A Core Outcome Set for Children With Feeding Tubes and Neurologic Impairment: A Systematic Review
CONTEXT: Uncertainty exists about the impacts of feeding tubes on neurologically impaired children. Core outcome sets (COS) standardize outcome selection, definition, measurement, and reporting.
OBJECTIVE: To synthesize an evidence base of qualitative data on all outcomes selected and/or reported for neurologically impaired children 0 to 18 years living with gastrostomy/gastrojejunostomy tubes.
DATA SOURCES: Medline, Embase, and Cochrane Register databases searched from inception to March 2014.
STUDY SELECTION: Articles examining health outcomes of neurologically impaired children living with feeding tubes.
DATA EXTRACTION: Outcomes were extracted and assigned to modified Outcome Measures in Rheumatology 2.0 Filter core areas; death, life impact, resource use, pathophysiological manifestations, growth and development.
RESULTS: We identified 120 unique outcomes with substantial heterogeneity in definition, measurement, and frequency of selection and/or reporting: “pathophysiological manifestation” outcomes (n = 83) in 79% of articles; “growth and development” outcomes (n = 13) in 55% of articles; “death” outcomes (n = 3) and “life impact” outcomes (n = 17) in 39% and 37% of articles, respectively; “resource use” outcomes (n = 4) in 14%. Weight (50%), gastroesophageal reflux (35%), and site infection (25%) were the most frequently reported outcomes.
LIMITATIONS: We were unable to investigate effect size of outcomes because quantitative data were not collected.
CONCLUSIONS: The paucity of outcomes assessed for life impact, resource use and death hinders meaningful evidence synthesis. A COS could help overcome the current wide heterogeneity in selection and definition. These results will form the basis of a consensus process to produce a final COS.
- COMET —
- Core Outcome Measures in Effectiveness Trials
- COS —
- core outcome set
- G —
- GJ —
- OMERACT —
- outcome measures in rheumatology
- PRISMA —
- preferred reporting items for systematic reviews and meta-analyses
- PROM —
- patient-reported outcome measure
Children with neurologic impairment, such as cerebral palsy or trisomy 21, often have complex health care needs and account for a large proportion of pediatric health care costs.1 These children are at increased risk of feeding problems owing to self-feeding impairment and oromotor dysfunction, often necessitating a reliance on gastrostomy (G) or gastrojejunostomy (GJ) tube feeding.2,3 Parent descriptions of child experiences include needing assistance with feeding (89%), choking with food (56%), vomiting (22%), aspiration pneumonia (31%), and prolonged feeding time (28%).2 Parents of many of these children also describe feeding as being a stressful and unenjoyable experience. Up to one-half of this population is significantly undernourished with resulting growth failure.4
Although permanent feeding tubes, such as G-/GJ-tubes, have been shown to improve weight gain among children with neurologic impairment, there is a degree of uncertainty regarding their impact on a variety of other outcomes. The literature has called for additional research into G-/GJ-tube outcomes, such as survival, aspiration and respiratory infections, health care use and cost, feeding times, caregiving burden, and child and primary caregiver quality of life.4–6 Substantial heterogeneity in outcome terminology, definition, and measurement among empirical studies contributes to this uncertainty about G-tube feeding efficacy.5–11 The lack of standardized outcomes and the associated inadequate information on the risks and benefits of treatments impair the decision-making process and may underlie variability in practice.
