Life-long effects of malnutrition using semi-quantitative EEG analysis

1 ABSTRACT The non-linear spatiotemporal features in the continuing EEG recordings could be helpful to infer the physio and pathological significance of early insults on the brain, such as early malnutrition and their long-term effects. A unique opportunity is opened with the Barbados Nutrition Study (BNS) dedicated to studying Protein-Energy Malnutrition (PEM) with two groups, children suffering an early PEM episode and their controls. We evaluated the resting-state EEG (N=108, PEM=46) in 1978, and we repeated the EEG (N=97, PEM=46) in 2018. We did a qualitative analysis of the EEG using a semi-quantitative scale (Grand Total EEG (GTE)) and an item response theory (IRT) approach to estimate a latent variable that is able to explain the subjacent neurophysiological status (NPS). Finally, we applied a mixed-effects model with a sensitivity index for ignorability to test differences between the controls and PEM groups while accounting for the missing data mechanisms ( nlme (Pinheiro J. 2020) and the ISNI package in R(Xie et al., 2018). The fixed effects were group, age, gender, and socioeconomic status; the random effect was the variability inherent to each participant and evaluator. Results The simple visual inspection of the 1978 EEG recordings detected 39 participants with abnormalities (28 PEM and 11 Controls; p<0.05); in 2018, a total of 63 participants showed abnormalities in the EEG recordings (35 PEM and 28 Controls; p<0.01)). The polytomous IRT analysis revealed that all items had been loaded well onto the latent factor, and the highest value of the Focal abnormality reached 0.97. The fixed effect of the groups (PEM vs. Control) was highly significant, with a p-value of 0 and the c index of 5.27 . Age was also significant with a p-value of 0.0093 and the c index of 14.793 , whereas Gender and SES were not significant. The contrasts at the two different time points (childhood (1978) mean age= 8.45, adulthood (2018) mean age=48.30) also showed highly significant differences between groups with a p-value of 0 . Conclusions EEG abnormalities were seen in both PEM and control groups during the school years and later in middle adulthood, with a higher proportion of abnormalities in the previously malnourished BNS participants at both ages. The statistical significance of these differences was confirmed through a latent variable approach and a linear mixed-effect model, which discriminated successfully against the long-term effects of early malnutrition on the brain up to 50 years after the onset of malnutrition in the first year of life.