(BioMed Central, 2021) Schröder, Helmut; Subirana, Isaac; Wärnberg, Julia; Medrano Echeverría, María; González Gross, Marcela; Gusi, Narcís; Aznar, Susana; Alcaraz, Pedro E.; González Valeiro, Miguel A.; Serra-Majem, Lluis; Terrados, Nicolás; Tur, Josep A.; Segú, Marta; Homs, Clara; García-Álvarez, Alicia; Benavente-Marín, Juan C.; Barón-López, F. Javier; Labayen Goñi, Idoia; Zapico, Augusto G.; Sánchez-Gómez, Jesús; Jiménez-Zazo, Fabio; Marín-Cascales, Elena; Sevilla-Sánchez, Marta; Herrera-Ramos, Estefanía; Pulgar, Susana; Bibiloni, María del Mar; Sistac-Sorigué, Clara; Gómez, Santiago F.; Institute on Innovation and Sustainable Development in Food Chain - ISFOOD
Background: Validation of self-reported tools, such as physical activity (PA) questionnaires, is crucial. The aim of this study was to determine test-retest reliability, internal consistency, and the concurrent, construct, and predictive validity of the short semi-quantitative Physical Activity Unit 7 item Screener (PAU-7S), using accelerometry as the reference measurement. The effect of linear calibration on PAU-7S validity was tested. Methods: A randomized sample of 321 healthy children aged 8–16 years (149 boys, 172 girls) from the nationwide representative PASOS study completed the PAU-7S before and after wearing an accelerometer for at least 7 consecutive days. Weight, height, and waist circumference were measured. Cronbach alpha was calculated for internal consistency. Test-retest reliability was determined by intra-class correlation (ICC). Concurrent validity was assessed by ICC and Spearman correlation coefficient between moderate to vigorous PA (MVPA) derived by the PAU-7S and by accelerometer. Concordance between both methods was analyzed by absolute agreement, weighted kappa, and Bland-Altman statistics. Multiple linear regression models were fitted for construct validity and predictive validity was determined by leave-one-out cross-validation.
