Objectives: Very low birth weight (<1500 g) infants frequently require
packed red blood cell transfusions, and transfusion rates vary among
neonatal intensive care units (NICUs). We analyzed transfusions and co
mpared outcomes among NICUs. Study design: In a 6-site prospective stu
dy, we abstracted all newborns weighing <1500 g (total = 825) born bet
ween October 1994 and September 1995. Transfusion frequency and volume
and phlebotomy number were analyzed by site and adjusted for birth we
ight and illness severity. We compared rates of intraventricular hemor
rhage, necrotizing enterocolitis, bronchopulmonary dysplasia, growth,
and length of stay between the high and low transfuser NICUs. Results:
Sites differed significantly in mean birth weight, illness severity,
number of transfusions, pretransfusion hematocrit, blood draws, and do
nor number. Multivariate adjustment for these risks showed that the hi
ghest transfusing NICU transfused an additional 24 cc/kg per baby duri
ng the first 14 days and 47 cc/kg per baby after 15 days, relative to
the lowest transfusing NICU. The presence of arterial catheters increa
sed the frequency of blood transfusions. The rates of intraventricular
hemorrhage, necrotizing enterocolitis, and bronchopulmonary dysplasia
were not higher in the 2 lowest transfusing NICUs, nor were there dif
ferences in 28-day weight gain or length of stay. Conclusions: Major d
ifferences in transfusion practices for very low birth weight infants
exist among NICUs. Because clinical outcomes were no different in lowe
r transfuser NICUs, it is likely that transfusion and phlebotomy guide
lines could result in fewer transfusions, fewer complications, and red
uced cost.