Relative Validity of a Semi-Quantitative Food Frequency Questionnaire for Adolescents

Abstract

<b><i>Introduction:</i></b> A valid food frequency questionnaire (FFQ) is a priority for epidemiological studies. We aimed to perform a relative validity of an FFQ for Portuguese adolescents. <b><i>Methods:</i></b> A semi-quantitative Portuguese FFQ (SQ-PortFood-FQ) was developed and the relative validity performed using the multiple-pass 24-h recall. Eighty adolescents (aged 10–17 years) completed both questionnaires to measure energy intake, macro- and specific micronutrient intake. We used the following methods for the relative validity: (1) Spearman’s correlation coefficient (<i>r</i>); (2) cross-classification into the same third (agreement) and into the opposite third (disagreement); (3) weighted κ-coefficient (κ_W); (4) and the Bland-Altman method. <b><i>Results:</i></b> We obtained acceptable correlation coefficients between questionnaires for energy (<i>r =</i>0.59;<i> p</i> &#x3c; 0.001), protein (<i>r =</i>0.60;<i> p</i>&#x3c; 0.001), fat (<i>r</i> = 0.55;<i> p</i> &#x3c; 0.001), carbohydrates (<i>r =</i>0.51;<i> p</i> &#x3c; 0.001) and sodium (<i>r =</i>0.51;<i> p</i> &#x3c; 0.001). Tertile agreement for energy, protein, fat, carbohydrates and vitamin C intakes classified correctly more than 50% and grossly misclassified less than 10% of participants. Energy and protein produced moderate agreement between methods with a κ_W &#x3e;0.40. The SQ-PortFood-FQ overestimated all analysed data when compared to the 24-h recall, although those estimations were &#x3c;10% for energy, macronutrients and cholesterol. The 95% limits of agreement showed that all the variables (energy, protein, fat, carbohydrates, fibre, calcium, sodium, iron, saturated fatty acids, monounsaturated fatty acids and cholesterol) were within the desirable limits of agreement of 50–200%, with the exception of vitamin C (76–296%) and polyunsaturated fatty acids (PUFA) (75–232%). <b><i>Discussion/Conclusion:</i></b> The SQ-PortFood-FQ showed acceptable relative validity at the individual level for energy, protein, fat, carbohydrates, fibre, calcium, sodium, iron, saturated fatty acids, monounsaturated fatty acids and cholesterol. Non-significant relationships were found for vitamin C and PUFA. Our results are in line with previous studies carried out amongst adolescents. As far as we know this is the first FFQ validated for Portuguese adolescents.