Objective: In epidemiologic studies, diabetes classification is typically based on lab measures (i.e., hemoglobin A1c (HbA1c), fasting or non-fasting glucose, 2-hour post-prandial glucose) and use of diabetes medication from a medication inventory or self-report. However, new medication classes initially intended for diabetes, such as glucagon-like peptide-1 (GLP-1) agonists and sodium-glucose cotransporter-2 (SGLT2) inhibitors now have multiple treatment indications such as cardiovascular disease and weight management, presenting a methodological challenge for diabetes classification in observational studies. Our objective was to develop and evaluate a new algorithm to classify diabetes status in cohort studies accounting for novel medication classes with multiple indications. Research Design and Methods: Using data from 1,767 Jackson Heart Study (JHS) participants who completed Exam 4 (2021-2025), we developed a new diabetes classification algorithm based on a combination of lab measures (HbA1c, glucose), identifying diabetes medication classes from the participant’s medication inventory, and using the participant’s self-reported use of medications for diabetes for confirmation of status when using newer medication classes (GLP-1, SGLT2). We compared this new algorithm to the traditional diabetes classification which assumes a single primary indication for diabetes medications. Modified Poisson regression was used to estimate crude associations between diabetes, comparing the newer definition and the traditional definition, and select cardiovascular risk factors. Results: The new definition classified 36.8% of participants as having diabetes compared to 37.0% using the traditional definition. There were 147 (8.3%) participants differentially classified including fewer participants with missing data in the new definition. The two definitions yielded similar prevalence ratios (PR) for the association of diabetes with obesity (New definition PR=1.44(1.27,1.63); Traditional definition PR=1.40(1.23,1.58)) and chronic kidney disease (New definition PR=1.14(1.01,1.29); Traditional definition PR=1.16(1.03,1.30). Conclusions: The new and traditional diabetes definitions led to similar associations in the JHS. However, as the use of newer diabetes medication classes continues to increase for other health conditions, there is greater potential for misclassification in observational studies and the new definition should be considered for use.
Sims et al. (Tue,) studied this question.