Does consistency between actual living arrangements and expectations (ALIVE) matter? A population-based longitudinal study of older adults Authors: Ting Chen, Sicheng Li, Tianjiao Lan, Wenqiang Zhang, TianFu Gao, Liyong Lu Published: 2025 ABSTRACT AbstractBackgroundEvidence indicates that living arrangements significantly influence the well-being and health of older adults. However, the psychological needs of this population have increasingly emerged as important public health concerns, with preferences regarding living arrangements serving as a key determinant. Ignoring this issue can lead to incomplete or inaccurate assessments of how various living arrangements affect their health. To better understand this impact, this study investigates the relationship between living arrangements and survival outcomes among older adults, while considering the alignment between Actual Living Arrangements and Expectations (ALivE).MethodsData for this study were sourced from the China Longitudinal Healthy Longevity Survey (CLHLS) covering the period from 2005 to 2018. Living arrangements were assessed based on participants’ self-reported expect living arrangement (indicating preference) and current situations in the questionnaire, categorized into “living alone (or with a spouse only)”, “co-residence with children”, or “living in an institution”. Survival status and date of death were determined through interviews with close family members during each survey round. Cox proportional hazards regression models were used for longitudinal analysis, and strategies for heterogeneity analysis were employed to identify vulnerable subgroups.ResultsThe study included 4,272 older adults, of whom 2,188 (51.22%) died during the follow-up period (2008 to 2018). The median survival time was 5 years. Compared to individuals with consistent living arrangements, those with discrepancies between actual and preferred arrangements had a hazard ratio (HR) of 1.36 (95% CI: 1.20–1.53), indicating a 36% increased risk of death. Sensitivity to these inconsistencies was greater among males, individuals with fixed occupations before age 60, those with average or difficult economic conditions, individuals with unfavorable dietary patterns, and those who are not currently exercising.ConclusionsThis study highlights the significance of aligning older adults’ living arrangements with their preferences. The association with health is more closely related to whether living arrangements match personal desires rather than the specific type of arrangement.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-025-26027-3.Keywords: Living arrangements, Senior care, Healthy aging, Geriatric, Longitudinal analysis Introduction The global population is experiencing an unprecedented and rapid trend of aging, which has significant implications across all societal sectors [1]. By 2030, the number of individuals aged 60 and older is expected to reach 1.4 billion, with projections rising to 2.1 billion by 2050 [2]. As the importance of aging issues increases, research has increasingly focused on factors influencing the health of older adults, including nutrition, lifestyle behaviors, economic status, and social environments. Among these, living arrangements—a component of the social environment—play a critical role in shaping health outcomes for older adults [3–5]. Many studies have examined the relationship between living arrangements and health, among older adults [4, 6–9]. However, focusing solely on the effects of living arrangements on health raises a crucial question: do preferences for specific living arrangements influence the health outcomes? Ignoring these preferences may result in incomplete or inaccurate assessments of how various living arrangements is related to older adults’ health. For instance, some research indicates that older adults living with others have lower mortality rates compared to those living alone [9–11]. Conversely, other studies suggest that living alone might benefit health [12, 13] or find no significant health differences between living alone and other arrangements [14–16]. These inconsistencies may stem from variations in study populations, regional social and cultural contexts, and methodological differences [5]. A critical but often overlooked factor is whether older adults’ preferred living arrangements align with their actual situations [17]. For example, some older adults may prefer living alone and find cohabitation undesirable, which could reduce the sense of loneliness, thereby possibly mitigating the health risks associated with living alone [18]. Conversely, if older individuals are forced to live alone despite preferring other arrangements, this mismatch may lead to greater dissatisfaction with the arrangement, resulting in even worse health outcomes [19]. Therefore, to thoroughly understand the impact of living arrangements on older adults’ health, it is essential to consider whether living arrangements preferences align with their actual arrangements. It is evident that more deliberate research is critically required, a groundbreaking study has employed a cross-sectional design, showing a negative association between inconsistent living arrangements and self-rated health among institutionalized and community-residing older adults [17]. To mitigate potential biases to some extent, such as endogeneity arising from reverse causality, this study conducted longitudinal cohort data and statistical methods to further explore the important and progressive health issues related to psychological preferences among older adults. Notably, this study