this impact; (3) clarify the implications of these findings for advancing sustainable well-being. Theoretically, this study enriches the sustainability and aging literature not by merely confirming that poverty harms cognition, but by revealing how these effects vary non-linearly across the cognitive distribution and through which psychosocial pathways they operate. Practically, it provides scientific evidence for reorienting poverty-alleviation strategies away from income-based targeting toward multidimensional, health-focused interventions for vulnerable elderly populations, thereby advancing the coordinated implementation of SDG1 and SDG3 and contributing to sustainable well-being in China.

2. Literature Review and Theoretical Hypotheses

2.1. Literature Review

As China experiences a rapidly aging population, the welfare and cognitive health of older adults have emerged as pressing concerns across policy, clinical, and research domains. Cognitive impairment in older adults, a neurological disorder, not only impacts their quality of life but places significant stress on families and society. As individuals age, their brain function and structure undergo changes that can lead to cognitive decline. Symptoms such as memory loss, confusion, and speech difficulties can severely affect the lives of elderly individuals and their families. According to the World Alzheimer’s Report [7], the number of elderly people with cognitive impairments is rising each year. In China, the rapid growth of the aging population presents an imminent risk of a substantial increase in the number of cognitive disorder patients. In 2023 [8], the elderly population (65 years and older) in China reached 217 million, representing 15.4% of the total population. Simultaneously, the prevalence of cognitive disorders among the elderly is rising. Current estimates suggest that there are approximately 53 million cognitive impairment patients aged 60 and above in China. This situation not only burdens patients and their families but also strains the nation’s medical and social security systems, necessitating urgent attention and comprehensive strategies. It is crucial to identify the potential factors influencing cognitive decline in advance and implement measures to minimize the risk of such conditions.

2.1.1. Cognitive Impairment: Research Status, Influencing Factors, and the Underexplored Role of Multidimensional Poverty in China

Research indicates that cognitive impairment represents a continuum, with amyloid and tau burden defining distinct disease stages that progress along a measurable trajectory [9,10]. Studies like Bermudez et al. (2024) demonstrate the utility of AI-based EKG age prediction in identifying associations with Alzheimer’s neuropathologic changes, further emphasizing the importance of innovative diagnostic tools in understanding cognitive decline [11]. The study by Jia L et al. focused on dementia in China, covering epidemiology, clinical management, and research progress. Their findings highlight a growing prevalence of dementia in China, correlated with an aging population, thereby underscoring the significance of early diagnosis and intervention [12]. Mackin RS and Areán PA focused on the cognitive impairment within a cohort of older adults with severe mental illness in a community mental health setting, indicating that the incidence of cognitive impairment in this group is high and often not fully documented [13]. Both studies suggest that the cognitive health of the elderly warrants close attention in different contexts, and that timely identification and intervention can meaningfully slow progression. Mild cognitive impairment can advance to dementia with age and declining cognitive abilities [14,15]. Therefore, it is essential to recognize early cognitive impairment and its influencing factors. Timely screening and preventive measures can significantly reduce the likelihood of progression to dementia. A recent longitudinal cohort study demonstrated that the lifetime absolute risk of cognitive impairment can be quantified through amyloid PET severity, underscoring the value of early biomarker detection in prevention strategies [16]. Furthermore, longitudinal modeling of plasma p-tau217, amyloid PET, tau PET, and cognition has revealed the temporal sequence in which these biomarkers evolve, offering a dynamic framework for tracking preclinical Alzheimer’s disease progression [17]. Existing research has examined the factors affecting cognitive impairment, such as the impact of depression, disabilities [15,18], and metabolic syndrome [19]. Notably, brain microstructural changes detected by NODDI—reflecting HbA1c levels and systolic blood pressure—have been shown to mediate predictions of cognitive decline, underscoring the biological pathway through which metabolic risk factors erode cognitive reserve [20]. Recent advancements in biomarker research highlight the predictive power of plasma and MRI biomarkers for post-mortem tau pathology, offering new avenues for early detection and intervention in Alzheimer’s disease [21,22]. Diffusion MRI-based biomarkers have additionally been shown to capture a prolonged pre-lesional phase of cerebral small vessel disease, suggesting that white matter damage may begin long before clinical manifestation and contribute to cognitive vulnerability [23]. Methodological challenges remain in neuroimaging research, as harmonization of white matter hyperintensity measurements across different MRI protocols remains an unresolved issue that may affect cross-study comparisons [24]. It has also explored the influence of ozone exposure [25] and pollution [26] from an environmental perspective, as well as the effects of social engagement [27] and socioeconomic status [28] from a subjective psychological viewpoint. Studies such as Hu et al. (2017) have further analyzed factors predicting cognitive decline in Alzheimer’s disease prevention, emphasizing the interplay between socioeconomic factors and cognitive health [29]. However, studies on the impact of multidimensional poverty on cognitive impairment remain limited to small, low-to-middle-income countries in Africa, Central Asia, West Asia and India [30,31,32]. Notably, despite being a typical middle-income country that eliminated absolute poverty in 2020, no scholars have examined the impact of multidimensional poverty on cognitive impairment in China. Unlike previous studies, we select China, a large middle-income country of substantial global relevance, as the focus of our research for the first time, which will enhance the study of regional inequalities related to cognitive function and multidimensional poverty.

2.1.2. Multidimensional Poverty: Theoretical Evolution, Measurement Innovations, and Its Application in Assessing Elderly Cognitive Health Contexts

