Efeito da variação do dia a dia da ingestão dietética na identificação de padrões alimentares

Abstract

Evaluating dietary intake is fundamental to identify risk groups and foods associated with the development of these diseases. Methods such as the 24-hour Food Recall and the Food Frequency Questionnaire are used to assess dietary intake, while Factor Analysis (FA) is applied to characterize dietary patterns and associate them with health. However, measurement errors in the collection instruments, especially random error, can weaken the correlations between foods and compromise the accurate identification of dietary patterns. This study sought to evaluate the effect of random error on the identification and definition of dietary patterns in the population, estimating usual consumption and point consumption (with information from two days of consumption) and comparing them. Dietary patterns were obtained through Factor Analysis, first using information from 20 days of consumption for the usual pattern and subsequently random combinations of 2 days for the point patterns. To compare the usual with the point patterns, alignment was performed to ensure the comparability of the factors obtained. The mean factor loadings of the 2-day combinations were weaker than those of the usual pattern. This implies that a given 2-day combination is likely to identify the usual dietary patterns, or at least most of the food items that make up the factors. However, the lower the variance explained by the factor, the greater the attenuation of the factor loadings of the two-day combinations, and thus they would have a small probability of being correctly identified in this pattern when using 2-day information. These results reinforce the importance of increasing the number of collection days to reduce the impact of random error and improve the accuracy of estimates. This recommendation is in line with previous studies, which have already highlighted the importance of multiple food records for more robust and reliable analyses, yet more studies are needed.