Gastrostomy Tube Use in Pediatrics: A Systematic Review

These guidelines apply throughout the process of enteral access decision-making, diagnostic workup, device insertion, and follow-up with patient and family well-being as the central unit of concern. They are intended to be broadly relevant to all health care providers (medical specialists, proceduralists, and primary care providers) and practice contexts (academic institutions, community hospitals). Management of gastroesophageal reflux disease tangential to enteral tube placement is out of scope. Further, the authors acknowledge that the provision of individualized care cannot be easily reflected in guideline recommendations but is paramount in every child with barriers to meeting their caloric needs by mouth.

The 3 primary authors (L.B., A.G., and R.B.) assembled a multidisciplinary group, including pediatric gastroenterology (A.S. and R.R.), neonatology (L.F. and J.C.), interventional radiology (IR) (J.A.), and general pediatrics (M.P. and M.C.). Allied health service providers also contributed to guideline generation: dietitian (J.L. and J.S.), speech/feeding therapist (M.M.), and advanced practice nurse (D.J.). This ad hoc group was assembled to form recommendations on the basis of the findings of the systematic review.

A modified Delphi needs assessment¹  of impactful questions regarding enteral access in children was performed with the membership of the American Pediatric Surgical Association Outcomes and Evidence-Based Practice committee. Four overarching questions and secondary subquestions were identified and formed the basis of individualized search strategies (Supplemental Information Methods 1).

A medical librarian oversaw a search of PubMed, Google Scholar, Medline, and Scopus between January 1, 1999, and January 31, 2020. Each question considered for review included a combination of the search terms: “gastrostomy,” “g-tube,” and “tube feeding.” Additional search terms were developed based on the evaluation and discussion of each of the 4 questions with content experts and are provided (Supplemental Information Methods 2). Results were limited to the pediatric population (birth–18 years of age) and English language articles. Inclusion/exclusion criteria were tailored for each section (Supplemental Information Methods 3); Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed.² 

Level of evidence was assigned to included manuscripts on the basis of Oxford Centre for Evidence-Based Medicine Levels of Evidence.³  Risk of bias was assessed using the ROBINS-1 tool for assessing bias in comparative nonrandomized studies and methodological index for nonrandomized studies for noncomparative literature.⁴  Meta-analysis was conducted when data afforded a quantitative analysis with RevMan (Review Manager Version 5.4, The Cochrane Collaboration, 2020).

Online video conference meetings were held with clinician stakeholders after relevant evidence had been precirculated for each of the 4 questions, as shown in Supplemental Information Methods 4. Both the recommendation content, as well as the accompanying level of evidence, were thoroughly reviewed. Each recommendation was voted upon through anonymous polling, with adoption once a consensus of >80% of members agreed. Grading of Recommendations, Assessment, Development, and Evaluations methodology was employed in synthesizing evidence into recommendations.⁵ 

Members of the guideline creation panel volunteered their time without compensation; no conflicts of interest were identified among any panelists.

Our literature search identified 2362 records, of which 139 had their full texts fully assessed. Fifty-eight articles were ultimately included in this review, half of which concerned the technique of GT placement. The PRISMA study flow diagram summarizes reasons for exclusion and the assignment of studies into questions (Fig 1). Most studies originated in the United States, were retrospective in nature, and involved a small cohort of pediatric patients. Consensus was achieved on a total of 17 recommendations for the 4 proposed themes (Table 1).

Our literature search identified 262 abstracts addressing this content area. Of these, 12 were identified for full-text review, and 7 were included (Table 2).

No articles supported the assumption that it is unsafe to send an infant home on NGFTs or that NGFTs hinder weaning to oral feeds. We specifically reviewed the literature for articles addressing symptoms, delay in weaning, adverse events, and readmissions. Data supports the notion that NGFTs provide adequate nutrition and that many children are able to wean from NGFTs successfully. Nasogastric feeding was demonstrated to improve growth parameters specifically among patients with congenital heart disease (CHD), neurologic impairment, prematurity, and metabolic diseases. More than half (54%) of patients with CHD weaned from nasogastric (NG) to oral feeds at a median of 31 days, and most weaned once the underlying cardiac disease was definitively corrected.¹⁰  Similarly, premature infants discharged from the hospital on NGFT weaned to oral feeds at a median of 9 days, suggesting that the majority of premature infants will wean completely within a few months of discharge.⁶  Reports of NGFT use in older children have demonstrated that rapid weaning is possible, and that long-term NGFT use is feasible (in 1 study of 304 children with a mean age of 2.5 years, 26% weaned off NGFT in <30 days and 4% were fed safely by NGFT for >5 years).¹²  Even in children with severe neurologic disease or other condition in which they would be unlikely ever to be fed by mouth, an NGFT trial optimizes nutrition before gastrostomy tube (GT) placement, helps the family become comfortable with supporting a child on tube feeds, and proves that a child will tolerate a prescribed feeding regimen.^13,14 

