Core Outcome Set for Necrotizing Enterocolitis Treatment Trials Free

This project was prospectively registered at the Core Outcome Measures in Effectiveness Trials Handbook initiative (Study #1920), with the protocol published prospectively.¹⁸  The involvement of parents and patient representatives in this COS is described using the Guidance for Reporting on Involvement of Patients and Public short-form checklist (Supplemental Information Methods 1).¹⁹  The current report was constructed according to the COS-STAndards for Reporting and Revised Standards for Quality Improvement Reporting Excellence recommendations.^20 ,21 

We formed a steering group to guide the COS development, comprising different disciplines and perspectives (Supplemental Information Methods 2). We aimed to include 3 stakeholder groups: A lay panel consisting of adult individuals with personal experience of NEC (ie, parents of infants diagnosed with NEC or adults who experienced NEC in infancy¹⁸ ), a neonatal panel consisting of clinicians and researchers involved in the neonatal period, and a nonneonatal panel consisting of clinicians and researchers involved in follow-up care for NEC after the neonatal period. We included nonneonatal experts as a separate panel because they might provide a unique view on long-term outcomes. Further information on stakeholder group formation can be found in our published protocol.¹⁸ 

For the lay panel, participants were approached through local NEC patient organizations, such as NEC United Kingdom and Instituto Pequenos Grandes Guerreiros (Brazil), and general neonatal organizations, such as Care4Neo (Netherlands), European Foundation for the Care of Newborn Infants (European), Tunisian Neonatal Association (Tunisia), Preterm Infant Parents Network (Uganda), and the Adult Preemie Advocacy group (United Kingdom).

For the second and third panels, we approached 146 clinicians and/or researchers from high-income countries and 126 from low–middle-income countries (defined by the Organization for Economic Co-operation and Development²² ) on the basis of their scientific output using SciVal, as described previously.¹⁸  After receiving a written explanation of the 3 panels, participants selected the one most fitting to them before the start of the first round. If experts were involved in both the neonatal and postneonatal periods, we arbitrarily included them only in the neonatal panel.

The outcome list was based on a systematic review of outcomes in recent NEC trials, with additions from the steering group resulting in 45 candidate outcomes (Supplemental Table 4).⁸  We focused on including outcomes measurable by an accepted tool or instrument. This process was described in detail in the published protocol.^8 ,18  Outcomes were categorized according to the Taxonomy for Outcomes in Medical Research into the following areas^23 ,24 : death, physiologic/clinical, life impact, resource use, and adverse events. Categorizing outcomes allows for the identification of areas that are underrepresented by current clinical trials. We aimed to include at least one outcome for every “core area.” Care was taken to avoid ambiguity of the language used in the list of outcomes, and clarifications for scientific terms were constructed by the steering group in collaboration with family representatives (Care4Neo).

Participants scored all 45 outcomes on a 9-point Likert scale, with 1 to 3 labeled “not important,” 4 to 6 “important but not critical,” and 7 to 9 “critical.” Participants were asked to add any outcomes that they considered important but not present in the initial outcome list at the end of round 1. These additional outcomes were added to the original outcome list after round 1 if they were proposed by at least 2 participants and deemed suitable by the steering group.

All participants who completed round 1 were invited to round 2. The participants were presented with their own scores and a graphical description of the scores of the 3 panels using bar graphs. After considering the views of their and other stakeholder groups, the participants were asked to rescore each outcome. Participants were also asked to score new outcomes added after round 1. For round 3, participants were asked to rescore the same outcomes as in round 2 except for any outcomes that met the criteria of “consensus out” (definition below) in a minimum of 2 panels. In this round, the graphical description of the scores included the preliminary consensus score (definition below) on the basis of the round 2 results.

For outcomes included in the COS, “consensus in” was defined as ≥70% of participants rating the outcome 7 to 9 and ≤15% rating it 1 to 3. Consensus out was defined as ≥70% participants rating an outcome 1 to 3 and ≤15% rating it 7 to 9. Outcomes not meeting these definitions were classified as “no consensus.”