Methods have been developed to address the lack of selection, measurement, and reporting of meaningful outcomes across many clinical areas, including the creation of a core outcome set (COS), a consensus-derived, standardized minimal collection of outcomes.12,13 Moreover, evidence suggests that patient-reported outcome measures (PROMs) enable patient-centered care that guides the clinical decision-making process.14,15 Therefore, COS should be a composite of both clinically observed indicators as well as PROMs, as recommended in a number of recent outcome initiatives, such as Outcome Measures in Rheumatology (OMERACT) and Core Outcome Measures in Effectiveness Trials (COMET).13,15
The OMERACT Filter 2.0 is a framework for researchers to determine what outcomes should be measured, how they should be measured, and for deciding which outcomes should be selected when creating a COS. Envisioned in 1992 and continually developed through consensus-driven discussion, OMERACT was established as a procedural framework for the identification and selection of valid, responsive, and feasible rheumatologic outcomes in clinical trials.16 In recent years, however, it has been made as generic as possible so that it can be used in other disease areas.15 The COMET initiative, whose goal is to spur COS development across all possible areas of health research and clinical decision-making, has since collaborated with and endorsed OMERACT.13,15 Key reasons for using OMERACT for COS development include the methodological strength of its evidence and consensus-based techniques and the paucity of empirical, consensus-driven alternatives.12,15 Together, these initiatives encourage the selection of outcomes that are important to a wide range of stakeholders, including clinicians, patients, and policy makers, such that the resulting research is relevant and meaningful.
The OMERACT and COMET initiatives both recommend that one of the initial steps to COS development is a systematic review to identify outcomes already selected and/or reported in the literature.13,15 In doing so, the systematic review can provide evidence for the need of a COS and identify a list of potential outcomes and their corresponding measurement tools.12,13,15 As the first stage of a process to develop a COS, we undertook a systematic review to synthesize qualitative data on all outcomes included in English-language publications investigating the health of children (≤18 years) living with neurologic impairment and dependent on G-/GJ-tubes.
This review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for systematic reviews17 and according to a defined protocol. The study has been registered with the COMET registry (http://www.comet-initiative.org/studies/details/736).
We included peer-reviewed articles focusing on health outcomes of children aged ≤18 years living with neurologic impairment who are dependent on G-/GJ-tubes. Neurologic impairment was defined as a “group of congenital or acquired long-term conditions that are attributed to impairment of the brain and/or neuromuscular system and create functional limitations. A specific diagnosis may not be identified. Conditions may vary over time, occur alone or in combination, and include a broad range of severity and complexity.”18
Gray literature, editorials, conference abstracts, and commentaries were excluded. Studies that focused on any other method of feeding and those with a lack of recorded outcomes after G or G-incorporated tube placement were excluded as well. Case studies (ie, studies of only 1 patient) were excluded from the review because of their small sample size.
In April of 2015, a literature search of peer-reviewed articles was performed by a research assistant (C.B.). Searches were conducted within the following databases: Medline (1946–April 2015), Embase (1947–April 2015), and Cochrane Central Register of Controlled Trials (1991–April 2015). Database-specific subject headings were targeted and exploded to focus on children with G-/GJ-tubes living with neurologic impairments. The 3 distinct concepts related to population (eg, “infant,” “child,” and “adolescent”), intervention (eg, “gastrostomy,” “feeding tube,” and “ostomy”), and diagnosis (eg, “neurologic disease,” “neurodisability,” and “nervous system disorder”) were combined with the Boolean “AND” in Medline, Embase, and Cochrane Central Register of Controlled Trials. No date or year restrictions were applied during the literature search stage, although the search was restricted to English language publications. Our search strategy is displayed within the Supplemental Information.
C.B. scanned titles and abstracts for potential outcomes associated with children with any form of feeding tubes and neurologic impairment. C.B. also screened the full set of abstracts, whereas 2 pediatricians with expertise in management of feeding tubes for children with neurologic impairment (E.C. and S.M.) joined C.B. as coreviewers in duplicate screening of a randomly selected proportion of abstracts (40%). The final set of articles that passed the screening round were reserved for full-text review (performed by C.B. and K.C.J.), which ensured that articles met the inclusion criteria. Articles that passed full-text inspection were retained for data extraction. When necessary, a third reviewer (M.Z.K.) resolved any disagreements that arose during this process.
