Acta Pædiatrica ISSN 0803-5253

REGULAR ARTICLE

Ultrasonography and radiography findings predicted the need for surgery

in patients with necrotising enterocolitis without pneumoperitoneum
Yu He1,2,3,4, Ying Zhong5, Jialin Yu (yujialin486@126.com)1,2,3,4, Chen Cheng1,2,3,4, Zhengli Wang1,2,3,4, Luquan Li1,2,3,4
1.Department of Neonatology, Children’s Hospital, Chongqing Medical University, Chongqing, China
2.Chongqing Key Laboratory of Paediatrics, Chongqing, China
3.Ministry of Education Key Laboratory of Child Development and Disorder, Children’s Hospital, Chongqing Medical University, Chongqing, China
4.China International Science and Technology Cooperation Base of Child Development and Critical Disorders, Chongqing, China
5.Department of Neonatology, The Children’s Hospital Zhejiang University School of Medicine, Chongqing, China

Keywords ABSTRACT
Imaging reports, Multivariate analysis, Necrotising Aim: The timing of surgical intervention in patients with necrotising enterocolitis (NEC)
enterocolitis, Risk factors, Surgery
without pneumoperitoneum remains unclear. This study aimed to identify the predictors
Correspondence associated with surgical intervention in such patients and to assess how effective imaging
Jialin Yu, Department of Neonatology, Children’s
Hospital, Chongqing Medical University, 136, reports were as an aid to surgical decision-making.
Zhongshan Road, Yuzhong District, Chongqing Methods: We collected clinical data, laboratory investigations and imaging findings on NEC
400014, China.
patients without pneumoperitoneum. A critical imaging report was defined as persistent
Tel: +86 023 6363 5567 |
Fax: +86 023 6363 5567 | dilation of bowel loops and evidence of portal venous gas on radiography and thickening of
Email: yujialin486@126.com the bowel wall, absent peristalsis and evidence of echogenic-free fluid on sonography.
Received Independent predictors of surgical NEC were identified using multivariate logistic regression
14 June 2015; revised 2 November 2015; analysis. Sensitivity and specificity analyses were performed for the imaging findings, and
accepted 11 December 2015.
receiver operator characteristic curve analysis was used to evaluate the predictive accuracy.
DOI:10.1111/apa.13315 Results: Of the 238 neonates studied, 54 (22.69%) required surgical intervention. The
multivariate logistic regression analysis showed that abdominal erythema, C-reactive
protein levels and the critical imaging report were independent predictors of the need for
surgical intervention. The critical imaging report was the most powerful predictor of surgical
NEC.
Conclusion: Both ultrasonography and radiography findings proved helpful in predicting
the need for surgery in NEC without pneumoperitoneum.

INTRODUCTION toneum. Early surgery, if indicated, is a potential life-saving

Necrotising enterocolitis (NEC) is a life-threatening intesti-
nal disease that occurs primarily in premature infants. Over
the last few decades, NEC has emerged as one of the most
common surgical emergencies in neonates, with high
associated morbidity and mortality (1). Advances in neona-
tal care, particularly resuscitation and ventilation support
technology, have resulted in increased survival rates among
premature infants and a concomitant increase in the
incidence of NEC (2). Up to 49% of neonates with NEC
may require surgical intervention (3), and in the absence of
definitive indications for surgical intervention, the care of
infants with NEC has been a contentious issue for paedi-
atric surgeons. This is a particular issue in the case of NEC
patients without radiographic evidence of pneumoperi-
Abbreviations
AUC, The area under the curve; CI, Confidence interval; IQR,
Interquartile range; NEC, Necrotising enterocolitis; OR, Odds
ratio; ROC, Receiver operating characteristic curve.
intervention in these patients and may improve their
prognosis (4).
The current and absolute indication for surgery in NEC
patients is evidence of perforation on the abdominal
radiograph (5), but more than 50% of infants who receive
Key notes
 The optimal timing for surgical intervention in patients
with necrotising enterocolitis (NEC) without pneu-
moperitoneum remains unclear.
