Necrotizing enterocolitis during the first week of life: a multicentered case–control and cohort comparison study

Article metrics



Necrotizing enterocolitis (NEC) is rare during the first week of life; most cases occur after 2 to 4 weeks. We hypothesized that when NEC develops in the first week, certain predisposing factors and feeding practices are identifiable. To test this, we sought to identify every case of NEC diagnosed during the first week within the Intermountain Healthcare system during the most recent 6-year period.

Study Design:

Data were collected from neonates admitted to any Intermountain Healthcare neonatal intensive care unit (NICU) with a date of birth from 1 January 2001 through 31 December 2006. Electronic and paper records were obtained for all with a diagnosis of NEC (Bell stage II) within the first 168 h. X-rays, physician notes, nursing records, laboratory reports and operative reports were subjected to critical review to reexamine the diagnosis of NEC. Among those with confirmed NEC, we recorded underlying conditions and every feeding given prior to the diagnosis of NEC. Comparisons were made with patients that did not develop NEC, yet were cared for in the same NICUs, during the same period of time, and of the same gestational ages.


A total of 28 neonates were identified electronically as having NEC during the first week. Critical review confirmed this in 21, but 5 were determined at laparotomy to have had spontaneous intestinal perforation, and 2 others were found on surgical reports to have had a congenital infarction of the colon. Total 20 of the 21 confirmed cases developed NEC while in a NICU being treated for another condition. The exception was a small-for-gestational-age neonate in a well baby nursery. Compared to 6100 controls, the 21 with early NEC were more likely to have had a meconium-positive test for illicit drug exposure (P<0.005), early onset sepsis (P<0.034) and respiratory distress (P<0.039). They were less likely than case–controls to have been fed human milk (P=0.003) and were more likely to have been fed formula exclusively (P=0.019). None who were fed human milk exclusively developed early NEC. Twelve of the twenty-one were fed (by gavage or bottle) amounts exceeding the upper limit of volumes taken by breastfed neonates.


We speculate that the prevalence of NEC during the first week could be reduced by identifying at-risk patients, feeding them human milk exclusively for the first week and using feeding volumes that do not exceed that taken by healthy breastfed neonates.


The prevalence of necrotizing enterocolitis (NEC) has not decreased appreciably during the past four decades.1 In part, this is explained by the capability of modern neonatology to sustain ever-smaller and more immature neonates, since NEC is more common among these patients.2 However, in part, this failure is due to the lack of successful NEC prevention strategies.3

Perhaps one reason prevention strategies have not been particularly successful is that NEC is a multifactorial disease. Consequently, no single approach to prevention is likely to pertain to all subtypes.3, 4 To focus our efforts on a small and potentially more homogeneous subtype of NEC, we conducted the present study of NEC occurring in the week following delivery. Using the data sets of Intermountain Healthcare, a healthcare delivery system in the western United States, we identified all patients in the past 6 years with Bell stage II NEC diagnosed during the first week. We examined the data from each case, and sought potential features and feeding practices that differed between these patients compared with others who did not develop NEC yet were cared for in the same hospitals, at the same time, and were of the same gestational ages.


Data were collected as a de-identified limited data set from archived Intermountain Healthcare records. The information collected was limited to the information displayed in the Results section of this report. Data were obtained for patients admitted to any Intermountain Healthcare hospital with the diagnosis of NEC, Bell stage II, during the first week (168 h) after birth. Data were limited to those with a date of birth between 1 January 2001 and 31 December 2006.

When NEC was identified from electronic records, the medical records (paper charts) were examined by two or more of the authors. All radiographic images, radiographic reports, physician's notes and nursing notes pertinent to the issue of NEC were reviewed. To document the occurrence of NEC, we used the criteria originally proposed by Bell et al.,5 as subsequently modified by Walsh and Kliegman,6 and adopted by the Vermont Oxford Network.7 This definition required the presence of one or more of the following three clinical signs; (1) bilious gastric aspirate or emesis, (2) abdominal distension, (3) occult or gross blood in stool (no fissure) and one or more of the following three radiographic findings; (1) pneumatosis intestinalis, (2) hepatobiliary gas, (3) pneumoperitoneum. If a patient had focal gastrointestinal perforation, based on visual inspection of the bowel at the time of surgery, the condition was not listed as NEC but as spontaneous intestinal perforation (SIP). We note that while our methods of distinguishing NEC from SIP are not fully consistent with the guidelines recently proposed by Gordon et al.,8 who pointed out that pneumoperitoneum without pneumatosis would be categorized as NEC if treated by abdominal drainage under the Vermont Oxford Network definitions, all infants with pneumoperitoneum receive laparotomy within our system, thereby circumventing this diagnostic pitfall.

