Comparison of Methods to Assess Adherence to Infant and Young Child Feeding Practices and Provision of Low-Aflatoxin...

Levels of adherence to recommended protocols in an intervention trial can affect outcomes and confound the results. To broaden the evidence about the selection and utility of adherence measures in varying contexts, we describe the level of adherence to the Mycotoxin Mitigation Trial (MMT) randomized intervention using caregiver-reported survey questions and compare inferences of adherence between multi-module surveys and interactive 24 h dietary recalls based on our program theory.

The MMT was a two-arm cluster-randomized trial conducted in 52 health facilities (clusters) in central Tanzania. Surveys were conducted with all trial participants at three time points and dietary recalls were conducted in a cohort at 12 mo.

The 12 mo survey was conducted with 2112 caregivers and the 18 mo survey was conducted with 2527 caregivers. A cohort of participants (n = 282, 20 clusters) was selected for dietary recalls, balanced by arm. Reported feeding of blended porridge flours, whether MMT-provided or own-sourced, was high at 12 and 18 mo, between 73 and 95%, with only slight differences between the surveys and recalls. Inferences were similar for continuation of breastfeeding, feeding frequency, and dietary diversity. Only the amount of porridge fed the previous day differed statistically by method, with higher amounts reported in the recalls compared to the survey.

Detailed analysis of reported behaviors, based on the MMT program theory, supports high adherence to the recommended trial behaviors. Survey data and 24 h dietary recalls were convergent for almost all indicators, strengthening the trial's conclusions and allowing for either method to be selected for similar research.