 Our study of 238 patients showed that elevated C-
reactive protein, the presence of abdominal erythema
and critical imaging findings correlated with surgical
intervention in such patients, and critical imaging was
the most powerful predictor.
 Radiography and ultrasonography may aid surgical
decision-making in patients with necrotising enterocol-
itis without pneumoperitoneum.

©2015 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2016 105, pp. e151–e155 e151
Surgical prediction for enterocolitis
surgery for NEC or die do not show any sign of perforation
(6). Risk factors that may be associated with poor prognosis
include metabolic acidosis (7), thrombocytopenia (8),
leukopenia, leukocytosis (9), portal venous gas (10), fixed
abdominal mass, persistently dilated loops of bowel or
erythema of the abdominal wall (11). However, whether
these risk factors could help to predict the need for surgical
intervention in patients without pneumoperitoneum is
unknown. A comprehensive multivariate analysis may
help to identify the risk factors for patients who do not
display any evidence of perforation, but do need surgical
intervention.
This study was a retrospective evaluation of 238 infants
with confirmed NEC who had no radiological evidence of
pneumoperitoneum. Patients were categorised into two
groups: a surgical group and a nonsurgical group. We
carried out a comprehensive analysis of factors such as the
patient’s history, clinical presentation, laboratory tests,
imaging findings and a number of special treatment aspects
identify those that were likely to predict surgical interven-
tion among these patients. The predictive values of specific
abdominal imaging findings in identifying surgical NEC
patients were also particularly analysed.
MATERIAL AND METHODS
The Department of Neonatology at the Children’s Hospital
of Chongqing Medical University, Chongqing, China, is a
national clinical specialty department. There are 215 beds,
including a 40-bed NICU and a 175-bed intermediate care
section, which comprise the premature infant ward and the
term infant ward.
Study population
We included neonates who were treated in our hospital
between January 2000 and December 2014 and had
confirmed NEC according to the modified Bell’s criteria.
Surgically identified NEC patients were also enrolled. The
inclusion and exclusion criteria are described in Figure S1.
Patients who did not respond to conservative management
and displayed a deteriorating general condition fulfilled the
criteria for surgical intervention and were assessed by the
paediatric surgeons. The surgical interventions included
laparotomy and peritoneal drainage. The medical manage-
ment for NEC included bowel decompression, discontinu-
ation of enteral feedings and close monitoring of abdominal
imaging reports.
Data collection
We collected data on demographic variables, clinical
characteristics and laboratory investigations, such as white
blood cell counts, platelet counts, C-reactive protein, pH,
base excess, glucose and sodium and potassium levels. The
data also included radiological and ultrasonic evidence of
NEC and treatment that the patient had received. The first
available laboratory test reports after either the clinical or
radiological diagnosis of NEC were included in the
analysis.
e152 ©2015 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2016 105, pp. e151–e155
He et al.
Imaging
During the reference period for this study, abdominal
radiography and grey-colour ultrasonography were rou-
tinely used to monitor NEC patients in our hospital. If a
NEC infant needed an abdominal X-ray, abdominal sonog-
raphy was also performed at the same time. The paediatric
radiologists and sonologists then reported on the imaging
findings. Any imaging reports taken immediately before
surgery were included in this analysis. The analysis also
included any data on the bowel wall, bowel movements,
portal venous gas pneumatosis and intra-abdominal fluid
that were identified by sonography and any evidence of gas
and dilatation of intestinal loops that were identified by
radiography.
Statistical analysis
Before we carried out the statistical analysis, we defined
several characteristics and transformed or interpreted the
data according to the protocols shown in Table S1. These
definitions were either derived from previous studies or the
cut-off points in our study. All analyses were performed by
SAS 9.0 (SAS Institute Inc, Cary, North Carolina, USA).
