- Email: [email protected]
Mortality and Management of Surgical Necrotizing Enterocolitis in Very Low Birth Weight Neonates: A Prospective Cohort Study
Mortality and Management of Surgical Necrotizing Enterocolitis in Very Low Birth Weight Neonates: A Prospective Cohort Study Melissa A Hull, MD, Jeremy G Fisher, MD, Ivan M Gutierrez, MD, Brian A Jones, MD, Kuang Horng Kang, MD, Michael Kenny, MS, David Zurakowski, PhD, Biren P Modi, MD, Jeffrey D Horbar, MD, Tom Jaksic, MD, PhD, FACS Necrotizing enterocolitis (NEC) is a leading cause of death in very low birth weight (VLBW) neonates. The overall mortality of NEC is well documented. However, those requiring surgery appear to have increased mortality compared with those managed medically. The objective of this study was to establish national birth-weight-based benchmarks for the mortality of surgical NEC and describe the use and mortality of laparotomy vs peritoneal drainage. STUDY DESIGN: There were 655 US centers that prospectively evaluated 188,703 VLBW neonates (401 to 1,500 g) between 2006 and 2010. Survival was defined as living in-hospital at 1-year or hospital discharge. RESULTS: There were 17,159 (9%) patients who had NEC, with mortality of 28%; 8,224 patients did not receive operations (medical NEC, mortality 21%) and 8,935 were operated on (mortality 35%). On multivariable regression, lower birth weight, laparotomy, and peritoneal drainage were independent predictors of mortality (p < 0.0001). In surgical NEC, a plateau mortality of around 30% persisted despite birth weights >750 g; medical NEC mortality fell consistently with increasing birth weight. For example, in neonates weighing 1,251 to 1,500 g, mortality was 27% in surgical vs 6% in medical NEC (odds ratio [OR] 6.10, 95% CI 4.58 to 8.12). Of those treated surgically, 6,131 (69%) underwent laparotomy only (mortality 31%), 1,283 received peritoneal drainage and a laparotomy (mortality 34%), and 1,521 had peritoneal drainage alone (mortality 50%). CONCLUSIONS: Fifty-two percent of VLBW neonates with NEC underwent surgery, which was accompanied by a substantial increase in mortality. Regardless of birth weight, surgical NEC showed a plateau in mortality at approximately 30%. Laparotomy was the more frequent method of treatment (69%) and of those managed by drainage, 46% also had a laparotomy. The laparotomy alone and drainage with laparotomy groups had similar mortalities, while the drainage alone treatment cohort was associated with the highest mortality. (J Am Coll Surg 2013;-:1e8. 2013 by the American College of Surgeons)
BACKGROUND:
is inversely related to birth weight, with the great preponderance of affected patients having very low birth weight (VLBW).2,4-6 It may be inferred that VLBW patients with NEC receiving an operation (surgical NEC) harbor a more severe form of the disease than those treated without surgery (medical NEC).7 However, there are no large-scale cohort studies directly evaluating the mortality of surgical and medical NEC by birth weight categories. Such data are important to surgeons because they potentially afford a more accurate determination of prognosis and may be used as a basis for future quality improvement efforts. Various strategies are available to treat surgical NEC: laparotomy and primary peritoneal drainage (PPD). Primary peritoneal drainage was initially proposed as a
Necrotizing enterocolitis (NEC) is the most common gastrointestinal emergency of newborns1 and remains a major cause of mortality.2,3 The incidence of NEC Disclosure information: Nothing to disclose. Drs Hull and Fisher contributed equally to this work. Presented at the New England Surgical Society 94th Annual Meeting, Hartford, CT, September 2013. Received September 27, 2013; Revised November 14, 2013; Accepted November 18, 2013. From the Department of Surgery, Center for Advanced Intestinal Rehabilitation, Boston Children’s Hospital and Harvard Medical School, Boston, MA (Hull, Fisher, Gutierrez, Jones, Kang, Zurakowski, Modi, Jaksic) and the Vermont Oxford Network, Burlington, VT (Kenny, Horbar). Correspondence address: Tom Jaksic, MD, PhD, FACS, 300 Longwood Ave, Fegan 3, Boston, MA 02115. email: [email protected]
ª 2013 by the American College of Surgeons Published by Elsevier Inc.
1
ISSN 1072-7515/13/$36.00 http://dx.doi.org/10.1016/j.jamcollsurg.2013.11.015
2
Hull et al
Surgical Necrotizing Enterocolitis Mortality
Abbreviations and Acronyms
NEC NICU PPD SIP VLBW VON
¼ ¼ ¼ ¼ ¼ ¼
necrotizing enterocolitis neonatal ICU primary peritoneal drainage spontaneous intestinal perforation very low birth weight Vermont Oxford Network
bridge to laparotomy in 19758 and subsequently has gained popularity as a potentially definitive therapy.9-11 Two randomized controlled trials comparing laparotomy to PPD for NEC demonstrated no significant difference with respect to mortality.12,13 Unfortunately, due to the small size of these trials, a Cochrane Review determined that alone or in combination, these investigations cannot be used to guide therapy.14 A third randomized control trial is currently enrolling patients, with a specific emphasis on evaluating neurologic outcomes.15 Given this incomplete evidence base, it is of interest to determine with what frequency surgeons choose PPD and laparotomy in the treatment of surgical NEC. This study used a large, prospectively collected cohort of VLBW neonates to establish birth weight-based mortality benchmarks for surgical NEC. These mortality rates are compared with those from a contemporaneous group of patients with rigorously defined medical NEC. The frequency of PPD and laparotomy use and the mortalities associated with these therapies, are also reported.
METHODS The Vermont Oxford Network (VON) is a nonprofit voluntary collaboration of health care professionals dedicated to improving the quality and safety of medical and surgical care for newborn infants and their families. The VON prospectively collects data on infants of birth weight 401 to 1,500 g, who are born at participating institutions or who are transferred to such an institution within 28 days of birth. These data are accrued until neonates are discharged from the hospital, die, or reach 1 year of age in the hospital. Death after transfer to a non-VON center is or readmission is tracked and included. Discharge home without readmission is considered survival. Data are collected by local staff using uniform definitions and then submitted electronically or on paper forms to the VON central office. Records are subjected to automated checks for quality and completeness and returned for correction if needed. This study was performed as part of an ongoing collaboration between VON and a group of pediatric surgeons. Compilation and quality review of data were performed
J Am Coll Surg
with surgical input, ensuring that data fields were completed and that any ambiguities in the entries were clarified. For example, surgeons evaluated any unclear surgical coding by hand. No protected health care information was collected. The VON is approved by the University of Vermont Institutional Review Board (#04-370) and exempted from review at Boston Children’s Hospital. Among the 655 United States VON centers, survey data indicate that 37% of centers were classified as having a type A NICU (restriction on assisted ventilation or only perform minor surgery). Forty-seven percent had a type B NICU (no restriction on assisted ventilation and perform major surgery, ie, repair of tracheoesophageal fistula/ esophageal atresia, or meningomyelocoele). The remaining 16% of centers had a type C NICU (no restriction on ventilation and perform cardiac surgery requiring bypass for neonates). For this cohort study, data were prospectively collected in the VON database from United States centers between January 2006 and December 2010 on newborns weighing between 401 and 1,500 g. Neonates with major birth defects, as listed in the VON Manual of Operations16 (anencephaly, holoprosencephaly, and major cardiac defects), and those with a length of stay of 3 or fewer days were excluded. Per VON Manual definition, NEC was diagnosed either by direct observation of intestine at operation or pathologic examination or by using a set of strict clinical criteria. A clinical diagnosis of NEC was made based on at least 1 physical finding (bilious gastric aspirate or emesis, abdominal distention, or occult/gross blood in the stool in the absence of anal fissures) and at least 1 radiographic finding (pneumatosis intestinalis, hepatobiliary gas, or pneumoperitoneum). Severity of NEC was not specifically coded in the dataset and the diagnosis of NEC can be coded only once. The VON Manual definition also distinguishes spontaneous intestinal perforation (SIP) from NEC if laparotomy is performed. Spontaneous intestinal perforation is defined as a single focal defect without other significant bowel pathology seen at laparotomy. Because SIP and NEC cannot be incontrovertibly differentiated without a laparotomy, the PPD group necessarily contains patients with both diagnoses. So for veracity of comparison, SIP and operative NEC are grouped together and referred to as “surgical NEC” in this article. Primary peritoneal drainage and laparotomy were identified by specific procedure codes. Patients with codes entered for PPD and laparotomy were considered to have undergone both procedures. Because the dates of procedures were not included, the order could not be specified.
