Main meal frequency measures in the Health Behaviour in School-aged Children study: agreement with 7-day 24-h recalls

To estimate agreement between questionnaire-based frequency measures from the Health Behaviour in School-aged Children study (HBSC) and 7-day 24-h recall measures of breakfast, lunch and evening meals among 11-15-year-olds, and examine whether disagreement between the two methods varied by socio-demographic factors. In one week 11-15-year-old Danish students completed HBSC questionnaires including meal frequency items. The following week they completed daily 24-h recall questionnaire about their meals (response rate 88.4 %, n = 412). Good to moderate agreement for the breakfast measure: per cent agreement 0.70-0.87, kappa 0.43-0.65. Fair agreement for the lunch measure: per cent agreement 0.53-0.84, kappa 0.26-0.54. High per cent agreement for the evening meal measure (0.83-0.95) but poor kappa agreement (0.14-0.19). Being immigrant predicted disagreement between the two methods for week day breakfast OR (95 % CI) 2.17 (1.16-4.04) and lunch 2.44 (1.33-4.48). We found good to moderate agreement between frequency and 7-day 24-h recall measures for breakfast, a fair agreement for lunch and for evening meal the two agreement methods provided different results. Migration status predicted disagreement between the two methods.