The program used for data collection was a modified subsystem of ‘clinical workstation’. 3M company (Minneapolis, MN) approved the structure and definitions of all data points for use within the program. Data were collected from the electronic medical record, case mix, pharmacy and laboratory systems. Trained and designated personnel entered and accessed data. The enteral feeding type (human milk vs formula) and the feeding method (breast vs bottle vs gavage) were not recorded in the electronic record. Therefore, to obtain this information a case–control substudy was undertaken. For each confirmed case of NEC (n=21), 10 controls were selected (n=210). The controls were matched with the cases on three elements; (1) they were selected from the same hospital, (2) they were of the same gestational age (±1 week) and (3) they were selected from the same period of time (±3 months). Feeding volumes from breastfed neonates were estimated based on weighing them immediately before and after breastfeeding (using a conversion of 1 g of weight gain=1 ml of milk). The Intermountain Healthcare Institutional Review Board approved the study.

Descriptive statistics were calculated using SPSS (v 13.0) for Windows. Between-group means were tested using independent samples t-tests when parametric assumptions were met, and with Wilcoxon rank-sum tests used for nonparametric comparisons. Proportions were compared between groups using the Fisher's exact test. Two-tailed tests were used and for all tests α was set at 0.05.


A total of 28 neonates were identified, using the electronic medical records, as having NEC, Bell stage II, with the diagnosis made within 168 h of birth. Critical review confirmed this diagnosis in 21 of the 28. Five had pneumoperitoneum but no radiographic evidence of pneumatosis or portal air. All five cases had a laparotomy, with operative and pathological reports describing an isolated perforation in the distal small bowel without NEC. Thus, these five were recategorized as having SIP, not NEC. Two other infants with congenital cardiac defects, one with hypoplastic right heart syndrome, and the other with tricuspid atresia, had pneumatosis intestinalis diagnosed on the second or third day. A laparotomy was performed immediately on both, and the operative and pathological reports identified these as congenital ischemic necrosis of the colon.

Two of the twenty-one with confirmed NEC were term infants. These 2 cases were previously published among 30 cases of NEC in term infants.9 Total 20 of the 21 with confirmed NEC had been admitted to a neonatal intensive care unit (NICU) for treatment of another condition. The other patient, a small-for-gestational-age (SGA) infant with a tight nuchal cord at delivery, developed NEC while in a well baby nursery.

Clinical features of the 21 neonates who developed NEC and the 6100 who did not are listed in Table 1. Those who developed NEC were more likely to have had a meconium-positive test for illicit drugs, early onset bacterial sepsis and respiratory distress syndrome. The specific drugs detected in meconium in the four cases were opiates (two cases), cocaine (one case) and cannabinoids (one case).

Table 1 Clinical features (mean±s.d. or %) of 21 neonates with NEC diagnosed during the first 7 days after birth

Feeding practices among the 21 neonates who developed NEC are shown in Table 3. Although the intent was to study 210 case–controls, 215 were actually studied. When compared with the case–controls, those who developed NEC were less likely to have received human milk exclusively. In fact none of the neonates who developed NEC were fed human milk exclusively and they were more likely to have been fed formula exclusively (Table 2).

Table 3 Features and outcomes of 21 neonates who had NEC diagnosed during the first 7 days following birth
Table 2 Feeding practices are compared between two groups; 21 who developed NEC, Bell stage II, during the first 7 days following birth vs 215 matched controls

The volume of enteral feedings given to each patient before NEC was diagnosed is shown in the Figure 1. Also shown on the figure is the range of feeding volumes taken by healthy breastfeeding term infants and by healthy bottlefeeding term infants during the first 2 days after birth.10, 11 At least 12 of the patients that developed NEC were fed volumes exceeding the upper limit of milk intake by breastfeeding infants.