adopted a more objective health outcome, the mortality, which is less influenced by information bias and holds particular significance in research involving the older populations, who always face greater healthy vulnerability and higher rates of comorbidities [17, 20]. Theoretically, the consistence of desired and actual living styles can improve the mental health (or reduce psychological problems) of older adults, which affects their long-term overall health and risk of death [17]. According to person-environment fit theory, individuals actively seek alignment between their actual needs and environment, and outcomes are most optimal [21], postulating that health and well-being are a function of the “fit” between a person’s competencies(such as preferences or needs, values, and abilities) and their environmental demands and characteristics (such as job, home or community). This theory emphasizes, an optimal alignment (such as between preferred living arrangement and the actuality) might promote a high quality of life, but a misalignment can produce adverse outcomes, including stress, diminished performance, or the motivation to disengage from that environment [22]. Self-determination theory emphasizes individuals exercising the degree of choice and control they desire over those areas of life they consider important, the goodness-of-fit between preference and actuality, can influences their autonomous motivation, engagement, and overall wellness in related behavior or performance [19, 23]. Empirical evidence supports that living preference modifies the associations of living arrangements with loneliness, for example, those who prefer living with children but live alone are more likely to feel lonely comparing those prefer living alone [18, 24]; and dissatisfaction with the living arrangement of older adults is negatively associated with health outcomes [25], although the evidence is still very limited. Furthermore, when facing inconsistence of living arrangement, older adults from different socio-demographic backgrounds may experience heterogeneous health impacts. A key underlying mechanism for this disparity might be differential levels of psychological resilience [26]. Evidence showed that people exposed to more unfavorable social circumstances are more vulnerable to poor mental health, indicating lower psychological resilience [27]. Reviewing the social determinants of mental health, it is generally agreed that race and ethnicity (minority), socioeconomic status (lower level), gender (female) and gender identity, unemployment and precarious employment, as well as social support (less connection) are crucial factors associated with lower psychological resilience [27–30]. For example, females usually experienced more difficulty in adaptation and had greater physical reactions to stimuli originating from their surroundings [31]; socioeconomic adversity might impose psychological strain, thereby undermining individual resilience [32]; and psychological resilience is higher among those who with stable income, regardless of their actual income level [33]. Also, some health-related behaviors, including diet, exercise and bad habits, are associated with mental well-being and key non-pharmacological interventions [34]. Overall, there is a significant lack of relevant evidence focusing on the healthy reactions of the older adults, particularly with different characteristics regarding the dissatisfaction with living arrangements, and exploratory subgroup analyses need to be constructed to further describe the patterns involved. In China, where owns the world’s largest ageing population, the living arrangement for older adults has undergone significant changes in recent years [35, 36]. For example, more older adults have been living in institutions (this proportion has always been relatively low in China due to traditional culture and institutional care system [37]), or became empty-nest household due to social policy and development, such as one-child policy and urbanization [36, 38]. While previous studies have shown that a relatively high consistency rate between actual and expected living arrangement in Chinese older adults, the converging trends in cognitive health and objective socioeconomic factors with developed nations suggest a potential decline in such consistency [17, 39]. With the fundamentally altering of the structure of family, the popularity of the Internet among older adult and the increasing attention to geriatric psychology [40], exploring the relationship between their consistency and health is the basis for further optimizing the health and the welfare of the older adults. To comprehensively analyze the associations of living arrangements and older adults’ health, researchers should employ a two-step approach: [1] Assess whether discrepancies between expected and actual living arrangements are related to mortality, a major health outcome; [2] Examine the health outcomes of various living arrangements for individuals whose actual arrangements align or do not align with their preferences. Considering that older adults’ preferences may shape the association between living arrangements and their health, policies aimed at promoting healthy aging should avoid a one-size-fits-all approach. Instead, recommendations should be customized to individual preferences to better align living arrangements with older adults’ choices. Currently, there is a scarcity of studies incorporating such nuanced analyses. To address this gap, this study utilizes longitudinal data from the China Longitudinal Healthy Longevity Survey (CLHLS) to