Reliability and validity of a modified MEDFICTS dietary fat screener in South African schoolchildren are determined by use and outcome measures

Abstract

BACKGROUND In South Africa, noncommunicable diseases and obesity are increasing and also affect children. No validated assessment tools for fat intake are available. Objective To determine testeretest reliability and relative validity of a pictorial modified meats, eggs, dairy, fried foods, fats in baked goods, convenience foods, table fats, and snacks (MEDFICTS) dietary fat screener. DESIGN We determined testeretest reliability and diagnostic accuracy with the modified MEDFICTS as the index test and a 3-day weighed food record and parental completion of the screener as primary and secondary reference methods, respectively. Participants/setting Grade-six learners (aged 12 years, 4 months) in an urban, middleclass school (n¼93) and their parents (n¼72). OUTCOME MEASURES Portion size, frequency of intake, final score, and classification of fat intake of the modified MEDFICTS, and percent energy from fat, saturated fatty acids, and cholesterol of the food record. Statistical analyses For categorical data agreement was based on kappa statistics, McNemar’s test for symmetry, and diagnostic performance parameters. Continuous data were analyzed with correlations, mean differences, the Bland-Altman method, and receiver operating characteristics. RESULTS The classification of fat intake by the modified MEDFICTSwas testeretest reliable. Final scores of the group did not differ between administrations (P¼0.86). The correlation of final scores between administrations was significant for girls only (r¼0.58; P¼0.01). Reliability of portion size and frequency of intake scores depended on the food category. For girls the screener final scorewas significantly (P<0.5) correlated to total, saturated fat, and cholesterol intakes (butnot topercentenergy fromfat andsaturatedfattyacids intakes).The sensitivity of the modified MEDFICTS was very high (>90%), but chance corrected agreement between the classifications was poor. Parents did not agree with their children. CONCLUSIONS Testeretest reliability and relative validity of a modified MEDFICTS dietary fat screener in South African schoolchildren depended on the use and outcome measures applied.