Management of Gastroschisis

Supplemental Digital Content is available in the text Objective: In infants with gastroschisis, outcomes were compared between those where operative reduction and fascial closure were attempted ≤24 hours of age (PC), and those who underwent planned closure of their defect >24 hours of age following reduction with a pre-formed silo (SR). Summary of Background Data: Inadequate evidence exists to determine how best to treat infants with gastroschisis. Methods: A secondary analysis was conducted of data collected 2006–2008 using the British Association of Pediatric Surgeons Congenital Anomalies Surveillance System, and 2005–2016 using the Canadian Pediatric Surgery Network. 28-day outcomes were compared between infants undergoing PC and SR. Primary outcome was number of gastrointestinal complications. Interactions were investigated between infant characteristics and treatment to determine whether intervention effect varied in sub-groups of infants. Results: Data from 341 British and Irish infants (27%) and 927 Canadian infants (73%) were used. 671 infants (42%) underwent PC and 597 (37%) underwent SR. The effect of SR on outcome varied according to the presence/absence of intestinal perforation, intestinal matting and intestinal necrosis. In infants without these features, SR was associated with fewer gastrointestinal complications [aIRR 0.25 (95% CI 0.09–0.67, P = 0.006)], more operations [aIRR 1.40 (95% CI 1.22–1.60, P < 0.001)], more days PN [aIRR 1.08 (95% CI 1.03–1.13, P < 0.001)], and a higher infection risk [aOR 2.06 (95% CI 1.10–3.87, P = 0.025)]. In infants with these features, SR was associated with a greater number of operations [aIRR 1.30 (95% CI 1.17–1.45, P < 0.001)], and more days PN [aIRR 1.06 (95% CI 1.02–1.10, P = 0.003)]. Conclusions: In infants without intestinal perforation, matting, or necrosis, the benefits of SR outweigh its drawbacks. In infants with these features, the opposite is true. Treatment choice should be based upon these features.

A ffecting approximately 1:3000 live-births in the British Isles, 1 and 1:2200 live births in the United States of America, 2,3 the congenital abdominal wall defect known as gastroschisis is one of the most common neonatal conditions managed by pediatric surgeons. Its etiology is unknown, however, there is widespread evidence of increasing prevalence across international jurisdictions over the last 20 years. 4 It is characterized by early gestational herniation of the abdominal organs through a paraumbilical defect, almost always to the right of the umbilical cord. Clinical outcomes are influenced by the severity of injury incurred by the exposed fetal intestine and features of ''complex'' gastroschisis include the development of intestinal necrosis, perforation, and atresia which increase the risk of short and long term morbidity and mortality. 5 In 2011, a National Confidential Enquiry into Patient Outcomes and Death review of the organizational and clinical aspects of children's surgery in the UK highlighted 2 key findings. First, when necrotizing enterocolitis was excluded, infants who died under the care of pediatric surgeons had most commonly undergone surgery for gastroschisis, exomphalos, or malrotation. 6,7 Second, 2 of the drivers for outcomes not being as good as they could be in key pediatric surgical conditions were difficulties and delays in the surgical decision-making process. 7 Such difficulties and delays stemmed in part from the lack of robust evidence that was available to support surgeons in their decision-making, and 1 condition where this is likely to be impacting outcomes, is gastroschisis. 8,9 In high income countries, the initial management of infants born with gastroschisis involves nasogastric decompression, antibiotics, intravenous fluid management, and prevention of hypothermia and evaporative fluid loss by bowel protection. Each of these steps is relatively uniformly performed regardless of where or by whom the infant is treated. There is, however, significant variation in how the eviscerated abdominal contents are reduced and the abdominal wall defect closed. The 2 most commonly used reduction and closure strategies are operative primary fascial closure (PC), and pre-formed silo placement with staged reduction and delayed closure (SR). 1,10 With PC, the abdominal contents are reduced on day 1 of life under a general anesthetic, before the abdominal wall defect is closed using fascial sutures. In SR, the abdominal contents are placed in a pre-formed, spring-loaded, silastic silo, the volume of which is reduced over several days to gradually return the intestines to the abdominal cavity. Silos can be placed and reduced in volume on the neonatal intensive care unit without requiring general anesthetic. Once the abdominal contents are fully reduced, the abdominal wall defect can then either be closed under general anesthetic using fascial sutures, or at the cot-side using a suture-less skin only closure. The final, less commonly used form of reduction and closure is wardbased reduction and sutureless closure, in which the abdominal contents are reduced on day 1 of life without general anesthetic, and a skin only closure achieved using steri-strips. Only 1 randomized controlled trial has been attempted comparing operative management strategies for infants born with gastroschisis, and whilst this showed no difference in ventilator days between the treatments compared, PC and SR, the study failed to recruit to target and was therefore under-powered to detect any differences. 11 The overall aim of our study was to investigate whether it was possible to use existing, prospectively collected data to identify which infants with gastroschisis should be treated using each of the 2 most common surgical strategies, PC and SR.