All participants who completed round 3 were invited to an online consensus meeting. Here, participants were presented with the “top 10” outcomes reaching consensus in in the lay panel and the top 10 outcomes reaching consensus in in the combined neonatal and nonneonatal panel, to ensure equal representation by both stakeholder groups and limit the size of the final COS. The discussion of each outcome was started by a family representative or patient summarizing the Delphi results, the panelists’ written comments, and sharing their views on the importance of the outcome. The scores from Delphi round 3 were shown per panel. During a moderated discussion, participants were encouraged to critically evaluate if the outcome was relevant, specific enough to be improved by an intervention and measurable. Outcomes reaching no consensus were only discussed after a unanimous decision by all attending participants. All discussed outcomes were then rescored anonymously on a 9-point Likert scale and outcomes reaching consensus in were included in the final COS.

Existing definitions and time points for outcomes included in this COS were extracted from NEC literature identified through the previously conducted systematic review and existing guidelines.^8 ,25  These definitions were discussed in a steering group meeting and unanimous agreement was reached for all included outcomes.

The Medical Ethics Review Board of the University Medical Center Groningen, the Netherlands, reviewed this study and waived ethical approval, because this is not clinical research with human subjects, as meant in the Dutch Medical Research Involving Human Subjects Act. Electronic informed consent was obtained from all participants after they were informed about the nature of the study, before voluntary registration. Participants could withdraw at any time.

We compared the median scores of the lay and combined neonatal and nonneonatal stakeholder groups, scores of the various clinical specialties, and scores from participants from high- and low–middle-income countries. Potential attrition bias was assessed by comparing median scores of those who completed all 3 rounds with those who did not. We used SPSS for Windows version 28.0 (IBM corp., Armonk, New York) for all analyses, and performed multiple Mann-Whitney U tests. We adjusted for multiple testing using a Bonferroni correction, setting the significance level at P < .006 (2-sided).

In the initial round of the Delphi study, 100 participants enrolled. The enrollment period, initially set at 4 weeks, was extended to 3 months to facilitate the inclusion of additional participants. Participant inclusion is summarized in Table 1 and Supplemental Table 5. At the conclusion of round 1, consensus in was reached on 3 outcomes: recurrent NEC, the need for reoperation, and quality of life. Participants suggested 67 outcomes to add to the initial selection (Supplemental Table 6). After removing overlapping outcomes, all outcomes were discussed with patient organizations Care4Neo and NEC United Kingdom. All outcomes mentioned at least twice were discussed by the steering group, including a family representative, and voted on anonymously using a 9-point Likert scale. Three outcomes reached consensus and were added to round 2: growth after hospital discharge, intestinal problems, and duration of parenteral nutrition at home (Supplemental Table 7). The full list of outcomes discussed, as well as the scores of the steering group members, is included (Supplemental Table 8).

The second round was open for 5 weeks instead of the originally specified 4 and 48 outcomes were scored (Supplemental Table 7). Eighty-eight of the initial 100 participants completed their entries. After round 2, 6 outcomes met the consensus in criteria: behavioral disorders, the need for reoperation, neurodevelopmental impairment, quality of life, recurrent NEC, and residual intestinal length. No outcomes were dropped after round 2, because none reached the consensus out criteria.

Because no candidate outcomes were removed, the same outcome list as in round 2 was used. Seventy-one participants completed this round, including 3 researchers, 12 pediatric surgeons, 3 pediatric gastroenterologists, 38 neonatologists, 3 (former) NEC patients, and 12 family representatives. These participants were from 25 different countries, with the majority (78%) originating from high-income countries and an equal distribution of male and female participants. In terms of professional experience, most participants reported having 11 to 20 years of experience working with NEC, and clinicians typically saw 11 to 20 cases each year. The group of NEC patients (n = 3) and family representatives (n = 12) was composed of individuals from the United Kingdom, the Netherlands, and Norway, with a significant majority (93%) being female. The full list of scores can be found in Supplemental Table 9. In the group of the NEC patients and family representatives, 16 outcomes reached consensus. Outcomes #9 to #13 reached the same score, with 80% ≥7. Because no distinction could be made between the 5 equal scores, all 13 outcomes were tabled for discussion in the consensus meeting. Outcomes #14 to #16 (behavioral disorder, intraventricular hemorrhage, stricture) achieved consensus with ≥7 from 73% of the lay panel. However, because our protocol stated we would include a maximum of 10 outcomes per stakeholder group, these were not discussed in the consensus meeting. In the combined expert group, including both neonatal and nonneonatal experts, 6 outcomes reached consensus; these outcomes showed a 100% overlap with the consensus in outcomes from the NEC patient and family representative group.

