This study aims to assess the feasibility and concurrent criterion validity of two innovative Information Communication Technology (ICT) tools - Life-logging wearable cameras with image-assisted recall (LLWC-IAR) and Interactive Voice Response Diaries (IVR) via mobile phones, using 15-hour direct observations as the criterion method. It builds on our previous research (IMMANA 2) conducted in Eastern Uganda, which adapted these ICT-tools for use in a low-income-country setting, and assessed their feasibility and criterion validity for assessing dietary diversity and time use patterns. In the proposed study, we will refine these tools; assess their feasibility of use in a very different socio-cultural context, such as among agro-pastoralist communities in Northern Uganda; and extend the behaviours measured to include hygiene practices which are important for understanding impacts of agriculture interventions on maternal and infant nutrition and health, and in a wider context, informing public health campaigns, about hand-washing practices during a global pandemic.
The study will be conducted in two phases. In the first phase, formative research will be done to refine the ICT-tools based on insights gained in the previous study, including how to reduce interaction time with participants, improve maternal message comprehension and image recognition, and successfully collect data when lighting is poor and electricity and connectively unstable. In the second phase, a cross-sectional study will be done to assess the criterion validity of the ICT-tools, for assessing maternal and infant dietary diversity and hygiene practices, and for the LLCW-IAR of measuring women’s time use patterns, childcare practices, and environmental sanitation. A 15-hour direct observation will be the criterion method. In this phase, women with a 6-11m old infant (n=144) in a contrasting context (such as participants in the AfrII/Farm Africa Livestock for Livelihoods project, Moroto and Nakapiripirit districts, Karamoja region) will be randomly selected to participate from those eligible. The behaviours and practices will be assessed, for the same day using direct 15-hour observation, and ICT-tools. Independent measures of these behaviours will be made using the ICT-tools on a day when observations are not done. Six patterns of data collection will account for biases related to chronology of data collection, day-of-the week effects on agreement, and influence of an observer on behavioural practices or ICT-tool data capture success. Questionnaires will be administered to capture information on the participant’s experiences with each of the tools and socio-demographic characteristics.
To assess criterion validity, limits of agreement between the estimated and observed variables and relative bias will be assessed using the Bland and Altman method of agreement, and inter-rater agreement will be assessed using the weighted Cohen's kappa coefficient.
The research will provide robust evidence on the feasibility, concurrent criterion validity and error structures of two innovative, interdisciplinary digital tools, which if successful, can be used to more accurately evaluate a range of impacts of agriculture interventions on maternal behaviours and practices than is currently provided by traditional paper-based recall tools. Such information is critical to ensure agriculture interventions have positive and not negative impacts on maternal and infant well-being.