Over the past two decades, the concept of multidimensional relative poverty has gained traction among scholars. Based on the data from the China Education Panel Survey, Chen J and Zhang Z investigated the impact of family poverty on adolescents’ cognitive and non-cognitive abilities. They found that family poverty has a negative impact on adolescents’ development [33]. Traditional poverty theories assert that poverty stems from income levels. However, Amartya Sen first argued that poverty measurement cannot be based solely on income. It requires evaluating multiple dimensions, including capabilities, freedoms, social opportunities, economic resources, transparency tools, and political liberties [34]. This concept transcends traditional poverty limitations, utilizing multiple dimensions to assess the poverty status of the elderly. The concept of multidimensional relative poverty expands upon absolute poverty by utilizing multiple dimensions to measure deprivation. Building on the theoretical innovations of predecessors, Alkire and others constructed a general model for measuring and decomposing multidimensional poverty [35,36]. This model, known as the AF multidimensional poverty measurement, suggests assessing the multidimensional poverty index through three dimensions: health, education, and living conditions. Butt, S.A. et al. subsequently applied this counting method to measure multidimensional poverty by the Alkire-Foster counting method. The study found an association between multidimensional poverty and atherosclerotic cardiovascular disease (ASCVD), revealing the link between poverty and health-related issues [37]. To better capture multidimensional poverty among the elderly, we systematically adapt the standard MPI by removing child-oriented indicators ill-suited for older populations, replacing them with geriatric-sensitive variables, and incorporating two additional dimensions that reflect the specific vulnerabilities of China’s aging population. Ultimately, the multidimensional poverty model utilized in this paper integrates five dimensions: income poverty, life quality poverty, health poverty, security poverty and education poverty, aiming to provide a more comprehensive assessment of poverty among older adults. In this context, exploring the effects of multidimensional poverty on cognitive function in older adults will enhance our understanding of the underlying causes and mechanisms of cognitive disorders. Furthermore, it will provide a scientific basis for developing effective intervention strategies. Beyond its comprehensiveness, the multidimensional poverty framework is theoretically significant because it captures the compounding nature of deprivation. When older adults simultaneously experience income poverty, health deprivation, and educational disadvantage, these dimensions do not operate independently—limited income restricts healthcare access, poor health undermines cognitive reserve, and low education reduces compensatory capacity, suggesting that poverty’s cognitive toll is substantially larger than any single deprivation measure would indicate. Those with lower cognitive reserve are least equipped to buffer against such compounding deprivations, rendering multidimensional poverty a uniquely powerful—and systematically underdetected—predictor of cognitive decline. We therefore propose:

H1. 

Family multidimensional poverty is negatively associated with the cognitive function of older adults.

2.1.3. Potential Mechanisms Linking Multidimensional Family Poverty to Elderly Cognitive Function

Multidimensional poverty can impact elderly individuals’ cognitive function from various aspects. Research indicates that poverty and multidimensional poverty can reduce subjective well-being [38,39]. Subjective well-being, or self-reported quality of life, as a potential risk factor, may influence the likelihood of Alzheimer’s disease and dementia [40,41]. Lower levels of subjective well-being correlate with an increased risk of both conditions. Furthermore, scholars have noted that multidimensional poverty and relative deprivation can impact the mental health of adolescents and the elderly, leading to symptoms of depression [42,43,44]. Pauls et al. and Kromydas et al. explored the relationship between employment status and mental health and concluded that there is a negative correlation between employment precariousness and mental health. The employment situation can accurately reflect the multidimensional poverty status [45,46]. Additionally, a positive mental health status not only lowers the risk of cardiovascular diseases [47] but also significantly reduces the risk of cognitive impairment in middle-aged and older adults [48,49]. In addition, in studies of other diseases such as cardiovascular diseases, some scholars have found that compromised self-reported physical and mental health status elevates the risk of vascular events and mortality, indicating that self-reported physical and mental health status is a major factor affecting health [50]. Based on this literature, we choose self-reported quality of life and mental health as mediating variables to deeply explore the underlying mechanisms through which multidimensional poverty affects cognitive function in older adults. While the overall direction of the poverty–cognition relationship is well established, whether mental health or self-reported quality of life plays a more dominant mediating role across different poverty dimensions remains empirically unresolved. Identifying these mechanisms is not merely an academic exercise; it determines which intervention points are most actionable for policymakers in China’s post-poverty-eradication context, where income-based poverty has been largely addressed but psychosocial deprivations persist. Thus, we hypothesize:

H2. 

Self-reported quality of life mediates the relationship between family multidimensional poverty and elderly cognitive function.

H3. 

Mental health mediates the relationship between family multidimensional poverty and elderly cognitive function.

In summary, to advance sustainable well-being of the elderly in China, this study examines multidimensional poverty as an explanatory variable, investigating its impact on cognitive function. To further explore the effects of multidimensional poverty across different household types, genders, and regions, this paper conducts a heterogeneous analysis categorizing household types into urban and rural, genders into male and female, and regions into eastern, central, western, and northeastern. Additionally, it utilizes mental health and subjective life evaluation as mediating variables to analyze the mechanisms by which multidimensional relative poverty impacts cognitive performance. Finally, the paper offers targeted suggestions and preventive measures based on the actual conditions in Chinese society to mitigate the risk of cognitive decline in the elderly.

3. Research Design and Methodology

3.1. Empirical Strategy and Model Specification

To comprehensively examine the relationship between family multidimensional poverty and cognitive function among older adults, this study employs a multi-staged and rigorous empirical strategy. First, we establish a baseline OLS regression model to estimate the average impact. Second, considering potential self-selection bias and endogeneity, we apply the Propensity Score Matching (PSM) method as a robust check to ensure the reliability of our findings. Third, to explore the heterogeneous effects across different levels of cognitive health, a quantile regression approach is utilized. Furthermore, we conduct a detailed heterogeneity analysis to identify disparities across urban-rural and regional dimensions. Finally, a mediation analysis framework is constructed to uncover the underlying mechanisms. The specific model specifications are detailed as follows.

3.1.1. Baseline Regression Model

To test Hypothesis H1 (family multidimensional poverty has a significantly negative impact on the cognitive function of older adults), this study constructs an OLS regression model with cognitive function score as the dependent variable and multidimensional poverty as the core explanatory variable:

$${CF}_i={\alpha }_0+{\alpha }_1{MP}_i+{\sum }_{k=2}^n{{\alpha }_{k}X}_{ik}+{\epsilon }_i$$

(1)

${CF}_i$ represents the cognitive function score of the i-th elderly individual, measured by the Mini-Mental State Examination (MMSE) score; ${MP}_{\mathrm{i}}$ is the multidimensional poverty index of the i-th elderly individual, the core explanatory variable of this study; $X_{ik}$ denotes a series of control variables affecting the cognitive function of the elderly, including gender, age, age squared, financial support, marital status, household type, economic status, childhood economic situation, and main occupation before the age of 60; ${\alpha }_0$ is the constant term, ${\alpha }_1$ is the coefficient of multidimensional poverty (the key parameter to be estimated, expected to be negative),${\alpha }_k$ are the coefficients of control variables; ${\epsilon }_i$ is the random error term, which follows a normal distribution with a mean of 0 and a constant variance.

