Identification of Macrosomic, normal and intrauterine growth retarded neonates using the modified Neonatal Growth Assessment Score.

OBJECTIVE: Separation of Macrosomic, normal and intrauterine growth retarded (IUGR) neonates using the modified Neonatal Growth Assessment Score (NGAS).
METHODS: A sample of 23 neonates with evidence of growth acceleration and 74 neonates classified as normal or IUGR in a previous study was used in this investigation. The prenatal growth of all neonates had been studied longitudinally with ultrasound. From the data collected, Rossavik models describing the growth trajectories of head circumference (HC), abdominal circumference (AC), thigh circumference (ThC), femur diaphysis length, head cube (A) and abdominal cube (B) were specified. These models were used to predict birth weight (WT), crown-heel length (CHL), HC, AC and ThC at birth. Actual birth measurements of WT, CHL, HC, AC and ThC were evaluated using age-specific normal size curves and compared to predicted measurements by means of the growth potential realization index (GPRI). GPRI values were evaluated by comparison to previously established normal ranges and used to calculate Neonatal Growth Assessment Scores (NGAS(5)). NGAS(5) values, together with assessments of anatomical measurements and GPRI values, were used to establish macrosomic, normal and IUGR groups. Principal components analysis was applied to the sets of GPRI values available for each neonate to provide a principal component score for separating macrosomic and normal neonates (m(2)NGAS(51)) or one to separate macrosomic, normal and IUGR neonates (m(3)NGAS(51)) using linear discriminant analysis. The groups identified by these multivariate methods were compared to the original classification and their characteristics evaluated.
RESULTS: The m(2)NGAS(51) and a boundary value of 207.5% separated macrosomic and normal neonates with an accuracy of 97.3%. The m(3)NGAS(51) and boundary values of 210.0 and 182.5% separated macrosomic, normal and IUGR neonates with an accuracy of 96.9%. No single GPRI value or anatomical measurement could achieve these levels of accuracy. All normal infants were AGA but only 45.5% of the IUGR group were SGA and only 60.9% of the macrosomic group were LGA. Thirteen different types of IUGR and eleven different types of macrosomia were identified based on GPRI values.
CONCLUSION: The modified NGAS accurately separates macrosomic, normal and IUGR neonates although growth abnormalities are expressed in different ways in different individuals. Copyright 2004 S. Karger AG, Basel