Data extraction was performed by 2 reviewers (C.B. and K.C.J.) independently and in duplicate. Any disagreements that arose during this process were resolved through discussion, with a third reviewer requested (M.Z.K.) if consensus could not be reached. The data collection form required the following data to be extracted from each article: year of publication, country of origin, study design, sample size, age range of study participants, diagnoses of study participants, type of tube procedure (ie, G or GJ), presence of buttons (ie, low-profile stoma-entry devices that can be sealed when not in use), all included (ie, selected and/or reported) outcomes and definitions, and who observed the outcome (ie, child, parent, or health professional/researcher).
Because the aim of our systematic review was to identify heterogeneity of reporting regardless of the study design (whether primary research or research synthesis), each article was treated as a unique study. Qualitative data on all outcomes were extracted, regardless of whether they were mentioned only in Methods, Results, or both sections of the article. All outcomes were recorded descriptively because our objective was to characterize the heterogeneity in outcome reporting rather than to perform meta-analyses on quantitative outcome data.
Because it was not always possible to determine whether outcomes were associated or caused by insertion of the G-/GJ-tube, any outcomes that were recorded after tube insertion were included. Key attention was paid to the existence of dichotomous outcomes relative to continuous outcomes (eg, obesity and malnutrition versus weight). Outcomes were assigned to be either primary or secondary outcomes. Primary outcomes were identified as those that were (1) stated as being a primary outcome in the article; (2) stated as being the outcome of interest when investigating the effects of an intervention or procedure; and (3) explicitly referred to by title or rationale. Those not fulfilling these criteria were assigned to the category of “secondary outcome.”
Given previous evidence from COS development in other populations (eg, children with appendicitis), we anticipated that extracted outcomes would vary in their terminology and how they were assessed.19 Therefore, to ensure consistency throughout our process, we merged outcomes with similar definitions or themes regardless of their terminology. For example, child growth as an outcome can be defined and measured through study-specific outcomes (eg, weight-for-age z scores), generic outcomes (eg, growth), and abstract constructs (eg, parent perception of their child’s weight). In such a situation, these outcomes would be reclassified and merged under terms that reflected the underlying concept behind the outcome (ie, weight or height).
Assignment of Outcome Terms to OMERACT Core Areas
The OMERACT Filter 2.0 encourages the comprehensiveness of a COS by specifying all key aspects of a health condition in each of 4 core areas, namely: (1) death, (2) life impact, (3) resource use, and (4) pathophysiological manifestations.15 Given our focus on pediatric populations, we added a fifth core area called (5) growth and development.20
The first 3 core areas (ie, death, life impact, and resource use) are meant to describe the impact of a health condition on a broad range of stakeholders, whereas pathophysiological manifestations account for the biophysiological impact of a health condition.15 The growth and development core area describes the impact of a health condition on the natural growth and development of a child.20 Outcome terms (what to measure) and their corresponding measurement tools (how to measure) were mapped onto each of the 5 core areas. For example, any outcome that was explicitly associated with child growth and development over time was placed in the growth and development core area (eg, physical growth [measured via measurements of weight, height, body part circumferences, and lengths], neurocognitive development [measured by anecdotal reports], and puberty [measured by the Tanner scale]). Outcome terms were assessed and mapped onto core areas independently and in duplicate by 2 reviewers (C.B. and K.C.J.) with a third clinician reviewer resolving discrepancies, if necessary (M.Z.K.).
Assessment of Outcome Reporting
To assess heterogeneity in outcome selection and/or reporting among articles rather than outcome data, we employed visual and qualitative inspection of an outcome matrix13,19 instead of parametric tests. Outcome matrices are graphical representations of outcome inclusion among a series of articles, whereby each reviewed article is stacked across the x-axis, and all outcomes are listed on the y-axis. In each instance where an outcome is selected or reported in an article, the cell is colored in blue (for a secondary outcome) or red (for a primary outcome). By browsing horizontally, one can visually see the frequency, consistency, and disparity of outcome inclusion across articles. By browsing vertically, one can see which outcomes are included in a single article.
