This study is the first to assess the associations between T2DM and a broad range of functional and structural network characteristics in adults. The study shows that more socially isolated individuals (smaller social network size) more frequently had newly diagnosed and previously diagnosed T2DM, while this association was not observed with pre-diabetes. In women, proximity and the type of relationship was associated with newly diagnosed and previously diagnosed T2DM. A lack of social participation was associated with pre-diabetes as well as with previously diagnosed T2DM in women, and with previously diagnosed T2DM in men. Living alone was associated with higher odds of previously diagnosed T2DM in men, but not in women. Less emotional support related to important decisions was associated with newly diagnosed T2DM in women, and both newly and previously diagnosed T2DM in men. Less practical support related to jobs was associated with previously diagnosed T2DM in women and newly diagnosed T2DM in men. Less practical support for sickness was associated with newly diagnosed and previously diagnosed T2DM in men and women. These associations were not observed in pre-diabetes.
All associations between social network characteristics and diabetes status were independent of BMI, educational level, employment status, alcohol consumption, smoking status, general health status and chronic conditions as prior CVD and hypertension.
Structural social network characteristics
The present study showed that social isolation, indicated by a smaller social network size, was associated with higher odds of newly diagnosed and previously diagnosed T2DM in men and women. This finding is in line with longitudinal analyses conducted by Altevers et al. (2015), and Lukaschek et al. (2017) who found that poor structural support (measured by Social Network index [SNI], including a measure of social network size) increased the risk of T2DM [9, 27]. In addition, our data show that a smaller social network size was only associated with T2DM, not with pre-diabetes. This is also consistent with longitudinal data, which did not find significant associations of social integration, including structural characteristics, with pre-diabetes . Furthermore, we as well as Gallo et al. (2015) observed associations between structural network characteristics and T2DM among both sexes , while Altevers et al. (2015) found this association among men, but not among women . A possible explanation for this discrepancy is that Altevers et al. (2015) limited the variability in their sample by dichotomizing the Social Network Index (SNI), while we and Gallo et al. (2015) used a continuous scale. Therefore, their non-significant findings in women may be attributable to low power .
In women, higher percentages of network members living within walking distance and higher percentages household members were associated with newly and previously diagnosed T2DM. Similarly, a network composed of fewer friends was associated with higher odds of previously diagnosed T2DM in women, suggesting that the smaller network size in T2DM is largely attributable to having less friends than those with NGM. The associations of proximity and the type of relationship with T2DM in women indicate that a network that is centralized to those with the closest relationships, with less network members at a social and geographical distance, is associated with T2DM. In men, we observed that higher percentages of household members were associated with newly diagnosed T2DM. Furthermore, these associations were again not observed in pre-diabetes.
As we are the first to address the composition of the social network in terms of proximity and type of relationship in relation to T2DM, and as significant associations have mostly been observed for women, further research is needed to corroborate our findings.
Living alone was associated with higher odds of newly diagnosed and previously diagnosed T2DM in men, but not in women. This finding is consistent with previous longitudinal studies that identified living alone as a risk factor for T2DM [11, 27], while having a partner decreases the risk for T2DM  in men but not in women. Moreover, similar to Hilding et al. (2015), we only found borderline significant associations between living alone and pre-diabetes . However, these non-significant risk estimates may be attributable to a low power, as we had a relatively small sample to address this association (less than 40 men with pre-diabetes were living alone).
The lack of social participation was associated with pre-diabetes in women and with previously diagnosed T2DM in both men and women. In longitudinal research, participation in social activities has been shown to decrease the risk of pre-diabetes and T2DM in women and the risk of pre-diabetes in men . However, in this cross-sectional study, we cannot exclude the possibility that early changes in glucose metabolism may cause non-specific complaints such as tiredness and feeling unwell, which may explain why individuals chose to limit their social participation. In either scenario, social participation may serve as a target for intervention or an indicator suitable for diabetes prevention strategies.
Functional social network characteristics
In the present study, we observed that less emotional support with important decisions was associated with newly diagnosed T2DM in women, and both newly and previously diagnosed T2DM in men. Less practical support with small jobs was associated with previously diagnosed T2DM in women and newly diagnosed T2DM in men. Less practical support for sickness was associated with newly diagnosed and previously diagnosed T2DM in men and women. Both Norberg et al. (2007) and Jones et al. (2015) showed that low emotional support was associated with T2DM in women  and older adults , although their methods used to assess functional support were less detailed. The longitudinal results from Norberg et al. (2007) suggest that low functional support increases the risk of T2DM .
To our knowledge, this study is the first to assess the association of a broad range of functional support measures with pre-diabetes, newly diagnosed T2DM and previously diagnosed T2DM. Our results indicate that emotional support in important decisions, and practical support with small jobs and in sickness were important characteristics that should be addressed in T2DM prevention strategies. However, in this cross-sectional study, we cannot assess whether participants received an absolutely lower level of functional support, or whether they perceive it as less adequate to their needs (that means relatively lower), and therefore, their satisfaction with functional support is lower. Recently, it has been shown that low social network satisfaction is associated with increased risk of T2DM .
Strengths & Limitations
A major strength of the current study was the measurement of structural and functional characteristics with the use of a name generator, one of the best known, most detailed and most widely used instruments to examine ego-centered network data . This resulted in a much broader range of structural and functional social network characteristics than assessed in previous studies. Next, we were able to examine the associations of structural and functional network characteristics in individuals with pre-diabetes, newly diagnosed and previously diagnosed T2DM compared to those with NGM. The associations of pre-diabetes and newly diagnosed T2DM have rarely been studied before. Moreover, we adjusted the analyses for several different variables, i.e. age, body mass index, educational level, employment status, smoking status, alcohol consumption, general health and chronic medical conditions, showing robust results, which makes residual confounding unlikely. Finally, the population-based design of The Maastricht Study and its size were key assets .
A few limitations should also be mentioned. The study is cross-sectional in nature, and therefore, the possibility of reverse causality cannot be excluded. Furthermore, as we performed multiple statistical tests, our analyses may include false positive results. However, the majority of significant associations had a p-value ≤0.01 or even ≤0.001, limiting the chance of false positive findings. Additionally, the present study population consisted of relatively healthy participants, as is common in population-based cohort studies, and it is possible that we did not include those in the population who were the most socially isolated. Therefore, we may have underestimated the effect sizes.
Targeting social network characteristics may prove a promising prevention strategy for T2DM. More socially isolated individuals (smaller network size) more often had T2DM. Broadening their network should be encouraged, as we have shown that a smaller social network size was associated with T2DM in both men and women. Moreover, social participation was associated with pre-diabetes and previously diagnosed T2DM, stimulating participants to became members of a club may also be considered in future intervention development. In addition, social participation may be used as an indicator in diabetes prevention strategies. Moreover, interventions aiming to generate behavioral change (e.g., physical activity) may also tailor to the social network of the participant, as it has been shown that network targeting can be used to increase the adoption of specific public health interventions . In addition, as men living alone seem to be at a higher risk for the development of T2DM, they should be indicated as high-risk group.
Moreover, targeting social network characteristics may also have benefits for other chronic conditions, as it has been shown that most of those with a long-term disorder are multimorbid , and social network characteristics have been found to associate with cardiovascular, endocrine, and immune function . In addition, social isolation and living alone have been found to increase the likelihood of mortality .