In addition, to further explore the impact of each sub-dimension of multidimensional poverty (income poverty, life quality poverty, health poverty, security poverty, education poverty) on cognitive function, the study replaces ${MP}_i$ with each sub-dimension index separately and re-estimates the model:

$${CF}_i={\beta }_0+{\beta }_1{MP}_{di}+{\sum }_{k=2}^n{{\beta }_{k}X}_{ik}+{\mu }_i$$

(2)

where ${MP}_{di}$ represents the d-th sub-dimension index of multidimensional poverty (d = 1, 2, 3, 4, 5), and ${\mu }_i$ is the random error term.

3.1.2. Mediating Effect Model

To test Hypotheses H2 and H3 (mental health and self-reported quality of life play mediating roles in the relationship between family multidimensional poverty and elderly cognitive function), this study adopts the three-step mediating effect test method proposed by Baron and Kenny (1986) [51] and constructs a chain mediating effect model:

$$M_i={\delta }_0+{\delta }_1{MP}_i+{\sum }_{k=2}^n{{\delta }_{k\tau }X}_{ik}+{\xi }_i$$

(3)

$${CF}_i={\phi }_0+{\phi }_1{MP}_i+{{\phi }_{2}M}_i+{\sum }_{k=2}^n{{\beta }_{k}X}_{ik}+{\omega }_i$$

(4)

$M_i$ is the mediating variable, including self-reported quality of life and mental health status of the i-th elderly individual, measured by the CES-D scale score; ${\xi }_i$, ${\omega }_i$ are random error terms.

The mediating effect is determined by the following criteria: (1) ${\alpha }_1$ in model (1) is significant (total effect exists); (2) ${\delta }_1$ in model (3) is significant, indicating that multidimensional poverty affects the mediating variable; (3) ${\phi }_1$ in model (4) is significant, and the absolute value of ${\phi }_1$ is smaller than that of ${\alpha }_1$, indicating a partial mediating effect; if ${\phi }_1$ is not significant, it indicates a complete mediating effect. In addition, this study uses the Bootstrap method and Sobel test to further verify the significance of the mediating effect, with 1000 bootstrap samples drawn to calculate the 95% confidence interval of the mediating effect.

3.2. Data Sources and Study Samples

The data were obtained from the 2018 China Longitudinal Healthy Longevity Survey (CLHLS), a prospective cohort study of elderly individuals conducted by the Centre for Healthy aging and Development at Peking University/National Institute for Development Studies (NIDS), covering 23 provinces, municipalities, and autonomous regions within China. We restricted the sample to participants aged 60 and older and applied sequential exclusion criteria to remove cases with missing outcome, exposure, or covariate data (detailed in Figure 1). We began with the initial sample of 15,874 observations from the 2018 CLHLS survey. We then applied sequential exclusion criteria to maintain the data quality: first, we excluded respondents aged below 60; second, we removed cases with missing data for the Mini-Mental State Examination (MMSE) scores; third, we excluded observations missing any of the 15 multidimensional poverty indicators; then, we performed listwise deletion for respondents with missing values in key socio-economic control variables, and finally we removed observations that had missing mediating variables. This rigorous screening resulted in a final analytical sample of 2930 valid observations for subsequent regression analysis.

3.3. Variable Measurement and Description

3.3.1. Dependent Variable: Cognitive Function

The CLHLS questionnaire measures cognitive function using the Mini-Mental State Examination (MMSE) score, which is a widely used tool in geriatric cognitive assessment. It consists of 24 questions with a total of 30 scores, covering five main areas: general ability, reactivity, attention and calculation, memory, language comprehension and self-coordination. Higher scores indicate higher cognitive function. It reflects the current cognitive status of the elderly and serves as a key indicator for predicting the long-term health trajectory of the aging population.

3.3.2. Explanatory Variable: Multidimensional Poverty

Sustainable well-being serves as a core anchor for this study’s variable design, with individual multidimensional poverty adopted as the independent variable. It draws on previous studies, particularly the dimensions and indicators employed in the 2010 Human Development Report by the United Nations Development Program. While the Global Multidimensional Poverty Index (MPI) developed by the UNDP and OPHI provides a foundational framework for assessing poverty, its application is primarily geared toward national-level monitoring of general populations. Recognizing the demographic shifts and the specific vulnerabilities of the elderly in China, our study constructs a modified Multidimensional Poverty Index (MPI) that is uniquely tailored to this aging demographic using the CLHLS dataset. This system encompasses five dimensions: income poverty, life quality poverty, health poverty, security poverty and education poverty, all of which are fundamental to safeguarding their sustainable well-being.

Compared to the standard global MPI, which encompasses three dimensions (health, education, and living standards) across ten indicators designed for cross-national comparisons of the general population, the index developed in this study departs in three theoretically motivated and empirically necessitated ways. First, the standard MPI proxies health deprivation primarily through child mortality and nutrition indicators, which carry limited discriminatory power for elderly populations. By contrast, our health poverty dimension incorporates six geriatric-sensitive indicators—self-assessed health status, individual medical expenses, BMI, eyesight, dental health, and nutrition— that more precisely capture the functional health deterioration known to precede cognitive decline in older adults. Second, we introduce a dedicated “security poverty” dimension encompassing old-age insurance, medical insurance, and access to timely medical treatment. This dimension is absent from the standard MPI, which was developed largely for contexts where formal social protection systems remain nascent. In post-poverty-eradication China, however, the gap between institutional coverage and actual accessibility of social security has emerged as a critical, yet underexplored, axis of elderly vulnerability. Third, we incorporate life satisfaction as a subjective indicator within the life quality poverty dimension, complementing the standard MPI’s purely objective material indicators. This addition aligns with Sen’s capability approach and the theoretical framework of sustainable well-being adopted in this study, allowing us to capture deprivation at both the material and subjective experiential levels. These adaptations are grounded in the specific socio-economic characteristics of China’s elderly population and constrained by the data availability of the CLHLS database, ensuring both theoretical coherence and empirical feasibility.

To quantify multidimensional poverty, we utilize the Alkire–Foster (AF) methodology, which identifies the poor based on a dual-cutoff approach. The calculation involves two systematic steps:

Step 1: Identification of Deprivation. For each individual i, we assess their status across d = 15 indicators. A deprivation status matrix $g^0$ is constructed, where an element $g_{ij}^0$ equals 1 if individual i is deprived in indicator j (i.e., $x_{ij}$ < $z_j$), and 0 otherwise.