Research Questions
1. Are there specific characteristics that can be used to determine which reduction and closure strategy should be used for a particular infant? 2. Do the 8 outcomes of importance identified in the previously developed NETS 1G gastroschisis core outcome 12 set differ at 28 days of age between infants treated using PC, and infants treated using SR?

Summary
This study comprised a secondary analysis of data collected prospectively by 3 population-based systems, the British Association of Pediatric Surgeons Congenital Anomalies Surveillance System (BAPS-CASS), 1 the Canadian Pediatric Surgery Network (CAP-SNet), 13 and the Canadian Neonatal Network (CNN). 14

Data Collection
Between October 2006 and March 2008, a British and Irish cohort study describing outcomes at 28 days of age for children born with gastroschisis was conducted using the BAPS-CASS infrastructure, 1 and from May 2005 onwards, CAPSNet collected demographic, early management, operative, and pre-discharge outcomes data for all infants diagnosed with gastroschisis in Canada. Infants whose gastroschisis data were collected through CAPSNet also had data relating to any neonatal intensive care unit admissions collected through the CNN. Data relating to all live-born infants diagnosed with gastroschisis in Canada between May 2005 and December 2016 were extracted from the CAPSNet/CNN databases and merged with data collected during the BAPS-CASS cohort study. Data collection methodology, duplicate checking, and chasing of missing data have been described elsewhere. 1,13,14 Further details of the database merging strategy, variable definitions, outcomes where direct equivalence did not exist, and sensitivity analyses conducted to investigate the effect of the variable mapping process are described in Supplementary

Interventions
Infants whose reduction and closure strategy did not meet one of the following definitions of PC or SR were excluded from the analysis. Intention to treat analyses were used throughout, such that all infants where PC was attempted were analyzed in this group even if closure was unsuccessful, due for example to raised intra-abdominal pressure or abdomino-visceral disproportion, and a silo, either pre-formed or custom (ie, fashioned from silastic sheeting and sutured to the defect margins), was placed instead.

CAPSNet Definitions
PC was defined as intended primary reduction of the eviscerated abdominal contents, with the first attempted reduction occurring within a day of birth, and sutured fascial closure being attempted. If the closure technique was unknown, the surgeon's intent to suture the fascia was assumed if the reduction and closure took place with an intubated or anesthetized infant.
SR was defined as use of a silo to facilitate delayed closure, with closure taking place more than 1 day after birth. Where the primary intention was to facilitate SR through use of a silo, and the type of silo was not known, the silo was assumed to be a pre-formed, spring loaded silo as opposed to a custom silo. This is because standard practice across Canada is to only use custom silos as a salvage operation where PC or use of a pre-formed silo has not been possible. It is therefore highly unlikely that any custom silos were used in cases where SR was the intended treatment strategy.

BAPS-CASS Definitions
For infants whose data were collected in the BAPS-CASS dataset, the first attempted reduction and closure strategy was classified as per the treating surgeon.

