Investigating the Minimal Clinically Important Difference for AzBio and CNC Speech Recognition Scores.

Abstract

OBJECTIVE To assess the minimal clinically important difference (MCID) values for cochlear implant-related speech recognition scores, which have not been previously reported. STUDY DESIGN Retrospective cohort. SETTING Tertiary referral center. PATIENTS Eight hundred sixty-three adult patients who underwent cochlear implantation between 2009 and 2022. MAIN OUTCOME MEASURES MCID values for consonant-nucleus-consonant (CNC) word scores and AzBio sentences in quiet and noise scores using distribution-based methods (half-standard deviation, standard error of measurement, Cohen's d, and minimum detectable change). RESULTS In this cohort, the mean preoperative CNC word score was 13.9% (SD, 15.6). The mean preoperative AzBio sentences in quiet score was 19.1% (SD, 22.1), and the mean preoperative AzBio sentences in noise score was 13.0% (SD, 12.0). The average MCID values of several distribution-based methods for CNC, AzBio in quiet, and AzBio in noise were 7.4%, 9.0%, and 4.9%, respectively. Anchor-based approaches with the Speech, Spatial, and Qualities of hearing patient-reported measure did not have strong classification accuracy across CNC or AzBio in quiet and noise scores (ROC areas under-the-curve ≤0.69), highlighting weak associations between improvements in speech recognition scores and subjective hearing-related abilities. CONCLUSIONS Our estimation of MCID values for CNC and AzBio in quiet and noise allows for enhanced patient counseling and clinical interpretation of past, current, and future research studies assessing cochlear implant outcomes.