In the current era, cross-institutional studies which attempt to characterize the quality of NICU care are becoming increasingly frequent. Ensuring comparability of data between facilities is essential to avoid mis-representations of an institution's quality of care. Since differences in timing and rates of transfer can lead to biased collection of outcome information, we examined transfer practices in six Level III NICUs participating in the SNAP-II project. Among 557 VLBW survivors, 248 (44.5%) were transferred to other facilities prior to their discharge home. Most transfers (87.9%) were to Level II facilities. Unit-specific transfer rates ranged from 6.8 to 80.6% (p<0.01). These differences were not accounted for by differences in birthweight, admission illness severity (as measured by SNAP), or inborn status. We obtained post-transfer length of stay data from the receiving units and found that the NICUs varied in the proportion of hospital days between the transferring and receiving facility as well as in the timing of transfer. Conclusions: Marked variability exists in the prevalence and timing of transfers out of Level III units. These differences have important implications for studies of both clinical outcomes that may occur late in the NICU stay (e.g. chronic lung disease, retinopathy of prematurity, cerebral white matter injury) and resource utilization (e.g. length of stay, costs, transfusion practices). Studies of NICU efficiency and quality of care should ascertain data for the full neonatal hospital course.Table

Table 1
Full size table