Novel subgroups of obesity and their association with outcomes: a data-driven cluster analysis

Abstract

Abstract Background Obesity is associated with various complications and decreased life expectancy, and substantial heterogeneity in complications and outcomes has been observed. However, the subgroups of obesity have not yet been clearly defined. This study aimed to identify the subgroups of obesity especially those for target of interventions by cluster analysis. Methods In this study, an unsupervised, data-driven cluster analysis of 9,494 individuals with obesity (body mass index ≥ 35 kg/m2) was performed using the data of ICD-10, drug, and medical procedure from the healthcare claims database. The prevalence and clinical characteristics of the complications such as diabetes in each cluster were evaluated using the prescription records. Additionally, renal and life prognoses were compared among the clusters. Results We identified seven clusters characterised by different combinations of complications and several complications were observed exclusively in each cluster. Notably, the poorest prognosis was observed in individuals who rarely visited a hospital after being diagnosed with obesity, followed by those with cardiovascular complications and diabetes. Conclusions In this study, we identified seven subgroups of individuals with obesity using population-based data-driven cluster analysis. We clearly demonstrated important target subgroups for intervention as well as a metabolically healthy obesity group.

Keywords

• Obesity
• Diabetes
• Cluster analysis
• Heterogeneity
• Prognosis