BACKGROUND: Prior studies have identified individual risk factors that are associated with necrotizing enterocolitis (NEC); however, the small sample sizes of these previous studies have not allowed the analysis of potential interaction between multiple variables and NEC. Our purpose was to describe the incidence and risk factors for NEC in premature neonates admitted for intensive care.
METHODS: We identified neonates as having NEC if they met accepted diagnostic criterion for necrotizing enterocolitis. Using a national database, we assessed the association between NEC and a battery of risk factors previously reported in peer-reviewed literature.
RESULTS: There were 15,072 neonates that met inclusion criteria; 14,682 did not have NEC, while 390 (2.6%) met criterion for NEC. Multivariate analysis showed that low birth weight was the most important risk factor for NEC. Other factors that were associated with an increased risk of NEC were exposure to antenatal glucocorticoids, vaginal delivery, need for mechanical ventilator support, exposure to both glucocorticoids and indomethacin during the first week of life, absence of an umbilical arterial catheter, and low Apgar score at 5 minutes. Length of hospital stay and mortality were higher in neonates with NEC than in neonates without NEC.
CONCLUSIONS: NEC remains an important cause of morbidity and mortality in prematurely born neonates. In contrast to previous studies, we found that exposure to antenatal glucocorticoids was associated with an increased risk for NEC independent of birth weight.
Necrotizing enterocolitis (NEC) is an acquired gastrointestinal disease associated with significant morbidity and mortality in prematurely born neonates. With improving care at the end of the presurfactant era, the incidence of NEC declined briefly, but increased after surfactant use became a standard of care.1 This reported increase is presumably because of the increased survival of extremely immature infants. In the last decade, there has been a gradual realization that not all clinically diagnosed cases of NEC fit the classic pathological description.2 For example, isolated small bowel perforation may be diagnosed as a form of NEC even though there is little bowel necrosis.2 Furthermore, not all types of NEC carry the same morbidity or mortality risks.
While many studies have identified individual risk factors related to the development of NEC, most studies include only a small number of neonates with NEC, are single-institution reports, or were done in the presurfactant era.3,4,5,6,7,8,9,10,11,12,13,14 Many authors have focused on a single factor rather than exploring the additive effects of several factors. Our review of the literature revealed no recent studies with sufficient cohort size to evaluate the relation between multiple factors for NEC. We have used our large national administrative data set to overcome the limitation of small numbers and to accomplish our primary goal of describing the incidence of NEC across birth weight and gestational age groups for premature neonates admitted for intensive care. Our secondary goal was to evaluate the interaction between multiple variables and NEC. Finally, in our data set, the diagnosis of NEC is divided into two groups: NEC-surgical and NEC-medical, which allowed us to pursue our third goal of describing the differences between neonates treated medically and those treated surgically.
We performed a retrospective review of an administrative (no unique patient identifiers) database to compare the demographic and outcome characteristics of neonates who developed NEC to those who did not. This database encompasses data from 98 neonatal intensive care units managed by Pediatrix Medical Group, Inc. These units are located in 24 different states across the United States and represent a geographically diverse population.15
Inclusion criteria were met if the neonate was born between 1 January 1998, and 31 January 2000, and was 23 to 34 weeks' estimated gestational age. Infants had to be born at the site-of-care and then admitted to the neonatal intensive care unit for ongoing care. Infants were excluded if they had major anomalies or were born at another facility.
Using a database derived from a computer-assisted tool that generates clinical progress notes, we reviewed data on birth weight, estimated gestational age, gender, site of birth and neonatal intensive care, type of delivery, and Apgar scores. Gestational age was determined by the best estimate of the neonatal and obstetrical care providers based upon physical examination of the neonate and menstrual and sonographic dating. We also reviewed reported use of antenatal glucocorticoids, postnatal intravenous glucocorticoids (dexamethasone or hydrocortisone), indomethacin, erythropoietin, surfactant, and umbilical (arterial or venous) catheters. We selected mechanical ventilation on day 1 of life as a proxy for severity of illness and breast milk on day 7 of life as a proxy for early commitment to the use of breast milk as nutritional source. We created three dichotomous (0/1) variables by assigning a value of one to neonates who received intravenous glucocorticoids (dexamethasone or hydrocortisone) and indomethacin during the first week of life; had any reported need for mechanical ventilation in the day 1 progress note; or had any reported use of breast milk in the day 7 progress note. For each of these three variables, we assigned a 0 value to neonates who had no evidence of exposure. In the database, the date the medication was started was collected information, so we also evaluated the timing of exposure for indomethacin and steroids. We defined early use of indomethacin as less than or equal to 3 days. Based on the work of Stark et al.,16 we also evaluated the interaction between the reported use of both indomethacin and steroids during the first week after birth. We evaluated discharge data including death before and age at discharge.
