Study design
This work adopted a cross-sectional study design nested in the project “Accessibility Evaluation of Health-Related Resources for the Elderly (AEHRR). All participants provided written informed consent before participation. The study was approved by the institutional review board at the School of Public Health, Zhejiang University.
Setting and participants
A total of 950 general residents aged ≥ 60 years were selected using a complex multistage sampling design. Sampling was conducted in 22 locations (urban/rural) in four provinces (Zhejiang, Heilongjiang, Xinjiang, and Chongqing) in China from 2019 to 2020. A population-based survey was conducted to assess the health and health-related resources of accessibility from multiple aspects: social, economic, behavioral and psychological, medical, and environmental among the general elderly. Subjects who were unable to complete the questionnaire were excluded.
Data collection
Data were collected using questionnaires provided during face-to-face interviews. Part of the family interviews was conducted by investigators who were guided by villagers proficient in the local language. The questionnaires consisted of 9 parts comprising 428 items. The main contents included the following: social resources, medical resources, community health service resources, psychological resources, environmental resources, and activities involving daily living assessment.
Measurements
Dependency scale was validated as a measure of dependency from the standardized Minnesota Multiphasic Personality Inventory-II in Chinese version. The dependency scale comprised 57 items. The raw score was converted into a standardized T-score. Dependency is defined as the standardized T-score greater than or equal to 60 points. Social resource status was assessed using the Chinese version questionnaires of Older American Resources and Services (OARS). The OARS social resource scale included three dimensions: social interaction, availability of social support and practical assistance, and interpersonal relation. The ratings were summed to yield a total score. High scores indicate high levels of social resource. Two potential variables were distinguished with four questions for the availability of social support and family assistance. The availability of social support was measured using two questions as follows: “Is there someone who would give you any help at all if you were sick or disabled?” and “Do you find yourself feeling lonely?” The family assistance included two questions: “Who do you live with at present?” and “Do you see your relatives and friends as often as you want to or not?”
Statistical analysis
Statistical analysis was restricted to the 913 participants with complete questionnaires and dependency assessment data. Descriptive statistics were used to describe the general characteristics of the study participants.
The logistic regression model was used to evaluate the association between dependency and social resources. The dependency score was treated as a binary variable. If the participant’s T-score of dependency scale is greater than or equal to 60 points, then they will be regarded as a dependent individual in the binary dependent variable of the logistic regression model expressed by “1”; otherwise, “0” if their score is lower than 60 points. As a continuous variable, the score of social resources is the sum of the scores of the three dimensions. High scores are considered to have more social resources.
In multiple linear regression analysis, the dependency score was included in the analyses as a continuous dependent variable. A separate standardized estimate was calculated for the four dimensions of social resources to evaluate the association between social resource level and dependency score from different dimensions. The scores were adjusted for age, gender, education level, marital status, individual income, smoking status, alcohol use, physical activity, and chronic disease status.
Analysis of covariance (ANCOVA) was performed two times by using the general linear model (GLM) procedure. In model 1, the change in dependency score was evaluated based on the four categories. The four categories were derived according to different chronic disease status and social support. The four categories were represented by groups. Group 1 is the reference group and consists of individuals who reported one or more chronic diseases and without social support. Group 2 consists of individuals without chronic diseases but without social support. Group 3 consists of individuals with chronic diseases and social support. Group 4 consists of individuals without chronic diseases but with social support. Variance homogeneity test and normality test were performed. The mean and standard error were calculated for dependency score among the four groups. The linear trend was assessed using a general linear model. Comparisons were conducted among the four groups by using an F test with a significance level of 0.05. All analyses were performed using SAS for Windows (version 9.4).
Results
The demographic characteristics of the study participants are shown in Table 1. The mean age of the participants was 68.5 years. Also, 10.3% of males and 14.4% of females were categorized as dependency in accordance with their response on the 57-item dependency scale.
Table 2 shows the significant negative association between the levels of social resource and dependency by binary logistic regression analyses. After adjusting for age, gender, chronic disease status, individual income, education level, physical activity, marital status, smoking and alcohol use, the odds ratio was 0.78 (95% CI, 0.73–0.84), p<0.001.
The β coefficients of social resources score in different dimensions and important independent variables for dependency score are shown in Table 3. The levels of social support were most strongly negatively associated with dependency score compared with those of the other three dimensions of social resources, such as social communication, family support, and interpersonal relationship in the multiple regression (p<0.001). The results of multiple linear regression analysis also showed that dependency score was significantly negatively associated with the individual income, education level, and physical activity but positively associated with the chronic disease status. The difference in the dependency score between males and females was significant. The association with dependency among males was significant than among women.
The differences in the mean dependency score between different chronic disease status and social support level are shown in Table 4. Group 2 (without chronic disease and without social support) had lower dependency score than group 1 (chronic disease and without social support). Group 3 (with chronic disease and with social support) had lower dependency score than group 2. Group 4 (without chronic disease and with social support) had lower dependency score than group 3. The test for linear trend of mean was significant. Similar results were obtained in other covariance analyses. Groups 1 (low income and without social support), 2 (high income and without social support), 3 (low income and with social support), and 4 (high income and with social support) had gradually decreasing mean dependency scores, resulting in a significant linear trend.