Only 1 article specifically addressed symptoms, which included emesis and gagging/retching.⁸  All described symptoms were essentially associated with the underlying medical condition rather than the NGFT itself. Three papers described adverse events.^6,9,10  None of the studies described adverse events related to home NGFTs, and, in fact, 1 paper described a protective effect of feeding at home. One of these 3 papers was a Cochrane review describing a single, quasi-randomized study and reported that patients in a home gavage feeding program had a lower relative risk of respiratory infections than the comparison group who remained in the hospital.⁹ 

Three papers explored readmissions because of NGFTs and parent satisfaction.^6,9,11  None found any risk of readmission because of home NGFTs. Nearly all parents (96%) felt home NGFTs were safe, 93% would repeat home NGFTs again,⁶  97% would recommend home NGFTs to others, 84% felt that it was beneficial overall, and 68% described feeling not stressed about NGFTs.¹¹  The Cochrane review reported no statistical difference between NGFT versus orally fed patients in parent comfort with handling infants’ feeding at discharge, taking responsibility for infants’ care, or anxiety scores.⁹ 

For any child unable either to meet nutritional goals or to take adequate nutrition by mouth safely, discharge home on nasogastric tube (NGT) feeds should be considered.

Children who are fed via NGT should have ongoing feeding therapy and nutritional monitoring throughout their hospitalizations and after discharge.

Only 2 papers specifically addressed the frequency of returning to the health care system for NGFT replacement. The Cochrane review prospectively captured health-service utilization: nonelective general practitioner visits, pediatric outpatient visits, or emergency department visits utilization, finding no difference between the control group hospitalized until reaching full oral feeds and the group discharged on the home gavage program.⁹  A survey of parents and home health care companies found that the majority of tubes were successfully replaced by the parent, and only 14% of respondents reported needing replacement by a health care agency.⁷ 

No study clearly described significant complications associated with parent placement of NGFTs. All parents demonstrated appropriate suspicion not to feed the lungs and seek expert help when they were unsure.¹¹  These data suggest that, with adequate exposure to NGFT replacement and maintenance over the length of the hospitalization, parents/caregivers can learn these skills in preparation for discharge. It is the consensus of the panel that this education, including troubleshooting, is superior to abrupt learning before discharge.

Before discharge from the hospital with NG feeding, parents/caregivers should be provided education about tube function, troubleshooting, and replacement.

Before discharge from the hospital with NG feeding, parents/caregivers should have a clear plan in place for how they will access the health care system if there are problems with the NG tube and/or NG feeding at home.

To quantify outcomes after GT placement and how they differ according to various patient characteristics, we screened 137 abstracts and 50 full-text articles and ultimately included 17 articles in the review (Table 3). These 17 articles described GT complications and outcomes for a total of 1771 GT patients. The studies encompassed patients with a wide range of diagnoses.

Proper identification and management of the underlying diagnosis (eg, diet modification for eosinophilic esophagitis) may obviate the need for GT placement and prevent unnecessary procedures. In addition, many of the very commonly described postoperative emergency department visits and readmissions that are attributed to GTs (such as feeding intolerance) are often driven by the child’s underlying disease process and could be prevented by optimal medical management.³⁰ 

GT placement should only be pursued after appropriate workup has been performed to investigate the underlying medical diagnosis.

Feeding/treatment plans are often not established until after placement of the GT. Doing so in advance results in a more streamlined postoperative course and is also likely to reduce returns to system.^13,14  Active engagement of parents/caregivers in the development and implementation of these plans is critical, enabling individualization of care with attention to the specific needs of the patient and family related to health literacy, socioeconomic status, and caregiver ability to manage feeding and tube care.

A feeding/treatment plan that meets both parental/caregiver and team goals should be established before GT placement.

The incidence of reported complications after GT placement ranged widely from 0% to 90% based on variable definitions. There were 5 studies reporting complication rates >50%.^{8,21,23,26,29}  Tube dislodgement was 1 of the most commonly reported complications, along with leak and granuloma. Studies describing GT outcomes for patients with CHD in general reported lower complication rates (or did not report any complications), but this was driven by the fact that only serious adverse events were defined as complications.^{15,16,18,19,31} 

Few studies reported length of GT use after placement, and no studies described lack of GT use or rapid weaning after placement. This likely reflects a publication bias because a series of patients with GTs that were never used is unlikely to be shared. Several members of the international panel reported experiences with patients who had GTs that were placed and never used after discharge or used only for brief periods of time.