investigate whether the alignment between actual living arrangements and expectations associates with survival outcomes among older adults. Additionally, the study examines the health differences across various living arrangements for individuals whose actual living conditions are consistent or inconsistent with their preferred arrangements and aims to identify vulnerable subgroups within the older adults who may be disproportionately affected by these factors. Methods Study participants The data for this study were obtained from the China Longitudinal Healthy Longevity Survey (CLHLS) from 2005 to 2018, which is a nationally representative cohort study of older Chinese adults [41, 42]. The waves of the CLHLS, which has involved over 20,000 participants across 23 out of 31 province-level units in mainland China, and standardized face-to-face interviews were conducted by trained interviewers with a structured questionnaire [42]. The ethical committee of Peking University granted ethical approval for the CLHLS study protocol (IRB00001052-13074), and written informed consent was obtained from all participants prior to their inclusion, and further survey details can be found in previous publications [43, 44]. A longitudinal design was conducted in this study, the actual and preferred living arrangements and covariates were measured at baseline, and the survival status was tracked from following waves in 2008, 2011, 2014 and 2018. For individuals entering the study in 2005, baseline information from 2005 was used, with survival data tracked from 2008 to 2018. For those entering in 2008, baseline information from 2008 was utilized, with survival data collected from 2011 to 2018, and so on. A total of 4,272 participants were included in the final analysis (Fig. 1), the exclusion criteria included: [1] Participants lacking information on living arrangements in any wave; [2] Individuals entering in 2018 were excluded from the analysis due to the lack of follow-up data on their survival status; [3] Participants whose living arrangements changed during the following waves, considering the individuals who change their living arrangement in this period may do so due to alterations in their health status, these individuals might also be more susceptible to adverse health outcomes, potentially increasing the likelihood of reverse causation and bias; [4] Participants who reported as dead in 2005 or entered in 2018; [5] Participants aged less than 65 years; [6] Participants lacking information on survival status and with erroneously coded survival outcomes (reference to previous literature [42]), or any covariates. Participants classified as “unknown” for any of the variables were treated as missing values and excluded from the final dataset. The diagram of data cleaning Living arrangement and survival outcome The actual living arrangements of participants were assessed using information from two sources. At baseline, trained interviewers asked participants, “Who are you currently living with?” The responses were categorized as follows: [1] living with family [2], living alone, or [3] living in an institution (e.g., senior center, assisted living facility, or welfare center). If participants indicated that they lived with family, further questions were posed to gather details about the identity and characteristics of other residents. Based on this information, the actual living arrangements were classified into three categories: [1] co-residence with children [2], living alone (or with a spouse only), and [3] living in an institution. To assess preferred living arrangements, the survey included the question, “Which living arrangement setting do you prefer?”. Participants could choose from five options, which were set up according to the generally expected living arrangements of the older adults in China: [1] living alone (or with a spouse only) regardless of the residential proximity to children [2], living alone (or with a spouse only) with children living nearby [3], co-residence with children (selected as long as there was at least one child living together) [4], living in an institution (e.g., senior center, assisted living facility, or welfare center), and [5] do not know. Responses categorized as “do not know” were treated as missing values and excluded from the analysis. To keep the information from the two survey questions of actual and expected living arrangement consistent, responses were further consolidated into three categories in the two variables: [1] living alone (or with a spouse only) [2], co-residence with children, and [3] living in an institution. Living arrangement concordance (coded as 1) was defined as when a participant’s preferred and actual living arrangements were the same. Discordance (coded as 0) was defined when the preferred and actual living arrangements differed. The primary outcome of interest was survival status, defined as the time from baseline until death from any cause, with follow-up concluding in 2018. Survival status and the date of death (year) were obtained through interviews with close family members conducted during each survey round. Participants were tracked from their initial interview until the occurrence of the outcome event or the latest interview. Participants who could not be contacted for follow-up after the baseline assessment were classified as “lost to follow-up” and censored in the survival analysis. Covariates Several potential confounding factors incorporated at baseline encompassing sociodemographic characteristics, health conditions, living conditions, lifestyle behaviors, and