We checked the normality of the data using the Kol-
mogorov–Smirnov test. Qualitative or categorical variables
are expressed as frequencies and proportions. Proportions
were compared using the chi-square or Fisher’s exact test,
as applicable. Data with non-normal distribution were
expressed as median and interquartile ranges (IQR), and
comparisons between the groups were assessed with the
Mann–Whitney test. Predictors of surgical intervention
were identified using univariate analysis. A multivariate
analysis for independent predictors was carried out using
stepwise logistic regression. Variables were included in
multivariate models if they demonstrated a significant
association with the univariate analyses. Results are pre-
sented as adjusted odds ratios (ORs), regression coeffi-
cients (b) and 95% confidence intervals (95% CIs). All
statistical tests were two-sided and performed at a
significance level of p < 0.05. The discriminative ability
of the risk factors and the comparative performance
between the surgical and nonsurgical groups were assessed
with the receiver operative curve (ROC) and the area
under the curve (AUC) using Medcalc 12.7 (MedCalc Soft-
ware, Ostend, Belgium). We also used sensitivity and
specificity and positive and negative likelihood ratios to
describe the predictive ability of each of the items in the
critical imaging report.
The study protocol was approved by the Institutional
Review Board of the Children’s Hospital, Chongqing
Medical University. As this was a retrospective review, no
parental consent was required.
RESULTS
Surgical strategies and pathologic findings of all 265 NEC
patients are described in Table S2. All NEC patients without
pneumoperitoneum were divided into two groups, with 54
in the surgical group and 184 in the nonsurgical group.
He et al. Surgical prediction for enterocolitis

All the patients’ characteristics, including general demo- on plain abdominal radiographs were associated with the
graphic information, maternal and perinatal history, clinical development of full-thickness necrosis and were a sign of
features, laboratory investigations, imaging investigations, imminent peritonitis (15). We believe that this radiological
comorbidities and special treatment, were fully described sign indicates a high risk of impending perforation. It was
and compared between the two groups (Table S3). Figure 1 interesting to note that finding portal venous gas on
shows the nine factors that were significantly associated radiography, rather than on ultrasonography, was signifi-
with surgical intervention on univariate analysis. cantly associated with surgical intervention, despite the fact
These nine factors were subjected to a stepwise logistic that the latter has been reported to be more sensitive in
regression model for multivariate analysis to identify inde- detecting this (16,17). Other studies have also reported an
pendent predictors for surgery in NEC patients without association between radiographic evidence of portal venous
pneumoperitoneum (Table 1). The likelihood ratio test
demonstrated a good model fit (p < 0.001). Abdominal
erythema, elevated C-reactive protein level and critical
imaging reports were found to be independently associated
with surgical intervention, and the critical imaging report Table 1 Multivariate analysis of predictors of surgical intervention in NEC patients
without pneumoperitoneum
was the most powerful independent predictor.
The AUC of the critical imaging report for predicting Multivariate analysis
surgical intervention among NEC patients without pneu- Adjusted 95% CI Regression
moperitoneum was 0.78, and this shows good predictive Variables OR p Value coefficient
ability. When the two parameters of the critical imaging
reports were present, the sensitivity of the critical imaging Abdominal erythema 7.50 1.54 35.51 0.01 2.02
report was 75.9% and the specificity was 68.5% (Fig. S2). Elevated C-reactive protein 7.57 3.02 18.39 <0.001 2.03
Critical imaging reports 10.04 1.54 36.51 0.03 2.31
We also compared the different parameters of the critical
imaging reports to evaluate the predictive ability of the NEC, Necrotising enterocolitis; CI, Confidence interval; OR, Odds ratio.
