Graphical exploration of dimensions of preterm infant growth in weight in association with biological, nutritional, and energy expenditure conditions.

The small sample sizes of studies involving preterm infants limit the use of statistics for examination of multivariate conditions contributing to clinically important growth dimensions of weight: rate of weight gain, body composition (fat-free and fat mass), and weight relative to reference infants (z score). The authors used graphical analyses, including multivariate proportional matrix, parallel coordinates, and bivariate plots with regression lines and splines, to explore specific variables derived from a theoretical model of biological, nutritional intake, and energy expenditure conditions influencing growth dimensions. The sample included 28 infants in 4 birth-weight categories: extremely low (<1,000 g), very low/smaller (1,000-1,249 g), very low/larger (1,250-1,499 g), and low (1,500-1,750 g). The authors examined the rate of weight gain before and after nipple feeding initiation. Fat-free mass was estimated with total body water and fetal reference data, and fat mass with skinfold thicknesses. Despite infants achieving the expected rate of weight gain for a fetus of the same postconceptional age, by hospital discharge 13 infants showed growth restriction with weight <10th centile. Infants with respiratory distress syndrome history were highest in negative z-score change from regain of birth weight to discharge, despite higher ordering on protein intake and fat-free mass. Graphical analyses provided visual patterns of distributions and orderings of measures of multiple variables that, taken together, identified potential influencing conditions and raised questions for further study. Other variables, including feeding protocols and practices, infant feeding competence, and health status, may contribute to variability in weight growth dimensions and influence relationships with biologic, nutritional, and energy expenditure conditions.