Social support linked to better health literacy in rural diabetes study

In a multicenter survey of rural patients with diabetes in Guangxi, China, greater social support was associated with higher health literacy both directly and indirectly through more active health information-seeking behavior. Researchers analyzed 2,178 valid questionnaires from adults receiving care at 15 county-level hospitals and found that health information-seeking behavior statistically accounted for 36.0% of the total association between social support and health literacy. Because the study was cross-sectional and did not adjust its structural model for several potential confounders, the pathway does not establish that support causes better literacy or that changing these behaviors will improve diabetes outcomes.

The survey was conducted from January to July 2022 and followed STROBE reporting guidance for observational studies. Participants were selected through multistage stratified random sampling: five cities representing the central, eastern, southern, western, and northern regions of Guangxi were included, three counties were selected from each city, and one hospital with an internal medicine department was chosen in every county. Of 2,280 recruited patients, 2,178, or 95.5%, supplied complete usable responses; 102 questionnaires were excluded because of missing or incomplete data. The final sample had a mean age of 63.25 years, with a standard deviation of 12.71, and a mean diabetes duration of 7.96 years, with a standard deviation of 4.07.

The investigators used three validated questionnaires to measure distinct aspects of the proposed relationship. The Social Support Rating Scale assessed subjective support, objective support, and the use of available support, with total scores ranging from 12 to 66. The 38-item Diabetes Health Information-Seeking Behavior Scale covered information needs, access through healthcare professionals, books, the internet, and interpersonal communication, as well as subjective, objective, and cultural barriers; scores ranged from 38 to 190. The 14-item Health Literacy Scale evaluated functional, communicative, and critical health literacy, with higher scores indicating greater literacy. Investigators trained staff at all 15 hospitals, provided on-site guidance, checked questionnaires for completeness, and obtained disease-related information from electronic medical records.

Patients’ mean health literacy score was 30.47 out of a possible 56, with a standard deviation of 9.36. Functional health literacy was the strongest dimension, scoring 11.17 ± 3.76, followed by communicative health literacy at 10.64 ± 3.92 and critical health literacy at 8.65 ± 3.07. The relatively low critical-literacy score suggests that many respondents may have had difficulty evaluating and applying information once they encountered it, although the observational design did not directly test that explanation. Mean health information-seeking behavior was 109.45 ± 18.50. Information needs were the highest-scoring dimension at 34.35 ± 6.44, while internet-based information channels scored lowest at 4.44 ± 1.61. Healthcare professionals were the most prominent channel among the non-need dimensions, scoring 18.92 ± 4.43.

Social support was comparatively limited. The median total score was 22, with an interquartile range of 22 to 30. Median scores were 8 for subjective support, 7 for objective support, and 7 for support utilization. The results also showed that only 22.7% of the analyzed group, 494 participants, had an HbA1c below 7.0%, while 77.3%, or 1,684 participants, had values of 7.0% or higher. The study did not test whether social support or health literacy was associated with glycemic control in its mediation model, so the HbA1c distribution should not be treated as evidence of a pathway from support to clinical outcomes.

All three principal measures were positively correlated. Health literacy was associated with social support at Spearman’s r = 0.482 and with health information-seeking behavior at r = 0.513, while social support and information-seeking behavior were associated at r = 0.503; each relationship had a p-value below 0.001. The researchers then constructed a structural equation model with social support as the exogenous variable, information-seeking behavior as a mediator, and health literacy as the outcome. Its fit was described as acceptable, with a goodness-of-fit index of 0.958, adjusted goodness-of-fit index of 0.935, comparative fit index of 0.950, and root mean square error of approximation of 0.072. Bootstrap mediation analysis used 5,000 resamples.

The model estimated a direct association between social support and health literacy with a standardized coefficient of 0.285, a 95% confidence interval of 0.247 to 0.322. Social support had a stronger standardized association with information-seeking behavior, β = 0.859, with a 95% confidence interval of 0.798 to 0.920. Information-seeking behavior, in turn, was associated with health literacy at β = 0.186, with a 95% confidence interval of 0.168 to 0.204. The estimated indirect effect through information-seeking was 0.160, with a 95% confidence interval of 0.142 to 0.179, and represented 36.0% of the modeled total effect. These are statistical relationships among variables measured at one point in time, not proof that receiving support changes behavior or that behavior subsequently improves literacy.

The model’s structure provides a possible explanation for why the three factors moved together. Social support may give patients greater confidence, access, or encouragement to seek diabetes information; that information may then help them understand and use health services more effectively. This interpretation is consistent with the Health Belief Model cited by the authors, in which external cues can motivate health-related action. The finding also echoes earlier research in the same rural population linking diabetes knowledge with information-seeking behavior, but the present study extends that work by explicitly modeling social support and health literacy together. Its mediation estimate nonetheless remains hypothesis-generating because temporal sequence and causation cannot be established.

Several design features strengthen the internal organization of the survey. It covered 15 hospitals in five geographically distributed cities, used multistage stratified random sampling, applied previously validated Chinese-language scales, and analyzed a large sample with an explicit structural model. The information-seeking scale had a reported content validity index of 0.92 and Cronbach’s alpha of 0.866; the health literacy instrument had a Cronbach’s alpha of 0.868, and the social support scale had reported alpha values of 0.89 to 0.94. Investigators also checked data entry, excluded incomplete questionnaires, assessed model assumptions, and used robust estimation for minor deviations from multivariate normality.

Important constraints limit how far the results can be generalized. All participants were hospital patients in one Chinese province, and people who rarely or never attend county-level hospitals—including some residents of remote villages—may have been underrepresented. Age, education, income, diabetes duration, HbA1c, and insulin use were not adjusted for in the structural equation model to preserve parsimony, leaving possible residual confounding. The variables were self-reported and collected at a single time point, creating the possibility of common-method and social-desirability biases, while the researchers did not statistically assess common-method bias. The information-needs dimension was also removed from the final model because of low factor loadings and nonsignificance, despite its high descriptive score. Future longitudinal and interventional studies would be needed to test whether strengthening support or specific information channels actually changes health literacy, and community-based recruitment could clarify how the pattern compares with patients outside hospital care.

Subject of Research: Cross-sectional relationships among social support, health information-seeking behavior, and health literacy among rural patients with diabetes in Guangxi, China.

Article Title: Study on the relationship between social support, health information-seeking behavior, and health literacy among rural patients with diabetes: a cross-sectional study

Article References: Chen, J., Lei, X., Bai, C., Han, J., Fu, G., & Zhang, Y. (2026). Study on the relationship between social support, health information-seeking behavior, and health literacy among rural patients with diabetes: a cross-sectional study. Frontiers in Endocrinology, 17, Article 1921954. https://doi.org/10.3389/fendo.2026.1921954

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1921954

Keywords: rural diabetes, health literacy, social support, health information seeking, cross-sectional study, Guangxi China, patient education, structural equation modeling, glycemic control, relationship, between, social