abdominal radiography and ultrasonography findings for
surgery. This showed that radiographic evidence of portal
venous gas had the highest specificity and absent peristalsis
on ultrasonography had the highest sensitivity (Table 2). Table 2 Predictive characteristics of critical imaging findings among NEC patients
without pneumoperitoneum who underwent surgical intervention
Positive Negative
DISCUSSION likelihood likelihood
Radiographic evidence has been shown to be critical when Sensitivity Specificity ratio ratio
it comes to evaluating the severity of NEC (12,13). In our
Radiography
study, persistent dilation of the bowel loops and portal
Portal venous gas 9.3% 97.3% 3.4 0.93
venous gas was predictors of surgical intervention and in Persistent dilation of 37.0% 78.8% 1.7 0.80
more than 80% of the patients with persistent dilation of the bowel loops
bowel loops, surgical intervention was eventually required. Ultrasonography
In the ten-scale assessment criteria developed by Courtney Echogenic fluid or/ 22.2% 88.6% 1.9 0.88
for standardising radiographic reporting in NEC patients, and focal fluid
fixed or persistent dilation of the bowel loops (14) was Parietal thickening 57.4% 77.7% 2.6 0.54
found to be highly associated with eventual surgical inter- Absent peristalsis 88.9% 45.1% 1.6 0.25
vention in suspected NEC, which is consistent with the our NEC, Necrotising enterocolitis.
findings (6). In another study, identifying fixed bowel loops

Figure 1 Univariate analysis: risk factors for surgery in NEC patients without pneumoperitoneum. Odds ratio and 95% confidence interval of various risk factors for
surgery in NEC patients without pneumoperitoneum on univariate analysis. Critical imaging reports: persistent dilation of bowel loops, portal venous gas on radiography
and bowel wall thickening, absent peristalsis, echoic free fluid or focal fluid on ultrasonography. CRP: C reactive protein.

©2015 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2016 105, pp. e151–e155 e153
Surgical prediction for enterocolitis
gas and a higher mortality rate or a high rate of near total
intestinal involvement (6,10,18,19).
Ultrasonography provided direct images of the bowel and
could detect the presence of fluid in the peritoneal cavity,
which helped decision-making (20). The presence of both
echogenic fluid and focal fluid has been reported to be
indicators for surgical intervention, but anechoic fluid was
more common in both of the groups in our study. Small
amounts of anechoic free fluid could be a normal finding in
healthy neonates (16,20), while the presence of echogenic
fluid has been shown to suggest intestinal perforation, even
if there was no evidence of free air in the peritoneal cavity
(21). Ultrasonography also helps to assess bowel wall
thickness and peristalsis, which are important factors in
monitoring NEC. We also assessed the value of abdominal
sonography in detecting bowel wall thickness. Parietal
thickening and absent peristalsis were found to be useful,
but bowel wall thinning failed to show its predictive ability.
In a study by Faingold et al., the sole finding of parietal
thickening by more than >2.6 mm did not distinguish
definite NEC patients from suspected NEC patients, as it
was a common finding in all NEC patients (20). In our
study, a cut-off value of 2.8 mm for bowel wall thickness
was helpful in predicting surgical intervention. Absent
peristalsis is quite common in advanced NEC patients,
and we found that it may also be a warning sign for surgical
intervention in NEC without pneumoperitoneum.
The present study also demonstrated that combining
abdominal radiography and ultrasonography to predict the
need for surgical intervention was superior to using just one
of these imaging modalities. The two imaging modalities
could not replace each other when it came to monitoring
the severity of NEC patients and guiding the clinical
decision-making.
Our finding of elevated C-reactive protein as an inde-
pendent risk factor for surgical intervention in NEC
patients was consistent with a previous study (22). The role
that erythema of the abdominal wall plays in predicting
intestinal gangrene in NEC patients has previously been
described (23,24). Although its low prevalence in NEC
patients limits its usefulness, it still looks promising when it
is combined with other predictors.
