ABSTRACT
BACKGROUND: diabetic complications and olfactory dysfunction (OD) in patients with type 2 diabetes mellitus (T2DM) seem related. This study aims to evaluate the prevalence of OD in T2DM patients and to analyze its relationship with diabetic complications. METHODS: 130 T2DM patients and 100 comparable controls were enrolled. Olfaction was evaluated using the Extended Smell Test (TDI) and the Italian brief Questionnaire of Olfactory Disorders - Brief-IT-QOD. T2DM patients were divided into: "Group 1", patients with no complications, and "Group 2", patients with at least one diabetic complication. Non-parametric tests were used. Machine learning algorithms were applied to explore which variables were most important in predicting the presence of OD in T2DM. RESULTS: The prevalence of OD was significantly higher in Group 2 than in controls (71.4% vs 30%) and in Group 1 (71.4% vs 43.3%). However, when comparing the TDI scores between Group 1 and 2 the only significant difference was found for the discrimination scale and not for the identification and threshold scales. Brief-IT-QOD scores were significantly higher in Group 2 than in controls. The Random Forest and variable importance algorithms highlighted the relevance of LDL, glycated hemoglobin, type of complication (macrovascular) and age in determining OD in T2DM. The last three variables were included in a nomogram for the prediction of OD risk in T2DM. CONCLUSIONS: T2DM patients with diabetic complications are more frequently affected by OD. Poor glycemic control, LDL values, age and presence of macrovascular complications are the more important factors in determining OD in T2DM patients.