Definition of NEC
In our database, a diagnosis of NEC is recorded if the neonate has one or more of the following clinical signs: bilious, gastric aspirate or emesis, abdominal distention, or occult or gross blood in stool without evidence of a rectal fissure; and has one or more of the following radiographic findings: pneumatosis intestinalis, hepatobiliary gas, or pneumoperitoneum. Selection of either NEC-medical or NEC-surgical may be chosen at data entry.
In order to include all cases of NEC, we searched for gastric, intestinal, and bowel perforation in the diagnosis field. We also reviewed the surgical procedure and text comment fields to identify neonates who had abdominal surgery for NEC or bowel perforation. All patients having a diagnosis of NEC-medical or NEC-surgical were included in our analysis of NEC. Neonates with a diagnosis of isolated perforation and no evidence of NEC were excluded (n=16). Neonates with a diagnosis of NEC-surgical or who had surgery for NEC were included in the surgical NEC group.
Before beginning our analysis, the available literature was reviewed to identify factors thought to be important in the development of NEC. Each of these prospectively identified factors was evaluated first by univariate analysis and then in multivariate analysis. During the univariate analysis all of the numeric data (birth weight, EGA, APGAR, and length of hospital stay) were evaluated using both parametric (ANOVA or two-sample t-test) and nonparametric test (Kruskal–Wallis nonparametric test, and Mann–Whitney test). When the data were nonparametric, we used the Kruskal–Wallis nonparametric test when making more than two comparisons and the Mann–Whitney test for two sample comparisons. As a result of the large sample size, differences between groups were highly significant (often p<0.0001) and both approaches (parametric and nonparametric) yielded similar results. All data analysis was done using the 2001 version of the statistical software package, “Number Cruncher Statistical Systems”, Kaysville, UT created by J. Hintze (www.NCSS.com). Categorical variables (e.g., race, gender, medications, and interventions) were evaluated using a two-tailed χ2 test.
After univariate analysis (results are displayed in Tables 1 and Table 2), multivariate logistic regression was used to identify factors independently associated with NEC (results are displayed in Tables 3 and Table 4). We developed our model by incorporating the variables that were found to have significant interactions (p≤0.1) with the rate of NEC. Variables were entered into the model using a stepwise selection (p-value for entry and retention ≤0.1). Cases with missing values for any of the independent variables were excluded from the analysis. Only variables with adjusted odds ratio 95% confidence intervals that did not cross 1 were considered to have an independent and significant association with NEC. The same approach was used to determine which factors were associated with surgically treated as compared to medically treated NEC. Differences in length of stay and weight at discharge were assessed using ANOVA with both estimated gestational age and a diagnosis of NEC as a covariate.
A total of 15,072 neonates were included in our analysis, of which 390 (2.6%) met the criteria for NEC and 14,682 (97.3%) did not. Of the 390 patients with NEC, 245 (63%) were treated medically (medical NEC group) and 145 (37%) had abdominal surgery (surgical NEC group). The diagnosis of NEC occurred earlier in the surgically treated group compared to the medically treated group (17±17 vs 21±14 days, p=0.02).
Univariate analysis showed that neonates who developed NEC had lower birth weights, a younger gestational age, and were more often black patients, than neonates who did not develop NEC (Table 1). Similarly, neonates in the surgically treated group had lower birth weights, a younger gestational age, and lower Apgar scores at 1 minute than neonates who were managed medically.
