OBJECTIVES: To examine the variation in the incidence and to identify the timing of the presentation of necrotizing enterocolitis (NEC) in a cohort of preterm infants within the Canadian Neonatal Network (CNN).
METHODS: This was a population-based cohort of 16 669 infants with gestational age (GA) <33 weeks, admitted to 25 NICUs participating in the CNN between January 1, 2003, and December 31, 2008. Variations in NEC incidence among the participating NICUs for the study period were examined. We categorized early-onset NEC as occurring at <14 days of age and late-onset NEC occurring at ≥14 days. Multivariate logistic regression analysis was performed to identify risk factors for early-onset NEC.
RESULTS: The overall incidence of NEC was 5.1%, with significant variation in the risk adjusted incidence among the participating NICUs in the CNN. Early-onset NEC occurred at a mean of 7 days compared with 32 days for late-onset NEC. Early-onset NEC infants had lower incidence of respiratory distress syndrome, patent ductus treated with indomethacin, less use of postnatal steroids, and shorter duration of ventilation days. Multivariate logistic regression analysis identified that greater GA and vaginal delivery were associated with increased risk of early-onset NEC.
CONCLUSIONS: Among infants <33 weeks’ gestation, NEC appears to present at mean age of 7 days in more mature infants, whereas onset of NEC is delayed to 32 days of age in smaller, lower GA infants. Further studies are required to understand the etiology of this disease process.
- BW —
- birth weight
- CNN —
- Canadian Neonatal Network
- GA —
- gestational age
- IVH —
- intraventricular hemorrhage
- NEC —
- necrotizing enterocolitis
- PDA —
- patent ductus arteriosus
- RDS —
- respiratory distress syndrome
- SGA —
- small for gestational age
- SIP —
- spontaneous intestinal perforation
- SNAP-II —
- Score for Neonatal Acute Physiology, version II
- VLBW —
- very low birth weight
What’s Known on This Subject:
Necrotizing enterocolitis (NEC) can present within the first week of life in term infants. In preterm infants, NEC usually appears after commencement of feeds and can occur between 2 and 3 weeks of life.
What This Study Adds:
Among infants <33 weeks’ gestation, NEC appears to occur at mean age of 7 days in more mature infants, whereas onset of NEC is delayed to 32 days of age in smaller, lower gestational age infants.
Necrotizing enterocolitis (NEC) is one of the leading causes of morbidity and mortality in preterm infants.1–3 The incidence of NEC has remained relatively stable over recent epochs in the very low birth weight (VLBW) infant populations. The Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network reported gestational age (GA) specific mean incidence of NEC of 3% to 11% between 1997 and 20004 and 5% to 15% between 2003 and 2007.5 Other single centers and neonatal networks have reported similar incidence of NEC.6–8 One group9 from Australia has reported a decrease in the incidence in NEC from 12% between 1992 and 1993 compared with 6% between 1998 and 1999 for infants 24 to 28 weeks’ gestation.9
Traditionally, NEC in the preterm infant is described in enterally fed infants occurring at several days, if not weeks of age.2,10 Lin and Stoll,1 in their review, state that “the disease [NEC] is especially poignant because it mainly affects premature infants who have survived the early neonatal period and subsequently face a disease with high morbidity and mortality.” The mortality for NEC can be as high as 50% and surgical treatment is necessary in almost 20% to 40%. The inverse relationship between GA and age of onset of NEC has been well documented,11,12 with term infants presenting with NEC in the first week of life. Based on recent clinical experience, we raise the question of whether there is an earlier onset of NEC emerging in the VLBW infant population as some infants appear to be presenting with NEC in the first week of life.
Sankaran et al13 reported that there was no significant variation in the risk adjusted incidence of NEC among Canadian NICUs. Since that publication, the number of participating NICUs in the Canadian Neonatal Network (CNN) has increased from 17 to 25. By using this expanded network cohort we undertook this study (1) to examine the variation in the incidence and (2) to identify the timing of the presentation of NEC in a national population-based cohort of infants <33 weeks’ GA admitted to participating NICUs of the CNN.
This population-based cohort included 16 669 infants with GA <33 weeks admitted to 25 NICUs participating in the CNN between January 1, 2003, and December 31, 2008.