Outcome Definition
Number of severe gastrointestinal complications in the first 28 days of life was investigated as the primary outcome. This was defined as per the recently developed NETS 1G gastroschisis core outcome set (COS) 12 to include intestinal perforation (identified after treatment), unplanned intestinal resection regardless of amount of bowel removed or the indication for the resection, mechanical intestinal obstruction requiring laparotomy, abdominal compartment syndrome, and enterocolitis. In CAPSNet, abdominal compartment syndrome was defined as ''an increase in intra-abdominal pressure requiring surgery to relieve pressure,'' whilst in BAPS-CASS, it was defined as per the treating clinician. Enterocolitis for both sets of infants was defined as per the treating clinician. Secondary outcomes are the remaining outcomes in the NETS 1G COS and are defined in Box 1.

Propensity Score Calculation
Propensity scores predicting an infant's probability of being treated using a particular reduction and closure strategy were calculated using logistic regression. The characteristics used to develop the propensity score were gestational age at birth, birth weight, weight <10th centile for gestational age at birth, year of birth, country of treatment, transfer in to a surgical center, antenatal diagnosis, sex, presence of an additional chromosomal or structural anomaly, intestinal necrosis, intestinal atresia, intestinal perforation, intestinal matting, Apgar score at 5 minutes, and the composite variable, ''complex'' gastroschisis. 5 Variables requiring further definition are described in Box 2.

Assessment of Interaction
Statistical interactions between reduction and closure strategy, and key infant characteristics, were investigated in a covariate and propensity score adjusted model describing the association between reduction and closure strategies and the number of incident severe gastrointestinal complications. Where clinically or statistically significant interactions were identified, these were used to define 2 subgroups of infants in which all subsequent analyses were conducted.
A statistically significant interaction was defined as one in which including the interaction term improved the fit of the covariate and propensity score adjusted model, as defined by a P-value of <0.05 on likelihood ratio testing. A clinically significant interaction was defined as one in which the treatment effects were in opposing directions at each level of the interaction, even if including the interaction term did not meet the level of statistical significance for improving the fit of the model.

Comparison of Reduction and Closure Strategies
The impact of treatment choice on outcome was investigated in each of the sub-groups using negative binomial regression, Poisson regression, Logistic regression, and linear regression as appropriate. Crude and propensity score plus covariate-adjusted estimates of effect were calculated. Covariates adjusted for were gestational age at birth, birthweight, intestinal perforation identified at presentation, and intestinal atresia, as these characteristics were demonstrated in this cohort to be associated with variation in outcome.

Infant Characteristics
A total of 1600 live-born infants with gastroschisis were identified during the defined reporting periods, 393 (25%) from Britain and Ireland, and 1207 (75%) from Canada. Of these, 671 (42%) underwent PC as their first attempted reduction and closure strategy, and 597 (37%) underwent SR. Of the remaining 332 infants who were excluded from the analysis, 183 (11%) underwent wardbased reduction with sutureless closure, 16 (1%) underwent SR facilitated by a custom silo, 71 (4%) underwent operative primary reduction with suture-less closure, and 11 infants (1%) had a mesh sutured in place at their first reduction and closure procedure. The first reduction and closure strategy was unknown for 51 infants (3%) (Fig. 1). Of the 671 infants where PC was attempted, 106 (16%) were unable to be closed primarily and therefore had a silo placed. Key characteristics of infants in the PC and SR treatment groups are shown in Table 1

Severe Gastrointestinal Complications
In the first 28 days of life, the 671 infants in the PC group experienced 105 severe gastrointestinal complications between them, whilst the 597 infants in the SR group experienced 68 severe gastrointestinal complications between them. Overall, in the PC group, 586 infants (88%) experienced no complications, 67 (10%) experienced 1, 13 (2%) experienced 2, and 4 (1%) experienced 3 or more. In comparison, in the SR group, 549 infants (92%) experienced no complications, 35 (6%) experienced 2, 9 (2%) experienced 2, and 4 (1%) experienced 3 or more. A breakdown of the types of severe gastrointestinal complications experienced by infants in each group is given in Table 2.