Our study had a large cohort in comparison with
previous studies, and the equipment we needed to assess
all the parameters we focused on is readily available in most
clinical settings, making it a useful approach in clinical
practice. The limitation of this research was that the
number of ultrasonographic signs we studied was inten-
tionally limited to the signs that were routinely used in our
clinical practice. The other sign, an absence of bowel wall
perfusion, which is helpful in determining the extent and
location of necrosis (20), was not included because Doppler
assessment was not routinely performed in our hospital.
CONCLUSION
This study identified several independent predictors of
surgical intervention in NEC patients without pneumoperi-
e154 ©2015 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2016 105, pp. e151–e155
He et al.
toneum. We found that abnormal ultrasonography and
radiography findings were the most powerful predictors of
the need for surgical intervention. These included persistent
dilation of the bowel loops and evidence of portal venous
gas, which were detected by radiography, and bowel wall
thickening, absent peristalsis and echogenic-free fluid or
focal fluid collection, which were detected by ultrasonog-
raphy.
ACKNOWLEDGEMENT
We are grateful to Shujuan He, Chuanyan Tao and Xin Tan,
students at the Children’s Hospital of Chongqing Medical
University, for their help with the data collection.
FUNDING SOURCE
All phases of this study were supported by the National
Natural. Science Foundation of China (No. 81370744),
Doctoral Degree Funding from the Chinese Ministry of
Education (No. 20135503110009), the subproject of
National Science & Technology Pillar Program during the
12th Five-year Plan Period in China (No. 2012BAI04B05),
the Scientific ResearchFoundation of Chongqing Municipal
Health Bureau (No. 2013-2-051), the Scientific Research
Foundation of the Science and Technology Commission of
Yuzhong District of Chongqing (No. 20140103) and the
State Key Clinic Discipline Project (No. 2011-873) in China.
CONFLICT OF INTEREST
The authors have no conflict of interest to declare.
References
1. Tepas J III, Sharma R, Leaphart CL, Celso BG, Pieper P,
Esquivia-Lee V. Timing of surgical intervention in necrotising
enterocolitis can be determined by trajectory of metabolic
derangement. J Pediatr Surg 2010; 45: 310–4.
2. Kafetzis DA, Skevaki C, Costalos C. Neonatal necrotising
enterocolitis: an overview. Curr Opin Infect Dis 2003; 16: 349–
55.
3. Rees C, Hall N, Eaton S, Pierro A. Surgical strategies for
necrotising enterocolitis: a survey of practice in the United
Kingdom. Arch Dis Child Fetal Neonatal Ed 2005; 90: F152–5.
4. Horwitz JR, Lally KP, Cheu HW, David Vazquez W, Grosfeld
JL, Ziegler MM. Complications after surgical intervention for
necrotising enterocolitis: a multicenter review. J Pediatr Surg
1995; 30: 994–9.
5. Neu J, Walker WA. Necrotising enterocolitis. N Engl J Med
2011; 364: 255–64.
6. Coursey CA, Hollingsworth CL, Wriston C, Beam C, Rice H,
Bisset G. Radiographic predictors of disease severity in
neonates and infants with necrotising enterocolitis. Am J
Roentgenol 2009; 193: 1408–13.
7. Kessler U, Mungnirandr A, Nelle M, Nimmo A, Zachariou Z,
Berger S. A simple presurgical necrotising enterocolitis-
mortality scoring system. J Perinatol 2006; 26: 764–8.
8. Rabinowitz SS, Dzakpasu P, Piecuch S, Leblanc P, Valencia G,
Kornecki E. Platelet-activating factor in infants at risk for
necrotising enterocolitis. J Pediatr 2001; 138: 81–6.
He et al.
9. Gupta SK, Burke G, Herson VC. Necrotising enterocolitis:
laboratory indicators of surgical disease. J Pediatr Surg 1994;
29: 1472–5.
10. Sharma R, Tepas JJ III, Hudak ML, Wludyka PS, Mollitt DL,
Garrison RD, et al. Portal venous gas and surgical outcome of
neonatal necrotising enterocolitis. J Pediatr Surg 2005; 40: 371–6.
