Precision and Sample Sizes Achieved for Infant and Young Child Feeding Indicators Evaluated in Anthropometry Assessments: A Secondary Analysis of Population-Representative Surveys in Refugee Settings.

Abstract

This study characterizes the sample size and precision for infant and young child feeding (IYCF) indicators evaluated in surveys conducted in refugee settings, characterized by sample sizes enabling timely information to inform humanitarian response. We analyzed surveys provided by the United Nations High Commissioner for Refugees conducted from 2013 to 2019. For eight IYCF indicators recommended in the current global guidance, we assessed achieved sample sizes and precision by evaluating the half-width of 95% confidence intervals (HWCIs) by survey design. Among cluster surveys, we characterized the intraclass correlation coefficients (ICCs) and design effects. Final analysis included 203 surveys from refugee settings in 15 countries. Ever breastfed and bottle feeding indicators were the most precise (median HWCI: 3.1% and 3.6%) due to inclusion of all children 0-23 months resulting in relatively large sample sizes and estimates of prevalence close to 100% and 0%, respectively. Exclusive breastfeeding under 6 months and the introduction of solid foods (6-8 months) had the lowest precision (median HWCI estimates: 12.6% and 18.2%). In cluster surveys, early initiation of breastfeeding and flesh food consumption stood out with markedly higher design effects (median estimates: 2.63 and 2.17) driven by relatively larger sample sizes and high ICCs (median estimates: 0.31 and 0.27); all other indicators had median design effects < 1.3. Evidence on expected variation in precision and heterogeneity by indicator can help design surveys with samples that allow for the rapid collection of data sufficient to inform emergency responses. High caution is needed when interpreting estimates of exclusive breastfeeding and introduction of solid foods in past surveys given lower precision.