Sisvan food intake markers from six to 23 months: critical analysis and validation

To explore the internal structure and analyze evidence of measurement invariance of the Sisvan form of food intake markers of the Food and Nutrition Surveillance System (Sisvan) for children aged six to 23 months.

Sisvan microdata from 2015 to 2019 were used. After analyzing sample adequacy, exploratory factor analysis was carried out considering factor loadings (≥ 0.30 and < 0.85), communality (> 0.20), model fit indices - Tucker-Lewis index (TLI) > 0.90, and root mean square error of approximation (RMSEA) < 0.08. A critical analysis of the form items was performed to adjust the parameters. Measurement stability was investigated between age groups, macro-regions and longitudinally by multi-group confirmatory factor analysis in configural, metric, and scalar invariance models. Configural invariance was accepted if RMSEA < 0.08 and TLI and comparative fit index (CFI) > 0.90. Metric and scalar invariances were accepted if ΔRMSEA < 0.015 and ΔCFI < 0.01, compared to the previous model.

After critical analysis, items were grouped (vegetables and leafy greens; meat or eggs and liver) and excluded (salty food; porridge with milk; rice, potatoes, yams, cassava, flour, or pasta). The exploratory model comprised three factors: healthy complementary feeding (fruit; vegetables; orange or dark green leafy vegetables or fruits; meat, offal, or egg; beans), ultra-processed (yogurt; hamburger and/or sausages; sweetened drinks; instant noodles, packet snacks, or salty cookies; sandwich cookies, sweets, or treats), and milk feeding (breast milk; other milk), with satisfactory factor loadings, communalities and fit (TLI: 0.918, RMSEA: 0.071). The reformulated instrument was stable in the invariance models tested.

With the incorporation of the modifications to the Sisvan form, the food intake markers subsidize a more adequate interpretation of the feeding situation of children aged six to 23 months.