Abstract
Longitudinal study, whose objective was to evaluate of the time trend in food consumption across the 2002-2015 period in schoolchildren aged 7 to 11 years, covered five food surveys in Florianopolis, southern Brazil. Methodological differences across the surveys (typical vs. previous day food consumption, pen-and-paper versus computer screen presentation) and some known risk factors, were adjusted for statistically. Offset by maximum food/beverage consumption per day allowed comparability of a varying number of food components in a food group over survey years. Significant reduction in some healthy (fruits and animal proteins) and unhealthy diet markers (sodas, sugary drinks, sweets) was observed over the 2013-2015 period compared with the 2002 and 2007 data. Computer screen presentation of the food questionnaire systematically reduced the reporting frequency compared to the pen-and-paper presentation. Converting frequency reports into rates and using contrast analysis allowed time trend estimation based on food questionnaires with different presentation modes, varying number of items within food group over survey years, and variations in probabilistic sampling procedures.
Key words:
Schoolchildren; Diet surveys; Food consumption; Longitudinal studies