The present study is based on data collected during the ‘Social well-being and determinants of health’ SWADES study Families of the SWADES cohort were invited to be part of the baseline questionnaire between April and May 2018 ( Saju et al., 2020a). The study catchment area was located in semi-rural region of Keezhmadu panchayat in Ernakulam, Kerala, India. The residents had mixed culture and socioeconomic background. Catchment area boundaries were precisely defined. Mapping was carried out to identify and locate all households. All the family members residing in that geographically located area aged 30 years and above were considered for the study. The objectives of the SWADES study are as follows: a) to monitor changes over time in physical, behavioural and social risk factors associated with chronic diseases and mental health comorbid conditions; b) to develop chronic disease risk prediction models and estimate the probability of having or developing a particular chronic disease within a specified period; and c) to scale up population and family health interventions and evaluate the impact. SWADES collected data pertaining to sociodemographic, physical, mental, functional, social cohesion, social support networks, behavioural risk factors, health services utilisation, cognitive function and risk of fall using respective scales or tools. A detailed description of the SWADES study and the recruitment of the sample population is presented in the SWADES protocol paper ( Saju et al., 2020b).

Participants provided written informed consent and the study was approved by the Institutional Review Board of Rajagiri Hospital (Study Reference Number: RAJH 18003).

The SWADES baseline data was used for the present study, and this study covered only two aspects of the SWADES full survey – Functional Ability and Social Cohesion.

We used a core minimum data set with cross-culturally validated assessments (mental conditions, physical health, demographics, extensive non-communicable disease risk factor questionnaires, disability/functioning and health service utilization). All study instruments were translated, back-translated, and assessed for acceptability and conceptual equivalence. Translations were done locally, by investigators fluent in English (the language of the instruments) and in the local language (Malayalam) to be used in the study. The local version was reviewed by experts in Kerala.

Sociodemographic and health conditions. Education level was ascertained, and coded as: no education, did not complete primary, completed primary, completed secondary and completed tertiary education. Type of family was described as: living alone, nuclear family, extended family and mixed family. Income was a continuous variable and was coded as quartile. We also recorded age, sex, marital status (single, married/cohabitating or divorced/widowed/separated), current occupational status, and family income.

Self-reported health conditions, such as stroke, diabetes, hypertension, heart disease, were reported by participants.

Measures of functional ability. The functional ability of each participant was measured by the World Health Organization Disability Assessment Schedule 2.0 (WHODAS 2.0), measuring health and disability ( Marti & Choi, 2020). The instrument covers six domains: 1) understanding and communicating with the world; 2) moving and getting around; 3) self-care; 4) getting along with people; 5) life activities; and 6) participation in society. Scores for each question range from 0 (no difficulty) to 4 (extreme difficulty/cannot do). The standardized global score ranges from 0 (non-disabled) to 100 (maximum disability) and this score was divided into quintiles. This measurement has been extensively validated in India and other low and middle-income countries ( Thomas et al., 2016).

Measures of social cohesion. Social cohesion was measured from the inner to the outer social circle of the participants through three items, including trust in neighbours and co-workers, community participation, perceived safety in the participant’s residential neighbourhood and general social trust. These three items were measured using a scale (see below) that has been extensively used in research conducted in low and middle income countries ( Fernández-Niño et al., 2019; Kulkarni & Shinde, 2015; Ramlagan et al., 2013).

Trust in neighbours and co-workers: The scale was standardized for convenience of interpretation. The subject responded to three questions about trust according to the response options ‘to a very great extent’, ‘to a great extent’, ‘neither great nor small extent’, ‘to a small extent’ and ‘to a very small extent’:

Trust in people in the participants’ neighbourhood;

Community participation: This item was measured with a standardized scale used in the Social Cohesion section of the SAGE questionnaire ( Kulkarni & Shinde, 2015), which indicates the frequency of involvement in community activities in the last 12 months. The subject responded to nine activities with the response options ‘never’, ‘once or twice per year’, ‘once or twice a month’, ‘once or twice a week’ or ‘daily’:

Attending public meetings in which there was discussion of local or school affairs;

Perceived safety in the participant’s residential neighbourhood: This was assessed based on two questions. The respondent answered to each according to the response options ‘completely safe’, ‘very safe’, ‘moderately safe’, ‘slightly safe’ or ‘not safe at all’:

How safe from crime and violence the participant feels when he or she is alone at home; and

All three variables were computed separately and standardized versions of three variables, trust, safety and participation were created using ‘egen’ command in STATA. Value range was then defined from 0 to 30. Based on this score, social cohesion was divided into five quintiles with lowest value in Quintile 1 (Q1) and highest value in Quintile 5 (Q5). Respondents in Q1 has lowest Social Cohesion and respondents in Q5 has highest Social Cohesion.

All statistical analysis was performed using STATA 14 and R version 3.6.3 We first compared the demographic characteristics of study participants using chi square tests, t-tests or Wilcoxon rank-sum tests, as appropriate to evaluate statistical significance. To evaluate the variables associated with the outcome of interest, we performed linear regression and calculated 95% confidence interval, and p-values to evaluate the statistical significance. Using R software, we plotted a jitter plot to study the association of disability with social cohesion. Finally, we performed path analysis to develop a hypothesis on probable causal relationship and recommendations for intervention development.