Interactions and Sub-group Definition
A clinically plausible interaction was identified between intestinal perforation identified at time of presentation, and treatment, with SR associated with an incidence rate ratio (IRR) of severe gastrointestinal complications of 0.65 in those without perforation, and an IRR of 1.96 in those with perforation. Similarly, a clinically plausible interaction was identified between presence of intestinal matting and treatment, with SR associated with an IRR of severe gastrointestinal complications of 0.31 in those without intestinal matting, 1.0 in those with mild matting, and 1.10 in those with severe matting. Finally, a clinically plausible interaction was also identified between intestinal necrosis and treatment, with SR associated with an IRR of severe gastrointestinal complications of 0.66 in those without intestinal necrosis, and 2.54 in those with intestinal necrosis. No interaction was identified between intestinal atresia and treatment type, with SR associated with an IRR of severe gastrointestinal complications of 0.70 in those without intestinal atresia, and 0.75 in those with intestinal atresia. No other interactions between key In BAPS-CASS, liver disease was diagnosed at the discretion of the treating clinician.
Using the available data, it was not possible to apply a single definition of liver disease uniformly for both British and Irish, and Canadian infants, and; therefore, to present the most severe estimate of the rate of liver disease, an infant was deemed to have been diagnosed with liver disease if they met the case definition for the study in which their data were originally used.

Box 2 Expanded definitions for propensity score variables
Additional chromosomal or structural anomalies Excluded those that were likely to be directly associated with evisceration of the abdominal contents, including intestinal atresia and undescended testes.
Single minor anomalies such as ''accessory skin tags, congenital'' were not included as an additional chromosomal or structural anomaly. A full list of additional structural and congenital anomalies is described in supplementary material 5 http://links.lww.com/SLA/C355.

Intestinal necrosis
Defined as per the treating surgeon In BAPS-CASS, categorised by surgeons as pink/healthy, meconium stained/healthy, dusky/ischaemic, or black/necrotic, In CAPSNet, categorised by abstractors using web-based pictorial guidance and text descriptors, as none, focal or diffuse.
Mapping of BAPS-CASS to CAPSNet categories is described in supplementary material 3 http://links.lww.com/SLA/C353, along with descriptions of the sensitivity analyses conducted.
characteristics and the effect of treatment on outcome were identified. The effect of treatment on outcome; therefore, varies according to the presence or absence of each of intestinal perforation, intestinal matting, and intestinal necrosis.
As the effect of SR on number of severe gastrointestinal complications differed between infants with and without intestinal perforation, intestinal matting, or intestinal necrosis, 2 sub-groups were defined. The first sub-group included infants without intestinal necrosis, perforation, or matting, and the second included infants with 1 or more of these features of bowel injury. All subsequent analyses were undertaken in these 2 sub-groups.

Infants Without Intestinal Necrosis, Perforation, or Matting
In the sub-group of 443 infants without intestinal necrosis, perforation, or matting, infants who underwent SR experienced statistically significantly fewer severe gastrointestinal complications in the first 28 days of life than those who underwent PC, adjusted IRR 0.25 (95% CI 0.09-0.67, P ¼ 0.006), but underwent a greater number of operations, adjusted IRR 1.40 (95% CI 1.22-1.60, P < 0.001), and required a marginally greater number of days of PN, adjusted IRR 1.08 (95% CI 1.03-1.13, P < 0.001), and were more likely to experience 1 or more infection, adjusted OR 2.06 (95% CI 1.10-3.87 P ¼ 0.025) ( Table 3 and Supplementary Table 2, http:// links.lww.com/SLA/C357). There were no differences in any other outcomes between infants who underwent SR, and those who underwent PC.

