|Home | About | Journals | Submit | Contact Us | Français|
We sought to determine the incidence of necrotizing enterocolitis (NEC) and spontaneous intestinal perforation (SIP) in surviving extremely low-birth-weight (ELBW, <1000 g birth weight) infants and to establish the impact of NEC on outcomes by hospital discharge and at 18 to 22 months adjusted age in a large, contemporary, population-based practice.
Hospital outcome data for all ELBW infants born in the greater Cincinnati region from 1998 to 2009 were extracted from the National Institute of Child Health Neonatal Research Network Database. Neurodevelopmental outcome at 18 to 22 months was assessed using Bayley Scales of Infant Development-II scores for Mental Developmental Index and Psychomotor Developmental Index. Multivariable logistic regression was used and adjusted odds ratios reported to control for confounders.
From 1998 to 2009, ELBW infants accounted for 0.5% of the 352 176 live-born infants in greater Cincinnati. The incidence of NEC was 12%, with a 50% case-fatality rate. Death before discharge, morbid complications of prematurity and neurodevelopmental impairment were all increased among infants diagnosed with NEC. Infants with surgical NEC and SIP had a higher incidence of death, but long-term neurodevelopmental outcomes were not different comparing surviving ELBW infants with medical NEC, surgical NEC and SIP.
Although ELBW infants comprise a very small proportion of live-born infants, those who develop NEC and SIP are at an increased risk for death, morbid complications of prematurity and neurodevelopmental impairment. No significant differences in neurodevelopmental outcomes were observed between the medical and surgical NEC and SIP groups.
Necrotizing enterocolitis (NEC) is a severe gastrointestinal disease with significant mortality and morbidity predominantly affecting preterm infants.1,2 Extremely low-birth-weight (ELBW) infants (birth weight ≤ 1000 g) and infants with gestation <28 weeks are most susceptible. Despite advances in neonatal and perinatal medicine that have enabled the survival of increasing numbers of ELBW infants, the incidence (5 to 11%) and mortality rate (15 to 25%) for ELBW infants with NEC has not improved over the past 30 years.2–5 Increasingly, extremely premature infants are surviving early problems of prematurity only to succumb to later complications, specifically NEC.6 Outcome studies indicate that many ELBW infants with NEC who receive medical treatment have longer-term outcomes similar to infants without NEC of matched gestational age. Infants with progressive disease requiring surgical intervention exhibit higher rates of mortality (30 to 40%) and suffer the preponderance of neurodevelopmental morbidities.2,7,8
Spontaneous intestinal perforation (SIP) is an acquired form of neonatal bowel disease that is distinct from NEC and has been documented within the low-birth-weight population for over 20 years.9 In the ELBW population, causal associations have been established between SIP and early postnatal steroids (use within the first week of life), early use of indomethacin (use within the first 3 postnatal days) and the synergistic combination of the two. These two risk factors are thought to play a causal role in the etiology of SIP through their effects on ileal trophism and motility. Other reported risk factors include two common pathogens (Candida and Staphylococcus epidermidis), early neonatal hypotension, umbilical artery catheters and dehydration.10 Surgical NEC and SIP have significant differences in presentation, demographics, morbidity and histopathology.11 In a cohort of 1357 ELBW infants, it was shown that intestinal perforation caused by NEC, as compared to SIP, is associated with worse neurodevelopmental outcome at 1 year.12
In a large, multicenter, retrospective analysis of ELBW infants born in the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network (NICHD NRN) centers from 1995 to 1998, Hintz et al.7 reported that surgical treatment of NEC is associated with poorer growth and neurodevelopmental outcomes compared with medical treatment. A systematic review by Schulzke and co-workers8 (n = 4239 very low-birth-weight infants born between 1977 and 2002) also found that survivors of NEC are at risk for long-term neurodevelopmental impairment, particularly if they require surgical treatment for NEC.
In a contemporary, population-based, neonatology practice, we sought to define the incidence of NEC and SIP, and to compare hospital newborn intensive care unit (NICU) outcomes and long-term growth and neurodevelopmental outcomes of ELBW infants with and without NEC or SIP, and to additionally compare these outcomes based on medical vs surgical treatment for NEC.
