Development and validation of a tool to assess appetite of children in low income settings

Abstract

Reliable and validated tools for measuring appetite of children in South Asia are not available. This study aimed to develop and validate a tool for assessing appetite level of under-five children. Based on literature review and findings from focus group discussions (FGDs), an initial 27-item interview-based tool, the "Early Childhood Appetite and Satiety Tool (ECAST)" was developed in Bangladesh. Fourteen FGDs were carried out in rural and urban settings and constructs for inclusion were derived from the themes and coding of FGDs and appetite assessment tools used in Western contexts. For structural validation, the ECAST-27-was administered on 150 mothers/primary caregivers of children aged 6-59 months, living in urban and rural areas. To validate the association with other variables, the ECAST was administered on mothers of children aged 12-24 months in the community (N = 50), and two groups of wasted, hospitalized children (Weight-for-length, Z score <-2SD) [group1: twenty acutely ill children aged 6-59 months; group 2: twenty children in nutritional rehabilitation aged 18-24 months]. Reliability of ECAST was estimated using Cronbach's alpha and Pearson's correlation coefficient. Kaiser-Meyer-Olkin = 0.73 and the Bartlett's test of sphericity, χ2(253) = 755.791, p < 0.001 indicated that the raw data were suitable. Given the convergence of the Scree plot, Kaiser's criterion and dropping of cross loading items, a 16-item ECAST was produced with three sub scales: Appetite cue; Food responsiveness and Emotion and preference, which were internally valid and had good test-retest reliability (Cronbach's alpha 0.6 and test-retest reliability 0.797). Total ECAST scores of wasted children with good appetite were significantly higher from those with poor appetite (p = 0.004 and 0.001 for two wasted groups respectively). Results suggest that ECAST may provide a useful measure to assess the appetite level of under-five children.