11. Roy A, Tayeb M, Khogeer SS, Al-Salem AH. Predictors of
gangrenous necrotising enterocolitis and extent of disease.
Early laparotomy versus peritoneal drainage. Saudi Med J
2005; 26: 447–52.
12. Thompson AM, Bizzarro MJ. Necrotising enterocolitis in
newborns: pathogenesis, prevention and management. Drugs
2008; 68: 1227–38.
13. Tam AL, Camberos A, Applebaum H. Surgical decision making
in necrotising enterocolitis and focal intestinal perforation:
predictive value of radiologic findings. J Pediatr Surg 2002; 37:
1688–91.
14. Coursey CA, Hollingsworth CL, Gaca AM, Maxfield C,
DeLong D, Bisset G. Radiologists’ agreement when using a 10-
point scale to report abdominal radiographic findings of
necrotising enterocolitis in neonates and infants. Am J
Roentgenol 2008; 191: 190–7.
15. Daneman A, Woodward S, de Silva M. The radiology of
neonatal necrotising enterocolitis (NEC). A review of 47 cases
and the literature. Pediatr Radiol 1978; 7: 70–7.
€ Figure S1 Description of the inclusion and exclusion
16. Dilli D, Suna O
ß , Erol R, Ozkan-Ulu
guz S H, Dumanlı H, Dilmen
U. Does abdominal sonography provide additional information
over abdominal plain radiography for diagnosis of necrotising
enterocolitis in neonates? Pediatr Surg Int 2011; 27: 321–7.
17. Epelman M, Daneman A, Navarro OM, Morag I, Moore AM,
Kim JH, et al. Necrotising Enterocolitis: Review of State-of-
the-Art Imaging Findings with Pathologic Correlation1.
Radiographics 2007; 27: 285–305.
18. Cikrit D, Mastandrea J, Grosfeld JL, West KW, Schreiner RL.
Significance of portal vein air in necrotising entercolitis:
analysis of 53 cases. J Pediatr Surg 1985; 20: 425–30.
19. Buras R, Guzzetta P, Avery G, Naulty C. Acidosis and hepatic
portal venous gas: indications for surgery in necrotising
enterocolitis. Pediatrics 1986; 78: 273–7.
©2015 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2016 105, pp. e151–e155
Surgical prediction for enterocolitis
20. Faingold R, Daneman A, Tomlinson G, Babyn PS, Manson DE,
Mohanta A, et al. Necrotising Enterocolitis: assessment of
Bowel Viability with Color Doppler US1. Radiology 2005; 235:
587–94.
21. Silva CT, Daneman A, Navarro OM, Moore AM, Moineddin R,
Gerstle JT, et al. Correlation of sonographic findings and
outcome in necrotising enterocolitis. Pediatr Radiol 2007; 37:
274–82.
€
ß etinkaya M, Ozkan
22. C H, Ko € u
€ ksal N, Akacı O, Ozg € r T.
Comparison of the efficacy of serum amyloid A, C-reactive
protein, and procalcitonin in the diagnosis and follow-up of
necrotising enterocolitis in premature infants. J Pediatr Surg
2011; 46: 1482–9.
23. Henry MC, Moss RL. Surgical therapy for necrotising
enterocolitis: bringing evidence to the bedside. Seminars in
pediatric surgery: Elsevier, 2005.
24. Kosloske AM. Indications for operation in necrotising
enterocolitis revisited. J Pediatr Surg 1994; 29: 663–6.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the
online version of this article:
criteria and subgroup division.
Figure S2 Receiver operating characteristic curve of the
critical imaging report for predicting surgical intervention
in NEC patients without pneumoperitoneum.
Table S1 Definitions used in the diagnosis and analysis for
predictors
Table S2 Surgical strategies and pathologic findings of two
subgroup of surgical NEC patients
Table S3 Characteristics of NEC patients with pneumoperi-
toneum in the two study groups
e155