Correction to the Prevalence the Physically Inactivity in the National Health Survey Chile 2009-10

Objective: The correction to the prevalence of the physical inactive in a 2009-10 Chilean National Health Survey (NHS) through the use of accelerometers.

Methods: Population older than 15 years old were recruited from different educational levels of the urban area of Santiago-Chile, who had answered the GPAQ ítem during 2009-10 NHS. Physical Activity (PA) was measured using ActiGraph GT3X for a period of 7 days long. Measures of agreement, sensitivity-specificity and discriminant analysis were evaluated in order to generate an adjustment model.

Results: 158 out 306 participants used the accelerometers ≥5 days (age=44.6 ±14.5 years, 55.7% females). Using the information of the accelerometers as a gold standard, GPAQ presented a sensitivity=0,44 to detect a individual as insufficiently active and specificity=0,80. Through discriminant analysis, the variables: PA minutes/day, sex, educational level and BMI, were identified as prognostic factors for a model that estimates the probability of being categorized as an insufficiently active. The predictive quality of the model has an acceptable discrimination (area under ROC curve=0,76), Sensitivity=0,74 / Specificity=0,61. The model correctly classifies 63,5% of the subjects. The model allows to correct the national prevalence of insufficiently active subjects from 27,1% to 53,5%.

Conclusion: The prevalence of adjusted model is comparable with the results reported in the Latin American countries. The GPAQ overestimates the PA. There is a missclassification error in the questionnaire, which can be moderately adjusted through a model that uses diary PA level reported by GPAQ, sex, BMI and educational level.