Portal venous gas and surgical outcome of neonatal necrotizing enterocolitis

Purpose: The prognostic significance of portal venous gas (PVG) in neonatal necrotizing enterocolitis (NEC) for operative intervention (OP), neonatal complications, and mortality remains uncertain. The authors designed a long-term prospective study to describe the natural history of PVG related to these outcomes and to test the hypothesis that PVG does not mandate OP. Methods: All infants admitted to a single center between October 1991 and February 2003 were evaluated weekly to identify all cases of NEC (defined as Bell stage 11 or higher). Demographic, radiological, surgical, and outcome data were abstracted prospectively. Radiographic studies were performed at the onset of illness and at subsequent 6- to 8-hour intervals or as clinically indicated. A single pediatric radiologist reviewed all radiographs. Values are expressed as mean +/- SD. Odds ratios and relative risk ratios are reported with 95% CIs. The level of significance was P less than or equal to.05. Results: After the exclusion of 24 infants with lethal diseases, major congenital or chromosomal anomalies, or recurrent episodes of NEC, 194 of 5891 infants developed NEC. The overall incidence of NEC was 3.7%. In 194 infants with NEC, the incidence of PVG was 33% (n = 64). Gestational age (30.8 +/- 4 vs 29.3 +/- 4.2 weeks; P =.02) but not birth weight (1609 +/- 761 vs 1434 +/- 810 g; P = NS) was greater in infants with PVG compared with infants without PVG (n = 130). Sixty-six (34%) infants with NEC underwent OP. Operative intervention occurred more frequently in infants with PVG compared with infants without PVG (OR, 2.5; CI, 1.37-4.76; P =.003)-only 48% of infants with PVG underwent OP. Among the variables, gestational age, severe NEC (Bell stage 111), severe intramural gas (in all 4 abdominal quadrants), and the presence of PVG, severe NEC was most highly associated with OP (OR, 77.47; CI, 10.36-580.16; P <.0001). Bell stage III NEC was present in 98% of infants who underwent OP compared with 40% of infants without OP (P <.0001). Of all infants with NEC, 37 (19%) died. Mortality was higher among infants who underwent OP (33% vs 12%; P < .0003). A multivariate regression model identified Bell stage III (OR, 3.74; Cl, 1.20-11.62; P =.02), but neither PVG nor OP, to be significantly associated with mortality. Of interest is that survival in infants with PVG was greater (but not significantly so) than in infants without PVG in both OP (74% vs 59%) and non-OP (91% vs 87%) groups. Furthermore, 30 of 64 (47%) infants with PVG survived without OP, and of all 33 infants with PVG who did not undergo OP, 30 (91%) infants survived. Conclusions: Decision for OP should be based on the severity of NEC and not on the presence of PVG alone because nearly half of infants with PVG survive without OP. Overall, the presence of PVG does not increase the risk of mortality among infants with NEC. Severe NEC, but not OP, is associated with higher mortality. (C) 2005 Elsevier Inc. All rights reserved.