The CNN maintains an established national database for the purposes of outcomes evaluation, benchmarking, and quality improvement. Data for CNN are collected from patient charts by trained data abstractors at each of the participating NICUs in accordance to the manual of standardized operational definitions for variables and outcomes defined by the CNN. Details of data collection and data management have been published elsewhere.14 Ethical approval was obtained from the institutional review board of all participating institutions.
Indices of neonatal outcome were defined according to the CNN Data Abstractor Manual.14 GA was calculated from the best obstetric estimate based on early prenatal ultrasound examination, obstetric examination, and obstetric history, if the postnatal pediatric estimate of gestation differed from the obstetric estimate by more than 2 weeks, the pediatric estimate was used. NEC was defined according to modified Bell’s criteria (≥ stage 2),15 and patent ductus arteriosus (PDA) was defined as clinical diagnosis plus treatment with indomethacin, surgical ligation, or both. Bronchopulmonary dysplasia was defined as requirement for oxygen at 36 weeks’ corrected GA16 or at discharge from the participating unit. Retinopathy of prematurity was diagnosed according to the international classification of retinopathy of prematurity.17 Diagnosis and severity of intraventricular hemorrhage (IVH) was based on the criteria of Papile.18 Severity of illness was measured by the Score for Neonatal Acute Physiology, version II (SNAP-II) as described by Richardson et al.19 Small for GA (SGA) was defined as birth weight (BW) <10th percentile for the given GA, and outborn status was defined if the infant was born at a hospital different from the one in which the participating NICU was located. Congenital anomalies were identified from a defined list within the CNN database.
We performed 2 separate analyses to identify: (1) risk factors for NEC, and (2) risk factors for early-onset NEC.
In the first analysis, univariate analyses were performed to describe the characteristics of the study population and to explore associations between population characteristics and NEC. Multivariate logistic regression model was used to identify risk factors for NEC. The variables shown to be significantly different (P < .05) between the infants diagnosed with NEC and those without NEC in the univariate analyses were selected as possible risk factors for NEC and entered in the model.
Variations in NEC incidence among the participating NICUs for the study period were examined by looking at crude and risk adjusted odds ratios for the NEC incidence, where the site with median incidence rate was chosen as the reference site. Variables included in the adjusted model were GA, Apgar score at 5 minutes, SNAP-II, girl, cesarean delivery, outborn infant, and SGA.
The histogram of the time to onset of NEC for the cohort suggested a bimodal distribution with the first peak around at 8 days of age and the second peak around at 19 days of age. The least overlap between the 2 distributions was noted to be at day 14. Based on this cutoff, we categorized early-onset NEC as occurring at <14 days of age and late-onset NEC occurring at ≥14 days. A second analysis was restricted to the subset of infants with NEC. In this second analysis, univariate analyses were performed to explore the characteristics of infants with early- and late-onset NEC. Multivariate logistic regression analysis was performed to identify risk factors for early-onset NEC. Significant variables (P < .05) identified on univariate analyses were entered into the regression model. Statistical significance was considered to be present at the level P < .05. Statistical analysis was performed by using SAS version 9.2 (SAS Institute, Inc, Cary, NC) software package.
NEC in the Cohort
Of the 16 669 preterm infants <33 weeks admitted during the study period, 858 (5.1%) infants had NEC ≥ stage 2. Baseline characteristics and clinical outcomes of infants with and without NEC are presented in Table 1. The neonatal factors significantly correlated with NEC were lower BW, GA, and 5-minute Apgar score, higher admission SNAP-II, SGA infant, outborn infant, use of narcotic in the first 3 days of life, and postnatal steroid use. Clinical morbidities significantly associated with NEC were respiratory distress syndrome (RDS), PDA treated with indomethacin after 24 hours of age, nosocomial infection, IVH ≥ grade 3, and more ventilation days. Mortality was significantly higher in infants with NEC compared with those without NEC. On multivariate logistic regression analysis, risk factors for NEC were lower GA, SGA infant, outborn infant, narcotic use during the first 3 days of life, postnatal steroid use, and presence of congenital anomalies (Table 2).