Medication and procedures
Univariate analysis showed that neonates who developed NEC were more often exposed to several different medications: pre- and postnatal glucocorticoids, surfactant, caffeine, erythropoietin, and indomethacin (Table 2). They were also more likely to have been exposed to invasive procedures such as umbilical vessel catheterization and mechanical ventilatory support than neonates who did not develop NEC were. Neonates in the surgically treated group were also more often exposed to several different medications (postnatal glucocorticoids, surfactant, and indomethacin) and invasive procedures (umbilical lines and mechanical ventilatory support) than neonates in the medically treated group. In addition, breast milk use was less frequent in neonates who developed NEC when compared to neonates who did not develop NEC. Similar results were found in comparing surgically treated neonates to medically treated neonates.
Multivariate logistic regression showed that the most important variable associated with the development of NEC was birth weight (Table 3, Figure 1b). When birth weight was included in the logistic regression, many of the factors found to be associated with an increased risk of NEC in our univariate analysis were no longer significant. Variables (other than birth weight) found to be independently associated with an increased risk for NEC were being on a ventilator on the first day of life, having received antenatal glucocorticoids and exposure to both glucocorticoids and indomethacin during the first week of life (Table 3). The interaction between antenatal glucocorticoids and birth weight is shown in Figure 2a. The interaction between birth weight and exposure to indomethacin and intravenous glucocorticoids during the first week is shown in Figure 2b. Neonates at highest risk for NEC weighed less than 1000 g and received both indomethacin and intravenous glucocorticoids during the first week of life. Being delivered by Cesarean section and a history of having had an umbilical artery catheter were associated with a decreased risk of NEC independent of birth weight (Table 3).
In evaluating the differences between surgically and medically treated NEC, logistic regression showed the most important variable associated with surgical NEC was having been on mechanical ventilation during the first day of life (Table 4). Birth weight and gestational age did not help to discriminate between neonates who developed surgical NEC and those who developed the medical disease. The other variable found to be independently associated with an increased risk for surgical NEC was exposure to both glucocorticoids and indomethacin during the first week of life. Delivery by Cesarean section and having received breast milk on day 7 of life were associated with a decreased risk (Table 4) for surgically treated NEC.
Compared to neonates without NEC, those with NEC were more likely to die (4 vs 12%, p<0.05) and to have prolonged hospital stays. Most (72%) of the mortality occurred in neonates who were treated surgically and mortality in neonates who were treated surgically was much higher than neonates who could be managed medically (23 vs 5%, p<0.05). In addition, compared to neonates without NEC, those with NEC were older at discharge to home (days) for all gestational age groups: 23, 24, 25 weeks (101±25 — no NEC vs 119±22 — NEC); 26, 27, 28 weeks (72±21 — no NEC vs 87±26 — NEC); 29, 30, 31 weeks (40±15 — no NEC vs 61±27 — NEC); 32, 33, 34 (17±11 — no NEC vs 40±17 — NEC) with an ANOVA p value of <0.01 for both covariates, NEC and estimated gestational age group.
Our data confirm previous reports1,18 that NEC is an important neonatal problem associated with significant mortality. In all 37% of the neonates with NEC received surgery and 24% of the surgically treated neonates died. Hospital stay was substantially longer for neonates with NEC within all birth-weight groups.
The incidence we report for NEC is similar to that reported previously.6,17 Previous studies have not made distinctions between surgically and medically treated NEC. Our data suggest that preterm neonates who required surgery for treatment of NEC were different from those who could be managed medically. They were diagnosed earlier, were smaller, and more often required mechanical ventilatory support during the first day after birth suggesting that they were sicker. Overall, these infants had worse outcomes than those treated medically. As a result of these differences, we suggest that a new set of terminology is needed to describe the broad spectrum of disease previously termed NEC. In our opinion, surgical NEC should be evaluated as distinct from NEC that can be managed medically and isolated spontaneous bowel perforation should be considered a separate and distinct diagnosis.
Similar to prior reports on the epidemiology of NEC,1,18 we found that the most important factor associated with NEC was low birth weight (Figure 1). In all 45% of the neonates with NEC weighed less than 1000 g and 61% of the neonates with surgical NEC weighed less than 1000 g.
