INTRODUCTION

This study aimed to identify prognostic factors and develop a classification model for predicting recovery in patients with peripheral facial palsy.

METHODS

Data from patients who received integrative medicine treatment with Bell's palsy and Ramsay-Hunt syndrome were collected. The change of House-Brackmann Grade (HB Grade) for 2 years from the onset and the factors presumed to be related to the prognosis were analyzed by reviewing electrical medical records retrospectively. The estimated recovery rates to HB Grade 2 and 1 were calculated by the Kaplan-Meier method. The factors affecting the prognosis were selected by using univariate Cox regression analysis. Subsequently, multivariate Cox regression analysis was performed on the selected factors. The factors derived from the Cox regression model were applied to the survival tree analysis model to establish the criteria for the classification of patients according to prognosis.

RESULTS

768 participants were included after screening. Based on the Kaplan-Meier method, the estimated recovery rates for HB Grade 2 and 1 for the 2 years were 98.2 and 83.3%, respectively. The univariate Cox regression analysis indicated that ten factors, including sex, diabetes, hemoglobin A1c, diagnosis, periauricular pain, hearing impairment, taste disorder, initial HB Grade, and average axonal loss (AAL) and maximum axonal loss (MAL) of nerve conduction study (NCS), affected prognosis. Finally, multivariate Cox regression showed that the AAL and MAL were related to prognosis. Five classification models predicting the 2-year estimated recovery rate established from the survival tree analysis were as follows: 100% (AAL < 70% and MAL < 80%), 87.1% (AAL < 70% and MAL ≥ 80%), 86.8% (70% ≤ AAL < 80%), 55.0% (80% ≤ AAL < 90%), and 24.2% (AAL ≥ 90).

CONCLUSION

The present results demonstrated that AAL and MAL of the NCS were significant factors in predicting the prognosis of peripheral facial palsy.