Vol.
-,
No.
-, -
2013
Statistical analyses were conducted using SAS version 9.3 (SAS Institute). The primary outcome was death before 1 year of age or hospital discharge. Multivariable logistic regression modeling was performed. Multivariate logistic regression modeling was performed with adjustments for clustering of infants within hospitals (generalized estimating equations were used to account for correlation between infants at a given hospital). Model covariates included gestational age, small for gestational age, maternal race, sex, multiple births, Apgar score at 1 minute, birth location, and vaginal delivery (because these are independent predictors of mortality on multivariable analysis). Additional multivariable analysis was performed to assess factors associated with treatment by PPD vs laparotomy. Values of p < 0.05 were considered significant. Categorical measures were compared using z-tests and continuous measures were assessed by unpaired t-test.
RESULTS There were 215,057 very low birth weight neonates identified from 655 VON centers in the United States (Fig. 1). A total of 26,354 were excluded: 10,821 with a major congenital malformation or unknown congenital
Hull et al
Surgical Necrotizing Enterocolitis Mortality
3
malformation status, 15,493 with length of stay 3 or fewer days or unknown, and 40 with unknown NEC status. Of the remaining 188,703 patients, 17,159 were diagnosed with NEC (incidence 9%). Demographic characteristics are listed in Table 1. There were small but significant differences between the medical and surgical NEC groups. Infants with surgical NEC had lower birth weights, younger gestational age, and lower Apgar scores. The overall mortality for all patients with NEC was 28%. Mortality declined with birth weight. There were 8,221 patients with medical NEC and 8,935 with surgical NEC. Medical NEC mortality was 21% overall, with significantly lower mortality in neonates of larger birth weight. Surgical NEC mortality was 35% overall (higher than medical NEC, p < 0.0001) and, unlike that of medical NEC, it did not consistently improve with larger birth weight (Fig. 2). The surgical group was subdivided into those undergoing laparotomy (n ¼ 6,126, 69%) and those receiving PPD (n ¼ 2,804, 31%). The PPD group was also divided into those with PPD alone (n ¼ 1,521) and those who underwent PPD and laparotomy (n ¼ 1,283). In other words, 46% of the PPD group also had a laparotomy. Mortality in the laparotomy only group was 31% overall
Figure 1. Study design schema. LOS, length of stay; NEC, necrotizing enterocolitis; PPD, primary peritoneal drain; VLBW, very low birth weight.
4
Hull et al
Table 1.
Demographic and Obstetric Characteristics
J Am Coll Surg
Surgical Necrotizing Enterocolitis Mortality
Characteristic
Male sex, n (%) Vaginal delivery Multiple birth Small for gestational age Birth location (inborn) Antenatal steroids Birth weight, g, mean SD Gestational age, wk, mean SD Apgar 1 min, mean SD Apgar 5 min, mean SD
Medical NEC (n ¼ 8,221)
Surgical NEC (n ¼ 8,935)
p Value
4,508 (54.8) 2,500 (30.4) 1,967 (23.9) 1,117 (13.6) 6,518 (79.3) 6,360 (79.0) 938 273 26.9 2.4 5.0 2.5 7.1 1.9
5,220 (58.4) 2,955 (33.1) 2,237 (25.0) 952 (10.7) 5,802 (64.9) 6,741 (76.3) 841 248 25.9 2.3 4.5 2.4 6.7 2.0
<0.001 <0.001 0.091 <0.001 <0.001 0.003 <0.001 <0.001 <0.001 <0.001
Statistical comparisons are made between medical and surgical necrotizing enterocolitis (NEC).
and demonstrated a plateau, mirroring the overall surgical group (401 to 500 g, 51%; 501 to 750 g, 36%; 751 to 1,000 g, 29%; 1,001 to 1,250 g, 25%; 1,251 to 1,500 g, 26%). Mortality in the PPD only group was 50% (higher than laparotomy only, p < 0.0001), and this mortality tended to remain high across the birth weight categories (401 to 500 g, 57%; 501 to 750 g, 54%; 751 to 1,000 g, 43%; 1,001 to 1,250 g, 43%; 1,251 to 1,500 g, 52%). Those who underwent both PPD and laparotomy had an overall mortality of 34% (similar to the laparotomy only group, p ¼ 0.42) and it declined with birth weight: 401 to 500 g, 50%; 501 to 750 g, 36%;
Figure 2. Mortality in surgical necrotizing enterocolitis (NEC) vs medical NEC by birth weight category. Odds ratios for mortality for surgical vs medical NEC are shown with 95% confidence intervals. *p < 0.0001. Black bar, surgical NEC; gray bar, medical NEC. Mortality in medical NEC: 401 to 500 g, 52% (n ¼ 211); 501 to 750 g, 36% (n ¼ 2,267); 751 to 1,000 g, 21% (n ¼ 2,501); 1,001 to 1,250 g, 11% (n ¼ 1,886); 1,251 to 1,500 g, 6% (n ¼ 1,356). Mortality in surgical NEC: 401 to 500 g, 52% (n ¼ 345); 501 to 750 g, 40% (n ¼ 3,608); 751 to1,000 g, 32% (n ¼ 2,823); 1,001 to 1,250 g, 27% (n ¼ 1,391); 1,251 to 1,500 g, 28% (n ¼ 768).
751 to 1,000 g, 30%; 1,001 to 1,250 g, 27%; 1,251 to 1,500 g, 33%. A comparison of mortality by surgical treatment type is found in Figure 3. Operative intervention rather than medical management was more common among lower birth weight infants. Figure 4 reflects results reported from categorical analysis, with the 1,251 to 1,500 g group as a reference. Furthermore, a clustered logistic regression was performed using previously listed covariates and birth weight as continuous measures. For each 100-g increase in birth weight, the odds of surgery dropped by 5%. On multivariable analysis, antenatal steroid exposure was associated with surgical intervention (odds ratio 1.17, 95% CI 1.06 to 1.29, p < 0.01). Cardiac compressions, endotracheal ventilation, use of epinephrine, and early bacterial sepsis were not associated with surgical intervention. Multivariable analysis was also performed to assess factors associated with the use of PPD as opposed to laparotomy. Those with lower birth weight were more likely to receive a drain when birth weight was evaluated as a continuous variable. When analyzed by category, the percentage
Figure 3. Mortality by surgical intervention. *Statistical difference from both other groups. Laparotomy only vs laparotomy and primary peritoneal drain (PPD), p ¼ 0.42. Laparotomy only vs PPD only, p < 0.0001. Laparotomy and PPD vs PPD only, p < 0.0001.
Vol.
-,
No.
-, -
2013
Figure 4. Percent of necrotizing enterocolitis patients treated surgically (laparotomy and/or primary peritoneal drain) by birth weight. *p < 0.05 when compared with 1,251 to 1,500 g group as a reference.
undergoing PPD decreased with each larger weight category (Fig. 5). Further, cardiac compressions or endotracheal ventilation at initial resuscitation were associated with PPD (odds ratio 1.17, 95% CI 1.00 to 1.36, p ¼ 0.05, and 1.29, CI 1.04 to 1.60, p ¼ 0.02, respectively). Use of epinephrine, early bacterial sepsis, and antenatal steroids were not associated with PPD.