Step 2: Aggregation and MPI Calculation. We compute the deprivation score ($c_i$) for each individual by summing the weighted deprivations:

$$c_i=\sum _{j=1}^d{w}_j{g}_{ij}^0$$

where $w_j$ denotes the specific weight assigned to indicator j. An individual is identified as multidimensionally poor if their deprivation score meets or exceeds the cross-dimensional cutoff k: ${MP}_i=1$ if $c_i \ge k$, and ${MP}_i=0$ if $c_i \ge k$.

Currently, no authoritative standards exist for determining the weights of dimensions and indicators. Therefore, this study adopts an equal weighting method based on previous research. It uses the multidimensional poverty threshold value of 0.3 provided in the Human Development Report to identify whether a sample qualifies as multidimensional poverty. This threshold implies that an elderly individual is classified as multidimensionally poor only if they suffer from a weighted deprivation index of at least 30%, ensuring the robustness of our identification of the truly vulnerable population. Details on the indicators and their dimensions are shown in

Table 1. The Index System for Multidimensional Poverty.

Target	Level 1 Indicators	Secondary Indicators	Weight
multidimensional poverty	Income poverty	Per capita family income	1/15
	Life quality poverty	Living conditions	1/15
		Drinking water	1/15
		Cooking fuel	1/15
		Life satisfaction	1/15
	Health poverty	Self-assessed health status	1/15
		Individual medical expenses	1/15
		BMI	1/15
		Eye-sight level	1/15
		Dental health	1/15
		Nutrition	1/15
	Security poverty	Old-age insurance	1/15
		Medical insurance	1/15
		Timely medical treatment	1/15
	Educational poverty	Educational attainment	1/15

Note: Weights are assigned equally among all 15 indicators ($w_j=1/15$).

.

3.3.3. Mediating Variables: Self-Reported Quality of Life and Mental Health

Unlike most prior studies, which focus on the direct association between poverty and cognition without examining mediating pathways, we constructed two mediating variables: self-reported quality of life and mental health to examine the pathways through which poverty erodes cognitive function. These mediating variables are core components of sustainable well-being, as self-reported quality of life reflects the subjective well-being of the elderly, and mental health is a key pillar of sustainable well-being, linking socioeconomic status to long-term cognitive health outcomes. We used the question “How do you rate your life at present?” in the Life Evaluation and Personality Scale to construct self-reported quality of life. Respondents rated their lives, with 1 being very bad and 5 being very good. Secondly, we used the 18 questions of Personality and Mood part and Depression (CESD) Scale in the Life Evaluation and Personality Scale to construct mental health. Higher total scores are associated with better mental health. The mediating effect model helps to uncover the transmission mechanism through which multidimensional poverty affects cognitive function, thereby enriching the theoretical understanding of sustainable well-being in the context of aging populations.

3.3.4. Control Variables

We selected gender, age, age squared, financial support, current marital status, household type, economic status, economic situation in childhood and main occupation before age 60 as control variables to control individual characteristics (

Table 2. The Descriptive Statistics for Multidimensional Poverty, Cognitive Function Score and All Control Variables.

Variable		N = 2930
Cognitive Function Score	Mean (SD)	28.30 (2.67)
Multidimensional poverty	Mean (SD)	0.58 (0.50)
Gender (male = 1, female = 0)	Male, N (%)	1685 (57.51%)
	Female, N (%)	1245 (42.49%)
Age	Mean (SD)	78.23 (9.95)
Age2	Mean (SD)	6218.82 (1624.46)
Is all of the financial support sufficient to pay for daily expenses?	Yes, N (%)	2632 (89.83%)
	No, N (%)	298 (10.17%)
Current marital status	Currently with spouse, N (%)	942 (32.15%)
	Others, N (%)	1988 (67.85%)
Household type	Urban, N (%)	1193 (40.72%)
	Rural, N (%)	1737 (59.28%)
How do you rate your economic status compared with other local people?	Very rich, N (%)	80 (2.73%)
	Rich, N (%)	602 (20.58%)
	So so, N (%)	2039 (69.59%)
	Poor, N (%)	187 (6.38%)
	Very poor, N (%)	21 (0.72%)
Often went to bed hungry as a child	Yes, N (%)	1930 (65.87%)
	No, N (%)	1000 (34.13%)
Main occupation before age 60	Agriculture, forestry, animal husbandry, fishery, N (%)	1441 (49.18)
	Others, N (%)	1489 (50.82%)

Note: Age² is included to capture the non-linear relationship between age and cognitive function.

).

4. Results and Analysis

4.1. Baseline Regression Results

Table 3. OLS regression analysis of multidimensional poverty on cognitive function.

	Cognitive Function Score (β, 95%CI)
Multidimensional poverty	−0.291 ***
	(−0.49, −0.97)
Income poverty		−0.483 ***
		(−0.72, −0.24)
Life quality poverty			−0.078
			(−0.28, −0.13)
Health poverty				−0.191 **
				(−0.38, −0.00)
Security poverty					0.114
					(−0.08, 0.31)
Education poverty						−0.275 **
						(−0.51, −0.04)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
N	2930	2930	2930	2930	2930	2930
r2_a	0.18	0.18	0.17	0.18	0.17	0.18
F	63.79	64.61	62.82	63.23	62.91	63.39

Statistics in parentheses ** p < 0.05, *** p < 0.01.

demonstrates that following adjustment for all covariates, multidimensional poverty exhibits a significant inverse association with cognitive impairment scores. Specifically, for every unit increase in the multidimensional poverty index, the cognitive function score decreases by 0.291 points, thereby providing empirical support for Hypothesis H1. Among the five sub-dimensions of multidimensional poverty, income poverty, health poverty, and education poverty also have a significant negative impact on cognitive function. Income poverty reduces the cognitive function score by 0.483 points, health poverty by 0.191 points, and education poverty by 0.275 points. However, life quality poverty has a non-significant negative impact, and security poverty shows a positive but non-significant association with cognitive function scores. This may be attributed to China’s extensive pension insurance system, which covers 95% of the population, reducing the impact of security-related poverty on cognitive function [52]. Overall, the baseline regression results confirm that multidimensional poverty is a significant risk factor for cognitive impairment in China, which has important implications for formulating policies to enhance the sustainable well-being of the aging population.

4.2. Robustness Test

To verify the robustness of the relationship between multidimensional poverty and the sustainable well-being of older adults, we conducted a robustness test using the Propensity Score Matching (PSM) method.

Table 4. The Propensity Score Matching (PSM) Result.