trauma experiences [42]. Sociodemographic characteristics include gender, year of birth (as a proxy for age), educational level, marital status, and work status prior to age 60, indicating the self-identified occupation before the general retirement age. To improve statistical power and distinguish whether the occupations were more likely to have stable welfare protection after retirement, such as medical insurance and pension insurance, which is related to the older adults’ living conditions and some health-related decisions, two categories were constructed: the first one named “stable employment” (including professional technical, governmental, institutional, managerial, commercial, service, industrial, and military personnel) and the other one named “unstable employment” (including self-employed, agriculture, forestry, animal husbandry, fishery, houseworker, never worked, and others). Health conditions involve health status rated by investigator, visual acuity, use of dentures, personal status of chronic diseases, cognitive competence, activities of daily living (ADL) capacity, and body mass index (BMI). Living conditions comprises type of birthplace, type of residence, source of drinking water, self-rated economic level, caregiver-supported status, and frequency of hospital visits. Lifestyle behaviors cover reading newspaper or books, participation in social activities, dietary diversity, staple food types, smoking and drinking habits, whether engaged in manual labor, frequency of exercise, and travel beyond the home county or city in the past two years. Trauma experiences include experiences of child loss, instances of frequently going to bed hungry as a child and parental survival status. In addition, the regression models included fixed effects for province of residence and year of cohort entry to control for unobserved regional and temporal confounders. Definitions of variables and details are shown in Texts S1 to S4 of the supplementary file. Data analysis Descriptive statistics of the involved participants were conducted, and the study comprises two main components to examine the relationship between living arrangements and survival outcomes: Firstly, cox proportional hazards regressions were employed to estimate hazard ratios (HRs) associated with consistency between actual and preferred living arrangements and survival outcomes. Compared to cross-sectional analyses, longitudinal designs and Cox regression, to some extent, provide greater potential for causal inference and robust effect estimates [42]. Covariates that violated the proportional hazards (PH) assumption were stratified in the regressions. A comprehensive description of the model specifications showed in Text S5. This component consisted of three parts: [1] We analyzed the consistency between actual and preferred living arrangements and survival outcomes among older adults. Also, to reduce the potential endogeneity inherent in observational studies where random assignment is not feasible, the results derived from Propensity Score Matching (PSM) and Inverse Probability Weighting (IPW) were analyzed, balancing out health and sociodemographic differences between the different consistence groups; [2] A stratified analytical strategy was employed to attain a nuanced understanding of the heterogeneous health effects associated with diverse living arrangements among older adults in consistent and inconsistent groups, respectively, thereby enabling the identification of underlying patterns and the relative vulnerability of specific subpopulations in living arrangement. We examined the association of different living arrangements (including living alone or with a spouse, co-residence with children, and living in an institution such as a senior center or welfare center) and survival outcomes for individuals whose actual living arrangements were consistent with their preferred arrangements. Next similarly, we investigated the association of different living arrangements and survival outcomes for individuals whose actual living arrangements were inconsistent with their preferred arrangements. Secondly, to further distinguish the vulnerable populations, subgroup analyses were performed. Potential modifiers were selected from a set of predefined variables, and participants were stratified based on selected potential modifiers: gender (male or female), residence type (rural or urban), work status before age 60 (stable or unstable employment), self-rated economic status (wealthy and generally satisfactory or difficult), dietary pattern (favorable, intermediate, or unfavorable), and exercise status (none, formerly exercised, or currently exercising). All data were managed in STATA 15.0 and analyzed using R 4.3.1. Statistical significance was defined as two-sided p < 0.05. Results General characteristics The final dataset included 4,272 older adults, with an average age over 80. More than 15% of participants had inconsistencies between their actual and preferred living arrangements. Males represented a slightly higher proportion of the sample (52.86%) compared to females (47.14%). Over 40% of participants resided in urban areas. More than 20% reported current smoking and drinking habits. Over 60% rated their household economic condition as average compared to others in their local area, and investigators assessed over 50% of participants’ health conditions as relatively good. Additionally, 62% rated their economic status as average relative to others. Detailed descriptive information about the variables used in this study is provided in Table 1. Table 1The descriptive