Comparison of the agreement between WeChat-based self-administered and interviewer-administered data on infant and young child feeding in China: A test-retest study.

Background: Measuring infant and young child feeding (IYCF) indicators is important in evaluating child health programs and making evidence-based decisions. With Internet and new media rapidly developing, communication apps such as WeChat (the most popular mobile social media platform in China) are widely used and can potentially be used as an alternative way to collect infant and young child feeding information. This study compares data agreement between a WeChat-based self-administered and an interviewer-administered survey on infant and young child feeding information. Methods: We recruited 297 mothers of children aged 6-23 months in Fenxi County, Shanxi Province, China. Using the Test-Retest method, we first collected data through a self-administered survey using a WeChat-based electronic questionnaire and asked 36 questions on breastfeeding and complementary feeding knowledge, practices, and information sources. We then conducted an interviewer-administered survey using the same questionnaire and compared the data agreement between the two survey methods during the same day. Cohen's kappa score (κ) and intraclass correlation coefficients (ICC) were used for data agreement analysis for all 36 questions and six key IYCF indicators. The McNemar test was used to identify differences between the two survey methods for the six key indicators. Results: There was substantial or almost perfect agreement for 33 questions (κ/ICC>0.60), and slight or fair agreement for the other 3 questions (κ/ICC<0.40). Agreement of all six key IYCF indicators was substantial or almost perfect (κ = 0.78-0.94), while two indicators showed statistical differences between the two survey methods (P = 0.03 for "Minimum meal frequency" and P = 0.001 for "Minimum accepted diet"). Analysis of reasons for inconsistencies showed that 43.6% of all the inconsistencies were not caused by the self-administered survey method. The cost of the interviewer-administered survey was much higher than that of the self-administered survey: ¥45.9 (US$6.8) vs ¥19.7 (US$2.9) per questionnaire. Conclusions: The WeChat-based self-administered method can be used for future data collection of infant and young child feeding information in China. Most of the questions and key indicators showed very good agreement without statistical differences between the two methods.