Risk factors for recurrence in benign paroxysmal positional vertigo patients and construction of a predictive nomogram

This study aims to investigate potential factors associated with recurrence in patients with benign paroxysmal positional vertigo (BPPV) and establish and validate a predictive model. A retrospective study was conducted on BPPV patients who visited our hospital from January 2022 to December 2024, and their complete clinical information was collected. Univariate and multifactorial logistic regression were used to identify independent variables associated with recurrence, and a nomogram was constructed to visualize the model. Model performance was evaluated using receiver operating characteristic curves and area under the curve, and model fit was assessed using the Hosmer-Lemeshow test. To enhance the robustness of validation, we introduced the Bootstrap method combined with 10-fold cross-validation for internal validation, and assessed its clinical practicality through calibration plots and decision curves. The recurrence of BPPV is closely associated with multiple factors. The established predictive tool exhibits good predictive accuracy and can assist clinicians in identifying high-risk individuals for recurrence and implementing intervention measures in advance. A total of 276 BPPV patients were included, among whom 70 experienced recurrence within 6 months, with a recurrence rate of 25.36%. Of the 276 patients, 112 were male and 164 were female, with an age range of 21 to 76 years (mean age: 58.0 ± 12.3 years). Among the 70 patients who experienced recurrence, 22 were male and 48 were female, with a mean age of 61.2 ± 11.7 years. Statistical analysis identified 5 significant independent risk factors: insomnia (OR = 3.593, 95% CI: 1.541-8.379), depressive state (OR = 3.800, 95% CI: 1.082-13.348), hypertension (OR = 4.315, 95% CI: 1.014-18.365), diabetes (OR = 3.216, 95% CI: 1.392-7.425), and hyperlipidemia (OR = 3.912, 95% CI: 1.412-10.840). The results of receiver operating characteristic analysis, Hosmer-Lemeshow test, and internal cross-validation indicated that the model demonstrated good discriminative ability and fitting, with stable performance and clinical reference value.