Risk factors and a nomogram model for recurrence of benign paroxysmal positional vertigo: a multicenter...

To explore factors influencing the recurrence of benign paroxysmal positional vertigo (BPPV) within 1 year after treatment with canalith repositioning procedure (CRP) and to construct a risk prediction model with a nomogram. This model aims to offer a scientific basis for of high-risk groups' early identification and to support the development of prevention strategies.

Data of BPPV patients diagnosed and successfully treated with CRP of 5 hospitals in Sichuan Province from Jan 2020 to Mar 2024 were retrospectively analyzed. The patients were divided into a training set of BPPV and a set of validation with a 7:3 ratio. After organizing the clinical data, the training set's patients were arranged into recurrent and non-recurrent subgroups based on the recurrence of BPPV within one-year post-treatment. Factors affecting BPPV recurrence were found using LASSO regression; then, multivariate logistic regression (MLR) analysis was used to build a nomogram-style risk prediction model. The "receiver operating characteristic" (ROC) curve, Hosmer-Lemeshow calibration curve, clinical "decision curve analysis" (DCA), and clinical impact curve were employed to assess the model's discrimination, accuracy, and clinical applicability.

600 patients in all were involved; 180 were in the validation set and 420 in the training set, following a 7:3 split. The overall recurrence rate of BPPV within 1 year was 39.17% (235/600 cases). The recurrence rates in the validation and training sets were 39.44% (71/180 cases) and 39.05% (164/420 cases), respectively. LASSO regression and logistic regression analyses identified hypertension, hyperglycemia, migraine, HADS-A, and 25(OH)D as significant factors for recurrence. On the basis of these factors, a nomogram prediction model for recurrence was established. The validation and training sets had an area under the ROC curve of 0.723 (95%CI: 0.645-0.801) and 0.728 (95%CI: 0.679-0.777), respectively. The "Hosmer-Lemeshow goodness-of-fit (HLGOF) test" indicated satisfactory calibration (training set: chi-square = 8.708, p = 0.368; validation set: chi-square = 13.303, p = 0.102). The analysis exhibited an excellent consistency between the model's predicted probabilities and actual outcomes. DCA and clinical impact curve analyses indicated a positive net clinical benefit at various threshold probability levels, affirming the clinical value of the nomogram model in forecasting BPPV recurrence.

BPPV patients treated with CRP exhibit a high recurrence rate within one-year post-treatment. Hypertension, hyperglycemia, migraine, HADS-A, and 25(OH)D levels were related with increased recurrence risk. The risk prediction model, presented as a nomogram, demonstrated good discrimination and calibration, effectively predicting the BPPV's recurrence risk within 1 year and offering significant clinical utility.