Figure 1

Daily enteral intake (feeding volume expressed in ml kg−1 birth weight), during the first week of life, among 21 patients that developed NEC during their first 7 days. Only the feedings preceding the diagnosis of NEC are shown. For comparison (gray bars) are shown the feeding volumes taken by healthy breastfed term neonates during the first 2 days following delivery (data obtained by weighing neonates before and after each breastfeeding).10, 11 Also for comparison (white bars) are shown the feeding volumes taken by healthy bottlefed term neonates during the first 2 days following delivery.11

Outcomes of those who developed early NEC are listed in Table 3. Eight were transported to a regional children's hospital for surgical management. Five of these underwent abdominal surgery, none had placement of a peritoneal drain. Of the five surgical cases, one had extensive bowel resection and later died of sepsis, and another had total bowel necrosis and support was withdrawn. These were the only two deaths (10% mortality rate); the other 19 were discharged to home following a hospital stay of 33.3±13.5 days.


Early NEC was described more than 20 years ago by Thilo et al.,12 who reported 13 cases from Children's Hospital of Denver. All the 13 cases had either respiratory distress or polycythemia before NEC occurred. Eleven of the thirteen were fed before NEC developed, but it is not clear whether the two who were never fed had what today we would define as NEC, or rather had another condition such as SIP13, 14, 15 or congenital infarction of the colon.15 In that report, Thilo et al.12 suggested that to reduce early cases of NEC, feedings should be given more cautiously to neonates who had respiratory distress or polycythemia.

It is possible that early NEC differs in fundamental ways from the classical variety that develops later. For instance, early NEC occurs almost exclusively in neonates born at >31 weeks’ gestation. Perhaps this is because those of higher gestation are more likely to receive substantial feedings during their first week, whereas those of earliest gestations are likely to receive fewer early feedings.16 This association is one of many links between feeding and NEC pathogenesis.2, 3, 4

Other differences are apparent between patients who develop early NEC vs classical NEC. For instance, Guthrie et al.16 examined patients with classical NEC vs controls who did not develop NEC, and found them to be more likely to be delivered vaginally, with lower 5-min Apgar scores, and with lower birth weights. In contrast, in our present series of early NEC, compared with controls, the mode of delivery, 5-min Apgar score, and birth weights were similar. In fact, we found little to prospectively differentiate between those who were at risk for developing early NEC from those who were not. The only predictive features we found were the presence of maternal drug abuse, early onset infection, respiratory distress, exclusive artificial formula feeding and perhaps feeding volumes in excess of what ‘nature’ may have intended.

We recently reported on another relatively rare subtype of NEC, namely cases developing among term infants.9 The 30 cases in that report, and the 21 cases in the present report, have several items in common. Actually, 2 of the cases of NEC in term neonates are included among the present 21 with early NEC, and therefore the total of the two reports is 49 distinct cases. Taken together, 48 of the 49 had been admitted to a NICU for treatment before NEC developed. The only exception was in a well baby nursery and was SGA with the history of a tight nuchal cord. We speculate that the underlying conditions of the 49 resulted in a common pathophysiology involving aberrant mesenteric blood flow. Whether this pathophysiology is transient or more chronic likely depends upon the underlying disorder associated with each precondition and may provide an explanation for disparate timing of NEC presentations between the two populations (early NEC vs term NEC). One example of a very transient precondition might be the effects of illicit maternal drug use. A maternal illicit drug history was not observed as a common feature among term infants with NEC,9 but was significant among those who developed early NEC.

We observed marked similarities in the feeding practices used in these 49 patients (term NEC and early NEC). Specifically, none were fed with human milk exclusively, yet exclusive human milk feeding was quite common among the case–controls. Also, exclusive formula feeding was significantly overrepresented among those who went on to develop NEC, while it was far less common among the controls.

The mortality rate was also similar among the 21 patients with early NEC (10%) and the 30 term neonates with NEC (13%).9 In fact, the mortality rate we recently reported among 131 neonates with NEC,17 most of which were classical cases, not early and not term neonates, was 6%. Statistically, there is no difference in mortality rate among these various groups of NEC patients.

