A predictive model for olfactory dysfunction in patients undergoing maintenance hemodialysis.

Abstract

Objective: To analyze the current status and risk factors of olfactory dysfunction in patients undergoing maintenance hemodialysis (MHD), establish a prediction model, and evaluate its effectiveness.

Materials and methods: The demographic characteristics and laboratory tests of MHD patients at the Second Affiliated Hospital of Soochow University's hemodialysis center from February to May 2024 were collected. Basic information from individuals with normal renal function at the physical examination center served as a normal control group. Olfactory function was assessed using Sniffin' Sticks, categorizing subjects as having normal, hyposmia, or anosmia based on their scores. Binary and ternary logistic regression analyses were conducted to ascertain risk factors for olfactory dysfunction occurrence and progression in MHD patients, establish predictive models, and assess their performance using receiver operating characteristic curve (ROC curve) and prediction probability graph.

Results: Among 53 individuals with normal renal function, 17 cases (32.1%) exhibited hyposmia, while 3 cases (5.7%) demonstrated anosmia. Among 159 MHD patients, 97 cases (61.0%) had hyposmia and 24 cases (15.1%) had anosmia. The prevalence of olfactory dysfunction in the MHD group was significantly higher than that observed in the normal control group (χ² = 26.434, p < 0.001). Logistic regression analysis revealed that advanced age (OR = 2.373, 95% CI: 1.143 - 4.928, p = 0.020), hypoalbuminemia (OR = 1.299, 95% CI: 1.051 - 1.607, p = 0.016), hypocholesterolemia (OR = 2.373, 95% CI: 1.143 - 4.928, p = 0.020), low education level (secondary or higher education/no education, OR = 0.065, 95% CI: 0.007 - 0.601, p = 0.016; primary education/no education: OR = 0.093, 95% CI: 0.019 - 0.457, p = 0.003) were identified as risk factors for anosmia development. Hypocholesterolemia was a risk factor for hyposmia (OR = 2.052, 95% CI: 1.228 - 3.429, p = 0.005). The prediction model of anosmia was In(P/1-P) = 14.97 + 0.997·age (years)/10 - 0.466·albumin (g/L) -1.942·cholesterol (mmol/L) +2.936·education level (P is the probability of anosmia). ROC curve and prediction probability graph indicated good performance of our prediction model.

Conclusion: The prevalence rate of olfactory dysfunction in MHD patients was found to be 76.1%. Advanced age, hypoalbuminemia, hypocholesterolemia, and low education level were identified as significant risk factors for anosmia. Advanced age, hypoalbuminemia, and lower education level also posed risks for deterioration of olfactory function.