characteristics of participantsCharacteristics Total (N = 4, 272) Outcome a, b Death No 2, 084 (48.78) Yes 2, 188 (51.22) Exposure AlivE Consistent 3, 598 (84.22) Inconsistent 674 (15.78) Actual living arrangements Co-residence with children 539 (12.62) Living alone (or with a spouse only) 3, 567 (83.50) Living in an institution 166 (3.89) Covariates Gender Female2, 014 (47.14) Male2, 258 (52.86) Year of birth 1922 (1914, 1931) Year of education 2.72 (3.94) Work status before 60 Stable employment1, 249 (29.24) Unstable employment3, 023 (70.76) Marry status Married2, 170 (50.8) Others2, 102 (49.2) Reading newspapers/books Almost everyday636 (14.89) Sometimes479 (11.21) Never3, 157 (73.9) Participating in social activities Almost everyday204 (4.78) Sometimes646 (15.12) Never3, 422 (80.1) Dietary score Unfavorable730 (17.09) Intermediate2, 362 (55.29) Favorable1, 180 (27.62) Staple food type Rice2, 593 (60.7) Corn148 (3.46) Wheat911 (21.32) Other620 (14.51) Smoking status Now969 (22.68) Former768 (17.98) Never2, 535 (59.34) Drinking status Now948 (22.19) Former607 (14.21) Never2, 717 (63.6) Physical activity No906 (21.21) Yes3, 366 (78.79) Exercise status Now2, 289 (53.58) Former431 (10.09) Never1, 552 (36.33) Travel time 0.21 (1.14) Vision status Health3, 021 (70.72) Problem or blind1, 251 (29.28) False teeth No2, 919 (68.33) Yes1, 353 (31.67) Health status rated by investigator Surprisingly healthy1, 270 (29.73) Relatively healthy2, 487 (58.22) Moderately ill468 (10.96) Very ill47 (1.10) Cognition score 24.70 (6.69) ADL Normal3, 731 (87.34) Unnormal541 (12.66) Chronic status No1, 580 (36.99) Yes2, 692 (63.01) BMI Normal2, 111 (49.41) Under1, 466 (34.32) Over695 (16.27) Residence Rural2, 355 (55.13) City or town1, 917 (44.87) Birthplace Rural3, 542 (82.91) City or town730 (17.09) Water type Unclean1, 574 (36.84) Clean2, 698 (63.16) Self-rated economic status Wealth 628 (14.70) Average 2, 851 (66.74) Difficulty 793 (18.56) Caregiver Spouse 1, 615 (37.8) Son or daughter 2, 008 (47) Others 649 (15.19) Go to hospital timely No 417 (9.76) Yes 3, 855 (90.24) Mother alive status No 4, 181 (97.87) Yes 91 (2.13) Father alive status No 4, 250 (99.49) Yes 22 (0.51) Has child died No 2, 989 (69.97) Yes 1, 283 (30.03) Starve No 1, 295 (30.31) Yes 2, 977 (69.69) Year of cohort entry 2005 2, 679 (62.71) 2008 1, 474 (34.5) 2011 119 (2.79) Total 4, 272aFellow-up until outcome occurring or lost fellow-up, instead of baseline characteristicbContinuous variables are reported as mean ± standard deviation (M ± SD), while categorical variables are presented as number and percentage (N (%)) The descriptive characteristics of participants aFellow-up until outcome occurring or lost fellow-up, instead of baseline characteristic bContinuous variables are reported as mean ± standard deviation (M ± SD), while categorical variables are presented as number and percentage (N (%)) During the follow-up period, 2,188 participants (51.22%) died. The median survival time was 5 years (95% CI: 5.00–6.00 years, the Confidence Limits for the values of median are based on the intersection of the horizontal line at 0.5 with the upper and lower limits for the survival curve). For individuals whose actual living arrangements were consistent with their preferred arrangements, the median survival time was 6 years, whereas it was 3 years for those whose actual arrangements differed from their preferences. The survival curve and median survival times are illustrated in Fig. 2. The survival curve median survival time by the consistency between actual arrangements and expected arrangements Regression analyses After adjusting for covariates that did not meet the proportional hazards (PH) assumption, all Cox proportional hazards models in this analysis satisfied the PH assumption for both the overall model and individual variables. The results of the PH tests are detailed in Table S1 of the supplementary file. First, we examined the association between the consistency between actual and preferred living arrangements and survival outcomes among older adults. The results, shown in Fig. 3, indicate that inconsistencies between preferred and actual living arrangements are positively correlated with an increased risk of death. Compared to individuals whose actual and preferred living arrangements were consistent, those with inconsistent arrangements had a hazard ratio (HR) of 1.36 (95% CI: 1.20–1.53), suggesting a 36% increased risk of death. The association between AlivE and survival outcomes Furthermore, considering the potential endogeneity that may arise from lack of randomized assignment in the real-word observation, we conducted the models with Propensity Score Matching (PSM) and Inverse Probability Weighting (IPW) to balance out health and sociodemographic differences between the different consistence groups. A 1:1 nearest-neighbor matching protocol was employed in the PSM, and the balance of covariates before and after matching, assessed using standardized mean differences (SMD), were presented in Figure S1. Compared to individuals who has consistent living arrangements those with inconsistent arrangements had a hazard ratio (HR) of 1.40(95% CI: 1.20–1.64) in the model with PSM, and 1.49(95% CI: 1.30–1.69) in the model with IPW (Fig. 4). The results of models with PSM and IPW Next, we assessed the association between different living arrangements—living alone (or with a spouse only), co-residence with children, and living in an institution (e.g., senior center, assisted living facility, or welfare center) and survival outcomes among individuals whose actual living arrangements were