Concordance and discordance between early objective and patient-reported voice outcomes after thyroid surgery

Abstract

Objective Objective voice measures and patient-reported impairment may not change in parallel after thyroid surgery. We assessed concordance between objective voice status on postoperative day 2 and patient-reported status on postoperative day 7 and explored characteristics of concordant and discordant profiles. Methods This prospective single-center cohort enrolled adults undergoing thyroid surgery from November 2025 to January 2026. Voice recordings were obtained preoperatively and on postoperative day 2; the 30-item Voice Handicap Index (VHI-30) and GRBAS ratings were obtained preoperatively and on day 7. Objective voice status was defined using prespecified changes in fundamental frequency, jitter, shimmer, harmonics-to-noise ratio, and maximum phonation time. Participants were assigned to four concordance profiles. Agreement statistics and exploratory multinomial logistic regression were applied. Supplementary analyses included smoothed cepstral peak prominence (CPPS), all five GRBAS dimensions, and objective classifications based on published reference values. Results Among 445 participants, 225 (50.6%) were concordant normal, 109 (24.5%) objective-only abnormal, 45 (10.1%) subjective-only abnormal, and 66 (14.8%) concordant abnormal. Under the primary definition, objective abnormality was more frequent than patient-reported abnormality (39.3% versus 24.9%). Overall, positive, and negative agreement were 65.4%, 46.2%, and 74.5%, respectively; Cohen’s kappa was 0.225 (95% confidence interval, 0.133–0.312), with significant marginal asymmetry (P<0.001). CPPS decline and GRBAS deterioration were most pronounced in the concordant abnormal group. Reference-based classification yielded 78.7% overall agreement and kappa of 0.414 (95% confidence interval, 0.313–0.509), with 21.3% discordance and McNemar P = 0.412. Exploratory regression associated objective-only abnormality mainly with lesion location and recurrent laryngeal nerve amplitude reduction, and subjective-only abnormality mainly with voice-use, voice-concern, and upper-airway factors. Conclusions Objective and patient-reported voice status showed incomplete, definition-dependent concordance, supporting complementary acoustic, perceptual, and patient-reported assessment. External validation remains necessary before these profiles guide clinical decisions.