RESUMO
Little is known about how health-related behaviours cluster across different populations and how lifestyle clusters are associated with type 2 diabetes (T2D) risk. We investigated lifestyle clusters and their association with T2D in a multi-ethnic population. 4396 Dutch, 2850 South-Asian Surinamese, 3814 African Surinamese, 2034 Ghanaian, 3328 Turkish, and 3661 Moroccan origin participants of the HELIUS study were included (2011-2015). K-medoids cluster analyses were used to identify lifestyle clusters. Logistic and cox regression analyses were performed to investigate the association of clusters with prevalent and incident T2D, respectively. Pooled analysis revealed three clusters: a 'healthy', 'somewhat healthy', and 'unhealthy' cluster. Most ethnic groups were unequally distributed: Dutch participants were mostly present in the 'healthy' cluster, Turkish and Moroccan participants in the 'somewhat healthy' cluster, while the Surinamese and Ghanaian participants were equally distributed across clusters. When stratified for ethnicity, analysis revealed three clusters per ethnic group. While the 'healthy' and 'somewhat healthy' clusters were similar to those of the pooled analysis, we observed considerable differences in the ethnic-specific 'unhealthy' clusters. Fruit consumption (3-4 days/week) was the only behaviour that was consistent across all ethnic-specific 'unhealthy' clusters. The pooled 'unhealthy' cluster was positively associated with prediabetes (OR: 1.34, 95%CI 1.21-1.48) and incident T2D (OR: 1.23, 95%CI 0.89-1.69), and negatively associated with prevalent T2D (OR: 0.80, 95%CI 0.69-0.93). Results were similar for most, but not all, ethnic-specific clusters. This illustrates that targeting multiple behaviours is relevant in prevention of T2D but that ethnic differences in lifestyle clusters should be taken into account.