This was an analysis of infants who had birth weights of 401 to 1000 g, were born from 1 January 1998 through 31 July 2009 and were in the Cincinnati Collaborative Outreach Program Database and the NICHD NRN registry. The NICHD NRN registry was developed to survey practice, assess morbidity and mortality, and provide information for planning of clinical trials. This was a population-based study with a single group of neonatologists that care for all infants requiring level II and level III care in the newborn NICU. All infants have been assessed by a stable team of follow-up examiners (TG and JS) throughout the course of the study. The numerator and denominator were known because all live births were entered into the Cincinnati Collaborative Outreach Program database, and all NEC and SIP cases in the region occurred or were referred to one of the three level III NICUs in Cincinnati. This allowed calculation of a true incidence for NEC for the Cincinnati region. The Institutional Review Boards at Cincinnati Children’s Hospital Medical Center, Good Samaritan Hospital and University of Cincinnati Hospital reviewed the data collection procedures and granted waiver of consent. Infants qualified for inclusion in the very-low-birth-weight registry when they were admitted within 14 days of birth to a Cincinnati NICU or were live born but died in the delivery room at a Cincinnati NICHD NRN participating hospital. Research nurses collected demographic, perinatal and infant data at each participating hospital using common definitions developed by the investigators and described in previous publications.3,13 Data were collected on diagnoses, treatments and in-hospital morbidities until death, discharge or 120 days; after 120 days or if the patient was transferred, data were collected with regard to death or discharge to home. Data with regard to NEC were collected for infants who survived 12 h and was defined as Modified Bell’s classification stage IIA or greater.14 Surgery for NEC included any surgical intervention (drain, laparotomy or both). SIP (spontaneous gastrointestinal perforation without proven NEC) was defined as an acute gastrointestinal perforation diagnosed by X-ray without classical radiographic findings of NEC or findings of NEC at surgery or autopsy, including pathological specimens.
Comprehensive follow-up at 18 to 22 months postmenstrual age was considered standard of care for ELBW infants at the Cincinnati NICHD NRN site. All neurological assessments were performed by one of two certified, masked developmental specialists. The neurological examination was based on the Amiel–Tison assessments,15 and the gross motor skills examination was developed from the work of Russell and Palisano.16,17 A single, experienced gold standard examiner (TG) administered the Bayley Scales of Infant Development-II (BSID-II). Mental Developmental Index (MDI) and Psychomotor Developmental Index (PDI) were obtained. BSID-II scores of 100±15 represent the mean±1 s.d. Scores <70 are 2 s.d.’s below the mean and represent significant developmental delay.18 Cerebral palsy was defined as a non-progressive central nervous system disorder characterized by abnormal muscle tone in at least 1 extremity and abnormal control of movement and posture. Deafness was defined as using hearing aids in both ears. Hearing impairment was defined as any restriction or lack of ability to perform within the range considered as normal, resulting from impairment, or if there was chronic otitis media associated with delayed speech skills. Blindness was defined as no useful vision in either eye. Visual impairment encompassed need for corrective lenses, blindness with some functional vision or blindness with no functional vision. Neurodevelopmental outcomes for the study were defined as the following: MDI <70, PDI <70, hearing impairment, visual impairment and cerebral palsy. A combined ‘any disability’ variable was also created, which included any of the above criteria. Statistical comparisons with regard to the above outcomes were conducted between NEC vs no NEC or SIP, medical NEC vs surgical NEC and SIP vs no NEC or SIP groups. Growth outcomes were defined as the weight, length and head circumference measured at the 18 to 22 month visit.
SAS V9.2 (Cary, NC, USA) was used to analyze the data. Unadjusted (crude) associations between group status for the hypothesis (NEC vs no NEC or SIP, medical NEC vs surgical NEC, surgical NEC vs SIP and SIP vs no NEC or SIP) and neonatal characteristics were explored through bivariable analysis using χ2, Fisher’s exact, Wilcoxon’s rank sum or t-test, where appropriate. The relationship between no NEC or SIP, NEC, surgical NEC, medical NEC and SIP groups with absolute and adjusted MDI and PDI were also analyzed.