DISCUSSION More than 80% of VLBW neonates born in the United States during a 5-year period were analyzed using a prospectively collected VON database with embedded surgical codes.17 Of 188,703 eligible patients, 17,159 were
Figure 5. Percent of surgical necrotizing enterocolitis patients treated with peritoneal drainage. Includes both “peritoneal drainage only” and “peritoneal drainage and laparotomy” groups. *p < 0.05 when compared with 1,251 to 1,500 g group as the reference.
Hull et al
Surgical Necrotizing Enterocolitis Mortality
5
diagnosed with NEC, for an incidence of 9% (Fig. 1). This yielded the largest VLBW NEC study cohort available to date.2,9,18 Generally, NEC mortality was inversely related to birth weight, with more than half of the patients undergoing surgical intervention for NEC (n ¼ 8,935, 52%). The mortality for surgical NEC was significantly higher than for medical NEC overall and for each birth weight category in excess of 750 g. There is great variation in the reported mortality associated with NEC. A recent review summarized studies available since 19779 with an overall mortality ranging from 12% to 39%.7,18-26 When neonates weighing less than 1,500 g were considered, using the 2000 Kids’ Inpatient Database, an overall mortality of 20% was calculated.18 Additional discrepancies in NEC mortality become evident when higher birth weight infants are separated from lower birth weight neonates (6% to 63%).18 This investigation demonstrated an overall NEC mortality of 28%, which fell within the broad range of these other reports. However, given its large sample size, a more detailed stratification of mortality was possible by birth weight categories, whether surgical therapy was used, and the type of operation performed. Medical NEC mortality in this study (21%) was higher than previously reported. Previous investigations used retrospective, administrative, or survey data and the definition of NEC was not uniform.7,9,27 The largest of these studies, combining data from both the National Inpatient Sample and Kids Inpatient Database, showed a mortality of 6.8% for medical NEC.27 However, this post hoc review of administrative data was not restricted to VLBW neonates and was defined NEC by ICD-9 code only, potentially resulting in up to 45% of the entire sample having Bell’s stage I disease.9 Bell’s stage I disease is a very low mortality group that encompasses all neonates with abdominal distension.9,28 In contradistinction, VON uses a more rigorous definition of NEC that includes only neonates with clearly established disease of Bell’s stage II and higher. Additionally, the VON’s capacity for long-term follow-up and limitation of the cohort to VLBW (birth weight less than 1,500 g) neonates also contributed to a higher and likely more accurate mortality estimate for VLBW neonates. In fact, an earlier surgical-VON collaborative investigation found that the baseline mortality for VLBW infants without a diagnosis of NEC was 10%, already higher than the quoted rates for medical NEC mortality in other studies.2 Although specific data are not available, it is possible that some of the medical NEC patients had severe disease or cormorbidities and were treated palliatively, which could have contributed to the higher overall mortality.
6
Hull et al
Surgical Necrotizing Enterocolitis Mortality
Patients receiving operations for NEC are commonly considered to have a greater burden of illness than those managed medically, though specific risk factors for progression of NEC have been elusive.29 The VON cohort in this study revealed an increased prevalence of disease severity markers in surgical NEC compared with medical NEC (lower birth weight, younger gestational age, lower Apgar scores, and male sex, all with p < 0.001). Even after accounting for these risk factors, surgical NEC was still independently predictive of death and showed a significantly higher mortality than medical NEC (35% vs 21%, p < 0.005). Further, in surgical NEC, unlike medical NEC, the reduction in mortality seen with greater birth weight was significantly blunted. This resulted in a plateau mortality of approximately 30% even for VLBW patients with birth weights above 750 g (Fig. 2). At the largest birth weight category (1,251 to 1,500 g), the difference in mortality was the most striking: 27% for surgical NEC as opposed to 6% for medical NEC (p < 0.0001). These birth weight- and treatment-based benchmark data may be used to more accurately apprise caregivers and families of ultimate prognosis and to facilitate valid comparisons of NEC mortality between centers. The relative frequency of surgical to medical NEC decreased progressively with greater birth weight (Fig. 4). Although birth weight is the single best prognosticator of mortality in the VON database (perhaps because of its inherent objectivity) it correlates well with estimated gestational age (in this dataset r ¼ 0.77). Therefore, it may be postulated that larger, more mature infants have an improved ability to respond to the pathologic stress of NEC and as a consequence, have a reduced requirement for surgery. Moreover, it was observed that although laparotomy was used more frequently than PPD across all birth weight categories, this predisposition was accentuated in larger neonates (Fig. 5). At least in part, this may have been because PPD, often done at bedside under local anesthesia, was more likely to be undertaken in those neonates with a perceived higher operative risk (ie, those with the lowest birth weights). Recognizing that the surgeons’ specific decisionmaking process is unknown and no cause and effect relationship between treatment selection and outcome can be concluded, this study nonetheless describes current practice regarding the use of PPD and laparotomy in the United States. Overall, there was a strong preference noted for laparotomy over PPD in neonates with surgical NEC (69% vs 31%, p < 0.001). Such a finding may extend beyond a bias regarding improved immediate outcomes after laparotomy. Necrotizing enterocolitis is a disease that can affect variable quantities of bowel, and the extent of small intestinal loss is a strong determinant of
J Am Coll Surg
long-term outcomes.30 So it is possible that surgeons chose laparotomy rather than PPD because it gave them a more complete assessment of viable residual intestinal length and therefore of prognosis. It is clear that surgeons in the United States are not in equipoise regarding the 2 operative treatment options and further research is warranted to ascertain the precise reasons for this observation. Necrotizing enterocolitis patients undergoing PPD alone (n ¼ 1,521) had a higher associated mortality than those receiving laparotomy alone (n ¼ 6,131) (50% vs 31%, p < 0.0001). The same trend was seen in a previous National Institute of Child Health and Human Development Neonatal Research Network cohort limited to neonates with birth weights less than 1,000 g.31,32 In this small but carefully conducted study, the mortality associated with PPD was 54% (43 of 80) vs 43% (33 of 76) for laparotomy. The lower mortality for laparotomy observed in the VON data may be explained by the inclusion of larger infants with birth weights up to 1,500 g. There is both a progressively greater use of laparotomy and significantly lower NEC mortality in these larger neonates (Figs. 2 and 5). Although risk stratification, as outlined in the Methods section, was used to make statistical comparisons between the PPD and laparotomy treatment groups, great caution must be exercised in interpreting these data. Post hoc analysis revealed that the PPD patients, as a whole, were likely more severely ill than the laparotomy alone group. Specifically, neonates who underwent PPD had a higher likelihood of receiving both cardiac compressions and endotracheal ventilation at birth (p < 0.05). Therefore, the documented association between increased mortality and PPD alone may still be attributed to the severity of underlying illness rather than a treatment effect. It is of considerable interest that neonates who had PPD and a laparotomy (46% of PPD patients) demonstrated a mortality that quite closely approximated that of the laparotomy only group (p ¼ 0.42, Fig. 3). Classically, PPD is used as a bridge to laparotomy,8 yet in this study, 27% of neonates undergoing PPD survived without further surgery. These data are congruent with those from other studies and suggest that patients receiving PPD alone represent a dichotomous group: those who do not survive to laparotomy and a subset who show clinical improvement and never require laparotomy.10,33,34 It may be surmised that the latter cohort is comprised of relatively physiologically stable neonates who have minimal NEC or spontaneous intestinal perforation.10 Future trials of PPD vs laparotomy may need to focus on these 2 very different VLBW groups: the very ill and the clinically stable. It is perhaps in these extremes where surgeons will have sufficient equipoise to conduct
Vol.
-,
No.
