• Zilma Maria Severino Silva e Costa, Rogério de Melo Costa Pinto, MendonçaTânia Maria da SilvaTMDS0000-0001-8392-672XUniversidade Federal de Uberlândia, Uberlândia, Brasil., and SilvaCarlos Henrique Martins daCHMD0000-0001-8634-6878Universidade Federal de Uberlândia, Uberlândia, Brasil..
• Universidade Federal de Uberlândia, Uberlândia, Brasil.
• Cad Saude Publica. 2020 Jan 1; 36 (6): e00228519.

AbstractThis study aimed to validate the Brazilian version of the item banks for Sleep Disturbance and Wake Disturbance from the Patient-Reported Outcomes Measurement Information System (PROMIS). Validation of these item banks used classical test theory, item response theory, calibrations of items according to the gradual response model proposed by Samejima, and analysis of Differential Item Functioning (DIF), with a sample consisting of 627 participants over 18 years of age. As for data quality, there were no missing data, five items from the Sleep Disturbance bank and four items from the WE bank showed a floor effect, and none of the banks showed a ceiling effect. The two banks displayed excellent reliability with homogeneous scale and good temporal stability. With confirmatory factor analysis, using absolute and incremental goodness-of-fit indices, both banks displayed good fit, showing that the instruments are valid and sufficiently unidimensional. The calibration showed that the items' psychometric properties were satisfactory with good coverage of the sleep-wake construct. As for DIF, both the Sleep Disturbance and Wake Disturbance banks proved to be homogeneous for the different test groups. In conclusion, the version in Brazilian Portuguese for the item banks on Sleep Disturbance and Wake Disturbance in the PROMIS proved to be a reliable, precise, and valid instrument with robust psychometric analyses.

11 keywords...

hide…