Multiple logistic regression models were developed to evaluate NEC-related risk for outcomes MDI <70, PDI <70 and any disability, while controlling for other covariates and confounders. Demographic, maternal and neonatal risk factors known to be associated with neurodevelopmental outcomes were entered into the regression models. These factors included use of antenatal glucocorticoids, rupture of membranes >24 h, birth weight, gender, race, surfactant therapy, intraventricular hemorrhage (IVH), bronchopulmonary dysplasia (BPD), cystic periventricular leukomalacia, sepsis and postnatal steroid treatment. A stepwise approach was taken to determine which variables remained in the final multivariable models. Variables were included in the regression models if found to be significantly related to the outcome at the P<0.2 level in unadjusted analysis. These variables remained in the adjusted models if they were significant at P<0.1. The Hosmer–Lemeshow test was conducted to determine the model’s goodness-of-fit. Logistic regression models could not be developed to compare SIP and surgical NEC or SIP and no NEC or SIP because of small sample size, which creates an unstable model. General linear models were constructed to compare mean MDI and PDI values (as continuous variables) between groups, while adjusting for the same confounders that were included in the logistic regression models. Adjusted means with 95% confidence intervals for both MDI and PDI were reported.
During the study period (infants born between 1 January 1998 and 31 July 2009 with hospital outcomes to 30 November 2009), a total of 352 176 live births were recorded in the Cincinnati Collaborative Outreach Program database, with 1959 infants (0.6%) in this birth cohort delivered at <1000 g. A total of 1722 ELBW infants in this birth cohort survived >12 h, and of these, 208 (12%) developed NEC, with 121 receiving surgical treatment and 87 receiving medical treatment. In total, 50 infants (41%) with surgical NEC and 53 infants (61%) with medical NEC survived to NICU discharge; 55 (3%) infants developed SIP and 26 (51%) survived to discharge (Figure 1). For infants born after 1 January 2006, the criteria for NICHD NRN high-risk infant follow-up was restricted to include only inborn infants <27 weeks gestational age, and the neurodevelopmental assessment tool was changed from the BSID-II to BSID-III.19 As a result, our 18- to 22-month follow-up analysis was truncated to 995 infants who survived to NICU discharge. During the time period between hospital discharge and follow-up, 20 infants died leaving 975 evaluable infants. A total of 865 infants were included in the 18- to 22-month neurodevelopmental analysis, yielding a follow-up rate of 89%. This long-term follow-up cohort included 785 infants without NEC or SIP, 30 infants with medical NEC, 32 infants with surgical NEC and 18 infants with SIP.
Characteristics of the ELBW infants included in this study are summarized in Table 1. Among infants with NEC, maternal treatment with antenatal antibiotics was significantly higher among infants who subsequently developed surgical NEC (P = 0.005 vs medical NEC). Infants who developed NEC had a lower birth weight (759±145 vs 783±144 g, P = 0.03) and were more likely to be Caucasian (P = 0.01). Otherwise, the NEC and no NEC or SIP groups did not differ significantly with regard to perinatal or neonatal factors. Infants with SIP were more likely to have been exposed to antenatal antibiotics (P = 0.004 vs no NEC or SIP), were more likely to have a lower birth weight (741±128 vs 783±144 g, P = 0.04) and lower gestational age (26.2±2 vs 25.1±1.4 weeks, P<0.0001). Infants with SIP had a lower gestational age (25.1±1.4 vs 25.7±2.1) when compared to infants with surgical NEC.
