RESUMO
OBJECTIVE: To examine circadian variation in deaths among infants < or =32 weeks' gestation admitted to Canadian neonatal intensive care units (NICU). STUDY DESIGN: We examined all infants (n=5192) between 24 and 32 weeks' gestation with complete data, who were admitted to 17 tertiary Canadian Neonatal Network NICUs from January 1996 to October 1997. Multivariable logistic regression was used to compare risk-adjusted early neonatal mortality rates (death within 7 days of NICU admission) of infants admitted during daytime (8 am to 5 pm) with infants admitted at night. RESULTS: Sixty percent (n=3131) of infants were admitted to the NICU at night. Patient risk factors significantly (P<.05) predictive of early neonatal death from multivariable logistic regression were male sex, outborn status, APGAR score <7 at 5 minutes, presence of congenital anomalies, low gestational age, and high admission Score for neonatal acute physiology, version II (SNAP-II). For inborn infants, in-house presence of a neonatal fellow or attending neonatologist at night (odds ratio, 0.6) and NICU admission at night (odds ratio, 1.6) were also predictive. CONCLUSIONS: Risk-adjusted early neonatal mortality odds was 60% higher among inborn infants < or =32 weeks' gestation admitted to NICUs at night compared with during daytime, equivalent to 29 excess deaths per 1000 infants.