We interviewed 997 participants, aged 30 years and above and residing in 573 households of the study area. The mean age of the population was 53.9 ± 14.2 years. The majority were married or cohabiting (82.7%). More than half (54%) of the population had completed primary education. Almost all (89%) the participants were born in a village. More than half of the women were housewives (50.5%) compared to 5% of men who were house husbands. Men who were employed with paid work were triple the number of women (61.4% vs. 18.5%), highlighting a patriarchal societal set-up. Two-fifths (41.8%) of the population had very low income. More than half (60.9%) of the population resided in a nuclear family (nuclear family consists of a couple and their dependent children; higher percentage of nuclear family imply lesser number of family carers for the care of people with disability, especially in the Indian context). The percentage distribution of the sample by selected socio-demographic variables and gender is presented in Table 1.

Among the study population, the prevalence of disability based on the WHO DAS score was 19.2%. Univariate analysis was employed to identify the crude association between disability and other exploratory variables ( Table 2). The prevalence of disability is twice the amount in women as compared to men (23.9% versus 11.0%). Disability prevalence increased with increase in age, lack of formal education, and presence of one or more physical or mental health condition. Participants who were widowed, divorced or separated had a higher prevalence of disability compared to married or unmarried participants. The disability scores were higher for the participants with lower economic status.

There was significant negative association between social cohesion and disability (-5.8 (-7.7 to -4), p=0.000). It was also observed that there was an independent association between people’s trust in neighbourhood (-0.57 (-0.83 to -0.30), p<0.001) and community participation (-0.38 (-0.47 to -0.29), p<0.001) with the disability.

The mean social cohesion score was 17.04 (± 6.70). Men had a higher chance of social cohesion compared to women (1.13 (0.80 to 1.47); p<0.001). Those above the age of 70 were found to have lower social cohesion compared to the other age groups (-1.04 (-1.61 to -0.47); p<0.001). While looking into the marital status of the respondents, married respondents had a socially active life compared to unmarried, widowed or divorced (1.37 (0.14 to 2.59); p=0.029). Compared to those who had no formal education, individuals with formal education took part in more social activities (1.27 (0.44 to 2.10), p= 0.003).

Figure 1 shows the association of disability with social cohesion. It shows that individuals with higher disability score reside in neighbourhoods with low social cohesion.

Multivariate analysis was performed to understand the association between social cohesion and disability after controlling for the confounding variables age, gender, education, marital status, occupation, income, physical condition and mental health condition ( Table 2). With the influence of the confounding variables, the coefficient value was reduced, but the results remained statistically significant. There was still a strong negative association between social cohesion and disability. This indicates that individuals with lower participation in the community, trust and safety exhibited higher levels of disability.

Structural model: First, we tested the direct effect of social cohesion (predictor variable) on disability (dependent variable) without mediators. The directly standardized path coefficient was not significant. Subsequently, the mediation model was tested, which included three mediators of mental health (depression, anxiety and stress) and a direct path from social cohesion to disability. The results showed that the model was a very good fit to the data. After adding the demographic variables, such as age, gender, income, education, as shown in Figure 2, the mediation model was tested again. The final meditation model showed a very good fit to the data.

Taken together, these results show the important role of social cohesion in the relationship between mental health (depression, anxiety and stress) and disability. The effect of social cohesion on disability through mental health is very high. The model revealed that demographic variables, such as, age, income, education and mental health, had a direct effect on disability. In addition, social cohesion had an indirect effect on disability through mental health.

Our study had a few limitations. This is a cross-sectional study, which limits our ability to make causal inferences, like whether disability is pushing individuals to a poorer neighbourhood or whether the poorer neighbourhood is the pertinent factor for disability. Second, since the sample comprised of participants from catchment areas of one state in India, the results cannot be generalised. Third, though the social cohesion scale is extensively used in low and middle income countries ( Kulkarni & Shinde, 2015; Fernández-Niño et al., 2019; Ramlagan et al., 2013), it is yet to be culturally validated in our population. Third, most participants were born in villages and nearly half of the population belonged to the low income category. Though the data was taken from respondents residing in that area for more than one year, it does not allow us to adjust for the duration participants were living in their respective neighbourhoods.

Nonetheless, our study adds to the literature by presenting evidence on the association of neighbourhood social cohesion and disability conducted in low and middle income countries. Data from our study gives insight to the extent by which the effect of personal and health characteristics on functional ability is mediated by social cohesion. Moreover, data was collected only from the individuals residing in the study area for more than one year, and this indicates a defined neighbourhood.

The study shows that there is a need to focus on neighbourhood or community level aspects along with individual level aspects to ensure increased quality of life for people with disability. Interventions targeting the entire community can bring about behaviour change in individuals, such as increased physical activity, as we can argue that those who receive positive support for physical activity tend to be more physically active. Our study also calls for a need for interventions offering community-based services, which enhance neighbourhood trust, safety and participation, need to be deployed. Government and policymakers should aim for strategies that enhance neighbourhood cohesion by integrating the physical, public and social health of its people. This will reduce accessibility, availability and affordability issues with respect to disability, as well as increase living standards of people.

Figshare: Relationship between neighbourhood cohesion and disability: findings from SWADES population-based survey, Kerala, India, https://doi.org/10.6084/m9.figshare.12610607.v3 ( Saju et al., 2020c).

Figshare: Relationship between neighbourhood cohesion and disability: findings from SWADES population-based survey, Kerala, India, https://doi.org/10.6084/m9.figshare.12610607.v3 ( Saju et al., 2020c).

This project contains the following extended data: -

Questionnaire consisting of questions used for this study (both English and local language - Malayalam)

STROBE checklist for ‘Relationship between neighbourhood cohesion and disability: findings from SWADES population-based survey, Kerala, India’, https://doi.org/10.6084/m9.figshare.12610607.v3 ( Saju et al., 2020c).

Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).