consistent with their preferred arrangements. The results, depicted in Fig. 5, show no significant differences in the risk of death among these living with children (HR = 0.94, P > 0.05) or in an institution (HR = 0.98, P > 0.05), comparing with those living alone (or with a spouse only) for individuals with consistent arrangements. Additional estimates of all covariates are presented in Figure S2 of the supplementary file. The association of living arrangements and survival outcomes in different AlivE groups. Notes: (1) HR is hazard ratio; 95%CI is the 95% confidence interval. (2) The reference group is individuals living alone (or with a spouse only); “Children” means co-residence with children; “Institution” means living in an institution Finally, we analyzed the association between different living arrangements (living alone or with a spouse as the reference group) and survival outcomes among individuals whose actual living arrangements were inconsistent with their preferred arrangements. The results, illustrated in Fig. 5, indicate that older adults living in institutions faced the highest risk of death among those with inconsistent arrangements (HR = 2.43, P < 0.05), but those who living with children had no significant differences in the risk of death (HR = 1.38, P > 0.05), comparing with the reference. Additional estimates of all covariates are provided in Figure S3 of the supplementary file. When older adults’ actual living arrangements are inconsistent with their preferences, residing in institutions appears to have the most crucial association with their health. Subpopulation analyses The results of the subgroup analysis are presented in Fig. 6. The key findings are summarized as follows: Males, individuals with a fixed occupation before age 60, those with an average or difficult household economic status, individuals with an unfavorable dietary pattern, and those who are not currently exercising are more sensitive to the associations of inconsistencies between actual and preferred living arrangements compared to females, individuals who unstably employed before age 60, those with a wealthier household economic status, individuals with a favorable or intermediate dietary pattern, and those who are currently exercising. Our study showed an exploratory subgroup analysis result between sub-populations; however, these subgroup analyses were not statistically significant by testing in models including interaction terms (P > 0.05). The results of subgroup analyses Robust test To mitigate the risk of inaccurate responses from participants with cognitive impairments, we excluded individuals with cognition scores below 18 and performed the regression analysis using the same methods as in the main analysis. In addition, considering older adults lost to follow-up (censored data) may differ systematically from those retained and bias the results, we constructed competing risk model. Also, excluding respondents who answered “do not know” to the question on preferred living arrangement might cause selection bias, we included these individuals to the main regression, treating as consistent or inconsistent, respectively. And considering the small sample sizes of those living inconsistently with expectations, particularly those living in institutions, which might be related to sample selection bias, we added inverse probability weighted regression results as robustness test, the details showed in Figure S3. The results, shown in Figure S4, were consistent with the main findings, thereby confirming the robustness of our results. Furthermore, considering potential overfitting concerns, a set of models with varying covariate combinations was constructed, and the Akaike Information Criterion (AIC) was employed, which indicated a robustness result of the main regression, shown in Figure S5. Discussion This study represents a pioneering large-scale, population-based longitudinal investigation into the relationship between concordance of living arrangements and survival outcomes among older adults. Our findings highlight the following: [1] Inconsistency between preferred and actual living arrangements is positively correlated with an increased risk of death [2]. For individuals whose actual living arrangements align with their preferences, there is no significant difference in survival risk across different types of living arrangements [3]. Among those with inconsistent arrangements, older adults residing in institutions face the highest risk of death. Neglecting the consistency between expected and actual living arrangements may contribute to these inconsistencies of results in previous study [4, 6–9, 17]. Additionally, males, individuals with a fixed occupation before age 60, those with an average or difficult household economic status, individuals with an unfavorable dietary pattern, and those who are not currently exercising are more sensitive to the associations of inconsistencies between actual and preferred living arrangements exercising. Meanwhile, some subgroups might be more prone to inconsistent situations. We further presented the relevant comparisons in Table S2, indicating that older individuals, those with lower economic status, women, and those with lower educational levels, who belong to socially disadvantaged groups, are less likely to have their preferences respected. More targeted in-depth research and intervention strategies study for vulnerable groups are needed to offer comprehensive policy recommendations for improving older adults’ living arrangements and promoting healthy aging. We found a significant positive