[The nomogram prediction model for the risk of dropout in sublingual immunotherapy of patients with allergic rhinitis].

Abstract

Objective: To develop and externally validate a nomogram prediction model for assessing the risk of treatment dropout in allergic rhinitis (AR) patients undergoing sublingual immunotherapy (SLIT). Methods: Between February 2016 and December 2019, data from 358 and 259 AR patients undergoing SLIT were collected from Guizhou Provincial People's Hospital and Huangshi Central Hospital, respectively. The data included general patient information, dust mite sIgE levels, allergen types, and 22 other clinical variables. Data from Guizhou Provincial People's Hospital were used as the training set, while data from Huangshi Central Hospital were served as the external validation set. A multivariable Cox regression model was used to identify independent factors associated with SLIT dropout and to develop a nomogram prediction model. Results: Multivariate Cox regression analysis identified several significant factors influencing SLIT dropout, including dust mite sIgE levels (Grade Ⅱ-Ⅳ; HR=1.48, 95%CI: 1.16-1.88), presence of other allergic diseases (HR=0.47, 95%CI: 0.37-0.61), Rhinoconjunctivitis Quality of Life Questionnaire (RQLQ) score (HR=0.98, 95%CI: 0.97-1.00), WeChat management (HR=0.77, 95%CI: 0.60-0.98), treatment efficacy (HR=0.72, 95%CI: 0.56-0.92), age (5-17 years, HR=0.50, 95%CI: 0.36-0.71;≥60 years, HR=1.42, 95%CI: 1.08-1.87), household income (<5 000 CNY, HR=1.44, 95%CI: 1.09-1.90;>20 000 CNY, HR=0.66, 95%CI: 0.44-0.99), allergen types (single dust mite, HR=0.70, 95%CI: 0.49-0.93; and combined pollen or mold, HR=1.45, 95%CI: 1.02-2.04), and time to efficacy <3 months (HR=0.73, 95%CI: 0.56-0.94), all P<0.05. At the third-year follow-up, the area under curve (AUC) for the nomogram model was 0.913 (95%CI: 0.881-0.943) in the training group and 0.886 (95%CI: 0.838-0.933) in the validation group. Calibration and decision curve analyses demonstrated the model's consistency with actual dropout rates and clinical benefit in both groups. Additionally, a Brier score of 0.29 further confirmed the model's predictive accuracy. Conclusion: We successfully develop a nomogram-based prediction model for SLIT dropout in AR patients, which could assist healthcare professionals in effectively identifying high-risk patients and facilitate the development of more personalized and timely treatment plans aimed at enhancing patient compliance.