The authors declare that they have no competing interests.
GB was responsible for study design. GB, MB, SS, BE, CS, KB, ÖE and EEB were responsible for data acquisition and analyses. ÖE wrote the initial manuscript, and all authors have been involved in drafting the manuscript and revising it critically for important intellectual content. All authors read and approved the final manuscript.
As the understanding of how different aspects of the physical activity (PA) pattern relate to health and disease, proper assessment is increasingly important. In clinical care, self-reports are the most commonly used assessment technique. However, systematic comparisons between questions regarding concurrent or criterion validity are rare, as are measures of predictive validity. The aim of the study was to examine the concurrent (using accelerometry as reference) and predictive validity (for metabolic syndrome) of five PA questions.
A sample of 948 middle-aged Swedish men and women reported their PA patterns via five different questions and wore an accelerometer (Actigraph GT3X) for a minimum of 4 days. Concurrent validity was assessed as correlations and ROC-analyses. Predictive validity was assessed using logistic regression, controlling for potential confounders.
Concurrent validity was low-to-moderate (r <0.35 and ROC AUC <0.7) with large misclassifications regarding time spent sitting/sedentary and in moderate-to vigorous PA. The predictive validity of the questions was good, and one question (PHAS) showed an 80 % decreased odds-ratio of having metabolic syndrome, after taking potential confounders into consideration.
In this mixed sample of adults, both concurrent and predictive validity vaired between items and between measures of the physical activity pattern. The PHAS and WALK items are proposed for assessment of adherence to PA recommendations.
Assessing PA patterns using self-report measures results in methodological problems when trying to predict individual risk for the metabolic syndrome, as the concurrent validity generally was low. However, several of the investigated questions may be useful for assessing risk at a group level, showing better predictive validity.