Type 2 diabetes mellitus (T2DM) has arrived at a pandemic level, affecting almost 537 million adults as reported in 2021 [1]. Psychosocial stress is recognized as a newly modifiable risk factor for T2DM [2]. Recently, social disconnection, an important source of psychosocial stress, was highlighted as a rising global problem affecting health, especially since the coronavirus disease 2019 (Covid-19) pandemic [3,4]. Social disconnection comprises the following two distinct but weakly correlated aspects: social isolation (objective aspect) and loneliness (subjective aspect). Social isolation is defined as an insufficient frequency or amount of objective social contact or being alone [5]. Loneliness generally denotes a painful feeling arising from the discrepancy between the perceived and actual levels of social contact [6]. Notably, adults with T2DM are generally susceptible to social isolation or loneliness [7]. Despite mounting evidence confirming a potentially close relationship between social disconnection and the risk of cardiovascular diseases [8,9] and mortality [10], it is unclear whether social isolation and loneliness are independent risk factors of incident T2DM.

The results of the literature investigating the relationship between social isolation or loneliness and T2DM have been controversial [11], [12], [13], [14], [15], [16], [17]. The discrepancies possibly arose from the heterogeneities in measurements of social disconnection across studies or cross-sectional designs. Another possible explanation for the controversy in the results is that all these studies, despite some of them having prospective designs, applied a single measurement of social disconnection. Recent studies using repeated measures have indicated that loneliness, specifically in a chronic pattern, significantly predicted dementia risk [18,19]. However, to the best of our knowledge, studies on how the changes in social isolation and loneliness affect the risk of incident T2DM are lacking.

Apart from the independent effects, investigating the joint effects of social isolation and loneliness on the risk of T2DM would be interesting, since the two are correlated but distinct [20]. Our previous study uncovered loneliness prominently modified the association of social isolation with cardiovascular risk [21]. However, since most previous studies concerning diabetes risk have only focused on one single aspect of social disconnection [11], [12], [13], [14], [15], [16], [17], whether social isolation and loneliness have joint effects on the risk of incident diabetes are still unclear. Alternatively, given the gene–environment interaction theory, whether these hypothesized associations can be modified by genetic risk for T2DM remains unknown.

Therefore, using individual data from the UK Biobank study, the present study aimed to investigate i) whether social isolation and loneliness were independently and jointly associated with the risk of incident T2DM; ii) which pattern of change in social isolation and loneliness increased the T2DM risk; and iii) whether social isolation and loneliness predicted the development of T2DM independently from the established genetic risk for diabetes.