Comorbidity clusters associated with newly treated type 2 diabetes mellitus: a Bayesian nonparametric analysis

Area of Science:

• Medical Informatics
• Epidemiology
• Computational Biology

Background:

• Type 2 diabetes mellitus (T2DM) is frequently accompanied by chronic comorbidities, leading to increased medication use and adverse events.
• Understanding comorbidity patterns is crucial for managing T2DM and its associated health complications.
• Existing research often focuses on individual comorbidities rather than complex, interconnected disease clusters.

Purpose of the Study:

• To identify prevalent comorbidity clusters in patients newly treated for T2DM.
• To analyze the temporal progression of these comorbidity clusters.
• To explore novel associations between diseases in the T2DM population.

Main Methods:

• Utilized IQVIA Medical Research Data (THIN database) of anonymized electronic health records.
• Included 175,383 patients with a first prescription for a non-insulin antidiabetic drug (NIAD) from 2006-2019.
• Applied Bayesian nonparametric models to identify 20 frequent comorbidity clusters and 14 latent features (LFs) and model their progression.

Main Results:

• Identified 20 distinct comorbidity clusters, characterized by 14 latent features (LFs).
• Observed significant associations between LFs, such as congestive heart failure (CHF) and chronic kidney disease (CKD).
• Demonstrated rapid progression in cardiovascular disease-related clusters and identified established and novel disease connections.

Conclusions:

• Bayesian nonparametric models effectively characterize complex comorbidity patterns in T2DM.
• The study revealed established T2DM complications and previously unrecognized disease interdependencies.
• Findings highlight potential for improved patient management through understanding comorbidity cluster dynamics.