Development and Validation of a Prognostic Model to Predict Hearing Recovery for Patients With Chronic Otitis Media

Abstract

Objective To explore the factors affecting postoperative hearing recovery in chronic otitis media (COM) patients, establish a clinical prediction model for hearing recovery, and verify the accuracy of the model. Methods Data of patients with COM who were admitted to our hospital between January 1, 2012 and September 30, 2020 were retrospectively analyzed. We collected data on relevant clinicopathological characteristics of patients. The patients were randomly divided into the development cohort and validation cohorts. A postoperative air-bone gap (ABG) ≤20 dB was defined as successful hearing recovery. Univariate and multivariable logistic regression analyses were used to investigate the association of several prognostic factors with hearing recovery. These factors were then used to establish a nomogram. The model was subjected to bootstrap internal validation and performance evaluation in terms of discrimination, calibration, and clinical validity. Results This study included 2146 patients with COM: the development cohort comprised 1610 patients (mean [standard deviation; SD] age, 44.1 [14.7] years; 733 men [45.5%]) and the validation cohort included 536 patients (mean [SD] age, 42.9 [14.4] years; 234 men [43.7%]). Multivariable logistic regression analysis showed that age, duration of onset, styles of surgery (tympanoplasty, canal wall up-CWU, or canal wall down-CWD), ossicular prosthesis, granulation or calcified blocks around the ossicular chain, ossicular chain integrity, duration of drilling, eustachian tube dysfunction, mixed hearing loss, semicircular canal fistula, and second surgery were associated with hearing recovery. A nomogram based on these variables was constructed. The area under the curve was 0.797 (95% confidence interval [CI], 0.778–0.812) in the development cohort and 0.798 (95% CI, 0.7605–0.8355) in the validation cohort. Conclusions This study demonstrated the various clinical factors correlated with hearing recovery in patients with COM. The nomogram developed with these data could provide personalized risk estimates of hearing recovery to enhance preoperative counseling and help to set realistic expectations in patients.