The SUPPORT study, a large multicenter study of EOL care in adult ICUs, showed that, even for very high-risk patients, doctors were unlikely to discuss DNR orders until 24-48h prior to death. We wondered whether there were similar patterns in discussions about EOL care or DNR orders in NICUs.METHODS: Babies were eligible for the study if they met ≥1 of the following criteria: <750g, congenital anomalies, grade IV IVH, asphyxia, chronic lung disease. We asked MDs to estimate each baby's chance of survival, anticipated quality of life on a 1-10 scale, and likelihood of developing cognitive or motor problems. We then asked whether there had been a discussion about DNR or EOL care prior to the baby's last day of life. RESULTS: 260/1204 (21%) admissions to the NICU met eligibility criteria. These babies accounted for 97% of deaths. We interviewed MDs in 138 (53%) of these cases. MDs had discussed EOL care prior to the last day of life in 46(33%) cases. Comparing babies about whom there was or was not such a discussion, there was no difference in MD assessments of the baby's likelihood of survival to discharge, of developing cognitive or motor problems, or of anticipated quality of life. 24/138 (17%) babies died. 17 (71%) babies who died did not receive CPR at the time of death. 76% of babies about whom there was a discussion of EOL options survived. There was discussion about EOL options about 11/24 (46%) of babies who died. CONCLUSIONS: For most babies who don't get CPR, there hasn't been discussion of DNR prior to the last day of life. Most babies about whom there is some discussion of DNR don't die. Prognostic estimates do not correlate well with discussions of EOL care in these 2 NICUs. In deciding which parents should discuss EOL treatment options, prognostic assessments are only part of the puzzle. Other factors seem to drive many such decisions.