-, -
2013
randomized controlled trials of sufficient size, and the design and primary end points of these studies may need to be quite different. Although the VON database is a powerful tool to assess the mortality and management of NEC, there are important limitations in its application. Because VON participation is voluntary, centers are not chosen at random or by criteria, so the resulting cohort is not strictly a population-based sample. However, this is mitigated by the fact that greater than 80% of VLBW neonates born in the United States are in the VON database and therefore were included for analysis. The VON collects information for each patient at discrete time points: hospital day 28, discharge or death, and transfer or readmission. This means that the specific codes entered (such as operative procedures) are not listed by date or necessarily in the order that they were actually performed. Further, although the operations undertaken are known, the specific indications and underlying decision-making remain unknown. Because this is an observational study across many centers, treatment of neonates (including feeding strategies, radiologic surveillance, and antibiotic therapy) was not standardized. Finally, it should also be mentioned that the VON data are an amalgam of results obtained from hospitals with heterogeneous resources, as outlined in the Methods section.
CONCLUSIONS In this very large United States cohort, 52% of VLBW neonates with NEC underwent operative intervention. Surgical NEC was associated with a significantly higher mortality than medical NEC both in aggregate and for all birth weight categories greater than 750 g. Furthermore, surgical NEC, unlike medical NEC, exhibited a plateau in mortality of approximately 30%, even in larger VLBW neonates. During the period of this study there was a marked preference exhibited for laparotomy rather than PPD in the surgical management of NEC. Neonates receiving laparotomy alone or PPD with laparotomy had similar mortalities. The highest mortality evident was in patients undergoing PPD alone, but this treatment may have been applied to neonates with the highest degree of overall illness. This national prospective cohort analysis provides surgeons with accurate mortality data for VLBW neonates with NEC and will be useful in the design and implementation of future randomized controlled trials. Author Contributions Study conception and design: Hull, Fisher, Gutierrez, Jones, Kang, Kenny, Zurakowski, Modi, Horbar, Jaksic
Hull et al
Surgical Necrotizing Enterocolitis Mortality
7
Acquisition of data: Kenny, Horbar Analysis and interpretation of data: Hull, Fisher, Gutierrez, Jones, Kang, Kenny, Zurakowski, Modi, Horbar, Jaksic Drafting of manuscript: Hull, Fisher, Gutierrez, Jones, Kang, Kenny, Zurakowski, Modi, Horbar, Jaksic Critical revision: Hull, Fisher, Gutierrez, Jones, Kang, Kenny, Zurakowski, Modi, Horbar, Jaksic REFERENCES 1. Lin PW, Stoll BJ. Necrotising enterocolitis. Lancet 2006;368: 1271e1283. 2. Fitzgibbons SC, Ching Y, Yu D, et al. Mortality of necrotizing enterocolitis expressed by birth weight categories. J Pediatr Surg 2009;44:1072e1075; discussion 1075e1076. 3. Neu J, Walker WA. Necrotizing enterocolitis. N Engl J Med 2011;364:255e264. 4. Cikrit D, Mastandrea J, West KW, et al. Necrotizing enterocolitis: factors affecting mortality in 101 surgical cases. Surgery 1984;96:648e655. 5. Llanos AR, Moss ME, Pinzon MC, et al. Epidemiology of neonatal necrotising enterocolitis: a population-based study. Paediatr Perinat Epidemiol 2002;16:342e349. 6. Ricketts RR, Jerles ML. Neonatal necrotizing enterocolitis: experience with 100 consecutive surgical patients. World J Surg 1990;14:600e605. 7. Guthrie SO, Gordon PV, Thomas V, et al. Necrotizing enterocolitis among neonates in the United States. J Perinatol 2003;23:278e285. 8. Ein SH, Marshall DG, Girvan D. Peritoneal drainage under local anesthesia for perforations from necrotizing enterocolitis. J Pediatr Surg 1977;12:963e967. 9. Rees CM, Eaton S, Pierro A. National prospective surveillance study of necrotizing enterocolitis in neonatal intensive care units. J Pediatr Surg 2010;45:1391e1397. 10. Cass DL, Brandt ML, Patel DL, et al. Peritoneal drainage as definitive treatment for neonates with isolated intestinal perforation. J Pediatr Surg 2000;35:1531e1536. 11. Lessin MS, Luks FI, Wesselhoeft CW Jr, et al. Peritoneal drainage as definitive treatment for intestinal perforation in infants with extremely low birth weight (<750 g). J Pediatr Surg 1998;33:370e372. 12. Moss RL, Dimmitt RA, Barnhart DC, et al. Laparotomy versus peritoneal drainage for necrotizing enterocolitis and perforation. N Engl J Med 2006;354:2225e2234. 13. Rees CM, Eaton S, Kiely EM, et al. Peritoneal drainage or laparotomy for neonatal bowel perforation? A randomized controlled trial. Ann Surg 2008;248:44e51. 14. Rao SC, Basani L, Simmer K, et al. Peritoneal drainage versus laparotomy as initial surgical treatment for perforated necrotizing enterocolitis or spontaneous intestinal perforation in preterm low birth weight infants. Cochrane Database Syst Rev 2011;[6]:CD006182. 15. National Institute of Child Health and Human Development (NICHD); National Center for Research Resources (NCRR). Laparotomy vs. Drainage for Infants with Necrotizing Enterocolitis (NEST). In: ClinicalTrials.gov [Internet]. Bethesda (MD): National Library of Medicine (US). 2009- [cited 2013 Jan 22]. Available
8
16. 17. 18. 19. 20.
21.
22. 23. 24. 25.
Hull et al
Surgical Necrotizing Enterocolitis Mortality
from: http://clinicaltrials.gov/show/NCT01029353 NLM Identifier NCT01029353. Accessed November 28, 2013. Vermont Oxford Network Database Manual of Operations for Infants Born in 2006 Vermont Oxford Network; Releases 9. 0e13.2. Horbar JD, Carpenter JH, Badger GJ, et al. Mortality and neonatal morbidity among infants 501 to 1500 grams from 2000 to 2009. Pediatrics 2012;129:1019e1026. Holman RC, Stoll BJ, Curns AT, et al. Necrotising enterocolitis hospitalisations among neonates in the United States. Paediatr Perinat Epidemiol 2006;20:498e506. Lucas A, Cole TJ. Breast milk and neonatal necrotising enterocolitis. Lancet 1990;336:1519e1523. Luig M, Lui K. Epidemiology of necrotizing enterocolitise Part II: Risks and susceptibility of premature infants during the surfactant era: a regional study. J Paediatr Child Health 2005;41:174e179. Moro M, Perez-Rodriguez J, Figueras-Aloy J, et al. Predischarge morbidities in extremely and very low-birth-weight infants in Spanish neonatal units. Am J Perinatol 2009;26: 335e343. Narang A, Rao R, Bhakoo ON. Neonatal necrotizing enterocolitis an epidemiological study. Indian Pediatr 1993;30: 1207e1214. Palmer SR, Biffin A, Gamsu HR. Outcome of neonatal necrotising enterocolitis: results of the BAPM/CDSC surveillance study, 1981-84. Arch Dis Child 1989;64:388e394. Pokorny WJ, Garcia-Prats JA, Barry YN. Necrotizing enterocolitis: incidence, operative care, and outcome. J Pediatr Surg 1986;21:1149e1154. Walther FJ, Verloove-Vanhorick SP, Brand R, et al. A prospective survey of necrotising enterocolitis in very low birthweight infants. Paediatr Perinat Epidemiol 1989;3: 53e61.