Identification of Included Articles
Figure 1 displays the process used to identify and select articles for this systematic review. The initial search identified a total of 9465 articles (2539 from Medline, 6805 from Embase, and 121 from the Cochrane Review database). After deduplication, the remaining 8790 records were screened on the basis of title, abstract, and availability of included outcomes, resulting in a total of 196 full-text articles to be assessed for eligibility for inclusion in the review.
Full-text assessment identified 104 articles5–7,9,21–120 for which data extraction was possible; the characteristics of which are summarized in Table 1. The overall rate of disagreement was low; a third reviewer was requested for <5% of the screened articles. The majority of these articles (78%) were cohort studies,5,6,9,21,24–27,29–43,45–47,50–52,54,57–64,66–70,72,73,75–79,82,84–88,90,92,94,98–118,120 followed by cross-sectional studies (9%),23,44,53,65,71,74,80,81,83 systematic reviews (5%),7,49,89,96,97 and controlled trials (5%).22,28,48,55,91 The 5 systematic reviews performed analyses on 0 to 25 articles each. Years of publication were from 1969 to 2014, and the median total sample size among our included texts was 51 (range, 3–24 920 children). A plurality of studies was conducted in the United States (26%), followed by the United Kingdom (21%) and Canada (14.5%).
Identification of Outcomes
Initially, 487 outcomes were observed among 104 articles. After combining outcomes with similar definitions, there were 82 different, unique outcome terms. Although complications were initially extracted as a singular outcome rather than a category of outcomes, the goal of this review was to identify all outcomes. Therefore, all outcomes grouped under complications were subsequently extracted, resulting in an additional 405 complication outcomes being extracted from 47 publications.
Combining complication outcomes with similar definitions resulted in the 76 outcome terms seen in Table 2, 38 of which had already been identified as outcomes, and an additional 38 of which were different and unique from the 82 outcomes identified earlier. Therefore, a total of 120 unique outcome terms were identified across all articles that were included in this review. All disagreements were resolved through consensus-by-discussion; a third reviewer was never needed.
Assignment of Outcomes to Core Areas and Assessment of its Heterogeneity
Each of the 120 unique outcome terms is displayed in Table 3. They have been assigned to the 5 core areas of the modified OMERACT Filter 2.0. The most populated core area was pathophysiological manifestations, with 83 outcome terms found in 79% (n = 82) of articles, 60% of which considered it as a primary outcome of interest. The majority of these outcomes had been previously categorized as complications by authors (Table 2). The growth and development core area, with 13 outcome terms, was included in 55% (n = 57) of articles, 46% of which considered it to be a primary outcome. The death core area, of which there were 3 outcome terms, was included in 39% (n = 41) of articles overall and considered a primary outcome in 42% of articles. The life impact core area, which included 17 outcome terms, was included in 37% (n = 38) of articles and considered a primary outcome in 55% of these. The majority of life impact outcomes (15 out of 17) were assessed by primary caregivers and/or patients. Finally, the resource use core area, of which there were 4 outcome terms, was included in 14% (n = 14) of articles and referred to as a primary outcome in 29% of these articles.
Assessment of Core Area Reporting Over Time
The distribution of primary and secondary outcome terms for the 5 core areas across the individual articles is displayed in the outcome matrices (Figs 2, 3, 4, 5 and 6). Articles have been sorted by publication date, with the oldest article (1969) to the left and the most recent articles (2014) to the right. The frequency at which that outcome was included across all articles is displayed in the right-most column. We hypothesized that trends across years would exist in 3 areas: (1) increase in overall frequency of outcome selection and/or reporting over time; (2) increase in reporting on whether they were included as primary or secondary outcomes over time; and (3) increase in number of core areas represented over time. No trends were found on visual inspection of the outcomes matrix in terms of frequency, classification as primary or secondary, and core area representation.
