Revista chilena de nutrición
versión On-line ISSN 0717-7518
HERRAN F, Oscar Fernando; QUINTERO L, Doris Cristina y ARDILA L, Maria Fernanda. SOURCES AND MAGNITUDE OF THE VARIATION IN THE DIET OF ADULTS FROM BUCARAMANGA, COLOMBIA. Rev. chil. nutr. [online]. 2006, vol.33, n.1, pp. 55-64. ISSN 0717-7518. http://dx.doi.org/10.4067/S0717-75182006000100006.
The incorrect measurement of the diet produces an attenuation or dilution of epidemiologic measures. The variation within-person and between-person are the main determinant of this effect. To describe the variability of the diet of a Colombian city, the record of the dietary intake survey of 167 adults carried out during seven days and in different seasons, during 1998-2003, was analyzed. Using analysis of variance (ANOVA), error within-person and between-person, was calculated the reason within-person/between-person, and the variation coefficients for the absolute values, transformed and adjusted by the energy using lineal regression. The reason within-person/between-person is not different for sex and smaller than 1 for all the nutritional variables, except for dietary fiber, iron and manganese. The error between-person was explained by the subjects, sex and the day of the week. With an error margin between 10% and 15%, it is viable in this population to study the diet correctly. The drop variability within-person will attenuate little the measures of absolute effect calculated in epidemic studies. However, the diet is homogeneous and characteristic of a poor country, what will hinder to find relationships between diet and illness
Palabras llave : Diet; ANOVA; coefficient of variation; sample size.