Feasibility of Using Tablet-Based Cognitive Assessments in a Large Randomized Trial in Ghana

• Published in: Current developments in nutrition Vol. 4; no. Supplement_2; p. 1110
• Main Authors: Yuan, Haiying, Ocansey, Maku, Oaks, Brietta, Sheridan, Margaret, Okronipa, Harriet, Hamoudi, Amar, Kumordzie, Sika, Adu-Afarwuah, Seth, Prado, Elizabeth
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01-06-2020; Oxford University Press
• Subjects: Maternal, Perinatal and Pediatric Nutrition
• ISSN: 2475-2991; 2475-2991
• DOI: 10.1093/cdn/nzaa054_182

We aimed to assess: 1) the feasibility of using three tablet-based tasks to measure the cognitive outcomes of a nutrition trial in Ghana, specifically the Paired Associate Learning (PAL), Serial Reaction Time (SRT), and Simon tasks from the Rapid Assessment of Cognitive and Emotional Regulation (RACER); 2) whether learning effects in the three tasks show expected patterns compared to published data; and 3) the sensitivity of the RACER scores to two nutritional indicators: length-for-age z-score (LAZ) and hemoglobin concentration (Hb).
966 children aged 4–6 years participated in the International Lipid-Based Nutrient Supplements trial preschool follow-up study in Ghana. Field staff measured anthropometry, child Hb, and administered the RACER tasks at a project office visit. We used repeated-measures regression controlling for pre-specified covariates to assess learning effects and Pearson’s correlations to assess the sensitivity of RACER scores to the two nutrition indicators.
The success rate for completing the PAL, Simon and SRT tasks was 99.0%, 99.3%, and 93.5% respectively. As expected, on the PAL task, children were more accurate in the delayed recall trials than in the training trials, demonstrating declarative learning. On the Simon task, children showed the expected pattern of inhibitory control by responding faster and closer to the target position in same-side trials than opposite-side trials. They also demonstrated procedural learning by responding faster in the ordered block than the random block in the SRT task. The correlations between RACER scores and the two nutrition indicators (LAZ and Hb) were not significant.
Tablet-based assessments can be used to evaluate the outcomes of large nutrition trials in low-resource settings with high success rate. The results showed expected patterns compared to results from children in high-income countries. Although we did not find expected correlations with nutrition indicators typically associated with cognitive development, future analysis will further explore these associations.
Bill & Melinda Gates Foundation, the Society for Research in Child Development, and USDA NIFA Hatch Project.