We did, however, find some new associations that are different from previous reports. Our most remarkable finding was the association between antenatal glucocorticoids and an increase in the incidence of NEC (Figure 2a). The vast majority of previous studies have shown the opposite relationship (i.e., antenatal glucocorticoids decrease the incidence of NEC).19,20,21,22 To examine whether increased survival in the most at-risk neonates might have influenced our results, we re-evaluated our statistics using only neonates who survived to 28 days of life. The association between antenatal glucocorticoids and NEC persisted and the estimated odds ratio was unchanged. Nearly identical to our findings, Lawrence et al.23 and Kamitsuka et al.24 have recently shown that exposure to antenatal glucocorticoids is associated with a two-fold increase in NEC.
We hypothesize there may be two reasons that our data on antenatal glucocorticoids are different from that reported in the previous decade. First, survival of more immature neonates has changed the at-risk population that is exposed to antenatal glucocorticoids. The protective effect of antenatal glucocorticoids might be birth-weight-specific; however, we did not find this to be true in our data set (Figure 2a). Second, antenatal glucocorticoids are much more commonly used today than in the late 1980s and early 1990s. The majority of our study population (59%) was exposed to antenatal glucocorticoids and some of these patients may have received multiple courses. Repetitive doses of glucocorticoids may have different morphologic effects upon gastrointestinal development when compared to a single course. Unfortunately, our database did not allow us to examine this hypothesis.
Glucocorticoid exposure is known to accelerate the maturation of the small bowel, but it does so in a selective fashion.25 Gordon et al.25 demonstrated that early postnatal dexamethasone administration to newborn mice results in accelerated mucosal growth, increased goblet cell numbers, and increased lumen diameter. In contrast, the surrounding smooth muscle fails to proliferate and is stretched to accommodate the growing lumen. This skewing of tissue growth may make the bowel wall more vulnerable to other factors, such as indomethacin.
Interestingly, our multivariate analysis did not show an association between postnatal intravenous glucocorticoids alone and NEC. This remained true even if we only evaluated neonates given intravenous glucocorticoids during the first week of life (Figure 2b) and therefore, we cannot infer that postnatal glucocorticoid use by itself increases the risk of NEC. However, our data provide an interesting parallel to the work by Stark et al.16 who found an increased incidence of intestinal perforation in infants weighing less than 1000 g who received both early dexamethasone and indomethacin. We found a similar proportion of extremely low-birth-weight infants who had increased risk of NEC when exposed to both indomethacin and early dexamethasone. While we consider intestinal perforations and NEC to be different disease entities, our data suggest that these diseases have common and synergistic risk factors within the smallest critically ill neonates.4
Our multivariate analysis identified two factors associated with a decreased risk for NEC — delivery by cesarean section and the presence of an umbilical arterial catheter (UAC). We speculate that extremely premature neonates delivered by cesarean section may have been less stressed by the birthing process. We cannot explain the “possible” protective role of a UAC, but our data do not support the widely held belief that the presence of a UAC increases the risk for NEC. As previously described, breast milk appeared to reduce the severity of NEC.2,6,26,27 Our multivariate analysis showed that in neonates who developed NEC, the reported use of breast milk decreased the likelihood that a neonate would require surgical treatment.
LIMITATIONS OF DATA
The strength of our study is that it utilizes a recent, national data set of a large number of premature neonates (≤34 weeks' gestational age), allowing us to take a complete and timely look at the epidemiology of NEC. The limitations of our study include the certainty of the diagnosis of NEC, which can be clinically confused with spontaneous intestinal perforations. Although we have excluded all diagnosed cases of perforation in this study, it is possible that a significant percentage of perforation cases have been misdiagnosed as surgical NEC. This is a dilemma that likely plagues all recent studies of NEC, since the diagnostic criteria for ileal perforations are not standardized and only one study has used histological criteria for the diagnosis of perforations in a clinical cohort.2 In addition, proxies may not adequately reflect the true severity of illness or the therapeutic approach. Retrospective studies are also limited by incomplete data, and evaluation of interactions between drugs (like steroids and indocin) is better done prospectively. Finally, this study is a retrospective analysis of clinical associations, which can be confounded by unmeasured related variables.
In summary, we have described the occurrence and spectrum of NEC in a large contemporary neonatal data set. Our findings add strength to the association between NEC and antenatal glucocorticoids in the postsurfactant era and broaden concerns for the gastrointestinal effects of indomethacin combined with postnatal glucocorticoids in the smallest and most critically ill neonates.
