Egocentric Social Network and Health Behaviors Among Older Adults with Diabetes: Analysis of National Social Life, Health, and Aging Project

PURPOSE: Managing diabetes is onerous responsibility for older adults and is often affected by social relations. By analyzing data collected with egocentric social network (ESN) measurement in National Social life, Health, and Aging Project (NSHAP), we aim to explore diabetes patients’ social network and its association with health behaviors. Specifically, this paper summarizes several indicators of ESN by age groups to navigate the impact of aging on ESN; explores whether the size of alters are correlated with several characteristics of nominated significant others (e.g., proportion of kinship) and quality of interactions; and examines whether selected indicators of ESN (e.g., proportion of kinship) are correlated with health behaviors, including exercise, smoking and alcohol use after controlling for respondent’s demographic characteristics.  

METHODS: We analyzed a subset of the NSHAP’s wave 1 data (N=3005) by selecting cases who had ever been diagnosed with diabetes in the past (N=640). This dataset was collected in 2005-2006 from nationally representative sample, whose age ranged between 57 and 85. The ESN collected three types of indicators of social network. First, size of significant people, hereafter alters, were surveyed with name rosters. Second, characteristics of alters were collected: relationship type (e.g., kinship), gender, and coresidency. Lastly, three contents of interaction, hereafter name interpreters, were questioned regarding emotional closeness, frequency of contact, and discussion of health matter. Health behaviors variables included frequency of physical activity, current smoking, and alcohol misuse. In analysis, ANOVA was conducted to explore correlation between ESN’s indicators and age groups (57-64/ 65-74/ 75-85). In addition, three logistic regression analyses were conducted to examine whether ESN’s indicators are correlated with health behaviors.

RESULTS: Age group was significantly correlated with the number of alters (p<01), number (p<.001) and proportion (p<.01) of alters living together, number (p<.01) and proportion (p<.05) of alters extremely emotionally closed, number of alters speaking everyday (p<.05), and number of alters at least somewhat talking about health matter (p<.05). In addition, we also found that network size was significantly associated with most of ESN’s indicators, except for proportion of alters at least very emotionally closed (p>.05). Lastly, we found that exercise frequency was associated with having less than high school degree (ref=some college and more) (odds ratio (OR)=0.39) and being Hispanic (ref=whites) (OR=2.33). Current smoking was associated with being female (OR=0.47), being in age group 57-64 (ref=75-85) (OR=5.00) or 65-74 (ref=75-85) (OR=3.01), having high school degree (ref=some college and more) (OR=2.40), and number of alters talking everyday (OR=1.42). Alcohol misuse was correlated with being female (OR=0.17).

IMPLICATIONS: We found that age is significantly associated with size of ESN, overall characteristics of alters, and quality of interactions with alters. Older diabetic patients with larger size of ESN appear to have social relations that can be used in mobilizing social supports. However, it was not clear how social network can contribute to health behaviors with this study. In future, with the NSHAP dataset, we recommend explanatory studies that directly test hypotheses involving specific indicators of ESN and health behaviors based on established health behavior theories.