Validação e calibração da versão em português do Banco de Itens Pain Interference do PROMIS adulto

Abstract

Objective: to translate, culturally adapt, validate and calibrate the Portuguese version of the PROMIS® Pain Impact Item Bank. Method: The Portuguese version of the PROMIS® Adult Pain Impact Item Bank was validated and calibrated, following PROMIS® methodology through analysis of psychometric properties based on the Classic Test Theory (CTT) and Item Response Theory (IRT). For validation and calibration, the 41 items in the database were applied to 605 patients and companions at the outpatient clinic at the Federal University of Uberlândia. The reliability analysis was assessed by the test-retest using the intraclass correlation coefficient (ICC) and through the internal consistency verified by the Cronbach alpha coefficient. The items' validity was measured from the convergent validity analysis (Spearman's coefficient) and construct validity using the Confirmatory Factor Analysis (CFA). For calibration, IRT parameters were estimated by Samejima Gradual Response method and the following assumptions were verified: unidimensionality, local independence and monotonicity. The analysis of the Item Differential Functioning of the item (IDF) was evaluated for different groups (sex, age, education, sick x healthy and American x Brazilian language). Results: Reliability was satisfactory, with alpha Cronbach = 0.984 and ICC = 0.993. The convergent validity was confirmed by evaluating Spearman's correlation coefficient, which ranged from 0.450 to 0.720 and the construct validity was satisfactory due to the strong correlation with the factorial model based on the global adjustment indexes of the CFA: Comparative Fit Index (CFI >0,95), Root Mean Square Error Approximation (RMSEA <0,06) e Tucker-Lewis Index (TLI >0,95). The IRT assumptions of unidimensionality, local identity and monotonicity were confirmed. The discrimination parameter ranged from 1.52 to 3.94, meeting the criterion of high discrimination standard, and the coverage of the construct was from -1.35 to 2.51. All 16 items that presented IDF had a negligible impact, considering Pseudo R² <0.13. Conclusion: The evaluation of the psychometric properties of the pain impact item bank proved to be accurate, reliable and valid for use in the Brazilian population.