In addition to avoiding GT placement altogether, routine NG trialing results in more successful outcomes after GT placement in those patients who do require enteral access; success with NGFT predicts success after GT.^13,14  Involvement of a multidisciplinary team, which may include a dietician and speech/feeding/occupational therapist depending upon locally available resources, is essential before and after GT placement. This aligns with the findings from a notable article that was published since our panel completed its work. Lagatta et al found that infants discharged with an NGFT had reduced NICU length of stay and fewer readmissions/emergency encounters compared with GT patients. Also, parents of infants discharged with NGFT reported health-related quality of life similar to parents of infants who were orally fed.³² 

For most children, GT placement should be pursued only after failure to wean off NG feeding and evaluation by a multidisciplinary team.

The typical complications discussed during surgical consent (bleeding, infection, injury to surrounding structures) are uncommon after GT placement, and the procedure tends to be considered low risk. Surgeons and medical providers referring patients for GT should communicate not just insertion risks, but also the risks of long-term GT management, including leak, granuloma, dislodgement, and feeding intolerance, which are not necessarily dangerous but can be very disruptive to quality of life. A comprehensive description of these complications is essential to the informed consent process.

In a study that described the parent perspective on caring for children with GT versus NGFT, the 2 main observations expressed by parents were (1) GT care is not easier than NGFT care, and (2) GT care does not get easier over time.⁸  By clearly describing the risks and benefits of both options, providers can facilitate shared decision-making. Parent/caregiver preferences are key in the determination of proper timing for a given patient. Clinicians should strive to present reasonable options with best available evidence to parents/caregivers and guide them toward a decision that is consistent with their values.

The informed consent process for GT insertion should describe the benefits and risks of the procedure, including major and minor complications and alternatives to placement, as well as enteral tube management.

Parents/caregivers should weigh benefits and risks of NG feeding and long-term NG tube management versus GT placement and long-term GT management and be encouraged to participate actively in decision-making.

A recent American Pediatric Surgical Association Outcomes systematic review identified minimal evidence in support of screening asymptomatic patients with a UGI in general (not specifically preoperative GT patients).³³  We reviewed whether all patients should have UGI before GT placements, which preoperative symptoms merit UGI, and what was the incidence of missed malrotation requiring future intervention if UGI was not obtained. We screened a total of 50 abstracts and performed full-text review on 5 articles, all of which were included. These 5 articles summarized outcomes for a total of 2689 patients who underwent UGI before GT placement (Table 4). All of the articles were case series; none reported a consistent approach to decision-making regarding UGI before GT placement.

A total of 52 patients (1.9%) of the 2689 undergoing pre-GT UGI had abnormal study results concerning for malrotation. Only 3 of the 5 studies, accounting for 10 of the 52 patients, reported intraoperative correlation with UGI.^34–36  In 7 out of 10, malrotation was confirmed and Ladd’s procedure performed. In the other 3, normal anatomy was found at laparoscopy, representing a false-positive UGI rate of 30%. For the 42 patients in the other 2 studies, intraoperative correlation was not reported, so it is unclear what proportion of these patients actually had malrotation (Supplemental Fig 2).^37,38  The aggregate incidence of malrotation of 1.9% is therefore likely a high estimate.

Two of the 5 studies reported incidence of missed malrotation in patients who did not have preoperative UGI.^34,36  Median length of follow-up in 1 of these studies was 9.6 months and was not reported in the other. One patient out of 349 had malrotation discovered at a later date and underwent Ladd’s procedure without significant morbidity, a missed malrotation rate of 0.3%. Missed malrotation resulting in need for further intervention is a rare event. Given the rarity of malrotation in the general population, and the lack of untoward events associated with missed malrotation, UGI should not be used as a routine screening for patients undergoing diagnostic workup for enteral access.

Do not routinely obtain UGI preoperatively for all patients undergoing GT placement.

A total of 44 of the 49 patients with confirmed malrotation (either confirmed in the operating room or seen on UGI without intraoperative confirmation) had a “significant” congenital anomaly (defined as being noticeable within days of birth), and 2 of the other 5 had cystic fibrosis. Even though malrotation was more frequently observed in children with congenital anomalies, it was noted to be a rare enough finding that the simple presence of an anomaly should not mandate UGI.

