Timing of surgical intervention in necrotizing enterocolitis can be determined by trajectory of metabolic derangement

Purpose: Seven metrics of metabolic derangement were evaluated as contributors to clinical decision support for operative intervention in infants with suspected necrotizing enterocolitis (NEC). Methods: Records of infants with suspected NEC without radiologic evidence of free air were queried for presence of 7 components of metabolic derangement (CMD), consisting of positive blood culture, acidosis, bandemia, thrombocytopenia, hyponatremia, hypotension, or neutropenia. Cases were stratified by clinical decision after each surgical evaluation as observation (OBS) or intervention (INT). Good outcome was defined as full enteric feeding by discharge and bad outcome as death or ongoing parenteral alimentation. Eleven infants undergoing operative intervention after an initial decision to observe were evaluated as matched pairs. Components of metabolic derangement/case and frequency of each CMD were determined for OBS and INT. Mann-Whitney U test was used to compare proportions of CMD in each group. Outcome was compared using chi(2). Observation was then stratified by outcome to determine whether 3 or more metabolic derangements warranting operative intervention would have changed initial clinical decision. The 11 matched cases were similarly analyzed using Wilcoxon-matched pairs. Results: Between March 2005 and July 2008, 35 infants with NEC received 53 surgical evaluations. A median of 1 CMD/case was defined in 32 instances of OBS. Surgical intervention was carried out in 19 infants with a median of 3 CMD/case. Mann-Whitney U test indicated significant difference in the frequencies of each CMD component in OBS vs INT (P = .04). Good outcome was achieved in 75% of OBS and 63% of INT (non-significant, NS). Analysis of OBS by outcome demonstrated a median 1 CMD/case of 25 with good outcome and 3 CMD/case in infants with bad outcome. Frequency of CMD was significantly higher in infants with bad outcome (P = .02). Wilcoxon-matched pair analysis of the 11 infants with paired evaluations demonstrated a similar distribution and frequency of CMD.