From the growth records of 276 rural Mexican children followed as part of a prospective study, individuals were separated into two groups: those with chronic malnutrition and those not displaying chronic malnutrition. Principal component analysis of a 38 X 38 correlation matrix of socioeconomic, demographic, morbidity, and maternal and paternal anthropometric variables yielded components of which 7 were retained: socioeconomic, demographic, morbidity, maternal body mass, paternal body mass, maternal linear dimensions, and paternal linear dimensions. Component scores were calculated for each subject and mean scores computed for malnourished and non-malnourished groups. Malnourished children showed mean component scores which were significantly less than zero for three components: socioeconomic, maternal linear, and paternal linear. This indicated that malnourished children came from poorer families with less well educated parents whose linear body measurements were smaller, relative to the community as a whole. A discriminant function analysis of the individual variables yielded five which contributed to the discrimination of malnourished and non-malnourished children: the height, biacromial and bicristal diameters, and the personal hygiene score of the father, and the bicristal diameter of the mother. Using these variables to calculate the discriminant function, 19 of 25 malnourished children were correctly identified, giving a sensitivity of 76%. This suggests that the use of parental variables is suitable as a screening device, given the high sensitivity.
Johnston, Francis E.; Scholl, Theresa O.; Newman, Bruce C.; and Cravioto, Joaquin
"An Analysis of Environmental Variables and Factors Associated With Growth Failure in a Mexican Village,"
4, Article 4.
Available at: https://digitalcommons.wayne.edu/humbiol/vol52/iss4/4