A sample of 1931 Israeli infants was measured for body weight (WT), length (HT) and head circumference (HC) for approximately 2 years. The Count model with 3 parameters was chosen as the best fitting and most parsimonious function to approximate growth of all 3 studied traits. In the model parameter a relates to birth indices, b--to velocity of growth, and c--to rapid early childhood growth, or acceleration. Assuming a difference in growth patterns in the periods of different length, the whole sample was divided into 3 groups: 1) infants with last measurement around the age of 12 months; 2) infants with last measurement around the age of 18 months, and 3) infants with last measurement around the age of 24 months. The individuals measured up to 12 months were presented in all three groups. 27 curve fitting parameters, corresponding to 3 different follow-up intervals for WT, HT and HC were computed for each individual. A high correlation was detected between the a parameters regardless of time interval for 3 measured traits. A negative correlation was found between b and c parameters within the same time interval. A consistent positive correlation was indicated between a and b parameters, especially for body length and head circumference. A principal component analysis extracted five independent factors explaining 88.1% of the total variance. Three first factors retained parameters b and c, describing growth rate and pattern of each trait separately, namely, F1 was responsible for head circumference, F2 was a body length factor, F3 was a body weight factor. F4 extracted all birth indices, observed (HC0, HT0 and WT0) and expected (parameters a). The composition of principal factors allowed us to assume that there might be a strong involvement of a pleiotropic genetic source in determination of birth size traits and an independent genetic source controlling the pattern of growth for each trait separately.
Read full abstract