All infants born between 23 weeks and
weeks of gestation, between March 2002 and August 2004, at 1 of the 14 institutions were eligible for inclusion in the ELGAN Study, a study designed to identify the antecedents of brain damage in preterm infants. Eighty-three percent of eligible infants were enrolled. The enrollment rates according to center ranged from 67% to 93%. However, the consent processes were different at different sites. At some sites, recruitment occurred when a mother was admitted with threatening preterm delivery. At other sites, recruitment was not attempted until after delivery or admission to the intensive care nursery. Institutional review boards at each institution approved the ELGAN Study, and written informed consent was obtained before enrollment. All variables and outcomes were defined a priori, and research personnel were trained before study start. Historical, demographic, and clinical data were abstracted from maternal and infant clinical charts, and placentas were examined for evidence of inflammation.
Data collected from the maternal charts included race, ethnicity, pregnancy complications, multifetal pregnancy, and duration of gestation. Gestational age was determined on the basis of (in order of decreasing preference) fetal ultrasound scans obtained before 14 weeks (63%), ultrasound scans obtained between 14 weeks and 18 weeks (17%), ultrasound scans obtained after 18 weeks with a consonant maternal report of the beginning of the last menstrual cycle (13%), and date of the last menstrual cycle only (7%). The presence of chorioamnionitis (defined on the basis of polymorphonuclear leukocytes in the chorion or chorioamnion) and funisitis (defined on the basis of polymorphonuclear leukocytes in the wall of a blood vessel in the umbilical cord or the chorionic plate) was determined by a pathologist who first engaged in training procedures to minimize inter-observer variability, was masked with respect to maternal history, and used the predefined operational definitions.
Neonatal blood pressure measurements are reported for postnatal day 0 through day 6. Because the first 24 hours of life represent a critical time period for ELGANs, day 0 began at the time of birth and continued through 24 hours of age. Day 1 began at the end of the first 24 hours of age and ended at midnight of that calendar day. For example, for an infant born at 1 pm on January 1, day 0 of life would be the next 24 hours, until 1 pm on January 2. Day 1 of life would start at 1 pm on January 2 and extend until midnight (11 hours). Therefore, day 1 varied in length and was always the “short day.” Days 2 through 6 began and ended at midnight. Mean blood pressure was measured either directly from an intraarterial catheter or from an automated blood pressure cuff. We did not record which oscillometric devices were used.
To identify blood pressures that prompted intervention, we compared the box plots of the lowest blood pressures on each day for infants who received any treatment for hypotension, including vasopressors (eg, dopamine, dobutamine, and epinephrine) or saline solution at >10 mL/kg per day, and infants who did not receive treatment.18
Saline therapy was recorded as a total amount received per day and included normal saline solution and Ringer's lactate solution; to avoid the cumulative amount of saline flushes, we considered a total amount of >10 mL/kg per day to be therapy for hypotension. Comparisons were made in 4 gestational age strata (ie, 23–24, 25, 26, and 27 weeks). Recognizing the possibility that some infants might have received saline therapy of >10 mL/kg per day because of abnormalities in skin perfusion, low urine output, or other indications that might or might not have been related directly to low blood pressure, we examined infants who received vasopressor treatment in analyses of risk factors for treatment and center variation. We did not include the administration of blood products as a treatment for hypotension because we could not distinguish this use from the treatment of a hematologic disorder (eg, red blood cell transfusion for the treatment of anemia). We did not include corticosteroids as a treatment for hypotension because all patients who received corticosteroids also received vasopressors, except for 10 patients. Those patients were treated at centers where randomized, controlled trials of corticosteroids were being conducted; the 10 patients were excluded because we had no way of knowing whether the patients received corticosteroids or placebo.
We examined candidate risk factors for any treatment and vasopressor treatment, including gender, gestational age, birth weight z
score, maternal race, receipt of prenatal steroid treatment, chorioamnionitis, funisitis, pregnancy complications, multifetal pregnancy, Apgar score, and Score for Neonatal Acute Physiology–II (SNAP-II) score.19
Birth weight z
score is the number of SDs (and direction) of the infant's birth weight from an assumed mean for gestational age.20
We considered these potential confounders when examining how centers differed in the use of any treatment and vasopressor treatment during the first 7 postnatal days.
We evaluated the associations between subject characteristics and nontreatment or treatment with saline boluses and/or vasopressor or between nontreatment and treatment with just vasopressor by using Fisher's exact test. Characteristics that distinguished the groups with a P
value of ≤.3021
were included in logistic regression analyses that assessed the tendency of each of the 14 sites to treat at any time during the first 7 postnatal days.