Methods: Criteria for transfusion in very premature infants are poorly grounded in physiology. Comparing transfusion practices among NICUs may identify alternative clinical strategies. We report red blood cell transfusions studied prospectively in 6 NICUs for 624 babies <1500g, adjusted for birth weight and illness severity (SNAP). Early deaths and transfers (≤1 day) are excluded (N=17).

Results: The percent receiving transfusions, mean number of transfusions (# tx) on days 1-14 and day 15+, and number of donor exposures differed significantly among NICUs as did mean number (#) blood draws on day 1 and mean pre-transfusion hematocrit (hct) at days 15+. Linear regression to adjust for baseline differences in BW and SNAP showed significant (p<.05*) differences between NICUs, measured as +/- cc per kg/baby relative to the reference NICU. Table

Table 1
Full size table

Conclusions: Large differences in transfusion practices are not explained by differences in illness severity and birth weight, and result in multiple donor exposures. These variations in practices must be further explored by comparison with NICU outcomes.