Holman RC, Stoll BJ, Clarke MJ, Glass RI . The epidemiology of necrotizing enterocolitis infant mortality in the United States. Am J Public Health 1997;87:2026–2031.
Gordon P, Rutledge J, Sawin R, Thomas S, Woodrum D . Early postnatal dexamethasone increases the risk of focal small bowel perforation in extremely low birth weight infants. J Perinatol 1999;19:573–577.
Ghidini A, Espada RA, Spong CY . Does exposure to magnesium sulfate in utero decrease the risk of necrotizing enterocolitis in premature infants? Acta Obstet Gynecol Scand 2001;80:126–129.
Gordon PV, Marshall DD, Stiles AD, Price WA . The clinical, morphologic, and molecular changes in the ileum associated with early postnatal dexamethasone administration: from the baby's bowel to the researcher's bench. Mol Genet Metab 2001;72:91–103.
Bury RG, Tudehope D . Enteral antibiotics for preventing necrotising enterocolitis in low birthweight or preterm infants. Cochrane Database Syst Rev 2000;CD000405.
Chandler JC, Hebra A . Necrotizing enterocolitis in infants with very low birth weight. Semin Pediatr Surg 2000;9:63–72.
De Laet MH, Dassonville M, Johansson A, et al. Small-bowel perforation in very low birth weight neonates treated with high-dose dexamethasone. Eur J Pediatr Surg 2000;10:323–327.
Kennedy KA, Tyson JE, Chamnanvanakij S . Rapid versus slow rate of advancement of feedings for promoting growth and preventing necrotizing enterocolitis in parenterally fed low-birth-weight infants. Cochrane Database Syst Rev 2000;CD001241.
Ledbetter DJ, Juul SE . Erythropoietin and the incidence of necrotizing enterocolitis in infants with very low birth weight. J Pediatr Surg 2000;35:178–181.
Parilla BV, Grobman WA, Holtzman RB, Thomas HA, Dooley SL . Indomethacin tocolysis and risk of necrotizing enterocolitis. Obstet Gynecol 2000;96:120–123.
Hoyos AB . Reduced incidence of necrotizing enterocolitis associated with enteral administration of Lactobacillus acidophilus and Bifidobacterium infantis to neonates in an intensive care unit. Int J Infect Dis 1999;3:197–202.
Neu J, Weiss MD . Necrotizing enterocolitis: pathophysiology and prevention. J Parenter Enteral Nutr 1999;23:S13-S17.
Hung FC, Huang CB, Huang SC, Hsieh CS, Chuang JH . Necrotizing enterocolitis in newborn: nine years' experience. Changgeng Yi Xue Za Zhi 1997;20:29–33.
Uauy RD, Fanaroff AA, Korones SB, Phillips EA, Phillips JB, Wright LL . Necrotizing enterocolitis in very low birth weight infants: ideographic and clinical correlates. National Institute of Child Health and Human Development Neonatal Research Network. J Pediatr 1991;119:630–638.
Clark RH, Auten RL, Peabody J . A comparison of the outcomes of neonates treated with two different natural surfactants. J Pediatr 2001;139:828–831.
Stark AR, Carlo WA, Tyson JE et al. Adverse effects of early dexamethasone in extremely-low-birth-weight infants. National Institute of Child Health and Human Development Neonatal Research Network. N Engl J Med 2001;344:95–101.
Lemons JA, Bauer CR, Oh W et al. Very low birth weight outcomes of the National Institute of Child health and human development neonatal research network, January 1995 through December 1996. NICHD Neonatal Research Network. Pediatrics 2001;107:E1.
Stoll BJ . Epidemiology of necrotizing enterocolitis. Clin Perinatol 1994;21:205–218.
Ballard RA, Ballard PL . Antenatal hormone therapy for improving the outcome of the preterm infant. J Perinatol 1996;16:390–396.
Bauer CR, Morrison JC, Poole WK, et al. A decreased incidence of necrotizing enterocolitis after prenatal glucocorticoid therapy. Pediatrics 1984;73:682–688.
Halac E, Halac J, Begue EF, et al. Prenatal and postnatal corticosteroid therapy to prevent neonatal necrotizing enterocolitis: a controlled trial. J Pediatr 1990;117:132–138.