Matching Methods	k-Nearest Neighbors Matching (n = 5)	Radius Matching	Nearest Neighbors Matching with Caliper	Kernel Matching
_treated	−0.28 **	−0.31 ***	−0.28 **	−0.31 ***
	(−2.32)	(−2.82)	(−2.32)	(−2.78)
N	2912	2912	2912	2912

** p < 0.05, *** p < 0.01.

presents the results of four matching methods of the Propensity Score Matching (PSM) method. As can be seen from the data, under different matching methods, the coefficients of the treatment group are all significantly negative. The coefficient is −0.28 in k-Nearest Neighbors Matching, −0.31 in Radius Matching, −0.28 in Nearest Neighbors Matching with Caliper, and −0.31 in Kernel Matching. This indicates that after the PSM treatment, multidimensional poverty still has a significant negative impact on cognitive function scores, suggesting that the research results remain robust after considering the sample selection bias. Meanwhile, the sample size under each matching method is 2912, ensuring the reliability of the results. Figure 2a shows the matching test results of the Propensity Score Matching (PSM) method, further verifying the impact of PSM on the research results. Figure 2b presents the distribution comparison of samples before and after matching, indicating that PSM effectively balances the sample characteristics and solves the bias problem caused by non-random sample selection, thus providing visual evidence for the robustness of the baseline findings and the relationship between multidimensional poverty and the sustainable well-being of the elderly.

4.3. Quantile Regression Results

To clarify how multidimensional poverty affects the sustainable well-being of older adults with different cognitive levels—by exploring the heterogeneous vulnerability of groups at varying cognitive stages, Figure 3 shows the results of the quantile regression. From the 10th to the 90th percentile, the impact of multidimensional poverty and its three significant sub-dimensions (income poverty, health poverty, and education poverty) on different groups of cognitive function scores was examined. At lower quantiles (10–60%), poverty dimensions have a significant negative impact on cognitive function, except for health poverty. For example, at the 20th quantile, the impact of income poverty on cognitive function was particularly strong. However, from the 60th quantile point onwards, the confidence intervals contained zero, indicating that the effects become progressively insignificant. This finding suggests that multidimensional poverty and its sub-dimensions have a more pronounced impact on individuals with poorer cognitive function, highlighting the vulnerability of this group. This threshold-sensitive pattern suggests that social inequality most severely constrains cognition among those already operating at lower cognitive capacity. Cognitive functions are essential abilities that individuals use to acquire knowledge, solve problems, and adapt to changes in their environment. The intensification of multidimensional poverty not only restricts the material resources available to individuals but may also deprive them of access to education and healthcare services, thereby hindering the development of their cognitive capabilities. This effect is particularly severe for those already operating at lower cognitive levels, as they may lack sufficient resources to cope with and alleviate the stress brought about by poverty. It has important implications for sustainable well-being. Targeted poverty alleviation for older adults with poorer cognitive function can reduce the “cumulative disadvantage” of socioeconomic deprivation on their long-term health. It also prevents cognitive health from worsening further and protects their ability to live independently and with dignity. This is essential for achieving inclusive and sustainable healthy aging.

4.4. Heterogeneity Analysis

To clarify how the heterogeneous impacts of multidimensional poverty on elderly cognitive function shape their sustainable well-being, we conducted a heterogeneity analysis to examine group differences by gender, household type, and region. Following the official geographic divisions of the National Bureau of Statistics of China, we categorized the sample into four regions: East, Central, West, and Northeast, with the results presented in

Table 5. Heterogeneous effects by gender, household type and region.

	Gender		Household Type		Region
	Female	Male	Rural Residents	Urban Residents	East	Central	West	Northeast
	Cognitive Function Score
Multi-dimensional poverty	−0.34 **	−0.28 **	−0.22	−0.46 ***	−0.30 **	−0.53 **	−0.11	−0.38
	(−0.67, −0.01)	(−0.51, −0.04)	(−0.49, 0.06)	(−0.72, −0.19)	(−0.57, −0.02)	(−0.97, −0.09)	(−0.51, 0.29)	(−1.20, 0.44)
_cons	24.54 ***	36.73 ***	25.31 ***	34.32 ***	27.00 ***	35.54 ***	32.27 ***	28.72 **
	(17.01, 34.06)	(30.02, 43.44)	(18.10, 32.52)	(26.57, 42.07)	(19.07, 34.92)	(22.84, 48.25)	(22.55, 42.00)	(6.31, 51.12)
N	1245	1685	1737	1193	1395	587	770	178
r2_a	0.23	0.11	0.20	0.14	0.17	0.19	0.19	0.19
F	42.18	23.38	48.44	22.32	29.34	14.90	19.29	5.03

** p < 0.05, *** p < 0.01.

. First, multidimensional poverty affected cognitive function scores significantly for both males and females, but the coefficient was larger for females. Secondly, in terms of household type, the cognitive function of urban residents was more significantly affected by multidimensional poverty than that of rural residents. Regionally, the central and eastern regions show a more significant impact of multidimensional poverty on cognitive function, while the impact in the western and northeastern regions is less pronounced. These heterogeneous results suggest that the relationship between poverty and cognitive health is context-dependent. The significant impact observed in urban and more developed regions might be attributed to the sensitivity effect, where individuals in these areas have higher expectations for social participation and healthcare, making them more sensitive to multidimensional deprivation. Conversely, the non-significant findings in rural and western regions may reflect a survival priority, where basic physiological and survival needs take precedence over non-monetary poverty dimensions in impacting cognitive well-being. This implies that the mechanisms through which poverty influences cognitive aging are not uniform and are strongly conditioned by the local socio-economic environment.

4.5. Mediation Effect Analysis

The mediation effect analysis, using a three-step approach, shows that self-reported quality of life and mental health play mediating roles in the relationship between multidimensional poverty and cognitive function.

4.5.1. Meditating Role of Self-Reported Quality of Life

Table A1. Mediation effect of self-reported quality of life in the association between multidimensional poverty and cognitive function.