J Am Coll Surg
26. Wilson R, Kanto WP Jr, McCarthy BJ, et al. Epidemiologic characteristics of necrotizing enterocolitis: a population-based study. Am J Epidemiol 1981;114:880e887. 27. Abdullah F, Zhang Y, Camp M, et al. Necrotizing enterocolitis in 20,822 infants: analysis of medical and surgical treatments. Clin Pediatr 2010;49:166e171. 28. Bell MJ, Ternberg JL, Feigin RD, et al. Neonatal necrotizing enterocolitis. Therapeutic decisions based upon clinical staging. Ann Surg 1978;187:1e7. 29. Moss RL, Kalish LA, Duggan C, et al. Clinical parameters do not adequately predict outcome in necrotizing enterocolitis: a multi-institutional study. J Perinatol 2008;28:665e674. 30. Andorsky DJ, Lund DP, Lillehei CW, et al. Nutritional and other postoperative management of neonates with short bowel syndrome correlates with clinical outcomes. J Pediatr 2001; 139:27e33. 31. Blakely ML, Lally KP, McDonald S, et al. Postoperative outcomes of extremely low birth-weight infants with necrotizing enterocolitis or isolated intestinal perforation: a prospective cohort study by the NICHD Neonatal Research Network. Ann Surg 2005;241:984e989; discussion 989e994. 32. Blakely ML, Tyson JE, Lally KP, et al. Laparotomy versus peritoneal drainage for necrotizing enterocolitis or isolated intestinal perforation in extremely low birth weight infants: outcomes through 18 months adjusted age. Pediatrics 2006; 117:e680ee687. 33. Dimmitt RA, Meier AH, Skarsgard ED, et al. Salvage laparotomy for failure of peritoneal drainage in necrotizing enterocolitis in infants with extremely low birth weight. J Pediatr Surg 2000;35:856e859. 34. Ein SH, Shandling B, Wesson D, et al. A 13-year experience with peritoneal drainage under local anesthesia for necrotizing enterocolitis perforation. J Pediatr Surg 1990;25:1034e1036; discussion 1036e1037.
BACKGROUND:
is inversely related to birth weight, with the great preponderance of affected patients having very low birth weight (VLBW).2,4-6 It may be inferred that VLBW patients with NEC receiving an operation (surgical NEC) harbor a more severe form of the disease than those treated without surgery (medical NEC).7 However, there are no large-scale cohort studies directly evaluating the mortality of surgical and medical NEC by birth weight categories. Such data are important to surgeons because they potentially afford a more accurate determination of prognosis and may be used as a basis for future quality improvement efforts. Various strategies are available to treat surgical NEC: laparotomy and primary peritoneal drainage (PPD). Primary peritoneal drainage was initially proposed as a
Necrotizing enterocolitis (NEC) is the most common gastrointestinal emergency of newborns1 and remains a major cause of mortality.2,3 The incidence of NEC Disclosure information: Nothing to disclose. Drs Hull and Fisher contributed equally to this work. Presented at the New England Surgical Society 94th Annual Meeting, Hartford, CT, September 2013. Received September 27, 2013; Revised November 14, 2013; Accepted November 18, 2013. From the Department of Surgery, Center for Advanced Intestinal Rehabilitation, Boston Children’s Hospital and Harvard Medical School, Boston, MA (Hull, Fisher, Gutierrez, Jones, Kang, Zurakowski, Modi, Jaksic) and the Vermont Oxford Network, Burlington, VT (Kenny, Horbar). Correspondence address: Tom Jaksic, MD, PhD, FACS, 300 Longwood Ave, Fegan 3, Boston, MA 02115. email: [email protected]
ª 2013 by the American College of Surgeons Published by Elsevier Inc.
1
ISSN 1072-7515/13/$36.00 http://dx.doi.org/10.1016/j.jamcollsurg.2013.11.015
2
Hull et al
Surgical Necrotizing Enterocolitis Mortality
Abbreviations and Acronyms
NEC NICU PPD SIP VLBW VON
¼ ¼ ¼ ¼ ¼ ¼
necrotizing enterocolitis neonatal ICU primary peritoneal drainage spontaneous intestinal perforation very low birth weight Vermont Oxford Network
bridge to laparotomy in 19758 and subsequently has gained popularity as a potentially definitive therapy.9-11 Two randomized controlled trials comparing laparotomy to PPD for NEC demonstrated no significant difference with respect to mortality.12,13 Unfortunately, due to the small size of these trials, a Cochrane Review determined that alone or in combination, these investigations cannot be used to guide therapy.14 A third randomized control trial is currently enrolling patients, with a specific emphasis on evaluating neurologic outcomes.15 Given this incomplete evidence base, it is of interest to determine with what frequency surgeons choose PPD and laparotomy in the treatment of surgical NEC. This study used a large, prospectively collected cohort of VLBW neonates to establish birth weight-based mortality benchmarks for surgical NEC. These mortality rates are compared with those from a contemporaneous group of patients with rigorously defined medical NEC. The frequency of PPD and laparotomy use and the mortalities associated with these therapies, are also reported.
METHODS The Vermont Oxford Network (VON) is a nonprofit voluntary collaboration of health care professionals dedicated to improving the quality and safety of medical and surgical care for newborn infants and their families. The VON prospectively collects data on infants of birth weight 401 to 1,500 g, who are born at participating institutions or who are transferred to such an institution within 28 days of birth. These data are accrued until neonates are discharged from the hospital, die, or reach 1 year of age in the hospital. Death after transfer to a non-VON center is or readmission is tracked and included. Discharge home without readmission is considered survival. Data are collected by local staff using uniform definitions and then submitted electronically or on paper forms to the VON central office. Records are subjected to automated checks for quality and completeness and returned for correction if needed. This study was performed as part of an ongoing collaboration between VON and a group of pediatric surgeons. Compilation and quality review of data were performed
J Am Coll Surg
with surgical input, ensuring that data fields were completed and that any ambiguities in the entries were clarified. For example, surgeons evaluated any unclear surgical coding by hand. No protected health care information was collected. The VON is approved by the University of Vermont Institutional Review Board (#04-370) and exempted from review at Boston Children’s Hospital. Among the 655 United States VON centers, survey data indicate that 37% of centers were classified as having a type A NICU (restriction on assisted ventilation or only perform minor surgery). Forty-seven percent had a type B NICU (no restriction on assisted ventilation and perform major surgery, ie, repair of tracheoesophageal fistula/ esophageal atresia, or meningomyelocoele). The remaining 16% of centers had a type C NICU (no restriction on ventilation and perform cardiac surgery requiring bypass for neonates). For this cohort study, data were prospectively collected in the VON database from United States centers between January 2006 and December 2010 on newborns weighing between 401 and 1,500 g. Neonates with major birth defects, as listed in the VON Manual of Operations16 (anencephaly, holoprosencephaly, and major cardiac defects), and those with a length of stay of 3 or fewer days were excluded. Per VON Manual definition, NEC was diagnosed either by direct observation of intestine at operation or pathologic examination or by using a set of strict clinical criteria. A clinical diagnosis of NEC was made based on at least 1 physical finding (bilious gastric aspirate or emesis, abdominal distention, or occult/gross blood in the stool in the absence of anal fissures) and at least 1 radiographic finding (pneumatosis intestinalis, hepatobiliary gas, or pneumoperitoneum). Severity of NEC was not specifically coded in the dataset and the diagnosis of NEC can be coded only once. The VON Manual definition also distinguishes spontaneous intestinal perforation (SIP) from NEC if laparotomy is performed. Spontaneous intestinal perforation is defined as a single focal defect without other significant bowel pathology seen at laparotomy. Because SIP and NEC cannot be incontrovertibly differentiated without a laparotomy, the PPD group necessarily contains patients with both diagnoses. So for veracity of comparison, SIP and operative NEC are grouped together and referred to as “surgical NEC” in this article. Primary peritoneal drainage and laparotomy were identified by specific procedure codes. Patients with codes entered for PPD and laparotomy were considered to have undergone both procedures. Because the dates of procedures were not included, the order could not be specified.
Vol.
-,
No.
-, -
2013
Statistical analyses were conducted using SAS version 9.3 (SAS Institute). The primary outcome was death before 1 year of age or hospital discharge. Multivariable logistic regression modeling was performed. Multivariate logistic regression modeling was performed with adjustments for clustering of infants within hospitals (generalized estimating equations were used to account for correlation between infants at a given hospital). Model covariates included gestational age, small for gestational age, maternal race, sex, multiple births, Apgar score at 1 minute, birth location, and vaginal delivery (because these are independent predictors of mortality on multivariable analysis). Additional multivariable analysis was performed to assess factors associated with treatment by PPD vs laparotomy. Values of p < 0.05 were considered significant. Categorical measures were compared using z-tests and continuous measures were assessed by unpaired t-test.
