Jesus Gilmar Mercês de, Assis Maria Alice Altenburg de, Kupek Emil
Universidade Estadual de Feira de Santana, Feira de Santana, Brasil.
Universidade Federal de Santa Catarina, Florianópolis, Brasil.
Cad Saude Publica. 2017 Jun 5;33(5):e00163016. doi: 10.1590/0102-311X00163016.
The study evaluated the validity and reproducibility of the food consumption section of the questionnaire Food Intake and Physical Activity of School Children (Web-CAAFE), an Internet-based software for the qualitative measurement of food consumption by recalling the previous day. A total of 390 students in grades 2 to 5 (7 to 15 years) of a semi-integral public school participated in the study. The validity was tested by comparing the report in the Web-CAAFE and the direct observation of food consumed in the school in the previous day. The reproducibility was evaluated in a sub-sample of 92 schoolchildren, by comparing repeated reports in the Web-CAAFE on the same day. Probabilities of accuracy in the Web-CAAFE report in relation to the observation (matches, omissions and intrusions and respective 95% confidence intervals) among seven food groups were estimated through multinomial logistic regression. The average for the match rate was 81.4% (variation: 62% sweets and 98% beans); for the omission rate was 16.2% (variation between 2.1% dairy products and 28.5% sweets); for the intrusion rate was 7.1% (variation between 1.3% beans and 13.8% cereals). Sweets, cereals and processed foods, snack foods and fried foods simultaneously exhibited higher rates of omission and intrusion. Students 10 years of age or older had lower probabilities of intruding food items. There were no significant variations in the accuracy of the report between repeated measures. The Web-CAAFE was a valid and reliable instrument for the evaluation of food consumption, when applied to students in grades 2 to 5 of public schools.
该研究评估了《学童食物摄入与身体活动问卷》（网络版CAAFE）中食物消费部分的有效性和可重复性，这是一款通过回忆前一天情况对食物消费进行定性测量的基于互联网的软件。一所半公立学校二至五年级（7至15岁）的390名学生参与了该研究。通过比较网络版CAAFE中的报告与前一天在学校直接观察到的食物消费情况来测试有效性。通过比较92名学童在同一天网络版CAAFE中的重复报告，对92名学童的子样本进行了可重复性评估。通过多项逻辑回归估计了七个食物组中网络版CAAFE报告相对于观察结果（匹配、遗漏和侵入以及各自的95%置信区间）的准确性概率。匹配率的平均值为81.4%（变化范围：糖果为62%，豆类为98%）；遗漏率为16.2%（乳制品为2.1%至糖果为28.5%之间变化）；侵入率为7.1%（豆类为1.3%至谷物为13.8%之间变化）。糖果、谷物、加工食品、休闲食品和油炸食品同时表现出较高的遗漏和侵入率。10岁及以上的学生侵入食物项目的概率较低。重复测量之间报告的准确性没有显著差异。当应用于公立学校二至五年级的学生时，网络版CAAFE是评估食物消费的有效且可靠的工具。