NICU outcomes were substantially different between ELBW infants in the no NEC or SIP and NEC groups. The NEC rate in ELBW infants has not improved over the 12 years of observation (median: 11.6%; range: 8.3 to 19.8%). The death rate has not changed substantially between 1998 and 2009 (median: 25.5%; range: 15.4 to 32.4%). The incidence of death was twofold higher among infants who developed NEC compared to those who did not (P<0.001) (Table 2). The mortality rate was also significantly higher among infants with surgical NEC compared to those with medical NEC. Infants who developed NEC had a lower rate of BPD, but a higher rate of other significant morbid complications of prematurity, including IVH, late-onset sepsis, cystic periventricular leukomalacia and retinopathy of prematurity when compared to infants who did not develop NEC or SIP. Further, infants who developed NEC required significantly increased use of intensive care resources, such as increased total duration of ventilation, duration of parenteral nutrition and length of stay, compared to ELBW infants who did not develop NEC or SIP. Infants with SIP had a higher incidence of death, IVH, patent ductus arteriosus, late-onset sepsis, longer duration of ventilation, parenteral nutrition and NICU length of stay when compared to infants without NEC or SIP. Infants with SIP had a significantly higher incidence of patent ductus arteriosus and duration of ventilation when compared to infants with surgical NEC.
Growth outcomes were similar between infants in no NEC, NEC, medical NEC, surgical NEC and SIP groups (Table 3). Mean body length at 18 months was 1 cm greater for infants in the no NEC vs NEC group, but body weights were not significantly different. No significant differences were observed in growth measurements between medical and surgical NEC groups.
Absolute mean MDI and PDI values were significantly greater among infants in the no NEC group than those with NEC as shown in Table 3. Even after adjustment for use of antenatal glucocorticoids, birth weight, gender, race, surfactant therapy, IVH, BPD, sepsis and postnatal steroid treatment, MDI and PDI values remained significantly different between no NEC and NEC groups. The mean adjusted PDI was also greater for infants with medical vs surgical NEC. Adjusted mean MDI and PDI values were significantly lower in infants with SIP when compared to infants without NEC or SIP.
We conducted multiple logistic regression analyses for neurodevelopmental outcomes using MDI <70, PDI <70 and any disability as dichotomous variables. After controlling for birth weight, race, gender, multiple births, antenatal steroids, surfactant, BPD, sepsis and IVH, infants with NEC were more likely to have a PDI <70 compared to infants without NEC or SIP (odds ratio 2.64, 95% confidence interval 1.18 to 5.91). Similarly, the odds ratio for having any disability in infants with NEC was 2.59 (95% confidence interval 1.44 to 4.66) when compared to infants without NEC or SIP (Table 4). There were no major differences in outcomes between the surgical and medical NEC groups or between the surgical NEC and SIP groups.
This study is the largest, population-based study providing NICU outcomes of ELBW infants with and without NEC or SIP. It is also the first to document long-term neurodevelopmental outcomes of a large cohort of infants with NEC with follow-up over the most recent decade. Defining aspects of this study include: (1) total live births in the catchment area are available from the Cincinnati Collaborative Outreach Program database; (2) all ELBW infants in the study have received care from a single group of neonatologists practicing at all regional level II and level III nurseries; (3) all three NICUs participate in the NICHD NRN registry and follow-up programs; and (4) follow-up assessments have been performed by two investigators over the entire study period with high follow-up rate.
To determine population-based hospital outcomes, we studied 1959 ELBW infants of 352 176 live-born infants delivered in the greater Cincinnati region over the 12-year period (incidence 5.6 per 1000 infants). Among ELBW infants surviving the early neonatal period, the incidence of NEC in our population was 12%, and survival to NICU discharge was 65%. Our study is similar in scope to national, population-based studies from Canada and Sweden. Lee and co-workers20 reported the Canadian NICU Network outcomes for 1601 infants <28 weeks gestational age admitted in 1996 and 1997. The incidence of NEC was 10% and the NICU survival rate for extremely premature infants was 75%. The overall denominator for this population was not available since this study included units that received out-born infants and did not report numbers for live-born extremely premature infants or early neonatal deaths occurring in their catchment areas. The EXPRESS Group in Sweden conducted a population-based, prospective, observational study analyzing data from 305 318 infants born between 2004 and 2007.18 The incidence of extremely premature birth (<27 weeks gestational age) was 2.3 per 1000 live-born infants.21 In their study, 73% of extremely premature infants survived the early neonatal period and nearly 6% developed NEC.