	Cognitive Disorders Score	Self-Reported Quality of Life	Cognitive Disorders Score
Panel A
multidimensional poverty	−0.268 ***	−0.105 ***	−0.237 **
	(−2.62)	(−3.37)	(−2.33)
self-reported quality of life			0.285 ***
			−4.66
Panel B
Income poverty	−0.525 ***	−0.157 ***	−0.482 ***
	(−4.11)	(−4.00)	(−3.77)
self-reported quality of life			0.276 ***
			−4.52
Panel C
Life quality poverty	−0.059	−0.113 ***	−0.025
	(−0.55)	(−3.45)	(−0.24)
self-reported quality of life			0.293 ***
			−4.78
Panel D
Health poverty	−0.188 *	−0.187 ***	−0.134
	(−1.91)	(−6.18)	(−1.36)
self-reported quality of life			0.285 ***
			−4.62
Panel E
Security poverty	0.12	−0.121 ***	0.157
	−1.18	(−3.85)	−1.53
self-reported quality of life			0.301 ***
			−4.91
Panel F
Education poverty	−0.281 **	0.076 **	−0.303 **
	(−2.28)	−2.01	(−2.47)
self-reported quality of life			0.300 ***
			−4.91
Control variables	yes	yes	yes
N	2930	2930	2930

* p < 0.1, ** p < 0.05, *** p < 0.01.

in the Appendix A focuses on the mediating effect of self-reported quality of life in the relationship between multidimensional poverty and cognitive function. From the perspective of multidimensional poverty as a whole, its direct effect on cognitive function scores is significantly negative, indicating that an increase in the degree of multidimensional poverty will reduce cognitive function scores. At the same time, multidimensional poverty has a significant negative impact on self-reported quality of life, meaning that an increase in poverty levels will lower self-reported quality of life. Self-reported quality of life has a significant positive impact on cognitive function scores, that is, the higher the self-reported quality of life, the higher the cognitive function scores. In terms of each poverty dimension, income poverty and health poverty have significant negative direct effects on cognitive function scores and also have significant negative impacts on self-reported quality of life. Life quality poverty has no significant effect on cognitive function scores but has a significant negative impact on self-reported quality of life. Security poverty has no significant effect on cognitive function scores and has a significant negative impact on self-reported quality of life. Education poverty has a significant negative direct effect on cognitive function scores and a significant positive impact on self-reported quality of life.

Table 6. Bootstrap test and Sobel test of self-reported quality of life.

Mediation Variables	Poverty Dimensions		z	p > |z|	Percentile 95% CI		Proportion of Mediation Effects	Sobel Z
self-reported quality of life	multidimensional poverty	Ind.	−2.73	0.006	−0.054	−0.010	11.3%	−2.729 ***
		Dir	−2.37	0.018	−0.426	−0.040
	Income poverty	Ind.	−2.96	0.003	−0.075	−0.019	8.3%	−2.998 ***
		Dir	−3.14	0.002	−0.784	−0.204
	Life quality poverty	Ind.	−2.61	0.009	−0.061	−0.012	56.69%	−2.801 ***
		Dir	−0.24	0.809	−0.222	0.180
	Health poverty	Ind.	−3.65	0.000	−0.082	−0.026	28.3%	−3.701 ***
		Dir	−1.35	0.175	−0.330	0.058
	Security poverty	Ind.	−2.99	0.003	−0.062	−0.016	−30.3%	−3.03 ***
		Dir	1.58	0.113	−0.026	0.349
	Education poverty	Ind.	1.76	0.079	0.001	0.050	−8.1%	1.858 *
		Dir	−2.96	0.003	−0.497	−0.084

* p < 0.1, *** p < 0.01.

further analyzes the mediating effect of self-reported quality of life, including the results of the Bootstrap test and the Sobel test. In the relationship between multidimensional poverty and cognitive function, the indirect effect of self-reported quality of life is significant, the proportion of the mediating effect is 11.3%, and the z-value of the Sobel test is −2.729, indicating a significant mediating effect, thus providing empirical support for Hypothesis H2. In the relationships between income poverty, health poverty, and cognitive function, self-reported quality of life also has significant mediating effects, with the proportions of the mediating effects being 8.3% and 28.3%, respectively. For life quality poverty, its indirect effect is significant, and the proportion of the mediating effect reaches 56.69%. The indirect effect of security poverty is significant, but the proportion of the mediating effect is −30.3%, showing a negative mediation. This suppression pattern suggests that self-reported quality of life partially offsets the direct cognitive impact of security poverty, consistent with China’s broad pension coverage reducing overt material distress while leaving psychosocial pathways less buffered. Education poverty has a masking effect of 8.1%, and its indirect effect is marginally significant through the Bootstrap test. This counterintuitive finding may reflect a low-expectation adaptation effect: elderly individuals with lower educational attainment may set lower benchmarks for life satisfaction, resulting in comparatively higher self-reported quality of life despite objective deprivation. This indicates that multidimensional poverty reduces the sustainable well-being of the elderly by lowering their subjective quality of life, which in turn increases the risk of cognitive impairment.

4.5.2. Meditating Role of Mental Health

Using the same approach,

Table A2. Mediation effect of mental health in the association between multidimensional poverty and cognitive function.

	Cognitive Disorders Score	Mental Health	Cognitive Disorders Score
Panel A
multidimensional poverty	−0.265 ***	−1.118 ***	−0.234 **
	(−2.61)	(−3.25)	(−2.30)
Mental health			0.028 ***
			−5.1
Panel B
Income poverty	−0.483 ***	−1.020 **	−0.497 ***
	(−3.91)	(−2.36)	(−3.90)
Mental health			0.028 ***
			−5.08
Panel C
Life quality poverty	−0.078	−1.051 ***	−0.026
	(−0.75)	(−2.93)	(−0.24)
Mental health			0.029 ***
			−5.22
Panel D
Health poverty	−0.191 **	−2.813 ***	−0.111
	(−1.99)	(−8.57)	(−1.12)
Mental health			0.028 ***
			−5
Panel E
Security poverty	0.114	0.688 **	0.145
	−1.15	−2	−1.43
Mental health			−0.030 ***
			(−5.29)
Panel F
Education poverty	−0.275 **	0.569	−0.303 **
	(−2.30)	−1.37	(−2.48)
Mental health			0.030 ***
			−5.31
Control variables	Yes	Yes	Yes
N	2930	2930	2930

** p < 0.05, *** p < 0.01.

in the Appendix A and

Table 7. Bootstrap test and Sobel test of mental health.