RESULTS There were 215,057 very low birth weight neonates identified from 655 VON centers in the United States (Fig. 1). A total of 26,354 were excluded: 10,821 with a major congenital malformation or unknown congenital
Hull et al
Surgical Necrotizing Enterocolitis Mortality
3
malformation status, 15,493 with length of stay 3 or fewer days or unknown, and 40 with unknown NEC status. Of the remaining 188,703 patients, 17,159 were diagnosed with NEC (incidence 9%). Demographic characteristics are listed in Table 1. There were small but significant differences between the medical and surgical NEC groups. Infants with surgical NEC had lower birth weights, younger gestational age, and lower Apgar scores. The overall mortality for all patients with NEC was 28%. Mortality declined with birth weight. There were 8,221 patients with medical NEC and 8,935 with surgical NEC. Medical NEC mortality was 21% overall, with significantly lower mortality in neonates of larger birth weight. Surgical NEC mortality was 35% overall (higher than medical NEC, p < 0.0001) and, unlike that of medical NEC, it did not consistently improve with larger birth weight (Fig. 2). The surgical group was subdivided into those undergoing laparotomy (n ¼ 6,126, 69%) and those receiving PPD (n ¼ 2,804, 31%). The PPD group was also divided into those with PPD alone (n ¼ 1,521) and those who underwent PPD and laparotomy (n ¼ 1,283). In other words, 46% of the PPD group also had a laparotomy. Mortality in the laparotomy only group was 31% overall
Figure 1. Study design schema. LOS, length of stay; NEC, necrotizing enterocolitis; PPD, primary peritoneal drain; VLBW, very low birth weight.
4
Hull et al
Table 1.
Demographic and Obstetric Characteristics
J Am Coll Surg
Surgical Necrotizing Enterocolitis Mortality
Characteristic
Male sex, n (%) Vaginal delivery Multiple birth Small for gestational age Birth location (inborn) Antenatal steroids Birth weight, g, mean SD Gestational age, wk, mean SD Apgar 1 min, mean SD Apgar 5 min, mean SD
Medical NEC (n ¼ 8,221)
Surgical NEC (n ¼ 8,935)
p Value
4,508 (54.8) 2,500 (30.4) 1,967 (23.9) 1,117 (13.6) 6,518 (79.3) 6,360 (79.0) 938 273 26.9 2.4 5.0 2.5 7.1 1.9
5,220 (58.4) 2,955 (33.1) 2,237 (25.0) 952 (10.7) 5,802 (64.9) 6,741 (76.3) 841 248 25.9 2.3 4.5 2.4 6.7 2.0
<0.001 <0.001 0.091 <0.001 <0.001 0.003 <0.001 <0.001 <0.001 <0.001
Statistical comparisons are made between medical and surgical necrotizing enterocolitis (NEC).
and demonstrated a plateau, mirroring the overall surgical group (401 to 500 g, 51%; 501 to 750 g, 36%; 751 to 1,000 g, 29%; 1,001 to 1,250 g, 25%; 1,251 to 1,500 g, 26%). Mortality in the PPD only group was 50% (higher than laparotomy only, p < 0.0001), and this mortality tended to remain high across the birth weight categories (401 to 500 g, 57%; 501 to 750 g, 54%; 751 to 1,000 g, 43%; 1,001 to 1,250 g, 43%; 1,251 to 1,500 g, 52%). Those who underwent both PPD and laparotomy had an overall mortality of 34% (similar to the laparotomy only group, p ¼ 0.42) and it declined with birth weight: 401 to 500 g, 50%; 501 to 750 g, 36%;
Figure 2. Mortality in surgical necrotizing enterocolitis (NEC) vs medical NEC by birth weight category. Odds ratios for mortality for surgical vs medical NEC are shown with 95% confidence intervals. *p < 0.0001. Black bar, surgical NEC; gray bar, medical NEC. Mortality in medical NEC: 401 to 500 g, 52% (n ¼ 211); 501 to 750 g, 36% (n ¼ 2,267); 751 to 1,000 g, 21% (n ¼ 2,501); 1,001 to 1,250 g, 11% (n ¼ 1,886); 1,251 to 1,500 g, 6% (n ¼ 1,356). Mortality in surgical NEC: 401 to 500 g, 52% (n ¼ 345); 501 to 750 g, 40% (n ¼ 3,608); 751 to1,000 g, 32% (n ¼ 2,823); 1,001 to 1,250 g, 27% (n ¼ 1,391); 1,251 to 1,500 g, 28% (n ¼ 768).
751 to 1,000 g, 30%; 1,001 to 1,250 g, 27%; 1,251 to 1,500 g, 33%. A comparison of mortality by surgical treatment type is found in Figure 3. Operative intervention rather than medical management was more common among lower birth weight infants. Figure 4 reflects results reported from categorical analysis, with the 1,251 to 1,500 g group as a reference. Furthermore, a clustered logistic regression was performed using previously listed covariates and birth weight as continuous measures. For each 100-g increase in birth weight, the odds of surgery dropped by 5%. On multivariable analysis, antenatal steroid exposure was associated with surgical intervention (odds ratio 1.17, 95% CI 1.06 to 1.29, p < 0.01). Cardiac compressions, endotracheal ventilation, use of epinephrine, and early bacterial sepsis were not associated with surgical intervention. Multivariable analysis was also performed to assess factors associated with the use of PPD as opposed to laparotomy. Those with lower birth weight were more likely to receive a drain when birth weight was evaluated as a continuous variable. When analyzed by category, the percentage
Figure 3. Mortality by surgical intervention. *Statistical difference from both other groups. Laparotomy only vs laparotomy and primary peritoneal drain (PPD), p ¼ 0.42. Laparotomy only vs PPD only, p < 0.0001. Laparotomy and PPD vs PPD only, p < 0.0001.
Vol.
-,
No.
-, -
2013
Figure 4. Percent of necrotizing enterocolitis patients treated surgically (laparotomy and/or primary peritoneal drain) by birth weight. *p < 0.05 when compared with 1,251 to 1,500 g group as a reference.
undergoing PPD decreased with each larger weight category (Fig. 5). Further, cardiac compressions or endotracheal ventilation at initial resuscitation were associated with PPD (odds ratio 1.17, 95% CI 1.00 to 1.36, p ¼ 0.05, and 1.29, CI 1.04 to 1.60, p ¼ 0.02, respectively). Use of epinephrine, early bacterial sepsis, and antenatal steroids were not associated with PPD.
DISCUSSION More than 80% of VLBW neonates born in the United States during a 5-year period were analyzed using a prospectively collected VON database with embedded surgical codes.17 Of 188,703 eligible patients, 17,159 were
Figure 5. Percent of surgical necrotizing enterocolitis patients treated with peritoneal drainage. Includes both “peritoneal drainage only” and “peritoneal drainage and laparotomy” groups. *p < 0.05 when compared with 1,251 to 1,500 g group as the reference.
