[Value of autopsy in a neonatal intensive care unit]

1) To identify the profile of the cases requested for autopsy; 2) to analyze the clinocopathological discordance; 3) to investigate predictive factors for unsuspected clinically relevant diagnoses.

All autopsies performed between January 1999 and December 2005 in a tertiary neonatal intensive care unit, were retrospectively reviewed. Clinicopathological concordance was assessed independently by two neonatologists and two pathologists, according to a modification of the Goldman classification. A comparison was made between newborns who had an autopsy performed and those who did not and predictive factors for unsuspected findings were investigated.

During the study period, there were 309 deaths, and autopsies were performed in 128 (41.4 %) of these cases. Autopsies were more common in newborns who had gestational age > 36 weeks (p < 0.001), birthweight > 1500 g (p < 0.001) and congenital defects (p < 0.007). However, the probability that the autopsy was granted decreased with increasing death age (p < 0.016). Unsuspected diagnoses were observed in 49.2 % of the autopsies, being a major finding in 21.1 % of the cases. A clinicopathological discordance involving the prognosis was found in four cases (3.1 %). Relevant unsuspected findings could not be predicted from the ante-mortem clinical diagnosis, gestational age, birthweight, sex, and death age.

The autopsy remains the "gold standard" method to reveal major and unsuspected diagnoses and there is no substitute for it. Postmortem examination should be requested systematically in every neonatal death. However, several factors such as gestational age, birthweight, presence of congenital defects and death age, influence the likelihood of autopsy being granted.