Mediation Variables	Poverty Dimensions		z	p > |z|	Percentile 95% CI		Proportion of Mediation Effects	Sobel Z
Mental health	multidimensional poverty	Ind.	−2.55	0.011	−0.056	−0.008	12.0%	−2.742 ***
		Dir	−2.66	0.008	−0.422	−0.046
	Income poverty	Ind.	−2.17	0.030	−0.055	−0.002	5.5%	−2.142 **
		Dir	−3.39	0.001	−0.783	−0.210
	Life quality poverty	Ind.	−2.34	0.019	−0.056	−0.005	54.5%	−2.553 **
		Dir	−0.25	0.804	−0.228	0.177
	Health poverty	Ind.	−3.70	0.000	−0.122	−0.037	41.7%	−4.317 ***
		Dir	−1.13	0.258	−0.304	0.081
	Security poverty	Ind.	−1.82	0.069	−0.042	0.001	−16.3%	−1.87 **
		Dir	1.47	0.142	−0.049	0.339
	Education poverty	Ind.	1.26	0.207	−0.009	0.043	−5.9%	1.328
		Dir	−2.80	0.005	−0.516	−0.091

** p < 0.05, *** p < 0.01.

indicate that mental health fully mediates the relationship between health poverty and cognitive function, and partially mediates the association between multidimensional poverty as a whole, income poverty and cognitive function. Table A2 presents the test results of the mediating effect of mental health in the relationship between multidimensional poverty and cognitive function. In terms of multidimensional poverty, its direct effect on cognitive function scores is significantly negative (−0.265 ***), meaning that the higher the degree of multidimensional poverty, the lower the cognitive function scores. Meanwhile, multidimensional poverty has a significant negative impact on mental health, indicating that an increase in poverty levels will deteriorate the mental health status. However, mental health has a significant positive impact on cognitive function, that is, the better the mental health status, the higher the cognitive function scores. In the dimension of income poverty, the direct effect on cognitive function scores is significantly negative, the effect on mental health is significantly negative, and the effect of mental health on cognitive function is significantly positive. The dimensions of life quality poverty, health poverty, security poverty, and education poverty also show corresponding impact coefficients, respectively, reflecting the relationships between poverty in each dimension, cognitive function, and mental health, and providing detailed data support for studying the psychological mechanism by which multidimensional poverty affects cognitive function.

Table 7 further conducts an in-depth analysis of the mediating effect of mental health in the relationship between multidimensional poverty and cognitive function. In addition to presenting direct and indirect effect coefficients similar to those in Table A2, it also provides the proportion of the mediating effect, the z-value, and the p-value of the Sobel test. The results show that mental health plays a partial mediating role in the relationship between multidimensional poverty and cognitive function, with a mediating effect proportion of 12.0% and a Sobel test z-value of −2.742, indicating a significant mediating effect, thereby providing empirical support for Hypothesis H3. In the relationships between income poverty, health poverty, and cognitive function, there is also a partial mediating role, with mediating effect proportions of 5.5% and 41.7%, respectively, and the Sobel test also shows a significant mediating effect. For the relationships between life quality poverty, security poverty, education poverty, and cognitive function, mental health plays a masking effect, reflecting the complex mechanism of mental health in the process of different poverty dimensions affecting cognitive function and providing more abundant information for comprehensively understanding the relationship between poverty and cognitive function.

Based on the aforementioned empirical findings, the hypothesis posited in the literature review is substantiated. The psychological and emotional well-being of the elderly is intricately linked to their cognitive health. These two unique perspectives provide innovative insights and empirical evidence for understanding the complex relationship between poverty and cognitive health in the elderly. This mediating mechanism reveals that multidimensional poverty affects cognitive function not only through direct material deprivation but also through indirect psychological and emotional channels, which underscores the importance of addressing both the objective material needs and subjective psychological well-being of the elderly to improve their sustainable well-being.

5. Discussions

This study redefines family multidimensional poverty as a systemic barrier to older adults’ cognitive health and the realization of a framework of sustainable well-being, moving beyond its traditional view as a purely economic issue. Our findings confirm that family multidimensional poverty has a significant negative impact on older adults’ cognitive function, and this effect operates through two key mediating pathways: mental health and self-reported quality of life. This fills the existing research gap where “multidimensional poverty” and “elderly cognitive health” have often been studied separately in China, while also enriching the theoretical framework of sustainable well-being in aging-related research.

The contribution of this study lies not in confirming that poverty harms cognition—a directional relationship that prior literature has established—but in revealing its non-linear structure, the specific dimensions that carry the greatest cognitive burden, and the psychosocial pathways through which these effects operate. Our quantile regression results show that poverty effects concentrate disproportionately among individuals at the lower end of the cognitive spectrum—a threshold-sensitive pattern consistent with cumulative disadvantage theory that cannot be derived from intuition alone. Meanwhile, the dual mediating pathways identified here, and the divergent roles of different poverty sub-dimensions within each pathway, underscore that poverty erodes cognition through mechanisms that are both empirically non-trivial and policy-actionable.

From a mechanistic perspective, the mediating role of mental health highlights the psychological chain of multidimensional poverty. Deprivations in core areas such as economic security, healthcare access, and social support—key components of family multidimensional poverty—lead to negative emotions like anxiety and loneliness among older adults. These psychological stressors further impair brain neuroplasticity and cognitive processing abilities, which aligns with previous research showing a vicious cycle between poverty and mental health. At the same time, the mediating effect of self-reported quality of life emphasizes the material and environmental foundations of cognitive health. Deficits caused by multidimensional poverty, such as inadequate living security, limited healthcare access, and reduced social engagement, lower the level of cognitive stimulation older adults receive, accelerating cognitive decline. This finding confirms that cognitive function is not just a physiological phenomenon but is closely tied to individuals’ overall living conditions, providing a new perspective for understanding how social inequality interacts with health.

The nonlinear pattern revealed by our quantile regression results carries important theoretical and policy implications. The disproportionately stronger poverty effects at lower cognitive quantiles suggest that individuals with weaker cognitive reserve are least equipped to buffer against multidimensional deprivation—multiple dimensions of poverty compound one another, amplifying harm precisely among the most vulnerable. This is consistent with cumulative disadvantage theory, while also highlighting a policy-critical insight: since this compounding vulnerability operates among those least visible to income-based screening, China’s post-poverty-eradication governance must adopt multidimensional monitoring systems capable of identifying elderly individuals simultaneously deprived across multiple non-monetary domains before cognitive deterioration becomes irreversible.

Interestingly, our heterogeneity analysis reveals a distinct pattern: multidimensional poverty exerts a significant impact on cognitive function in urban and eastern-central regions, while this effect is not statistically significant in rural and western/northwestern regions. This contrast provides a nuanced insight into the poverty–cognition nexus in the post-poverty-eradication era. One plausible explanation is the survival priority in less developed regions: in extremely resource-poor areas, the struggle for basic survival and physiological needs may overshadow the marginal cognitive impact of multidimensional poverty indicators. Conversely, in urban and more developed areas, where basic survival needs are largely met, the variations in non-monetary poverty dimensions become more sensitive determinants of cognitive maintenance. This suggests that the pathway from poverty to cognitive health is context-dependent and mediated by the level of socio-economic development.

