Prediction of estimated fetal weight in extremely low birth weight neonates (500-1000 g)

This retrospective analysis of 76 extremely low birth weight neonates (500-1000 g) compared the actual birth weight with the estimated fetal weight derived from 20 published formulas. In our population, Rose's formula (natural log of birth weight = 0.143 x (biparietal diameter + mean abdominal diameter + femur length) + 4.198] had the smallest standard deviation and the best r2 (69 g and 0.780, respectively). The maximum estimated fetal weight underestimate and overestimate were 95 and 159 g, respectively. Using this formula, 46 of 63 (73%) of the estimated fetal weights were within 10% of the birth weight and 56 of 63 (89%) were within (+/-) 100 g of the birth weight. No formula was found to be statistically different by pooled estimate of differences of slopes. When only the biparietal diameter and abdominal wall circumference were available, the Shepard revision of the Warsof equation gave standard deviation 81.7 g and r2 of 0.748; 53 of 67 estimated weights (79%) were within 100 g of the birth weight. When only femur length and mean abdominal diameter were available, the Rose 3 formula [natural log of birth weight = 0.2053 (femur length + mean abdominal diameter) + 4.3726] gave a standard deviation of 89.9 g and r2 of 0.658. Fifty-seven of 71 estimated weights (80%) were within 100 g of the actual birth weight. This study, the largest comparison of formulas to estimate the birth weight of 500-1000-g fetuses, found that no formula estimated fetal weight significantly more accurately than any other.