Relationship Between Social Network, Social Support and Health Behaviour in People With Type 1 and Type 2 Diabetes

Cross-Sectional Studies

Nana F. Hempler; Lene E. Joensen; Ingrid Willaing

Disclosures

BMC Public Health. 2016;16(198) 

In This Article

Methods

We used the cross-sectional research design in two surveys, selecting 3500 people from Steno Diabetes Center, a specialist clinic in the Copenhagen area in Denmark (people with type 1 diabetes; N = 2419 and people with type 2 diabetes; N = 1081). The methods in the survey conform to the Declaration of Helsinki criteria and the studies were approved by the Danish Data Protection Agency (2011-41-6443 and 2010-41-4863).

Steno Diabetes Center: Population With Type 1 Diabetes

In 2011, 3626 people with type 1 diabetes from the center received a letter with a questionnaire concerning self-management behaviours, social networks, health, and psychosocial aspects such as diabetes distress and social support.[21] The letter also included a prepaid envelope. Two weeks after the questionnaire was sent out, the first reminder was sent to non-respondents, followed by a second reminder at four weeks. Both reminders included a new questionnaire and a prepaid envelope. An email and telephone service operated by a diabetes nurse was available for patients during the data collection. Data collection ended after 6 weeks.

Steno Diabetes Center: Population With Type 2 Diabetes

In 2010, 2045 people with type 2 diabetes received a letter with a questionnaire focusing on preferences for patient education, social networks and self-management behaviours.[22] The letter contained information about the survey and a web address for the online questionnaire. Two weeks after the first mailing a reminder was sent. Two weeks after the reminder was mailed, a third letter was mailed to those who had not completed the questionnaire, which included a paper version of the questionnaire and a prepaid return envelope. Data collection ended after 6 weeks.

Measurements

We included validated questions about health behaviour, education level, social support and social network in both questionnaires. We used two items from a validated revised version of the Summary of Diabetes Self-care Activities measure to assess health behaviour related to physical activity and diet.[23] The items assessed general diet and general exercise: 'On how many of the last seven days did you participate in at least 30 min of physical activity?' and 'How many of the last seven days have you followed a healthful eating plan'? Information concerning education level was measured in a single item: 1) primary and lower secondary education, 2) upper secondary and vocational school, 3) medium higher education (3–4 years), 4) long higher education (≥5 years). Data on sociodemographic data (sex and age) and diabetes type were obtained from the electronic patient record at Steno Diabetes Center.

We distinguished social relations by structure (social network) and function (social support), as described by Due et al..[24] To measure social relations, we used validated questions from the Danish Health and Morbidity Survey.[25] Questions about social network included: 1) cohabitation status (living without a partner or spouse, yes/no); 2) frequency of contacts with friends (almost every day/once or twice a week/once or twice a month vs. less than once a month/never); 3) frequency of contacts with family (almost every day/once or twice a week/once or twice a month vs. less than once a month/never). Perceived social support was measured by a validated item from the Danish Health and Morbidity Survey, assessing to what extent participants were certain of counting on help in case of severe illness (definitely/maybe vs. no).[25]

Statistical Analysis

Using logistic regression models, we compared social network and social support by diabetes type. Initially, we explored if education level and gender might modify the effect of the association between diabetes type and social relations by calculating the relative excess risk due to interaction (RERI), using a method outlined by Anderson et al..[26] We applied this method because interpretation of interaction effects in logistic regression models differs from linear regression models.[27] In accordance with the method, education level was dichotomized into primary/lower secondary education and upper secondary/medium/higher education. In case of no positive or negative RERI scores with associated confidence intervals above or below 0, we included education level and gender as a possible confounder in the models. All analyses were adjusted for age and diabetes duration. Additional covariates were added in the models based on assumptions of causal relations. For example, cohabitation status was assumed to influence contacts with the social network and therefore we adjusted for cohabitation status when we explored the association between diabetes type and social network. Furthermore, perceived social support was expected to be influenced by contact with social network and analyses exploring the association between diabetes type and social support were adjusted for social network variables.

Generalised linear regression models with 95 % confidence intervals (CI) were applied to analyse the association between diabetes type and diet/physical activity. Initially, we explored interaction terms between education level and diabetes type and between gender and diabetes type with respect to health behaviour. If interaction terms were significant (P < .05), analyses were stratified according to the specific variable. If interaction terms were non-significant, we included education level and gender as covariates in the models. Social network variables and social support variables were included stepwise in the models based on assumptions of causal relations. If education level and social network variables were significantly associated with the variable of interest, they were included in subsequent steps.

All analyses were adjusted for age and diabetes duration. Data were analysed using SAS statistical version 9.2 (SAS Institute Inc., Cary, NC, USA).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....