correlation between the alignment of actual and preferred living arrangements and decreased mortality risk. This also suggests that considering older adults’ preferences when arranging their living conditions is crucial. When expected arrangements align with actual ones, it can enhance life satisfaction and overall well-being [17, 20]. A positive outlook and good mental health contribute to overall health [45, 46]. Living preference might modifies the associations of living arrangements with loneliness, those who prefer living with children but live alone are more likely to increase the loneliness and undermined health [18, 24]. Although no research has fully explored the biological mechanisms through which alignment between expected and actual living arrangements affects health, our study provides more evidence to limited empirical study [17] that living preference might be related besides other factors such as physical health deterioration or care needs. When older adults’ preferences are met, they are more likely to engage proactively in life rather than passively accept their circumstances. Our results provided empirical evidence fitting person-environment fit theory [21] and Social-network theory [47], which also emphasizes the impact of social environment and community relations, and is conducive to building a more efficient and humanistic social pension pattern. And more in-depth mechanism research is urgently needed for better understanding and policy optimization. When actual living arrangements align with preferences, there is no significant difference in mortality risk among different living arrangements. This suggests that the living arrangement itself may not be a significant health factor; rather, the discrepancy between expected and actual arrangements may be the primary determinant of health outcomes. This may explain the inconsistent findings in previous research regarding the association between living arrangements and health. Our findings confirm that when living arrangements match preferences, whether living with family, alone, or in a care facility, there is no increased health risk as long as the individual prefers the arrangement. However, when there is a discrepancy between expected and actual living arrangements, living in institutions is related to the greatest health association. Among the different combinations of inconsistent living arrangements, the group of older adults who preferred living with children but living alone or with spouse accounting is the largest, accounting for 73% (490 individuals); inversely, 11% (76 individuals) wished to live alone but lived with children. Furthermore, 8.3% (56 individuals) individuals expected living in institutions but with children or alone. And around 8% (52 individuals) older adults expected to live with children and alone (or spouse) but living in institutions, although it is not the largest group in terms of proportion, special attention needs to be paid to the health conditions of this group of people. Previous studies not considering this discrepancy mostly have found that institutional living often correlates with poorer health outcomes on some indicators [48–50]. Institutional conditions in China, such as living environment and dietary conditions, may contribute to higher health risks [51]. Additionally, institutional living may lead to feelings of abandonment [52], negatively affecting mental health and overall well-being. Our study also found that men are more susceptible to the association of inconsistencies between actual and expected living arrangements compared to women, which is contrary to the expected hypothesis that being female is one of the negative determinants of mental health [53]. In Chinese culture, men traditionally assume family decision-making roles [54], and unmet expectations or passive acceptance of arrangements can create significant psychological stress, affecting men more due to their stronger desire for control [55]. The result in this study also indicates individuals stable employed before age 60 are also more affected, likely due to more predictable retirement income and a lower willingness to accept passive arrangements compared to those with less secure retirement prospects. And consistent with the research hypothesis, those reporting poorer family economic conditions are more vulnerable to adverse effects, as robust finances can mitigate health impairment by ensuring basic needs are met. Maintaining healthy lifestyle habits, such as a balanced diet and regular exercise, might also help alleviate health damage caused by inconsistencies between preferred and actual living arrangements [56, 57]. This study has several strengths. It considers older adults’ preferences when assessing the association between living arrangements and health and examines whether different living arrangements is related to health when preferences were met or unmet. This provides new evidence to address previous controversies and offers a fresh perspective on inconsistent research findings. The longitudinal design reduces reverse causality risk and enhances evidence strength. Furthermore, the study controls for a wealth of covariates, improving accuracy and credibility. However, the study has limitations. Unobserved confounders and non-random assignment may persist, complicating causal interpretation, although we used some statistical methods trying to control for confounding variables in observational studies by creating a pseudo-randomized experiment. Self-reported data may contain inaccuracies, although sensitivity analyses were performed. Healthier