We speculate one reason NEC remains a largely enigmatic condition is that it is actually several different entities. SIP is an example of a condition sometimes confused with NEC.13, 14, 15 Human milk feeding programs are not likely to reduce the prevalence of SIP, since many or most cases of SIP occur before the first feeding.4 Similarly, viral gastroenteritis might be an entity distinct from classical NEC.4 This illness might also be refractory to prevention using human milk. Thus, we maintain that as NEC prevention schemes are tested,3 it will be useful to attempt to categorize the NEC cases into component entities. The present data and that in our previous report9 lead us to conclude that term and early NECs are pathogenically similar. Finally, we speculate that the prevalence of early NEC will fall if clinicians will systematically identify at-risk neonates in their NICU and then institute human milk feedings (mother's own milk or banked human milk) in conjunction with conservative feeding guidelines for the first week of life.


  1. 1

    Moss RL, Dimmitt RA, Barnhart DC, Sylvester KG, Brown RL, Powell DM et al. Laparotomy versus peritoneal drainage for necrotizing enterocolitis and perforation. N Engl J Med 2006; 354: 2225–2234.

  2. 2

    Neu J . Neonatal necrotizing enterocolitis: an update. Acta Paediatr Suppl 2005; 94: 100–105.

  3. 3

    Grave GD, Nelson SA, Walker WA, Moss RL, Dvorak B, Hamilton FA et al. New therapies and preventive approaches for necrotizing enterocolitis: report of a research planning workshop. Pediatr Res 2007; 62: 510–514.

  4. 4

    Gordon PV . The little database that could: Intermountain Healthcare and the uphill quest for prevention of term necrotizing enterocolitis. J Perinatol 2007; 27: 397–398.

  5. 5

    Bell MJ, Shackelford P, Feigin RD, Ternberg JL, Brotherton T . Epidemiologic and bacteriologic evaluation of neonatal necrotizing enterocolitis. J Pediat Surg 1997; 14: 1–4.

  6. 6

    Walsh MC, Kliegman RM . Necrotizing enterocolitis: treatment based on staging criteria. Pediatr Clin North Am 1986; 33: 179–202.

  7. 7

    Vermont Oxford Network Database. Manual of Operations,Release 10.0. Burlington, VT, 2005, pp 77, 78.

  8. 8

    Gordon PV, Swanson JR, Attridge JT, Clark R . Emerging trends in acquired neonatal intestinal disease: is it time to abandon Bell's criteria? J Perinatol 2007; 27: 661–671.

  9. 9

    Lambert DK, Christensen RD, Henry E, Besner GE, Baer VL, Wiedmeier SE et al. Necrotizing enterocolitis in term neonates: data from a multihospital healthcare system. J Perinatol 2007; 27: 437–443.

  10. 10

    Dollberg S, Lahav S, Minouni FB . A comparison of intakes of breast-fed and bottle-fed infants during the first two days of life. J Am Coll Nutr 2001; 20: 209–211.

  11. 11

    Evans KC, Evans RG, Royal R, Esterman AJ, James SL . Effect of caesarean section on breast milk transfer to the normal term newborn over the first week of life. Arch Dis Child Fetal Neonatal Ed 2003; 88: F380–F382.

  12. 12

    Thilo EH, Lazarte RA, Hernandez JA . Necrotizing enterocolitis in the first 24 h of life. Pediatrics 1984; 73: 476–480.

  13. 13

    Attridge JT, Clark R, Walker MW, Gordon PV . New insights into spontaneous intestinal perforation using a national data set (1): SIP is associated with early indomethacin exposure. J Perinatol 2006; 26: 93–99.

  14. 14

    Attridge JT, Clark R, Walker MW, Gordon PV . New insights into spontaneous intestinal perforation using a national data set (2): two populations of patients with perforations. J Perinatol 2006; 26: 185–188.

  15. 15

    Attridge JT, Clark R, Gordon PV . New insights into spontaneous intestinal perforation using a national data set (3): antenatal steroids have no adverse association with spontaneous intestinal perforation. J Perinatol 2006; 26: 667–670.

  16. 16

    Guthrie SO, Gordon PV, Thomas V, Thorp JA, Peabody J, Clark RH . Necrotizing enterocolitis among neonates in the United States. J Perinatol 2003; 23: 278–285.

  17. 17

    Wiedmeier SE, Henry E, Baer VL, Stoddard RA, Eggert LD, Lambert DK et al. Center differences in NEC within one healthcare system may depend on feeding protocol. Am J Perinatol 2008; 25: 5–12.

Download references

Author information

Correspondence to R D Christensen.

Rights and permissions

Reprints and Permissions

About this article


  • NEC
  • early NEC
  • feeding
  • human milk
  • formula
  • risk factors

Further reading