Hull et al
Surgical Necrotizing Enterocolitis Mortality
5
diagnosed with NEC, for an incidence of 9% (Fig. 1). This yielded the largest VLBW NEC study cohort available to date.2,9,18 Generally, NEC mortality was inversely related to birth weight, with more than half of the patients undergoing surgical intervention for NEC (n ¼ 8,935, 52%). The mortality for surgical NEC was significantly higher than for medical NEC overall and for each birth weight category in excess of 750 g. There is great variation in the reported mortality associated with NEC. A recent review summarized studies available since 19779 with an overall mortality ranging from 12% to 39%.7,18-26 When neonates weighing less than 1,500 g were considered, using the 2000 Kids’ Inpatient Database, an overall mortality of 20% was calculated.18 Additional discrepancies in NEC mortality become evident when higher birth weight infants are separated from lower birth weight neonates (6% to 63%).18 This investigation demonstrated an overall NEC mortality of 28%, which fell within the broad range of these other reports. However, given its large sample size, a more detailed stratification of mortality was possible by birth weight categories, whether surgical therapy was used, and the type of operation performed. Medical NEC mortality in this study (21%) was higher than previously reported. Previous investigations used retrospective, administrative, or survey data and the definition of NEC was not uniform.7,9,27 The largest of these studies, combining data from both the National Inpatient Sample and Kids Inpatient Database, showed a mortality of 6.8% for medical NEC.27 However, this post hoc review of administrative data was not restricted to VLBW neonates and was defined NEC by ICD-9 code only, potentially resulting in up to 45% of the entire sample having Bell’s stage I disease.9 Bell’s stage I disease is a very low mortality group that encompasses all neonates with abdominal distension.9,28 In contradistinction, VON uses a more rigorous definition of NEC that includes only neonates with clearly established disease of Bell’s stage II and higher. Additionally, the VON’s capacity for long-term follow-up and limitation of the cohort to VLBW (birth weight less than 1,500 g) neonates also contributed to a higher and likely more accurate mortality estimate for VLBW neonates. In fact, an earlier surgical-VON collaborative investigation found that the baseline mortality for VLBW infants without a diagnosis of NEC was 10%, already higher than the quoted rates for medical NEC mortality in other studies.2 Although specific data are not available, it is possible that some of the medical NEC patients had severe disease or cormorbidities and were treated palliatively, which could have contributed to the higher overall mortality.
6
Hull et al
Surgical Necrotizing Enterocolitis Mortality
Patients receiving operations for NEC are commonly considered to have a greater burden of illness than those managed medically, though specific risk factors for progression of NEC have been elusive.29 The VON cohort in this study revealed an increased prevalence of disease severity markers in surgical NEC compared with medical NEC (lower birth weight, younger gestational age, lower Apgar scores, and male sex, all with p < 0.001). Even after accounting for these risk factors, surgical NEC was still independently predictive of death and showed a significantly higher mortality than medical NEC (35% vs 21%, p < 0.005). Further, in surgical NEC, unlike medical NEC, the reduction in mortality seen with greater birth weight was significantly blunted. This resulted in a plateau mortality of approximately 30% even for VLBW patients with birth weights above 750 g (Fig. 2). At the largest birth weight category (1,251 to 1,500 g), the difference in mortality was the most striking: 27% for surgical NEC as opposed to 6% for medical NEC (p < 0.0001). These birth weight- and treatment-based benchmark data may be used to more accurately apprise caregivers and families of ultimate prognosis and to facilitate valid comparisons of NEC mortality between centers. The relative frequency of surgical to medical NEC decreased progressively with greater birth weight (Fig. 4). Although birth weight is the single best prognosticator of mortality in the VON database (perhaps because of its inherent objectivity) it correlates well with estimated gestational age (in this dataset r ¼ 0.77). Therefore, it may be postulated that larger, more mature infants have an improved ability to respond to the pathologic stress of NEC and as a consequence, have a reduced requirement for surgery. Moreover, it was observed that although laparotomy was used more frequently than PPD across all birth weight categories, this predisposition was accentuated in larger neonates (Fig. 5). At least in part, this may have been because PPD, often done at bedside under local anesthesia, was more likely to be undertaken in those neonates with a perceived higher operative risk (ie, those with the lowest birth weights). Recognizing that the surgeons’ specific decisionmaking process is unknown and no cause and effect relationship between treatment selection and outcome can be concluded, this study nonetheless describes current practice regarding the use of PPD and laparotomy in the United States. Overall, there was a strong preference noted for laparotomy over PPD in neonates with surgical NEC (69% vs 31%, p < 0.001). Such a finding may extend beyond a bias regarding improved immediate outcomes after laparotomy. Necrotizing enterocolitis is a disease that can affect variable quantities of bowel, and the extent of small intestinal loss is a strong determinant of
J Am Coll Surg
long-term outcomes.30 So it is possible that surgeons chose laparotomy rather than PPD because it gave them a more complete assessment of viable residual intestinal length and therefore of prognosis. It is clear that surgeons in the United States are not in equipoise regarding the 2 operative treatment options and further research is warranted to ascertain the precise reasons for this observation. Necrotizing enterocolitis patients undergoing PPD alone (n ¼ 1,521) had a higher associated mortality than those receiving laparotomy alone (n ¼ 6,131) (50% vs 31%, p < 0.0001). The same trend was seen in a previous National Institute of Child Health and Human Development Neonatal Research Network cohort limited to neonates with birth weights less than 1,000 g.31,32 In this small but carefully conducted study, the mortality associated with PPD was 54% (43 of 80) vs 43% (33 of 76) for laparotomy. The lower mortality for laparotomy observed in the VON data may be explained by the inclusion of larger infants with birth weights up to 1,500 g. There is both a progressively greater use of laparotomy and significantly lower NEC mortality in these larger neonates (Figs. 2 and 5). Although risk stratification, as outlined in the Methods section, was used to make statistical comparisons between the PPD and laparotomy treatment groups, great caution must be exercised in interpreting these data. Post hoc analysis revealed that the PPD patients, as a whole, were likely more severely ill than the laparotomy alone group. Specifically, neonates who underwent PPD had a higher likelihood of receiving both cardiac compressions and endotracheal ventilation at birth (p < 0.05). Therefore, the documented association between increased mortality and PPD alone may still be attributed to the severity of underlying illness rather than a treatment effect. It is of considerable interest that neonates who had PPD and a laparotomy (46% of PPD patients) demonstrated a mortality that quite closely approximated that of the laparotomy only group (p ¼ 0.42, Fig. 3). Classically, PPD is used as a bridge to laparotomy,8 yet in this study, 27% of neonates undergoing PPD survived without further surgery. These data are congruent with those from other studies and suggest that patients receiving PPD alone represent a dichotomous group: those who do not survive to laparotomy and a subset who show clinical improvement and never require laparotomy.10,33,34 It may be surmised that the latter cohort is comprised of relatively physiologically stable neonates who have minimal NEC or spontaneous intestinal perforation.10 Future trials of PPD vs laparotomy may need to focus on these 2 very different VLBW groups: the very ill and the clinically stable. It is perhaps in these extremes where surgeons will have sufficient equipoise to conduct
Vol.
-,
No.
-, -
2013
randomized controlled trials of sufficient size, and the design and primary end points of these studies may need to be quite different. Although the VON database is a powerful tool to assess the mortality and management of NEC, there are important limitations in its application. Because VON participation is voluntary, centers are not chosen at random or by criteria, so the resulting cohort is not strictly a population-based sample. However, this is mitigated by the fact that greater than 80% of VLBW neonates born in the United States are in the VON database and therefore were included for analysis. The VON collects information for each patient at discrete time points: hospital day 28, discharge or death, and transfer or readmission. This means that the specific codes entered (such as operative procedures) are not listed by date or necessarily in the order that they were actually performed. Further, although the operations undertaken are known, the specific indications and underlying decision-making remain unknown. Because this is an observational study across many centers, treatment of neonates (including feeding strategies, radiologic surveillance, and antibiotic therapy) was not standardized. Finally, it should also be mentioned that the VON data are an amalgam of results obtained from hospitals with heterogeneous resources, as outlined in the Methods section.