The attrition rate for this COS was 29%, with a similar distribution among stakeholder groups (Table 1). The mean round 1 scores of participants who did not complete subsequent rounds were higher compared with participants who did complete all rounds (mean rank 3073 vs 2905, P < .001). However, no differences in scores were observed when comparing participants within stakeholder groups. The mean round 2 scores of participants who did not complete subsequent rounds were lower compared with participants who did complete all rounds (mean rank 2299 vs 2697, P < .001), with a similar pattern within the neonatal stakeholder group (mean rank 1745 vs 2023, P < .001).

Initially, 13 outcomes were eligible for discussion in the consensus meeting (Fig 1, Supplemental Table 10). Despite not reaching consensus in the third Delphi round, 2 additional outcomes were included in the discussions after unanimous agreement during the consensus meeting: growth and time to full enteral feeding.

The consensus meeting was held with 27 participants, including surgeons, neonatologists, a pediatric gastroenterologist, family representatives, and a NEC patient (Table 1). Participants from both high- and low–middle-income countries were present. Of the 15 outcomes discussed in the meeting, 5 outcomes reached consensus after a final vote and were included in the final core outcome set (Fig 2). The abbreviated notes from this meeting can be found in Supplemental Information Methods 3.

The COS for NEC consists of 5 outcomes from 3 areas: mortality, NEC-related mortality, short bowel syndrome, neurodevelopmental impairment, and quality of life. Recommendations on how to define these outcomes and when to measure them are included in Table 2.

Mean scores were calculated on the basis of the combined scores for all outcomes in round 3. Family representatives and patients had a higher mean score than the combined expert group (mean rank 2031 vs 1630, P < .001). Within the combined expert group, neonatologists scored similar compared with pediatric surgeons (mean rank 1200 vs 1183, P = .604), but pediatric surgeons also reached consensus on the need for reoperation. Seventeen participants (24%) were from low–middle-income countries, and their mean outcome scores were higher compared to participants from high-income countries (mean rank 1804 vs 1687, P = .003). Their most important outcomes also differed from those rated highly by participants from high-income countries (Table 3) and included all outcomes in the final COS except for quality of life. Instead, they also reached consensus on receiving breast milk, residual intestinal length, need for reoperation, and sepsis. No other groups were compared because of low sample size.

This comprehensive COS for NEC includes outcomes important to both patients and physicians, from diagnosis to adulthood. It consists of 5 outcomes we recommend including in all future NEC treatment trials: mortality, NEC-related mortality, short bowel syndrome, quality of life, and neurodevelopmental impairment. In developing this COS, we addressed and improved upon shortcomings identified in previous pediatric COS initiatives by actively incorporating the perspectives of infants and their families, prioritizing transparency, and adhering to reporting guidelines.^26 ,27  This COS includes outcomes from 3 core areas.^23 ,24 

The most reported outcome in recent NEC trials is mortality, but prolonged total parenteral nutrition use and neurodevelopmental impairment have also been reported with high frequencies.^8 ,9  Quality of life was notably absent in recent NEC trial reporting, underscoring the need for the inclusion of long-term outcomes.^9 ,11  Our collaboration with many neonatal and NEC societies ensured the active involvement of patients and family members in developing this COS. Their unique insights contributed to more outcomes discussed in the consensus meeting and the inclusion of 3 predominantly long-term outcomes. This aligns with a recent qualitative study by the NEC Society urging for more long-term studies,¹¹  as well as other pediatric COS efforts that include a significant number of long-term outcomes.^28 ,29  This COS includes short bowel syndrome, which aligns with NEC trials’ frequent reporting of intestinal failure and duration of parenteral nutrition.^8 ,9  Recent trials do not often report NEC-related mortality. However, its consideration as a separate outcome is crucial, given the high baseline mortality risk of the preterm population, and that the interventions aim to reduce NEC-related mortality.⁸  Various definitions of NEC-related mortality are currently in use. For this COS, outcome definitions were taken from NEC literature, identified in a systematic review and recent guidelines, for all outcomes.^{8 ,11 ,25 ,30} 

