Full text
| |
| Author(s): |
Eliseu Verly Junior
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | São Paulo. |
| Institution: | Universidade de São Paulo (USP). Faculdade de Saúde Pública (FSP/CIR) |
| Defense date: | 2012-01-11 |
| Examining board members: |
Dirce Maria Lobo Marchioni;
Valéria Troncoso Baltar;
Rafael Moreira Claro;
Jose Eduardo Corrente;
Edna Massae Yokoo
|
| Advisor: | Dirce Maria Lobo Marchioni |
| Abstract | |
Introduction: The estimation of the habitual food intake requires the use of proper statistical techniques. For some foods, it is possible to find consumption equal to zero in one or several days, even to individuals who usually consume them. Then, methods must be able to deal with two concerns: the zero-inflated distribution; and the correlation between probability and amount of consumption. Objective: To investigate methodological issues of a method to estimate habitual food intake to use it to analyze dietary data from a population. Methods: First, using the National Health and Nutrition Examination Survey 2007-2008, it was estimated the habitual dark green vegetable intake from a samples with different size and percentual of individuals with two 24-hour recall (24hr). After, it was used a subsample of the Inquerito de Saúde de São Paulo, a population-based cross-sectional study (n=716), in which the participants answered two 24hr and one food frequency questionnaire (FFQ). It was verified the effect of the FFQ as covariate on the prediction improvement of the models to estimate habitual food intake. Finally, using a bivariate modeling, it was estimated the habitual serving consumption of the food groups with recommendation of consumption in the Brazilian Food Guide. The percentual of individual who did not meet the recommendations were calculated as well. To all purposes, the habitual food consumption was estimated using the National Cancer Institute method. Results: The precision of the usual intake estimates decreases when low replication rates are used. For studies that aim to estimate the average or the percentual of individual with consumption bellow the cut-offs, the loss of precision may not represent an important bias. Considering foods with moderate to high percentual of non consumers, the simultaneously inclusion of the FFQ and a indicator of consumption variable (consume yes or not) was enough. The percentual of individuals with consumption bellow the recomandation were: 88 per cent for cereals, tubers and roots; 43 per cent for beans; 95 per cent for fruits, including fresh juices; 72 per cent for vegetables; 100 per cent for milk and dairy products; 11 per cent for meat and egg; 93 per cent for oils, fats and oil seeds; and 83 per cent for sugar and sweets (AU) | |
| FAPESP's process: | 09/11239-9 - Habitual food intake among individuals of São Paulo: populational based survey |
| Grantee: | Eliseu Verly Junior |
| Support Opportunities: | Scholarships in Brazil - Doctorate |