Predicting population: development and validation of a new predictive nomogram for evaluating medication nonadherence risk in a type 2 diabetes.

Background: Diabetes mellitus is a growing global health challenge and affects patients of all ages. Treatment aims to keep blood glucose levels close to normal and to prevent or delay complications. However, adherence to antidiabetic medicines is often unsatisfactory. Purpose: Here, we established and internally validated a medication nonadherence risk nomogram for use in Chinese type 2 diabetes mellitus (T2DM) patients. Methods: This cross-sectional study was carried out from July-December 2020 on randomly selected T2DM patients visiting a diabetes clinic and included 753 participants. Adherence was analyzed based on an eight-item Morisky Medication Adherence Scale (MMAS-8). Other data, including patient demographics, treatment, complications, and comorbidities, were also collected on questionnaires. Optimization of feature selection to develop the medication nonadherence risk model was achieved using the least absolute shrinkage and selection operator regression model (LASSO). A prediction model comprising features selected from LASSO model was designed by applying multivariable logistic regression analysis. The decision curve analysis, calibration plot, and Results: The prediction nomogram comprised several factors including sex, marital status, education level, employment, distance, self-monitoringofbloodglucose, disease duration, and dosing frequency of daily hypoglycemics (pills, insulin, or glucagon-like peptide-1). The model exhibited good calibration and good discrimination ( Conclusion: The number of patients who adhere to anti-diabetes therapy was small. Being single male, having no formal education, employed, far from hospital, long disease duration, and taking antidiabetics twice or thrice daily, had significant negative correlation with medication adherence. Thus, strategies for improving adherence are urgently needed.