| Sumario: | Obesity is a heterogeneous condition encompassing different phenotypes that may respond differently to bariatric surgery. We sought to identify homogeneous groups of patients who might require specific management adapted to their characteristics. We use the Severe Obesity Outcome Network (SOON) prospective cohort of subjects with class II and III obesity to identify groups of people with homogenous characteristics. An unsupervised probabilistic clustering method, latent class analysis, was used. The clusters obtained with this analysis were compared for their outcomes after bariatric surgery up to 5-years postoperatively. 965 subjects were analyzed, enabling five clusters to be described. Cluster 1: middle-aged women with few comorbidities. Cluster 2: young, healthy women. Cluster 3: men and women with the highest BMI and frequent comorbidities. Cluster 4: relatively healthy young men. Cluster 5: older men and women with very high frequency of comorbidities. Cluster 2 had the best total body weight loss. Cluster 3, which had the most severe medical and socio-economic profile, had also the smallest total body weight loss after bariatric surgery. Surgery complications' rates were not different between clusters. Subjects with obesity can be attributed to five different phenotypes that can be identified in routine clinical practice. Subjects of cluster 2, young healthy women, had the best weight loss after bariatric surgery. Cluster 3, comprising men and women with the highest BMI and frequent comorbidities, had poorer results after bariatric surgery. Subjects in this cluster should therefore benefit from more intensive management. ClinicalTrials.gov NCT02264431.
|
|---|