Variations in the Incidence of NEC
The overall incidence of NEC was 858 of 16 669 (5.1%) with crude incidence varying from 1.3% to 12.9% (median 4.6%). Figure 1 shows that there was significant variation in the risk adjusted incidences of NEC among the participating NICUs. Six NICUs had significantly increased incidence of NEC compared with the reference site with the median incidence rate, whereas 2 sites had lower incidence of NEC.
Early- Versus Late-Onset NEC
The time of diagnosis of NEC (in days) was documented in 841 of 858 (98%) of infants. Of the 841 infants, 336 (40%) infants had early-onset NEC, whereas 505 infants had late-onset NEC. The mean (SD) age of diagnosis in the early-onset NEC group was 7.6 (3.1) days as compared with 32 (17.2) days among the late-onset NEC group. The incidence of surgical NEC was significantly higher in early-onset NEC group (40%) compared with late-onset NEC group (28%; P < .001). Figure 2 shows the distribution of cases according to GA at birth and the postmenstrual age at the time of NEC diagnosis. The peak onset of NEC in this cohort was ∼32 weeks’ postmenstrual age.
As shown in Table 3, early-onset NEC infants were of greater BW and GA, had higher 5-minute Apgar score, were more likely to be delivered vaginally, had lower SNAP-II, and had a lower rate of congenital anomalies on univariate analysis. Compared with late-onset NEC infants, the early-onset NEC infants had lower incidence of RDS, PDA treated with indomethacin after 24 hours of age, less use of postnatal steroids, and shorter duration of ventilation days (Table 4). Multivariate logistic regression analysis identified that greater GA and vaginal delivery were associated with increased risk of early-onset NEC. Postnatal steroid use was not associated with early-onset NEC. There was a trend toward reduction in risk of early-onset NEC in infants with PDA treated with indomethacin (Table 5).
The overall incidence of NEC was 5.1% in this national population cohort of infants <33 weeks’ GA, which is within the range reported by other large network databases and single centers.4,5,7,8 There was a significant variation in the risk adjusted incidence of NEC among the 25 NICUs within the CNN. Our findings are consistent with the Eunice Kennedy Shriver National Institute of Child Health and Human Development neonatal network who report persistent variation across sites in the incidence of NEC in VLBW infants.4,5
The authors of previous epidemiologic studies11,12,20 have described early- and late-onset NEC with early-onset NEC presenting in the first week of life in term infants. Studies of lower GA and lower BW infants reveal mean or median age of occurrence of NEC ranging from 13 to 23 days.11,12,20–22 In his review, Neu2 describes “classic” NEC occurring in preterm infants after 8 to 10 days of age. The peak onset of NEC at postmenstrual age of 32 weeks is similar to that reported previously.11
Our data suggest that a different profile and presentation of NEC has emerged in this population of preterm infants <33 weeks’ gestation. There remains an inverse relationship between GA/BW and onset of NEC, but the timeline has shifted such that the preterm infants with BW >1000 g present with NEC at a mean of 7 days (early onset) and NEC is delayed until a mean of 32 days (late onset) in lower GA infants with BW <1000 g. Grosfeld et al23 reported on a 25-year cohort of infants with surgical NEC and identified average age at diagnosis of 13 days for infants with BW >1000 g and 21 days for infants with BW <1000 g.
In the late 1980s and early 1990s, reports of focal intestinal perforation began emerging.24–27 The authors of subsequent studies began to make the association between this entity and use of indomethacin and postnatal steroids.28,29 The term spontaneous intestinal perforation (SIP) has also been used to describe this lesion, which occurs in preterm infants, receiving minimal enteral feeds or before receiving any enteral feeds and occurring in the first week of life. The incidence of SIP has been difficult to ascertain because it can only be confirmed histologically with laparotomy.30 Surgical management of intestinal perforation with intraperitoneal drain cannot reliably confirm the diagnosis.
Because most neonatal networks and single centers,4–8 including CNN, have not reported an overall increase in NEC ≥ stage 2 and 1 group reported a decrease,9 this suggests that if SIP is misclassified and disproportionately contributing to the incidence of stage 3 surgical NEC, the incidence of stage 2 nonsurgical NEC must be decreasing globally. Alternatively, if SIP contributes marginally to the overall incidence of NEC, we should consider the possibility of existence of an early-onset NEC that occurs in VLBW preterm infants presenting in the first 2 weeks of life at a mean of 7 days postnatal.