Our findings confirm that family multidimensional poverty acts as a systemic threat to cognitive health, a phenomenon that resonates with findings in diverse socio-economic settings. For example, Trani et al. (2022), in their study of older adults in South Africa, demonstrated that multidimensional poverty significantly increases the risk of dementia, highlighting that material and social deprivations are shared global drivers of cognitive decline [31]. In contrast, Olivera and Tournier (2016) found in Peru that multidimensional poverty was associated with lower levels of successful ageing, suggesting that even in other developing economy contexts, poverty operates through multiple non-income pathways to undermine later-life health outcomes [53].

When compared with the South African context, our results in China reveal both commonalities and specific regional complexities. While both studies underscore that cognitive resilience is deeply embedded in socio-economic conditions, the transmission chain—the way poverty erodes cognitive function—appears to be sensitive to a country’s developmental stage. In the South African context, where extreme material deprivation often remains the primary challenge, cognitive health is directly impacted by basic survival needs [53]. In contrast, our analysis within China suggests that as a nation transitions to a post-poverty-alleviation era, the cognitive impact of poverty has become increasingly mediated by more complex psychological and social-environmental pathways. These cross-national comparisons suggest that while cognitive decline is a global concern, the specific mechanisms linking poverty to cognition—and thus the most effective intervention points—depend on a country’s development stage.

From a sustainability perspective, this study connects micro-level family poverty issues to the macro-level goal of sustainable well-being. Cognitive impairment resulting from family multidimensional poverty not only reduces older adults’ quality of life but also increases the burden of family care and social healthcare systems, which undermines the long-term sustainability of social structures. Our findings confirm that alleviating family multidimensional poverty constitutes a necessary, though not sufficient, condition for realizing sustainable well-being among older adults, particularly as non-monetary deprivations persist even after income poverty has been formally eradicated. This expands the application of the United Nations Sustainable Development Goals (SDGs) in aging research, providing empirical support for the coordinated advancement of SDG1 (No Poverty) and SDG3 (Good Health and Well-being).

This study has several limitations. First, the cross-sectional research design limits our ability to fully establish a causal relationship between family multidimensional poverty and older adults’ cognitive function. Future research should use longitudinal data to track dynamic changes in both variables over time. Second, the measurement of multidimensional poverty mainly focuses on objective dimensions such as economic status, healthcare, and education, with insufficient attention to older adults’ subjective experiences (e.g., sense of deprivation). Additionally, the study does not explore the moderating role of informal support, such as family care and community assistance, which deserves further investigation in future work.

6. Conclusions and Policy Implications

Using data from Chinese older adults, this study investigates the impact of family multidimensional poverty on older adults’ cognitive function, its mechanisms, and relevance to sustainable well-being.

Family multidimensional poverty—characterized by comprehensive deprivation in economic security, healthcare access, educational support, and living environment—exerts a significant negative effect on older adults’ cognitive function, with more profound impacts than single-dimensional economic poverty. This finding underscores that multidimensional poverty is not merely an economic issue but a systemic threat to the framework of sustainable well-being for older adults. Mental health and self-reported quality of life act as partial mediators in this process. Together, they form a transmission chain linking poverty to psychological distress, reduced subjective well-being, and ultimately cognitive impairment—revealing that poverty erodes sustainable well-being through both material and psychological channels.

Our results reveal that the pathway from poverty to cognitive impairment is deeply embedded in regional and demographic disparities. Whether it is the ‘survival-level’ vulnerability in resource-constrained regions or the high sensitivity to social-environmental deprivations in more developed centers, cognitive health is a sensitive barometer for the broader socio-economic structure. This study suggests that advancing the sustainable well-being of an aging society requires a shift in perspective: cognitive resilience should be viewed as a core pillar of development. Understanding the psychological and social wear caused by multidimensional poverty is thus critical for future sustainability strategies, not just to alleviate current poverty, but to ensure that the elderly population can maintain cognitive longevity and functional independence in a rapidly changing social landscape.

These findings carry particular urgency in the context of China’s post-poverty-eradication transition. With absolute poverty officially eliminated in 2020, income-based interventions have largely fulfilled their mandate, yet our results demonstrate that income poverty alone fails to capture the full cognitive burden of deprivation, as non-monetary dimensions including healthcare access, education, and psychosocial well-being exert additional cognitive costs that income-based metrics systematically overlook. Given that health poverty and income poverty drive the largest cognitive penalties in our results, post-eradication policy should redirect resources toward elderly populations with concurrent healthcare and economic deprivation, rather than maintaining universal income-based transfers alone. First, cognitive health screening should be integrated into existing community-based elderly care programs, enabling early identification of at-risk individuals, particularly those suffering from health and educational deprivation, which our results identify as the most cognitively damaging dimensions—before impairment becomes irreversible. Second, targeted psychosocial interventions—such as community mental health programs and social engagement initiatives—should be designed to specifically break the “poverty → psychological distress → cognitive decline” transmission chain identified in our mediation analysis, with priority given to older females and urban elderly populations where our heterogeneity analysis reveals the greatest vulnerability.

Our findings challenge the feasibility of a one-size-fits-all policy framework for cognitive health. The cognitive decline we observed is not merely a physiological endpoint; it is the culmination of long-term socio-economic constraints. In resource-constrained rural and Western regions, where fundamental survival often remains the primary challenge, cognitive health is largely tied to basic welfare stability—such as pension coverage and rural medical insurance. Here, stabilizing the economic baseline is the most effective cognitive-protective intervention.

In contrast, in the more developed Eastern and Central urban centers, where elderly cognitive function is highly sensitive to the nuances of social participation and healthcare access, policy focus must shift toward targeted, multidimensional cognitive health interventions. This entails a transition from passive financial relief to active social-environmental engineering—specifically breaking the ‘psychological distress—social withdrawal’ feedback loop identified in our mechanism analysis. Furthermore, the disproportionate impact on older females highlights an urgent need for gender-sensitive support systems that move beyond generic community services to target specific psychological stressors. Only by coordinating poverty alleviation, healthcare, and geriatric services through a unified, long-term monitoring mechanism can we address both the material and behavioral root causes of cognitive aging in China’s diverse developmental landscape.