The incidence of NEC in our population is similar to that reported in other North American populations, but twofold higher than that observed in Sweden.5,20,21 The reason for the marked discrepancy between the incidences of NEC in North American vs Swedish cohorts is uncertain. Clearly, the pathogenesis of NEC is multifaceted, with both genetic and environmental factors being implicated. Indeed, Bhandari and co-workers22 demonstrated a significant genetic component to risk of NEC and other complications of prematurity. Also, Morrow et al.23 identified the low or absent fucosyltransferase 2 gene expression to be predictive of severe outcomes, including severe NEC, in premature infants. Environmental factors including duration of antibiotic exposure and variability in NICU feeding practices may also underlie differences in regional and national NEC rates.24,25 The rates of death, morbid complications of extreme prematurity including late-onset sepsis, IVH, PVL, and retinopathy of prematurity and measures of neurodevelopmental outcome were significantly higher among infants afflicted with NEC compared to infants without NEC. The BPD rate was lower among infants with NEC, despite the longer duration of mechanical ventilation. This finding likely reflects the high mortality rate among infants with NEC. The burden of NEC on hospital resources is also substantial. Total days of ventilation during the hospital course were higher among infants afflicted with NEC than infants without NEC. The duration of parenteral nutrition was twofold longer among infants with NEC than infants without NEC. Infants requiring surgery for NEC require significantly longer duration of support with parenteral nutrition and mechanical ventilation and have a higher death rate when compared to infants with medical NEC. Furthermore, infants diagnosed with NEC had a significantly longer length of hospital stay, especially those infants requiring surgical treatment. These findings paralleled the length of stay and disease burden reported by other investigators.20
This is the first study comparing neurodevelopmental outcomes between SIP vs surgical NEC and SIP vs no NEC or SIP in ELBW infants. We did not find any differences in long-term neurodevelopmental outcomes between SIP and surgical NEC, but this may be owing to the limitation of this being a single center study with a limited sample size. Our long-term neurodevelopmental outcomes were similar to results reported by Hintz et al.7 and Schulzke et al.8 We found that the diagnosis of stage II or greater NEC is associated with an increased risk of adverse long-term neurodevelopmental outcome in preterm ELBW infants. Unlike previous studies, we observed that neurodevelopmental outcomes were similar between ELBW infants with medical NEC compared to those with surgical NEC. This disparity in results might be attributable to the high mortality rate of infants with surgical NEC at our center. Although our rate of mortality due to NEC was high, other investigators have also observed a trend toward an increased mortality attributable to NEC in recent years.6 Differences between infants with medical vs surgical NEC in growth outcomes at 18 to 22 months corrected age were modest in our cohort compared to those studied by Hintz and co-workers.7 We found a significant reduction in length (P = 0.05) and a trend toward reduction in head circumference (P = 0.06), but no difference in weight between infants with NEC and without NEC.
This study provides unique contemporary data for the incidence, hospital outcome and neurodevelopmental outcome of infants with NEC and SIP from a large, population-based neonatology practice in the United States. Infants with NEC are at high risk for adverse neurodevelopmental outcomes, irrespective of whether they received medical or surgical treatment. This study provides a baseline for conducting hypothesis-driven randomized trials and quality improvement initiatives aimed at the prevention of NEC.
We thank the following individuals and hospital staff who contributed to this research: Eunice Kennedy Shriver NICHD NRN administrator (Estelle Fischer) and research coordinators and staff: (Kate Bridges, Cathy Grisby, Jody Hessling, Lenora Jackson, Kristin Kirker, Holly Mincey, Greg Muthig, Stacie Tepe). The NICU staffs at Cincinnati Children’s Hospital Medical Center, Good Samaritan Hospital and University Hospital, Cincinnati. This work was supported in part by a grant from the National Institute of Child Health and Human Development Eunice Kennedy Shriver Neonatal Research Network (U10 HD 027853).
Conflict of interest
The authors declare no conflict of interest.