Cerebellar vermis measurement at cranial sonography for assessing gestational age in the newborn weighing less than 2000 grams

Prognosis of the very low birth weight infant depends more on gestational age than birth weight, but clinical assessment of gestational age of very low birth weight infants is inaccurate. We wished to determine how well cerebellar vermis dimensions might predict gestational age in infants of birth weight less than 2000 g. We obtained suitable midline sagittal images of the cerebellar vermis at cranial sonography, performed via the anterior or posterior fontanelle, in 41 infants, from the regional neonatal intensive care unit whose gestational age was known. We measured the cerebellar vermis area and diameter on the hard-copy image provided the margins of the vermis were clearly visible, the cerebellar tonsils were excluded from the image, and the anterior and posterior divisions of the corpus medullare were visible on the image. Vermis diameter was measured from the base of the fourth ventricle to the junction of folium and tuber vermis. Vermis area was calculated using a stereological method using a test system of regularly spaced points randomly placed over a magnified image of the cerebellar vermis. We generated regression equations for estimating gestational age using combinations of birth weight, vermis area, or vermis diameter for the 26 infants with birth weight of less than 2000 g for whom the cerebellar vermis measurements were obtained within one week of birth, Vermis area and diameter correlated very highly. They both can be used for predicting gestational age. The addition of either vermis area or diameter to birth weight improves accuracy of gestational age assessment. If birth weight was presumed to be unknown, cerebellar vermis area or diameter allow prediction of gestational age to within +/- 1.3 weeks (1 standard error) or +/- 2.5 weeks, using a 95% prediction interval. If the same method of reporting is applied to the New Ballard Score, the New Ballard Score predicts gestational age +/- 1.7 weeks (1 standard error) or +/- 3.4 weeks, using a 95% prediction interval.