Infants With Intestinal Necrosis, Perforation, or Matting
When outcomes for the group of 697 infants with intestinal necrosis, perforation, or matting were compared between PC and SR, 2 statistically significant differences were identified. Use of SR was associated with a greater number of operations in the first 28 days of life, adjusted IRR 1.30 (95% CI 1.17-1.45, P < 0.001), and a marginally greater number of days on which PN was used in the first 28 days of life, adjusted IRR 1.06 (95% CI 1.02-1.10, P ¼ 0.003) (

Types of Operation Performed
Overall, in the group of 671 infants who underwent PC, 1537 operations were performed, whilst in the 597 infants who underwent SR, 1839 operations were performed (Table 5).

DISCUSSION
In comparison to PC, use of SR in infants without intestinal necrosis, perforation, or matting was associated with an approximately 75% reduction in the incidence of severe gastrointestinal complications in the first 28 days of life, but at the expense of a 40% increase in number of operations, a doubling in the risk of experiencing 1 or more infections, and potentially an 8% increase in number of days on which PN was received over the same time period. In contrast to its effect in infants without intestinal perforation, necrosis or matting, the use of SR in infants with any of these features was associated with a 30% increase in the number of operations infants undergo in the first 28 days of life, and potentially a 6% increase in the number of days on which they receive PN, but no reduction in number of severe gastrointestinal complications. The presence of necrosis seems to have a major contribution to the differential impact of treatment with SR on outcome seen between these 2 groups of infants, but matting and perforation also contribute.
Three key strengths of this study are the size of the cohort, the patient-centered nature of the outcomes investigated, and the detailed, prospective data collection methodology utilized. The collection of nuanced data relating to an infant's physiology, degree of bowel injury and operative management allowed estimates of effect to be adjusted for infant's propensity scores and previously identified confounding factors, thereby ensuring that the results presented are as close as possible using observational data to those that would be seen in a randomized controlled trial. As the outcomes investigated were identified in the NETS 1G COS as important to key stakeholders, the results of the comparison between PC and SR hold direct relevance to clinical practice. Three limitations affected this work. The first, is that definitions of some outcomes varied between the databases, making it difficult to ensure comparability when creating the unified dataset.
The second, is that allocation to treatment has been confounded by intention. However, the use of covariate and propensity score adjusted analyses should have, if sufficient factors were taken account of and accurately measured, accounted for much of this confounding. [15][16][17] One potentially important confounding factor that was knowingly not adjusted for was hospital of treatment. Hospital of treatment was not adjusted for, as the ratio of centers to cases was too high such that adjustment would have unacceptably reduced the statistical power of the analyses, and the limitations introduced by this would, we believe, have outweighed the benefits of the further adjustment. The third overall limitation is that data collection was at 28 days of age, and therefore only early outcomes are presented. This is particularly relevant for the outcome PN usage, where, as a large proportion of infants remained on PN at 28 days of age, it is difficult to define the clinical significance of the statistically significant differences identified between PC and SR. Because not all infants were fully enterally fed at the time of outcome reporting, there is also the potential that some severe gastrointestinal complications were not identified that would have been if the cohort were followed up until enteral autonomy. However, as the proportion of infants requiring PN at 28 days of age is similar between the treatment groups, we would not expect this to significantly affect the conclusions of the study. If the cohort were followed up to later time-points, it is also possible that a different pattern of outcomes would have been seen when the 2 treatment groups were compared. As 28 days of age is relatively early in the course of treatment of many infants born with gastroschisis, some caution must; therefore, be exercised when using these results to guide treatment decision making.