Congenital anomalies alone are not an absolute indication for UGI before GT placement.

None of the 5 studies described the presence or absence of preoperative symptoms. One study recommended that no UGI was indicated if the patient was tolerating gastric feeds without emesis, but this strategy was not prospectively implemented or tested in any of the studies. The consensus of the expert panel was that UGI should be pursued as part of the diagnostic workup for feeding intolerance in children with emesis, but it should not be used as a screening test in asymptomatic patients.

Consider UGI before GT placement for patients who are not achieving adequate enteral nutrition because of emesis.

This section of the evidence review was constrained to comparative literature only, updating a similar systematic review from 2015.³⁹  Major complications (need for unplanned operation, perforated viscous, hemorrhage) were quantified; minor complications were summarized when available.

We performed a subanalysis of available gastrostomy types (Supplemental Fig 3) distinct from insertion techniques, as well as a review of technical variations in the laparoscopic insertion technique Supplemental Fig 4). The search strategy identified 137 candidate abstracts, of which 73 were fully screened. Twenty-nine articles were included in the review of comparative techniques; 2 articles compared outcomes on the basis of gastrostomy type.

Seven publications directly compared outcomes among pediatric patients undergoing an open gastrostomy (N = 1471) or a percutaneous endoscopic gastrostomy (PEG) (N = 679) (Table 5). All reports were retrospective reviews, 1 was multiinstitutional, and all had critical biases identified because of nonrandom patient selection and substantial confounding (Supplemental Fig 5). Most reports separately evaluated patients undergoing a concomitant antireflux surgery. No significant differences in major complications were noted between open gastrostomy (3.2%) and PEG (4.1%, P = .35) (Supplemental Fig 6).

Seven reports directly compared outcomes of pediatric patients undergoing either an open or laparoscopic gastrostomy insertion comprising >3000 patients (1535 open, 1596 laparoscopic) (Table 6). Risk of bias was critical for all included studies (Supplemental Fig 7). The aggregate rate of major complications was not significantly different between the open and the laparoscopic approach (3.2% vs 0.9%, P = .08), although the magnitude of difference does seem clinically significant. There was significant heterogeneity identified among studies and clear selection bias (Supplemental Fig 8). Based on the aggregate data, a number needed to treat (NNT) of 44 patients undergoing laparoscopic instead of open GT would prevent 1 major complication.

Twelve studies directly compared outcomes between the laparoscopic insertion technique and PEG; 2108 patients were aggregated (1321 PEG, 787 laparoscopic) (Table 7). Risk of bias was critical for all included studies (Supplemental Fig 9). Complication rates of laparoscopic approach versus PEG significantly favored the laparoscopic group (1.2% vs 5.4%, P < .0001) (Supplemental Fig 10). Based on the aggregate data, an NNT with laparoscopic GT to prevent 1 major complication from PEG is calculated to be 24.

Three papers directly compared outcomes between IR-placed GTs and PEGs in the pediatric population, including a well-designed randomized controlled trial from the United Kingdom (Table 8). Patients were well balanced between the 2 groups (IR = 321, PEG = 417), and risk of bias was critical for the retrospective reports but not present in the randomized trial (Supplemental Fig 11). Major complication rates were 1.9% in PEG patients and 3.7% in IR patients (P = .04), yielding an NNT of 55 PEG patients to prevent 1 major IR complication (Supplemental Fig 12). A sensitivity analysis of higher-level evidence (the single randomized controlled trial); however, yields equipoise between the 2 techniques. Importantly, direct comparisons of laparoscopic with different IR procedures, including antegrade or retrograde approaches, have not established a preferred technique. Local expertise should continue to direct institutional practices with respect to the choice between laparoscopic versus IR approach.

For most pediatric patients requiring enteral access, the laparoscopic approach is preferred to PEG.

The decision regarding procedural approach (open/laparoscopic, PEG, IR, or hybrid) may be influenced by individual patient, parent/caregiver, socioeconomic, and/or institutional factors.

GTs can broadly be categorized on the basis of their retaining design and their external appearance. Retaining options include a bumper (eg, Bard [Arizona]), flange (eg, Entristar [McKesson, Texas]), or balloon (eg, MicKey [Avanos, GA]). External design can either be low profile (eg, MicKey) or elongated (traditional PEG tube). Two studies directly compared outcomes of each of these design types, summarized in Supplemental Fig 3.^43,45  No significant differences in complications were noted among types, though Conlon et al noted the additional long-term risk of ∼2% of buried bumper syndrome associated with the Bard/Genie/PEG model. The reported concern of bowel obstruction after bumper-cutting was also acknowledged, as has been previously reported.⁶³  An additional anesthetic was commonly reported for patients receiving a PEG to change over their model to a low-profile device. Ultimately, individual patient needs may dictate precise tube specifications; for example, the long-term security of the tube after placement in patients prone to self-dislodgement, which may favor a low-profile, “bumper-style” device or need for improved venting for which an elongated design may be preferred.

