The construction and validation of prognostic prediction model for sudden sensorineural hearing loss in middle-aged and elderly people

Abstract

ObjectiveIdiopathic sudden sensorineural hearing loss (ISSNHL), as an otologic emergency, is commonly encountered and its prevalence has been climbing every year recently. To our knowledge, the prognosis of middle-aged and elderly patients is worse than that of young patients. Previous researches mainly focused on the adult population, which was considered as prognostic models who performed hearing recovery in ISSNHL. However, few studies regarding the middle-aged and elderly population who are regarded as prognostic models have been reported. Therefore, we aim to construct and validate a nomogram-based prognostic prediction model, which can provide a reference for the prognostic assessment in the middle-aged and elderly patients with ISSNHL. MethodA total of 371 middle-aged and elderly ISSNHL patients who were admitted to the Department of Otolaryngology-Head and Neck Surgery, Yanbian Hospital, Yanbian University, from April 2018 to April 2023 were enrolled in the study. All subjects were randomly divided into two groups including training group (n = 263) and validation group (n = 108). Lasso regression and multi-factor logistic regression were jointly utilized to screen out prognosis-related independent risk factors and establish a nomogram-based risk prediction model. The accuracy and clinical application value of the model were evaluated by combining the Bootstrapping method and k-fold cross-validation, plotting the receiver operating characteristic (ROC) curve, calculating the area under the ROC curve (AUC), plotting the decision curve analysis (DCA), and the calibrating curve. ResultWe used the method of lasso regression combined with multivariate logistic regression and finally screened out eight predictors (including age, number of affected ears, degree of hearing loss, type of hearing curve, duration of disease, presence of vertigo, diabetes, and lacunar cerebral infarction) that were included into the nomogram. The C-index were 0.823 [95% CI (0.725, 0.921)] and 0.851 [95% CI (0.701, 1.000)], and the AUC values were 0.812 and 0.823 for the training and validation groups, respectively. The calibration curve for the validation group was approximately conformed to that for the modeling group, indicating favorable model calibration. The DCA results revealed the modeling group (3%-86%) and the validation group (2%-92%) showed significant net clinical benefit under the majority of thresholds. ConclusionThis study developed and validated a nomogram-based prognostic prediction model which based on the eight independent risk factors mentioned above. The predictors are conveniently accessible and may assist clinicians in formulating individualized treatment strategies.