Heterogeneity in the Reporting of Individual Outcomes
The most frequently selected and/or reported outcome was weight (n = 52), which was reported as a primary outcome in 44% of these articles. The second- and third-most frequently included outcomes were gastroesophageal reflux (n = 36) and height (n = 27): considered as primary outcomes in 67% and 56% of these articles, respectively. Inclusion of the most frequently observed outcomes together in any 1 article was uncommon. For example, only 7 of 104 articles (7%) recorded all 3 of the most frequently included terms (ie, weight, gastroesophageal reflux, and height).
Heterogeneity in Outcome Measurement and Definition
We evaluated the measurements and definitions for outcomes assigned to the life impact, growth and development, and resource use core areas (Tables 4, 5, and 6). Here, we did not further evaluate death or pathophysiological manifestation outcomes. This was because death was a standalone objective measure; any additional definitions and measurement tools were not mentioned in the articles. Furthermore, pathophysiological manifestation outcomes were generally laboratory measurements; operationalization of each was beyond the scope of this review.
Examination of Tables 4 through 6 demonstrates that among outcomes, there was a high degree of variability in how they were measured and/or defined. For example, weight gain/loss was measured using 14 different outcome measurement tools (eg, weight-for-age, weight/height ratio, semistructured interview). Body part lengths were also observed by using 14 different outcome measurement tools (eg, right tricep circumference, tibial length, head circumference). It was common for outcomes in the life impact and growth and development core areas to be measured and defined in multiple ways.
We have analyzed outcome reporting from a large number of articles relating to the treatment of children with neurologic impairment living with G-/GJ-tubes. We have identified a substantial degree of heterogeneity in the reporting of core areas, the reporting of individual outcomes, and in outcome measurement and definition. Our systematic review found that articles mainly focus on pathophysiological manifestation and growth and development outcomes, whereas death, life impact, and resource use outcomes are underrepresented. The inconsistency in outcome selection, measurement, and reporting disallows direct comparison among articles and makes it difficult to determine the overall impact of G-/GJ-tubes in this population.
Representation of Core Areas
The underrepresentation of the life impact and resource use core areas suggests that all stakeholder interests, including patients, families, and other nonclinical experts, are not being collected and subsequently addressed. This finding suggests that although these core areas encompass issues that families feel are important to them, clinicians and researchers may in fact be undervaluing these issues altogether. It may also be because of the paucity of measurement tool standardization identified by our review.
Life impact outcomes are important because they represent the quality of life of the child and family, an integral component of a child’s health and well-being. Improvements in patient care can occur through observing these outcomes and addressing prominent themes throughout the care process.3 Given that patient and primary caregiver perceptions and perspectives do not always coincide with that of health care providers, it is imperative that life impact outcomes be measured through PROMs.5,121,122
Resource use outcomes are crucial because they describe the extent to which the existence of a feeding tube impacts the individual, their family, and the health system.15 For example, in North America, the cost of home care by the primary caregiver for a child with a G-/GJ-tube ($37 232) is more than twice the cost of that associated with caring for a child without a G-/GJ-tube ($15 004).123 In addition, the presence of multiple chronic complex care conditions can increase costs markedly, reaching up to $86 181.1 Because our target population requires substantial resources, more work needs to be done to identify additional hidden societal costs and determine which interventions and resources can maximize child, family, and health care system outcomes in a cost-effective manner.1,123
In contrast, growth and development outcomes were well represented. Over half of the articles included outcomes assigned to this proposed core area, suggesting that growth and development should be considered as a fifth core area in the pediatric articles by using the OMERACT filter. This core area captures growth and development not just associated with feeding tubes, but also the natural biophysical and neurologic development that occurs with age progression in children. Altogether, this suggests that there may be specialty-specific or age-group–specific factors that impact the validity of the OMERACT filter when employed in fields other than that for which it was originally proposed (ie, rheumatology). In this context, the monitoring of nutrition status, growth, and body composition allows health care providers to assess and address the needs of this population, while minimizing malnutrition and obesity.11,124
Outcome Definitions, Measurement, and Reporting Patterns
Variability in how frequently outcomes are observed, defined, and measured has broad implications for understanding the impact of G-/GJ-tubes for children with neurologic impairment, their families, and the health care system at large. This finding has been described in other populations, including children with appendicitis,19 esophagectomy,125 inguinal hernia repair,126 and pediatric acute diarrheal diseases.127 Taken together, this heterogeneity and variability provide a strong rationale for the creation of a COS for children with neurologic impairment living with G-/GJ-tubes. Without standardized reporting guidelines, the literature will continue to be nebulous, making it difficult to compare and contrast results and to evaluate the impacts of G-/GJ-tube placement on a variety of outcomes.
