Data source
The present study utilized data from the Longitudinal Ageing Study in India (LASI) wave 1 (2017–18), a nationwide and state-representative survey of aging and health. The first wave of the LASI surveyed 72,250 samples of individuals aged 45 and above, covering all 35 Indian states and union territories [30]. The main objective of the LASI survey is to provide longitudinal valid, reliable data on the geriatric population’s socioeconomic and health status, program and policy coverage status, and others. To arrive at the final units of observation, the LASI used a multistage stratified area probability cluster sampling design. LASI used a three-stage sample design in rural areas, while in urban areas, they used a four-stage sample design. The national report of LASI, wave 1, 2017–18, India, contains detailed information on the sampling framework and sample size selection [30].
Study sample
The present study used secondary data, i.e., LASI Wave 1, which includes a total sample of 72,250 people aged 45 and above and their spouses, regardless of age, with no missing values in age reporting. The participants were selected using a multistage stratified area probability cluster sampling design. The face-to-face interviews were used to interview the respondents in their households [30].
The participants were older individuals in our study, aged 60 and above, who were currently married, ever married, or unmarried. The final sample size of the study was 30,663 older individuals after excluding the respondents aged 59 years and below (n = 40,786), those who were in a live-in relationship (n = 170), and those who did not respond to self-rated health (n = 661). The details of the inclusion and exclusion criteria of the study sample are presented in Fig. 1. Since a live-in relationship is not treated as a married or unmarried status in India, we have removed it from the dataset considering Indian culture.
Variable description
Outcome variable
Self-rated health
In the individual schedule, a question was asked to the respondents, "Overall, how is your health in general?" with responses of "Very good," "Good," "Fair," "Poor," and "Very poor." The outcome variable, i.e., self-rated health, is binary in nature in the present study. We considered fair, poor, and very poor as poor (coded as 1), whereas very good and good are considered good (coded as 0) [31].
Explanatory variables
Marital status and living arrangement
Contemporary evidence has categorized marital status into several categories: single, married, widowed, divorced, and separated [32]. However, our study aims at marriage and its role in the subjective health of an individual. It does not focus on other non-married categories despite knowing that the association across different categories of marriage may differ. Thus, our study has categorized marital status as binary classification with “1” those who responded married as “currently married” and all other categories as “2” those who responded widowed, never married, separated, divorced, and deserted as “currently unmarried.” The previous study suggests that living arrangement is a key determining factor of subjective health at later stage of life [33]. Therefore, the current study also included living arrangements as a key explanatory variable of SRH among the older population. Thus, the living arrangements of older adults have been categorized as binary classification with “1” those who responded living with spouse / or others, living with spouse and children, living with children and others, or living with others as “Co-residing” and “2” those who responded living alone has recorded as “Living alone” [34] (Fig. 2).
Covariates
The analysis included and adjusted other sociodemographic, economic, and health-related characteristics (Fig. 2). Age was categorized as “60–69” years, “70–79” years, and “80 + ” years. Sex was categorized as male and female. Education was categorized as no schooling, up to the primary, up to secondary, and secondary & above. Working status was categorized as working and not working. Social participation was categorized as yes and no. Social participation was measured through the question, “Are you a member of any of the organizations, religious groups, clubs, or societies”? The response was categorized as yes and no. Life satisfaction was assessed among older adults using question a. In most ways, my life is close to ideal; b. The condition of my life is excellent; c. I am satisfied with my life; d. So far, I have got the important things I want in my life, e. If I could live my life again, I would change almost nothing. The responses were categorized as strongly disagree, somewhat disagree, slightly disagree, neither agree nor disagree, slightly agree, somewhat agree, and strongly agree. Using five statements, the life satisfaction scale was constructed as a ‘score of 5–20 as low satisfaction’, ‘score of 21–25 as medium satisfaction’, and ‘score of 26–35 as high satisfaction [35].
The six basic daily self-care activities that makeup activities of daily living include getting dressed, putting on chappals or shoes, walking across a room, bathing, eating, getting in or out of bed, and using the toilet, which includes getting up and down. Combining these six ADLs, a single variable was generated that was recorded as "no ADL" if the respondent had no difficulties performing any ADL and "ADL" if they had [35]. Additionally, IADLs included seven instrumental activity-related difficulties that were consistently performed. For example, preparing a hot meal (cooking and serving), shopping for groceries, making calls, taking medications, working in the garden or house, managing money by paying bills and keeping track of expenses, getting around or finding the address in a strange place were all taken into account when determining how well an individual could perform their instrumental activities of daily living (IADLs). IADLs were recorded as "no IADL" and "IADL," much like ADLs [35]. Morbidity status was categorized as 0, “no morbidity,” 1 as “single morbidity,” and 2 + as “multi-morbidity.” In the present study, we have measured financial condition based on the monthly per capita consumption expenditure (MPCE) computed and used as the summary measure of household expenditures. Sets of 11 and 29 questions on the expenditures on food and non-food items, respectively, were used to canvas the sample households. Food expenditure was collected based on a reference period of seven days, and non-food expenditure was collected based on reference periods of 30 days and 365 days. Food and non-food expenditures have been standardized to the 30-day reference period. Monthly per Capita Consumption Expenditure (MPCE) was coded as five quintiles, i.e., poorest, poorer, middle, richer, and richest [30]. Religion was coded as Hindu, Muslim, Christian, and others) [36]. Social group (Caste/Class) was coded as Scheduled Caste (SC), Scheduled Tribe (ST), Other Backward Class (OBC), and others. Caste is a well-known social stratification that leads to social prejudice afflicting lower castes (SCs, STs, and various sub-castes under OBCs) [36]. Caste-based discrimination is banned by legislation adopted under the Indian constitution. However, the practice of caste-based social exclusion is pervasive in India, which leads to caste-based inequalities in social and health well-being [37]. The place of residence was categorized as rural and urban. The region was coded as North, Northeast, Central, East, South, and West [30].
Statistical approach
Descriptive statistics and bivariate analysis were used in this study to evaluate the prevalence of subjective health in the country based on socioeconomic status and other characteristics. The significance level of the bivariate correlation was determined using the Chi-square test. In addition, binary logistic regression analysis was used to examine the association between marital status, living arrangements, and subjective health in older people. The equation of the logistic regression is as follows:
$$\mathrm{Logit}(\mathrm Y)=\ln(\mathrm p/(1-\mathrm p))=\alpha+{\mathrm\beta}_1{\mathrm X}_1+{\mathrm\beta}_2{\mathrm X}_2+{\mathrm\beta}_3{\mathrm X}_3\dots...\beta_{\mathrm k}{\mathrm X}_{\mathrm k}+\;\mathrm\varepsilon\;$$
The regression coefficients in this example were β1, β2… … …βk, and they showed the relative effect of explanatory variables and sociodemographic and health behavioral factors on the dependent variable, with the coefficients changing depending on the context of the studies. The results from the adjusted odds ratio estimated the interaction effects of marital status and living arrangements on subjective health in older individuals in India. Interaction estimates have been adjusted for all other factors. The interaction effects (Marital status # Living arrangement) were used for the outcome variable and key explanatory variables, and the independent effects of marital status and living arrangement on subjective health were computed.