Introduction

Necrotising enterocolitis (NEC) is an inflammatory gastrointestinal disease arising in neonates; it is one of the most common causes of mortality and morbidity in babies born prematurely [1]. The UK incidence of NEC in premature infants is 27.9 cases per 100,000 live births, with an overall mortality rate of 23.5% [2]. One in four babies with NEC will undergo surgical intervention, with mortality rates as high as 50.9% in infants with birthweights under 1000 g [3, 4]. It is known that outcomes, including neurodevelopmental outcomes, are worse in infants with NEC than those without, and even worse in those treated surgically compared to those whose NEC is managed non-operatively [5, 6]. Neurodisability following surgery for NEC occurs more frequently than intestinal failure, in as many as 60% of cases [4]. Despite this association between surgical NEC (sNEC) and worse neurodevelopmental outcomes being clearly established, there is limited understanding of the factors that contribute to these worse outcomes. These factors may be non-modifiable and related to the babies’ intrinsic condition or status, or they may be potentially modifiable, for example timing of surgery, and offer potential to improve outcomes in sNEC cohorts [7,8,9]. Previous reviews have examined outcomes following NEC overall, but a gap remains in reviewing the impact of sNEC on neurodevelopment [4].

Hence, the aims of our study were to:

  • Identify studies reporting neurodevelopmental outcomes in sNEC infants;

  • Investigate which factors in the management of sNEC are associated with neurodevelopmental outcomes and are potentially modifiable, so representing areas where changes in management could lead to improved outcomes in future sNEC cohorts.

Methods

Protocol and registration

A systematic review protocol was registered with PROSPERO [Code: CRD42022370309] on 31/10/2022.

Data sources and search strategy

The PubMed and Embase databases were interrogated on 01/11/2022, using key terms including: “Infant”, “Necrotising enterocolitis”, “Surgical”, “Neurodevelopmental”, and “Outcomes”. The full search strategy can be found in Appendix 1. The inclusion and exclusion criteria are as listed in Table 1, publication after the year 2000 was used in an attempt to ensure that included studies were relevant to current practice.

Table 1 Inclusion and exclusion criteria

Study selection

Records identified by the search strategy were collated and deduplicated. The resulting articles were screened by title and abstract against the inclusion/exclusion criteria. The remaining articles underwent full-text review by at least two authors, and any disagreements on eligibility for inclusion were resolved by discussion and consensus. Although review articles were excluded, the references were hand searched for any papers meeting our inclusion criteria.

Quality of included studies

The Quality in Prognosis Studies (QuIPS) tool was used to assess the quality of studies included [10]. Six domains were assessed: (1) study participation, (2) study attrition, (3) prognostic factor measurement, (4) outcome measurement, (5) study confounding, and (6) statistical analysis and reporting. For all papers, each domain was assessed as either: “sufficiently reported”, “moderately reported”, or “not reported”. Based on this, each paper and domain were graded overall as being either: “low risk”, “medium risk”, or “high risk”, in terms of potential bias.

Results

Study identification

The initial search strategy yielded 1170 articles. After deduplication and screening against inclusion and exclusion criteria, 22 remained to form the review (Fig. 1).

Fig. 1
figure 1

Prisma flowchart

Study findings

Of the 22 studies included, there were 15 retrospective, 6 prospective, and 1 prognosis study. The study characteristics are shown in Table 2, organised by study type (prospective first), and cohort size. We have not reported in detail the outcomes presented in the included studies as this fell outside the primary aim of our review, to identify factors associated with neurodevelopmental outcomes following sNEC.

Table 2 Characteristics of included studies

Quality of included studies

The majority of papers had only low or medium risk of bias. Two studies (Cuzilla et al., 2014, Ganapathy et al., 2013) were assessed to be of high risk of bias due to incomplete studies and follow-up [11, 12].

Included studies reported a variety of measures of neurodevelopmental outcome. These included formal developmental assessments, including the Bayley score, the diagnosis of a neurodisability (such as cerebral palsy) or surrogate markers, such as the presence or absence of white matter brain injury (WMBI) on imaging. Where comparisons were made, infants with NEC undergoing surgical intervention consistently had an increased risk of neurodevelopmental impairment compared to those treated without surgery [5, 8, 9, 11, 13]. Further, infants undergoing surgery had higher rates of intraventricular haemorrhage (IVH) and periventricular leukomalacia (PVL) compared to medically managed groups [8, 14,15,16,17,18,19,20].

Several pre-operative factors were universally associated with neurodevelopmental outcomes. These included non-modifiable factors related to the babies’ underlying condition or status, such as: gestational age at birth, birth weight, and age of NEC onset (Table 3). Extreme prematurity (< 28 week gestation at birth) and lower birth weight were associated with worse neurodevelopmental outcomes [6, 8, 13, 30]. Two studies reported that an earlier age of NEC onset from birthdate was significantly associated with subsequent development of WMBI [6, 9].

Table 3 Non-modifiable and modifiable factors identified

A recurring finding was that the majority of neonates undergoing laparotomy did not subsequently reach age-appropriate milestones, and potentially modifiable factors in the surgical management of NEC were reported to be associated with this [6, 21]. The use of an intra-peritoneal drain pre- or post-laparotomy was associated with increased WBMI in an sNEC cohort [6]. Following bowel resection, infants undergoing enterostomy formation had significantly worse neurodevelopmental outcomes when compared to infants undergoing primary anastomosis reported as both lower IQ score and a higher incidence of WMBI [20].

