Versão brasileira do Thyroidectomy- Related Voice and Symptom Questionnaire (TVSQ): uma análise baseada na teoria de resposta ao item

Abstract

Thyroidectomy is one of the main interventions for thyroid gland disorders. Like any surgical procedure, there is a possibility of sequelae, mainly affecting voice and swallowing. Therefore, it is important to recognize the significance of these disorders and enable effective screening for those who require evaluation. The Thyroidectomy- Related Voice and Symptom Questionnaire (TVSQ) is an instrument capable of screening for alterations related to voice complaints in general, vocal fold paralysis, and swallowing-related symptoms. The TVSQ has already undergone translation and adaptation processes for Brazilian Portuguese and has obtained validity evidence based on content, response processes, and internal consistency, but it has not yet been evaluated according to more contemporary theories that could complement the validation process and make it more robust. The objective of this study is to apply Item Response Theory (IRT) to the Brazilian version of the TVSQ. This is a validation study, restricted to the investigation of latent traits using IRT, with an observational, crosssectional, and documentary character, with a quantitative approach. This study is based on a database used in a previous study composed of 395 individuals residing in all regions of Brazil who responded to the TVSQ. After analyzing the data using Item Response Theory (IRT), it is possible to state that the TVSQ is composed of 3 factors, all of which present satisfactory discrimination parameters and difficulties, with a general Cronbach's alpha for each factor greater than 0.7. The items with the greatest discriminatory power for factors 1 (vocal symptoms), 2 (oropharyngolaryngeal symptoms), and 3 (discomfort in swallowing, breathing, and in the cervico-thoracic region), respectively, were Q9 ("Do you feel your voice has become weak?" - 𝛼𝛼 = 4.120), Q11 ("Do you feel a lot of mucus in your throat?" - 𝛼𝛼 = 4.288), and Q17 ("Do you feel discomfort or pain due to numbness in your chest?" - 𝛼𝛼 = 3.853). In turn, the most difficult items were Q10 (“Do you feel pain or discomfort after talking?” – 𝑏𝑏 = 1.040), Q14 (“Do you cough after eating or lying down?” – 𝑏𝑏 = 0.984) and Q19 (“Do you feel discomfort when eating or drinking?” – 𝑏𝑏 = 1.276). It is concluded that the application of IRT showed that the three factors of the instrument present satisfactory discrimination and difficulty indices, in addition to efficiently measuring the latent trait.