Smith LM, Qureshi N, Chao CR . Effects of single and multiple courses of antenatal glucocorticoids in preterm newborns less than 30 weeks' gestation. J Matern Fetal Med 2000;9:131–135.
Lawrence D, Brewer D, Hornung R, Mersmann M, Donovan D . Antenatal glucocorticoid use, not perinatal antibiotics, may result in increased risk of necrotizing enterocolitis in very-low-birthweight Infants. Pediatr Res 2001;49:Abstract 1798.
Kamitsuka MD, Horton MK, Williams MA . The incidence of necrotizing enterocolitis after introducing standardized feeding schedules for infants between 1250 and 2500 grams and less than 35 weeks of gestation. Pediatrics 2000;105:379–384.
Gordon PV, Price WA, Stiles AD . Dexamethasone administration to newborn mice alters mucosal and muscular morphology in the ileum and modulates IGF-I localization. Pediatr Res 2001;49:93–100.
Lucas A, Cole TJ . Breast milk and neonatal necrotising enterocolitis. Lancet 1990;336:1519–1523.
Schanler RJ, Shulman RJ, Lau C . Feeding strategies for premature infants: beneficial outcomes of feeding fortified human milk versus preterm formula. Pediatrics 1999;103:1150–1157.
In addition to the authors, the following physicians participated in this study: Harrisburg, PA — K. Lorah; Utica, NY — M. Siriwardena; Wichita, KS — E. Otero, C. Smart; Ft. Lauderdale, FL — S. Haskins; Boca Raton, FL — F. Miller; Coral Springs, FL — G. Melnick; Boynton Beach, FL — L. Whetstine; Denver, CO — E. Berman, D. Eichorst, J. Toney, P. Honeyfield; Houston, TX — R. Rivas, H. Pierantoni, E. O'Donnell; Englewood, CO — K. Zarlengo; West Palm Beach, FL — D. Kanter; Virginia Beach, VA — R. Balcom, E. Bollerup; Frederickburg, VA — J. Amin; Spartanburg, SC — V. Iskersky; Watertown, NY — K. Komar; Tarzana, CA — J. Banks; Ventura, CA — J. van Houten; Hoboken, NJ — M. Dyan; Stratford, NJ — J. Coleman; Trenton, NJ — K. Weiss, R. Axelrod; Fountain Valley, CA — A. Esparza; Covina, CA — V. Chundu, G. Martin; San Luis Obispo, CA — J. Martin; Newport Beach, CA — L. Wickham, B. Hannam; Riverside, CA — M. Leitner; Las Vegas, NV — M. Kaneta; Alexandria, VA — M. Dwyer, L. Goldberg; Albuquerque, NM — R. Nederhoff, S. Swetnam; Aurora, CO — M. Brown; Phoenix, AZ — M. McQueen; Dallas, TX — J. Whitfield, T. Brannon; Roanoke, VA — R. Allen; Dayton, OH — N. Kantor; Ogden, UT — N. Harper; Columbia, SC — S. Ellis; Panama City, FL — D. Sprague; Pensacola, FL — J. Nagel; Reno, NV — G. Yup; Tacoma, WA — J. Mulligan, G. Jordan, R. Knudson; Ponce, PR — E. Ochoa; Barrington, IL — F. Uraizee; Fort Worth, TX — R. Sidebottom, D. Turbeville, M. Stanley; Charleston, WV — S. Maxwell; San Juan, PR — A. Rivera, M. Ortega; Austin, TX — J. Courtney, D. Wermer, J. Scharnberg; San Jose, CA — E. Alderete; Rock Hill, SC — A. Payne; South Bend, IN — R. White; Kansas City, MO — B. Heimes, J. Anderson; Huntington, CA — R. Liberman; Elmira, NY — W. Helmuth, J. Felix.
From The Pediatrix-Obstetrix Center for Research and Education.
About this article
Developing routinely recorded clinical data from electronic patient records as a national resource to improve neonatal health care: the Medicines for Neonates research programme
Programme Grants for Applied Research (2019)
Current Problems in Surgery (2019)
Current Problems in Surgery (2019)
Current Pediatric Reviews (2019)