Potentially modifiable but not directly surgical factors in the management of babies with sNEC included the use of pharmacological agents. The postnatal use of steroids was associated with higher incidence of WMBI and the prophylactic use of miconazole was associated with a reduction in the incidence of bowel perforation in NEC infants, subsequently reducing the need for surgical intervention [6, 21]. Other factors reported to be associated with worse neurodevelopmental outcomes included red cell transfusion and a lack of skin-to-skin contact post-operatively [6, 21] (Table 3).

Data aggregation

Due to the heterogeneity of both the patients and neurodevelopmental outcomes reported in the studies included in the review, no data aggregation or further analysis was possible.

Discussion

We present a systematic review of the current evidence for associations between both patient and treatment-related factors and neurodevelopmental outcomes in infants with sNEC. In keeping with previous work, this confirms worse neurodevelopmental outcomes for infants undergoing surgery for NEC than those managed without surgery. The review also identifies a number of aspects of babies’ care that relate to later neurodevelopmental outcomes. Several are patient related and not amenable to modification such as lower gestational age and birth weight. Others do offer potential for change though, including the use of drains and enterostomy.

Whilst all studies in our review included data on modifiable and non-modifiable factors, few analysed their association with neurodevelopmental outcomes specific to surgical cohorts. In addition, the 22 studies included in the review reported very heterogeneous results regarding both patient characteristics and details of treatment delivered, making comparisons challenging.

Considering non-modifiable risk factors, it is unsurprising to see universally worse neurodevelopmental outcomes in association with lower gestational age and birth weight. A smaller and less mature baby will have less reserve, and hence greater vulnerability to the effects of NEC. As we continue to see more babies of extremely low gestational age undergo surgery, it seems likely that we will see more long-term neurodevelopmental morbidity in survivors [32]. This will have health service planning implications as these babies grow into childhood and beyond.

The surgical strategies for NEC that are reported as being associated with worse neurodevelopmental outcomes (the use of drains and enterostomies rather than primary anastomosis) potentially offer an opportunity to change practice to improve outcomes [6, 20]. The underlying state of the baby may well be a significant confounding factor in both these examples, with drains being reserved for the sickest babies and primary anastomosis only being possible in a more stable and robust baby. Nevertheless, enterostomies can lead to more challenging fluid and electrolyte management in neonates, and it is known that their presence in neonates is associated with poor somatic growth during a crucial phase in brain development, especially in the presence of inflammation [33, 34]. It is therefore plausible that avoiding an enterostomy could help protect brain growth and improve neurodevelopmental outcomes. Alternative surgical strategies such as damage limitation and second look laparotomies that avoid enterostomies may have potential to lead to better neurodevelopmental outcomes [35].

Non-surgical treatment factors reported to be associated with worse outcomes are very likely to be confounded by the characteristics of the babies in question. For example, postnatal steroids are associated with worse outcomes but are reserved for the sickest babies [6]. It cannot be concluded that steroids should be avoided to improve neurodevelopmental outcomes in sNEC. Likewise, the association of worse outcomes with blood transfusion and an inability to have skin-to-skin contact post-operatively almost certainly reflects the condition of the babies rather than a direct causal effect [6, 21].

The chief strength of our review lies in its size. Using broad search terms, we identified a large number of papers for screening and so can be confident that we have included all relevant articles. The articles included in the review present data on a total of 49,426 babies from a wide geographical spread, making it likely that conclusions will be broadly applicable. However, only 6 of the 22 papers included used prospectively collected data, and the remainder were retrospective, with the exception of 1 prognosis study. The review is further hampered by the heterogeneity of the studies included, both in the factors that may affect outcome that they included and in the outcome tools or measures that they use. Several did not use a formal neurodevelopment assessment tool and those that did used a variety of different ones. This has precluded aggregation of the data into a meta-analysis. The follow-up in some studies is also rather limited, and only 4 of the 22 studies followed children into their school age years, where the impact of neurodevelopmental issues is more fully exhibited [15, 18, 24, 29]. Bias, however, was not a major concern, with only two included papers being at high risk of bias due to inadequate follow-up.

Further research is needed to determine modifiable factors that influence neurodevelopmental outcome in sNEC cohorts. Ideally, these studies should explore the influence of factors, including blood transfusion, steroid, and antifungal use. In addition, a large-scale, multicentre, prospective study investigating the impact of different surgical strategies on neurodevelopmental outcomes in NEC infants would be of great interest to surgeons. These will need to include detailed data collection on the characteristics of the babies treated as well as the interventions carried out to be able to identify confounding variables. Follow-up of at least 2 years is required to adequately show neurodevelopmental impacts of NEC, but ideally 5 year follow-up would be completed to give a picture of longer term neurodevelopment. The use of standardised outcomes in future studies would greatly assist the comprehension and assessment of reported findings. Although not designed specifically for use in babies undergoing surgery, and indeed including the development of NEC as an outcome in itself, the Core Outcomes in Neonatology outcome set published in 2020 is a helpful tool for researchers [36]. It provides a range of relevant and measurable outcomes including longer term neurodevelopmental outcomes, and whilst it may be that future work specifically on surgical intervention in NEC requires modifications or additions to this twelve point, it should form the foundation of any future work in the area.