Despite its strengths, this COS is not without limitations. Notably, there was a lower percentage of parental participation than aimed for. This phenomenon has been observed in other COS development efforts, highlighting the challenge of enrolling and sustaining lay stakeholder engagement.^25 ,31 ,32  To overcome this, we recommend a focus group before the start of Delphi round 1, in addition to the early involvement of neonatal societies. This focus group may also be involved in formulating future recommendations for core outcome definitions and for how and when to measure these; only our steering group members and one family representative worked on this to date. We also notice the lack of nonneonatal expert involvement, prompting the fusion of the neonatal and nonneonatal groups before the consensus meeting. Yet, considering the various different health care systems worldwide, many neonatal experts are also involved in long-term care for infants with NEC, potentially defeating the purpose of distinguishing between a true neonatal and nonneonatal group. Finally, the mean round 2 scores of participants who did not complete the subsequent round were significantly lower compared with participants who completed all rounds, alluding to potential attrition bias. We believe that this attrition did not result in the loss of critical outcomes discussed during the consensus meeting because the composition of stakeholder groups remained consistent throughout the Delphi rounds and the outcome selection for the consensus meeting was based on the highest-scoring outcomes from both the lay and combined expert groups. Mean outcome scores from participants from low–middle-income countries were higher and more short-term outcomes reached consensus in this group. However, there was limited involvement from these regions, particularly among family representatives, potentially because of language barriers. This highlights the need to explore how to further enable participation in COS development in low–middle income settings and ensure a COS that better reflects the burden of NEC in these contexts.³³  Finally, 2 outcome areas were not represented: Resource use and adverse events.²⁴  As a supplement, other COS could be used to include outcomes from these areas, such as the intestinal failure or neonatal research COS.^25 ,32 

Future efforts should aim to enhance the inclusivity of this NEC COS by incorporating input from family representatives in low–middle-income settings and engaging a broader spectrum of stakeholders, including industry representatives and policymakers.²⁷ 

This COS for NEC represents a first step in standardizing outcome reporting. It highlights the need for long-term follow-up and ensuring that the experiences of patients and their families are considered. When implemented by the research community in future treatment trials, we hope this COS will facilitate meaningful comparison between studies, enable meta-analyses, and foster the development of evidence-based guidelines, to eventually positively impact the care of infants with NEC.

First, we thank all NEC and neonatal societies for their help in facilitating patient and family representative involvement in this study: NEC United Kingdom Charity (M. Spruce) and Brazil Instituto Pequenos Grandes Guerreiros (S. Rosito), the Dutch Neonatal Patient and Parent Advocacy Organization Care4Neo (M.J. Vermeulen and S.A. Obermann-Borst), the European Foundation for the Care of Newborn Infants (S. Mader and C. Tischer), Tunisian Neonatal Association (A. Ksia), Uganda Preterm Infant Parents Network (K. Bazilio), and the Adult Preemie Advocacy group (L. Ingledow and J. Lee). Second, we thank our participants (names only published with individual’s permission): Ruth Guinsburg, Daynia Elizabeth Ballot, Peter Davis, Patrick J Javid, Jan Hulscher, Kristine Lindemann, Shripada Rao, Fumihiko Namba, Mikko Pakarinen, Brian Reichman, Claire Radford, Janet Berrington, Christoph Binder, Amine Ksia, Anne Synnes, Jessie M Hulst, Larissa Eliana Genes, Merit Tabbers, Andrea Conforti, Elzbieta Verdigi, Giacomo Cavallaro, Melantha Coetzee, Simon Lam, Roel Bakx, Praveen Kumar, Nigel Hall, Satoshi Kusuda, Yaron Avitzur, Sanjay Keshav Patole, Marie Jose Butel, H. Tolga Çelik, Ingo Jester, Jeanette Wiltshire, Linseigh Green, Ee-Kyung Kim, Keith Barrington, Arend Bos, Kate Smedley, Arianna Aceti, Augusto Zani, Marie Spruce, Walusa Assad Gonçalves-Ferri, Hala Chaaban, Aloka Patel, Rebecca Parry, Elisabeth Kooi, Beatrice Ezenwa, Neena Modi, David Bader, Daniëlle Wouters, Lotte Becker, Simon Eaton, Martin Offringa, Lizelle Van Wyk, Omer Erdeve, Setya Dewi Lusyati, Martin Lacher, and Thomas Spruce.

COS

core outcome set

NEC

necrotizing enterocolitis