Strengths and Limitations
Our work has several strengths. First, it is based on an extensive literature search that employed numerous bibliographical databases (ie, Medline, Embase, Cochrane). Moreover, by not discriminating based on publication date, we were able to include publications from 1969 to 2014. In addition, the high reviewer agreement in our review suggests that there was a large degree of consistency throughout the evidence synthesis process. This study also captured a wide range of outcomes, as well as the ways in which they were both measured and defined. Additionally, by following an established systematic review methodology (ie, PRISMA), we were able to identify all relevant pediatric articles needed to extract the necessary information. Finally, the usage of the OMERACT Filter 2.0 allowed us to generate a comprehensive list of outcomes to inform the next stages of COS development.
The goal of our study was to extract selected and/or reported outcomes, so quantitative data collection was not within the scope of the work. We therefore used frequency counts and qualitative assessment of outcome matrices to assess heterogeneity, rather than the I2 statistic. The lack of quantitative data limited our ability to assess outcome severity, as well as our ability to characterize the strength of association or clinical significance of outcomes relative to G-/GJ-tubes. Even if we had collected quantitative data, we feel that the heterogeneity in outcome reporting, definitions, and measurement tools would have made quantitative pooling of the data unlikely. This therefore precludes the ability to draw conclusions about the strength of association between G-/GJ-tubes (versus no G) and potential outcomes.
Frequency of outcome selection and/or reporting should not be considered synonymous with the severity or importance of an outcome. For example, skin irritation was 5 times more likely to be selected and/or reported within included articles than perforation of stomach wall, despite the fact that the latter would have a larger impact across all core areas. We plan to address this shortcoming by having multiple stakeholders quantitatively rank the importance of outcomes in the second phase of this program of research.
In addition, the literature did not often contain neurologically normal comparison groups or explicit information on whether included outcomes were the rationale for G-/GJ-tube insertion or the product of insertion. We therefore recorded any outcomes, provided that they were recorded after the insertion of the G-/GJ-tube. Finally, by restricting our search strategy to English-only articles, we may not have been able to capture outcomes that are commonly included in a broader sample of articles.
Our findings suggest that there is a need for increased emphasis on life impact and resource use outcomes. It reinforces calls for comprehensive approaches to health outcomes assessment (ie, including life impact outcomes) as per World Health Organization standards for comprehensive health evaluation.128 It also provides support for calls to bolster medical education with awareness and information on the economic realities facing patients and families throughout their care.129 In the context of our target population, we believe that the COS can further the agenda of comprehensive health outcomes assessment and education.
This article was the first step in the creation of a COS for feeding tube–dependent children with neurologic impairment. Similar to COS development in other populations, the OMERACT Filter 2.0 has been identified as a strong framework for the extraction of outcomes to be used in a COS, as well as to organize and present the final COS.19 Now that the evidence synthesis is complete, all 120 outcomes extracted through this systematic review will form the foundation of the development of a COS.
