Objective: To obtain evidence of construct validity of the Voice-Related Quality of Life
questionnaire (V-RQOL) using Item Response Theory (IRT). Methods: This was a cross-
sectional study including 561 participants, divided into a Voice Disorders Group (VDG;
n=366) and a Vocally Healthy Group (VHG; n=195), defined through a multidimensional
voice assessment (self-assessment, acoustic analysis, auditory-perceptual judgment, and laryngeal examination). Participants completed the Vocal Screening Protocol (VSP) and the V-RQOL. Structural validation was conducted using IRT (2PL model), obtaining item discrimination (a) and difficulty (b) parameters, Item Characteristic Curves (ICC), and Test Information Curve. Hypothesis testing was performed using ROC curve analysis, with
calculation of accuracy, sensitivity, specificity, cutoff point (Youden index), and area under
the curve (AUC). The reference criterion was the presence of dysphonia. Results: ICCs
revealed a low frequency of intermediate response categories, and the Test Information Curve showed higher precision around θ = 0, supporting the dichotomization of response options. The item “I am sometimes anxious or frustrated (because of my voice)” showed the highest discrimination (a=4.151), whereas “I have difficulty speaking loudly or being heard in noisy places” was the least discriminative (a=1.890). The most difficult item was “I avoid going out socially (because of my voice)” (b=1.496), selected mainly by individuals with greater impairment, while the easiest was “I have difficulty speaking loudly…” (b=0.129), chosen even by individuals with lower impairment. The ROC curve indicated an AUC of 0.885 (95%CI: 0.855–0.917). The optimal cutoff point was θ = –0.801, with 91% sensitivity and 70.11% specificity. Conclusion: The V-RQOL-IRT showed adequate psychometric properties. Items related to emotional aspects were the most discriminative, while those concerning socialization were the most difficult to endorse. Low informativeness of intermediate categories supported dichotomization. ROC analysis demonstrated excellent accuracy, with a cutoff point able to differentiate individuals with and without voice-related quality of life impairment. θ values above –0.801 indicate a higher probability of reduced voice-related quality of life due to a possible voice disorder, according to the clinical reference criterion.