older adults may prefer to live alone and be more likely to choose the living arrangement themselves, although investigator-rated health status was controlled, the influence of selection bias may not be completely avoided. The study cannot differentiate between those who prefer to live alone versus with a spouse, or specify whether older adults live with sons or daughters, and those who expect to live with other relatives or non-relatives are not included, potentially limiting the analysis. Future research could collect more detailed data for further analysis, as well as the studies using other outcomes (such as psychological well-being or quality of life) also worthy of further attention and better inference. Also, the missing data of variables (around 4% of total sample size) may cause a certain systematic bias, which needs further attention. Additionally, some baseline data may change over time, though efforts were made to mitigate this by including birth year instead of age and conducting proportional hazards tests. Despite these measures, some time-dependent variables’ impacts cannot be entirely eliminated. Conclusion The findings suggest that aligning older adults’ living arrangements with their preferences is crucial for health. The study underscores that the discrepancy between expected and actual arrangements, rather than the arrangements themselves, may be the primary determining factor of health outcomes. Furthermore, it highlights that institutional living presents the greatest health association when there is a mismatch between expected and actual arrangements. Supplementary Information Supplementary Material 1: Text S1 Assessment of dietary diversity. Text S2 Assessment of cognitive ability. Text S3 Assessment of activities of daily living (ADLs) ability. Text S4 Details of covariates. Text S5 Cox proportional hazards model. Figure S1 The balance of covariates before and after PSM, assessed using standardized mean difference. Figure S2 The impact of living arrangements on survival outcomes among individuals whose actual living arrangements match their preferences. Figure S3 The impact of living arrangements on survival outcomes among individuals whose actual living arrangements differ from their preferences. Figure S4 The results for robust analysis. Figure S5 The results for models with different control variables. Table S1 The results of PH Test of Cox proportional hazard regression model. Table S2 The descriptive characteristics of participants in different AlivE Acknowledgements The data used in this study are from China Longitudinal Healthy Longevity Survey (CLHLS). The authors are grateful for the permission to use the data here. Thanks for all the participants and staffs of the China Longitudinal Healthy Longevity Survey (CLHLS) team. Abbreviations Actual Living Arrangements and Expectations the China Longitudinal Healthy Longevity Survey Activities of Daily Living Body Mass Index Hazard Ratios Proportional Hazards Propensity Score Matching Inverse Probability Weighting Authors' contributions LL designed and supervised the study. TC drafted the manuscript. TC, WZ, and TG analyzed the data. All author revised the manuscript. All authors read and approved the final manuscript. Funding This research is supported by the National Natural Science Foundation of China (Grant No. 72404173) and Natural Science Foundation of Shandong Province, China (Grant No. ZR2024QG156). The funders had no involvement in the study design, data collection, data analysis, interpretation, or writing. Data availability The CLHLS datasets, which analyzed during the current study, are publicly available at Peking University Open Research data platform (https://opendata.pku.edu.cn/dataverse/CHADS) and can be obtained after submitting a data use agreement to the CLHLS team. Declarations Ethics approval and consent to participate This study used data from the CLHLS. Ethics approval was not required to analyze these data. The CLHLS study protocol was approved by the ethics committee of Peking University (IRB00001052-13074), and written informed consent was obtained from all participants. Additional details of the CLHLS can be found in previous studies. This research was performed in compliance with the World Medical Association’s Declaration of Helsinki, which is a statement of ethical principles for medical research involving human participants. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1: Text S1 Assessment of dietary diversity. Text S2 Assessment of cognitive ability. Text S3 Assessment of activities of daily living (ADLs) ability. Text S4 Details of covariates. Text S5 Cox proportional hazards model. Figure S1 The balance of covariates before and after PSM, assessed using standardized mean difference. Figure S2 The impact of living arrangements on survival outcomes among individuals whose actual living arrangements match their preferences. Figure S3 The impact of living arrangements on survival outcomes among individuals whose actual living arrangements differ from their preferences. Figure S4 The results for robust analysis. Figure S5 The results for models with different control variables. Table S1 The results of PH Test of Cox proportional hazard regression model. Table S2 The descriptive characteristics of participants in different AlivE Data Availability Statement The CLHLS datasets, which analyzed during the current study, are publicly available at Peking University Open Research data platform (https://opendata.pku.edu.cn/dataverse/CHADS) and can be obtained after submitting a data use agreement to the CLHLS team.