In our study, early-onset NEC occurred in 40% of the cohort. From multivariate analysis, the independent risk factors for early-onset NEC were greater GA and vaginal delivery. There was a trend to lower risk of NEC in infants with PDA treated with indomethacin. This association of decreased NEC with PDA treatment was described previously by Sharma et al.31 In our study, postnatal steroid use and SGA infant were not independently associated with early-onset NEC. Therefore, even if early-onset NEC represented misclassified SIP, the described risk factors of indomethacin and postnatal steroids treatment were not significantly associated with early-onset NEC in our cohort. Our data do not permit us to distinguish between early postnatal steroid use for treatment of hypotension compared with later treatment of bronchopulmonary dysplasia and this may explain the lack of association between postnatal steroids and early-onset NEC.
We do not know why vaginal delivery is a risk factor for early onset NEC. We speculate that the early-onset NEC group were more mature, larger BW infants and probably more likely to be delivered vaginally. The “apparent paradox” of these vaginally delivered infants developing early-onset NEC may be related to feeding practices. Enteral feeding has been associated with NEC as most infants developing NEC have been fed. We speculate that more mature and potentially more stable infants could have been given enteral feeds earlier with more aggressive volume advancement. There is conflicting evidence as to whether these are risk factors associated with NEC. Recent meta-analyses32,33 do not confirm that early trophic feeds or rate of feeding advancements are associated with NEC. Exclusive breast milk feeds have been identified as protective for development of NEC compared with mixed feeds or formula feeds.34,35 More aggressive advancement of feeds with suboptimal breast milk volumes may have resulted in supplementation with formula feeds.36 In a recent study, only 30% of mothers were able to provide sufficient breast milk to their extremely premature infants.37
The authors of recent reviews1–3,10 discuss the multifactorial pathogenesis of NEC and suggest the complex interplay between immature gastrointestinal tract with associated immature mucosal barrier, immune defense, circulatory regulation, altered microbiota, inflammatory response, and genetics along with the more traditional putative factors of hypoxic-ischemic injury and enteral feeding in the host preterm infant. This raises the possibility that NEC is a developmental phenomenon with GA interacting with perinatal events, such as bacterial colonization and use of perinatal antibiotics that may influence the microbiota of the immature gut.
It is possible that NEC represents 2 or more different conditions, in differing patient populations, with differing etiologies. Gordon et al30 has suggested that NEC represents a disease entity within a spectrum of conditions affecting the neonatal gut, which they have termed “acquired neonatal intestinal disease.” SIP may be part of this spectrum of conditions. The association between surgical NEC and poorer neurodevelopmental outcome as compared with no NEC has been reported.38 Others have reported poorer neurodevelopmental outcome in patients with surgical NEC as compared with SIP.39 The expression of NEC and the long-term outcomes may depend on the underlying etiology in the population affected.
Strengths of this study include the large size cohort, the standardization of definitions of neonatal morbidities, and the quality of the CNN database in a regionalized healthcare system.
The definition of early- versus late-onset NEC derived from dichotomizing the cohort at 14 days is not based on any physiologic principle but on an inspection of our data. The distinction between stage 3 surgical NEC with intestinal perforation and SIP requiring surgical intervention cannot be delineated in our network population cohort. Data are not available to describe the relationship between type, timing, and rate of advancement of feeds and development of NEC. These are significant limitations of this study. Our study data do not allow for reporting of long-term neurodevelopmental outcomes in infants with NEC. Network databases, including the CNN, may have to consider stratification of these disease entities to better identify risk factors and predict long-term outcomes.
The overall incidence of NEC within the CNN is similar to other centers and networks. Within a cohort of preterm infants <33 weeks’ GA, NEC appears to present at a mean of 7 days (early onset) in more mature preterm infants and is delayed to 32 days (late onset) in lower BW, lower GA preterm infants. This early-onset NEC represents a shift in the timing of traditional preterm NEC. Further studies are required to understand the etiology of this disease process.