Two recently conducted systematic reviews 8,9 reported that multiple outcomes were better for infants who underwent PC than for those who underwent SR. However, there were significant limitations with the primary data upon which those conclusions were based, and it is; therefore, difficult to use these reviews to inform clinical decision making. It is also impossible to draw any robust conclusions from the 1 RCT that has been conducted comparing PC to SR 11 as the trial was abandoned due to a failure to recruit to target. 11 In contrast to the existing literature, this study has been able to at least in part address issues arising from selection bias and confounding that weaken the published systematic reviews, 8,9 and also achieve sufficient statistical power to detect differences in meaningful outcomes. 11 Many surgeons opt to use SR over PC due to the fact that it is felt to reduce the number of operations that children required. Our study has suggested the opposite to be true. This difference is likely due to the definition of operation used. Traditionally, many surgeons class an operation as a procedure performed under a general anesthetic, and therefore count infants who have had an uncomplicated  silo repair and sutureless abdominal wall closure, as undergoing zero operations. However, when the gastroschisis COS was developed, the clear message from all stakeholder groups was that the definition of an operation should include all silo placements, replacements, and abdominal wall closures, regardless of whether these were performed under general anesthetic or not. This definition did not however include episodes of silo volume reduction. 12 This expanded definition of operation likely accounts for the discrepancy seen between the results of this work and established thinking. We hypothesize that the benefit seen in this study to treating infants with intestinal necrosis, perforation or matting using PC is seen because undergoing general anesthetic allows a thorough assessment of the infant's intestines and abdomino-visceral ratio, and then based on those findings, individual tailoring of the reduction and closure strategy. In contrast, in those infants with less complex disease, that is, no necrosis, perforation, or matting, where individualized treatment is less necessary, we hypothesize that the gradual reduction of the abdominal contents that is associated with SR reduces the risk of increased intra-abdominal pressure and development of the severe gastrointestinal complications associated with it, but the staged nature of the procedure increases the number of operations that infants require, and the greater period of time over which the bowel is exposed increases the risk of systemic infections.
Based upon the results of this study, parents of infants born with gastroschisis who do not have intestinal perforation, intestinal necrosis or matted bowel at delivery should be counseled that treatment with SR has both pros and cons. However, as the magnitude of reduction in number of severe gastrointestinal complications outweighs the drawbacks of treatment with SR, we propose that these parents should be counseled that SR is associated with improved short-term outcomes, and on this basis, may be the most appropriate treatment for their children. This counseling must however be framed with the caveat that currently only short-term outcomes are known, and in particular, the long-term impact of potentially greater PN use in children treated with SR, is unknown. In contrast to parents of children without significant bowel injury, parents of children with intestinal necrosis, perforation or matting should be advised that at present, it seems that PC is the operation of choice, as there are demonstrable benefits to its use, but no demonstrable drawbacks. Parents must however also be informed that centers or individual practitioners within a center may have different levels of expertise with each approach, and therefore until there is a transition towards more standardized care, counseling should be contextualized in a center or practitioner-specific manner.
Whilst this work has provided evidence that can be utilized to begin developing a decision-making pathway for infants born with gastroschisis, further research is still required, focussing particularly on longer-term outcomes, and where appropriate, additional outcomes such as short bowel syndrome, which have not been reported as part of this core set of outcomes. We would advocate for conduct of a joint British, Irish and Canadian RCT comparing PC to SR in infants without intestinal necrosis, perforation or matting. Such an RCT should include a cost-utility analysis, and would require approximately 250 infants to be randomized to demonstrate a statistically significant difference between treatments in the number of severe gastrointestinal complications. We believe that such a trial is feasible and necessary to conduct. Further work is also necessary to ascertain the relative merits of other treatment strategies including ward-based reduction and sutureless closure, which were not specifically addressed in this study. However, until the point that such trial data, or high-quality observational data reporting long-term outcomes, are available, we propose that infants without intestinal necrosis, perforation or matting are managed using placement of a pre-formed silo, SR and delayed closure, whilst those infants with intestinal necrosis, perforation or matting undergo examination under general anesthetic, appropriate management of the injured bowel, and primary closure if feasible, safe, and not precluded by signs of raised intra-abdominal pressure. Percentage of those with complete data. y Adjusted for propensity score, gestational age at birth, birthweight, intestinal atresia, and intestinal perforation at presentation. CI indicates confidence interval; IQR, interquartile range; IRR, incidence rate ratio; OR, odds ratio.