CONCLUSIONS In this very large United States cohort, 52% of VLBW neonates with NEC underwent operative intervention. Surgical NEC was associated with a significantly higher mortality than medical NEC both in aggregate and for all birth weight categories greater than 750 g. Furthermore, surgical NEC, unlike medical NEC, exhibited a plateau in mortality of approximately 30%, even in larger VLBW neonates. During the period of this study there was a marked preference exhibited for laparotomy rather than PPD in the surgical management of NEC. Neonates receiving laparotomy alone or PPD with laparotomy had similar mortalities. The highest mortality evident was in patients undergoing PPD alone, but this treatment may have been applied to neonates with the highest degree of overall illness. This national prospective cohort analysis provides surgeons with accurate mortality data for VLBW neonates with NEC and will be useful in the design and implementation of future randomized controlled trials. Author Contributions Study conception and design: Hull, Fisher, Gutierrez, Jones, Kang, Kenny, Zurakowski, Modi, Horbar, Jaksic
Hull et al
Surgical Necrotizing Enterocolitis Mortality
7
Acquisition of data: Kenny, Horbar Analysis and interpretation of data: Hull, Fisher, Gutierrez, Jones, Kang, Kenny, Zurakowski, Modi, Horbar, Jaksic Drafting of manuscript: Hull, Fisher, Gutierrez, Jones, Kang, Kenny, Zurakowski, Modi, Horbar, Jaksic Critical revision: Hull, Fisher, Gutierrez, Jones, Kang, Kenny, Zurakowski, Modi, Horbar, Jaksic REFERENCES 1. Lin PW, Stoll BJ. Necrotising enterocolitis. Lancet 2006;368: 1271e1283. 2. Fitzgibbons SC, Ching Y, Yu D, et al. Mortality of necrotizing enterocolitis expressed by birth weight categories. J Pediatr Surg 2009;44:1072e1075; discussion 1075e1076. 3. Neu J, Walker WA. Necrotizing enterocolitis. N Engl J Med 2011;364:255e264. 4. Cikrit D, Mastandrea J, West KW, et al. Necrotizing enterocolitis: factors affecting mortality in 101 surgical cases. Surgery 1984;96:648e655. 5. Llanos AR, Moss ME, Pinzon MC, et al. Epidemiology of neonatal necrotising enterocolitis: a population-based study. Paediatr Perinat Epidemiol 2002;16:342e349. 6. Ricketts RR, Jerles ML. Neonatal necrotizing enterocolitis: experience with 100 consecutive surgical patients. World J Surg 1990;14:600e605. 7. Guthrie SO, Gordon PV, Thomas V, et al. Necrotizing enterocolitis among neonates in the United States. J Perinatol 2003;23:278e285. 8. Ein SH, Marshall DG, Girvan D. Peritoneal drainage under local anesthesia for perforations from necrotizing enterocolitis. J Pediatr Surg 1977;12:963e967. 9. Rees CM, Eaton S, Pierro A. National prospective surveillance study of necrotizing enterocolitis in neonatal intensive care units. J Pediatr Surg 2010;45:1391e1397. 10. Cass DL, Brandt ML, Patel DL, et al. Peritoneal drainage as definitive treatment for neonates with isolated intestinal perforation. J Pediatr Surg 2000;35:1531e1536. 11. Lessin MS, Luks FI, Wesselhoeft CW Jr, et al. Peritoneal drainage as definitive treatment for intestinal perforation in infants with extremely low birth weight (<750 g). J Pediatr Surg 1998;33:370e372. 12. Moss RL, Dimmitt RA, Barnhart DC, et al. Laparotomy versus peritoneal drainage for necrotizing enterocolitis and perforation. N Engl J Med 2006;354:2225e2234. 13. Rees CM, Eaton S, Kiely EM, et al. Peritoneal drainage or laparotomy for neonatal bowel perforation? A randomized controlled trial. Ann Surg 2008;248:44e51. 14. Rao SC, Basani L, Simmer K, et al. Peritoneal drainage versus laparotomy as initial surgical treatment for perforated necrotizing enterocolitis or spontaneous intestinal perforation in preterm low birth weight infants. Cochrane Database Syst Rev 2011;[6]:CD006182. 15. National Institute of Child Health and Human Development (NICHD); National Center for Research Resources (NCRR). Laparotomy vs. Drainage for Infants with Necrotizing Enterocolitis (NEST). In: ClinicalTrials.gov [Internet]. Bethesda (MD): National Library of Medicine (US). 2009- [cited 2013 Jan 22]. Available
8
16. 17. 18. 19. 20.
21.
22. 23. 24. 25.
Hull et al
Surgical Necrotizing Enterocolitis Mortality
from: http://clinicaltrials.gov/show/NCT01029353 NLM Identifier NCT01029353. Accessed November 28, 2013. Vermont Oxford Network Database Manual of Operations for Infants Born in 2006 Vermont Oxford Network; Releases 9. 0e13.2. Horbar JD, Carpenter JH, Badger GJ, et al. Mortality and neonatal morbidity among infants 501 to 1500 grams from 2000 to 2009. Pediatrics 2012;129:1019e1026. Holman RC, Stoll BJ, Curns AT, et al. Necrotising enterocolitis hospitalisations among neonates in the United States. Paediatr Perinat Epidemiol 2006;20:498e506. Lucas A, Cole TJ. Breast milk and neonatal necrotising enterocolitis. Lancet 1990;336:1519e1523. Luig M, Lui K. Epidemiology of necrotizing enterocolitise Part II: Risks and susceptibility of premature infants during the surfactant era: a regional study. J Paediatr Child Health 2005;41:174e179. Moro M, Perez-Rodriguez J, Figueras-Aloy J, et al. Predischarge morbidities in extremely and very low-birth-weight infants in Spanish neonatal units. Am J Perinatol 2009;26: 335e343. Narang A, Rao R, Bhakoo ON. Neonatal necrotizing enterocolitis an epidemiological study. Indian Pediatr 1993;30: 1207e1214. Palmer SR, Biffin A, Gamsu HR. Outcome of neonatal necrotising enterocolitis: results of the BAPM/CDSC surveillance study, 1981-84. Arch Dis Child 1989;64:388e394. Pokorny WJ, Garcia-Prats JA, Barry YN. Necrotizing enterocolitis: incidence, operative care, and outcome. J Pediatr Surg 1986;21:1149e1154. Walther FJ, Verloove-Vanhorick SP, Brand R, et al. A prospective survey of necrotising enterocolitis in very low birthweight infants. Paediatr Perinat Epidemiol 1989;3: 53e61.
J Am Coll Surg
26. Wilson R, Kanto WP Jr, McCarthy BJ, et al. Epidemiologic characteristics of necrotizing enterocolitis: a population-based study. Am J Epidemiol 1981;114:880e887. 27. Abdullah F, Zhang Y, Camp M, et al. Necrotizing enterocolitis in 20,822 infants: analysis of medical and surgical treatments. Clin Pediatr 2010;49:166e171. 28. Bell MJ, Ternberg JL, Feigin RD, et al. Neonatal necrotizing enterocolitis. Therapeutic decisions based upon clinical staging. Ann Surg 1978;187:1e7. 29. Moss RL, Kalish LA, Duggan C, et al. Clinical parameters do not adequately predict outcome in necrotizing enterocolitis: a multi-institutional study. J Perinatol 2008;28:665e674. 30. Andorsky DJ, Lund DP, Lillehei CW, et al. Nutritional and other postoperative management of neonates with short bowel syndrome correlates with clinical outcomes. J Pediatr 2001; 139:27e33. 31. Blakely ML, Lally KP, McDonald S, et al. Postoperative outcomes of extremely low birth-weight infants with necrotizing enterocolitis or isolated intestinal perforation: a prospective cohort study by the NICHD Neonatal Research Network. Ann Surg 2005;241:984e989; discussion 989e994. 32. Blakely ML, Tyson JE, Lally KP, et al. Laparotomy versus peritoneal drainage for necrotizing enterocolitis or isolated intestinal perforation in extremely low birth weight infants: outcomes through 18 months adjusted age. Pediatrics 2006; 117:e680ee687. 33. Dimmitt RA, Meier AH, Skarsgard ED, et al. Salvage laparotomy for failure of peritoneal drainage in necrotizing enterocolitis in infants with extremely low birth weight. J Pediatr Surg 2000;35:856e859. 34. Ein SH, Shandling B, Wesson D, et al. A 13-year experience with peritoneal drainage under local anesthesia for necrotizing enterocolitis perforation. J Pediatr Surg 1990;25:1034e1036; discussion 1036e1037.