E-KINDEX: A Dietary Screening Tool to Assess Children's Obesogenic Dietary Habits

Background: To date, no published dietary indices have evaluated the overall effects of various dietary components, beliefs, and practices on the development of obesity in children. Objective: To develop a dietary index that contains dietary characteristics and practices implicated in the development of obesity. Methods: The proposed index (i.e., the E-KINDEX [Electronic Kids Dietary Index]) incorporates 3 subindices: (1) a food groups intake index (13 items), (2) an index related to eating beliefs and behaviors (8 items), and (3) an index that evaluates dietary practices (9 items). The theoretical overall score ranges from 1 (worst) to 87 (best). For validation of the E-KINDEX, multiple linear and logistic regression analyses were applied that had as dependent outcomes various body composition indices of 622 children (9-13 years) from the CYKIDS (Cyprus Kids) study. In all models, adjustments were made for age, gender, physical activity level, TV viewing time, socioeconomic status, breastfeeding, and parental obesity status. Results: The highest E-KINDEX category (>60 points) was associated with 85% less likelihood of a child being obese or overweight (odds ratio [OR], 0.15; 95% confidence interval [CI], 0.05-0.41) and 86% less likelihood of having a waist circumference >= 75th percentile (OR, 0.14; 95% CI, 0.05-0.43). The correct classification rate of E-KINDEX (C-statistic) against excess body fat percentage was 84% (CI, 0.74-0.94). Moreover, for a single SD increase in the index score (i.e., 7.81 units), a decrease of 2.31 +/- 0.23 kg/m(2) in body mass index (BMI), of 2.23 +/- 0.35 in percent of body fat (BF%), and of 2.16 +/- 0.61 cm in waist circumference was observed. Similar results were found during 1-year follow-up of study participants. In particular, baseline E-KINDEX levels were inversely associated with children's BMI at 1-year follow-up (p = 0.024). Latent class analysis showed that the index has good discriminative value for only 68% of the total sample. Conclusion: The proposed E-KINDEX could be helpful in dietary assessment and in further research in the field of childhood obesity.