For most pediatric patients requiring enteral access, an external low-profile design with a balloon retention system is preferred.

Although a low-profile balloon type tube is preferred, patient-related factors (medical and behavioral), parent/caregiver preference, socioeconomic factors, and/or institutional factors should be considered.

Four recent publications discretely compared different techniques of laparoscopic insertion of gastrostomy tubes in children.^64–67 Supplemental Fig 4 summarizes the findings of these reports, including the major differences between a Seldinger approach with fascial fixation sutures, Seldinger with full-thickness temporary fixation sutures, and a “mini-open technique” where the stomach is exteriorized. No differences in major complications were appreciated.

For pediatric patients undergoing a laparoscopic enteral device insertion, the fixation technique (eg, use of transabdominal wall sutures versus semi-open technique) should be left to operator discretion.

We present a systematic review of the literature and a multidisciplinary, panel-driven consensus for pediatric patients being considered for surgical enteral access. Impactful recommendations include the safety of an NG feeding trial in most patients, the primacy of a deliberate decision-making process, the futility of a screening UGI study, and the superior safety profile of the laparoscopic placement technique.

Although our review suggests that NGT feeding is safe, there is a relative paucity of data regarding health-related quality of life and whether growth failure is more likely for patients who are NGT versus GT fed. Nevertheless, the panel concluded unanimously that discharge from the hospital on NGT feeds should be considered, and that GT placement should be considered only after failure to wean off NGT feeding. The vast majority of pediatric patients can safely be fed orally, even patients with some degree of aspiration because of oropharyngeal dysphagia. Swallowing dysfunction tends to improve over time in children, and permanent enteral support is rarely needed on a population level.⁶⁸ 

Our review demonstrates no benefit in screening asymptomatic patients with a UGI before GT placement. Reducing/eliminating radiologic tests for this patient population potentially addresses expenditures and resource utilization, with gastrostomy insertions currently accounting for 5.8% of variation across the United States.⁶⁹  More importantly, eliminating this practice reduces the exposure of vulnerable children to radiation and its attendant concerns of long-term malignancy.⁷⁰ 

Significant variability in complication rates for children undergoing GT placement exists, largely driven by differences in how complications are defined. This highlights the need for standard definitions and core outcome sets for studies reporting GT outcomes. The more common minor complications have a clear impact on quality of life after GT placement, and it is essential to capture these in addition to the more serious complications that are seen. Targets for quality improvement efforts for patients undergoing GT placement should include outcomes such as leak, dislodgement, granuloma, and parent/caregiver satisfaction, in addition to reoperation and revisits. Nonetheless, substantial comparative literature detailing major complication rates favors the laparoscopic approach for most patients in most clinical care contexts.

Acknowledged limitations include the fact that recommendations were generated almost exclusively from level 4 and 5 evidence, with few studies describing direct comparisons between GT placement and prolonged NGFT trial. Even those that did make this comparison did not use a standardized approach to decision-making regarding which patients had GT and which had NGFT. To define the optimal time frame for NGFT trial, studies that apply a standardized approach to decision-making regarding timing of GT placement are required.

Children who are unable to take adequate feeds by mouth and require supplemental tube feeds are high utilizers of the health care system. These recommendations have the potential to make a significant impact on health care utilization for this population through:

1. avoiding GT placement in children who are likely to learn how to eat by mouth by embracing the concept of prolonged NGT feeding use;

2. eliminating unnecessary preoperative workup, including routine use of UGI; and

3. optimizing placement techniques to minimize postoperative complications.

We thank the Outcomes and Evidence-Based Practice Committee of the American Pediatric Surgery Association for their work in the question–generation phase of the systematic review, and we acknowledge Kristina Flathers for her work in completing the extensive literature search for the systematic review. We also thank Theresa Michel for her expertise in preparing the manuscript for publication.

CHD

congenital heart disease

GT

gastrostomy tube

IR

interventional radiology

NG

nasogastric

NGT

nasogastric tube

NGFT

nasogastric feeding tube

NNT

number needed to treat

PEG

percutaneous endoscopic gastrostomy

UGI

upper gastrointestinal contrast study