The Maternal-Infant Care Research Center is supported by the Ministry of Health and Long-term Care, Ontario, Canada.
Site Investigators of the Canadian Neonatal Network were as follows: Shoo K. Lee (Director, Canadian Neonatal Network); Prakesh S. Shah (Associate Director, Canadian Neonatal Network); Wayne Andrews (Janeway Children’s Health and Rehabilitation Centre, St John’s, NF, Canada); Francine Lefebvre (Sainte Justine Hospital, Montreal, QC, Canada); Nalini Singhal (Foothills Medical Centre, Calgary, AB, Canada); Barbara Bullied (Everett Chalmers Hospital, Fredericton, NB, Canada); Rody Canning (Moncton Hospital, Moncton, NB, Canada); Gerarda Cronin (St Boniface General Hospital, Winnipeg, MB, Canada); Kimberly Dow (Kingston General Hospital, Kingston, ON, Canada); Michael Dunn (Sunnybrook Health Sciences Centre, Toronto, ON, Canada); Adele Harrison (Victoria General Hospital, Victoria, BC, Canada); Andrew James (Hospital for Sick Children, Toronto, ON, Canada); Zarin Kalapesi (Regina General Hospital, Regina, SK, Canada); Lajos Kovacs (Jewish General Hospital, Montreal, QC, Canada); David Lee (St Joseph’s Health Centre, London, ON, Canada); Douglas D. McMillan (Izaak Walton Killam Health Centre, Halifax, NS, Canada); Prakesh Shah (Mount Sinai Hospital, Toronto, ON, Canada); Cecil Ojah (St John Regional Hospital, St John, NB, Canada); Khalid Aziz (Royal Alexandra Hospital, Edmonton, AB, Canada); Bruno Piedboeuf (Centre Hospitalier Universitaire de Québec, Sainte Foy, QC, Canada); Patricia Riley (Montreal Children’s Hospital, Montreal, QC, Canada); Daniel Faucher (Royal Victoria Hospital, Montreal, QC, Canada); Nicole Rouvinez-Bouali (Children’s Hospital of Eastern Ontario, Ottawa, ON, Canada); Koravangattu Sankaran (Royal University Hospital, Saskatoon, SK, Canada); Mary Seshia (Health Sciences Centre, Winnipeg, MB, Canada); Sandesh Shivananda (Hamilton Health Sciences Centre, Hamilton, ON, Canada); Todd Sorokan (Royal Columbian Hospital, New Westminster, BC, Canada); and Anne Synnes (Children’s and Women’s Health Centre of British Columbia, Vancouver, BC, Canada).
- Accepted October 11, 2011.
- Address correspondence to Wendy H. Yee, MD, FRCPC, MSc, Rockyview General Hospital, 63-7007 14 St SW, Calgary, AB, Canada T2V 1P9. E-mail:
Drs Yee and Soraisham were involved in the study conception, design, interpretation of data, and drafting the article; Drs Shah and Aziz contributed to the study conception, design, interpretation of data, and critical review of the article; and Drs Yoon and Lee were involved in acquisition of data, data analysis, interpretation of data, and critical review of the article.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- Stoll BJ,
- Hansen NI,
- Bell EF,
- et al.,
- Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network
- Horbar JD,
- Badger GJ,
- Carpenter JH,
- et al.,
- Members of the Vermont Oxford Network
- Wilson-Costello D,
- Friedman H,
- Minich N,
- et al
- Lee SK,
- McMillan DD,
- Ohlsson A,
- et al
- Shennan AT,
- Dunn MS,
- Ohlsson A,
- Lennox K,
- Hoskins EM
- Teasdale F,
- Le Guennec JC,
- Bard H,
- Perreault G,
- Doray B
- Uauy RD,
- Fanaroff AA,
- Korones SB,
- Phillips EA,
- Phillips JB,
- Wright LL,
- National Institute of Child Health and Human Development Neonatal Research Network
- Buchheit JQ,
- Stewart DL
- Schanler RJ,
- Lau C,
- Hurst NM,
- Smith EO
- Hintz SR,
- Kendrick DE,
- Stoll BJ,
- et al.,
- NICHD Neonatal Research Network
- Copyright © 2012 by the American Academy of Pediatrics