Calibração e validação da versão brasileira do banco de itens relações sociais do Patient Related Outcomes Measures Information System (PROMIS®) pediátrico

Purposes: In recent years, health evaluation had profound transformations The development of the Item Response Theory (TRI) coupled with technological evolution has contributed significantly to the evolution of its measurement instruments. PROMIS® represents an advance in this area and allows practical and dynamic evaluations of Patient Related Outcome (PRO) Measures. The objective in this study was to verify the psychometric properties of the Brazilian Version of the pediatric PROMIS® Peer Relationships Item Bank, perform its calibration and obtain version parameters. Method: The 15 items in this bank were responded by 1202 people. To verify the theoretical assumptions about Item Response Theory, the confirmatory and exploratory factorial analysis were utilized. The Gradual Response Model was utilized to calibrate the items. The Differential Item Function (DIF) was investigated for sex (male X female), age range (childs x adolescents) and language, through logistic regression models and the impact through the pseudo R2 by Nagelkerke. Results: From the total, 14 items were calibrated and presented satisfactory validity and reliability properties. The discrimination parameter varied from -0.19 to 2.61, and the difficulty parameter, from -3.14 to 1.11. DIF was detected in 10 items, but with very low impacts. Conclusion: It was conclude that the pediatric PROMIS® Peer Relationships Item Bank presented 14 items with satisfactory psychometric properties which confer quality and validity in Brazil.