As detailed in the COMET registry, the next step will be to perform a 2-stage Delphi survey, where health care providers, methodologists, and primary caregivers (eg, parents) will assist in identifying a smaller set of meaningful outcomes across all OMERACT core areas.12 The final step of this process will be a meeting of stakeholders who reach consensus on a minimal set of outcomes to measure in children with neurologic impairment who are dependent on G-/GJ-tubes. This final COS will provide a list of standardized outcomes and measurement tools to guide both research and clinical decisions. It will also encourage clinicians and researchers to observe outcomes relevant to all stakeholders (eg, life impact and resource use outcomes) during the clinical decision-making and study design processes.
We found 120 unique outcomes across 104 publications. There was a large degree of heterogeneity, inconsistencies, and deficiencies in outcome selection, definition, measurement, and reporting. Existing outcome reporting is biased toward clinician and researcher foci of interest rather than relevant outcomes for patients/parents. Additionally, a large number of outcomes fell outside the scope of the OMERACT 2.0 framework, suggesting the need for a growth and development core area when assessing outcomes in this population. Our results provide a strong basis for the construction of a COS for feeding tube–dependent children with neurologic impairment.
- Accepted April 27, 2016.
- Address correspondence to Mufiza Kapadia, MD, MSc, PhD, Child Health Evaluative Sciences, Peter Gilgan Centre for Research and Learning, The Hospital for Sick Children, 686 Bay St, Toronto, ON M5G 0A4, Canada. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Funded by grant 1000041135 from the Norman Saunders Complex Care Initiative Grant Competition 2014. Dr Guttmann receives salary support from a Canadian Institutes for Health Research Applied Chair in Child Health Services Research.
POTENTIAL CONFLICT OF INTEREST: The authors have no indicated they have no potential conflicts of interest to disclose.
- Cohen E,
- Berry JG,
- Camacho X,
- Anderson G,
- Wodchis W,
- Guttmann A
- Nelson KE,
- Lacombe-Duncan A,
- Cohen E, et al.
- Mahant S,
- Friedman JN,
- Connolly B,
- Goia C,
- Macarthur C
- Gantasala S,
- Sullivan PB,
- Thomas AG
- Johansen MA,
- Berntsen GKR,
- Schuster T,
- Henriksen E,
- Horsch A
- Åvitsland TL,
- Birketvedt K,
- Bjørnland K,
- Emblem R
- Cantez MS,
- Gerenli N,
- Ertekin V,
- Durmaz O
- Cheung KM,
- Lo CH,
- Tse PWT,
- Chan KH
- Friedman JN,
- Ahmed S,
- Connolly B,
- Chait P,
- Mahant S
- Kang A,
- Zamora SA,
- Scott RB,
- Parsons HG
- Kehar M,
- Wadhwa N,
- Upadhyay S
- Kong CK,
- Wong HS
- Launay V,
- Gottrand F,
- Turck D,
- Michaud L,
- Ategbo S,
- Farriaux JP
- Martínez-Costa C,
- Calderón C,
- Pedrón-Giner C,
- Borraz S,
- Gómez-López L
- Mollitt DL,
- Golladay ES,
- Seibert JJ
- Naureckas SM,
- Christoffel KK
- Rempel GR,
- Colwell SO,
- Nelson RP
- Seguy D,
- Michaud L,
- Guimber D, et al
- Sharma R,
- Williams AN,
- Zaw W
- Sleigh G,
- Brocklehurst P
- Strauss D,
- Kastner T,
- Ashwal S,
- White J
- Sullivan PB,
- Morrice JS,
- Vernon-Roberts A,
- Grant H,
- Eltumi M,
- Thomas AG
- Toporowska-Kowalska E,
- Gębora-Kowalska B,
- Jabłoński J,
- Fendler W,
- Wąsowska-Królikowska K
- Townsend JL,
- Craig G,
- Lawson M,
- Reilly S,
- Spitz L
- Morris C,
- Janssens A,
- Allard A, et al
- Mascarenhas MR,
- Meyers R,
- Konek S
- Johnston BC,
- Shamseer L,
- da Costa BR,
- Tsuyuki RT,
- Vohra S
- Copyright © 2016 by the American Academy of Pediatrics