GERD management in neonatal ICU infants: study protocol for a randomized controlled trial comparing maturation, added rice starch formula and proton pump inhibitors

Abstract

Background

The prevalence of gastroesophageal reflux disease (GERD) diagnosis among infants in the neonatal intensive care unit (NICU) is high and varies across USA NICUs. GERD diagnosis is challenging as symptoms can be physiological and association is difficult to determine without objective testing methods. Precision diagnostics using 24-hour pH-impedance (pH-Imp) studies with symptom correlation to reflux events have been developed and validated for this population. Current empiric-based, non-standardized GERD management practices are highly variable and are widely untested for safety and efficacy in high-risk NICU infants. The rationale for this randomized controlled trial (RCT) is to compare the efficacy and safety of three widely used interventions for infants that meet the criteria for acid-mediated GERD, objectively diagnosed by 24-hour pH-Imp.

Methods

This is a prospective, single-center RCT aiming to enroll convalescing NICU infants between 35- and 47-weeks postmenstrual age with objectively diagnosed GERD using a unique parallel 3-arm design comparing the short-term effects of three common management strategies: natural maturation vs. proton pump inhibitors (PPI) vs. added rice starch (AR) formula. Using established 24-hour pH-Imp inclusion criteria, subjects will be randomized to one of the three arms for a 4-week treatment course allocated 1:1:1, stratified by acid reflux index (ARI) (3–7% and > 7%) and feeding method at inception (full oral versus any tube feeds). Primary outcome measures include maintenance or improvement in oral feeding and symptom improvement measured at 4 weeks or discharge whichever is sooner. A second pH-Imp study will be done on treatment to examine treatment efficacy on GERD metrics. The primary clinical endpoints are oral feeding improvement and absence of troublesome symptoms. The mechanistic hypotheses that pH-Imp mechanisms are distinct between assigned therapies, and between clinical success and failure will also be tested.

Discussion

The goal of this study is to refine novel diagnostic pH-Imp criteria for infant GERD diagnosis and compare the efficacy of the three unproven therapies, examine pathophysiological mechanisms for clinical outcomes, define duration of therapy, and assess consequences of interventions.

Trial registration

ClinicalTrials.gov. Brief Title: GERD Infant Feeding Therapeutics Trial (GIFT Trial). NCT Number: NCT06114836. Unique Protocol ID: STUDY00003300.

Background

Pediatric gastroesophageal reflux disease (GERD) is defined as the reflux of gastric contents leading to troublesome symptoms (not defined) and/or complications (e.g., failure to thrive, eating difficulties, vomiting, coughing, or esophagitis) that affect daily functioning [1–3]. This symptom-based definition is even more ambiguous in the neonatal intensive care unit (NICU) infant, as most infants are symptomatic from multi-systemic illness which makes clinical observation unreliable to identify or diagnose GERD [4, 5]. Convalescing infants have widely prevalent feeding difficulties and attribution of symptoms to GERD or esophagitis without objective correlation is not appropriate. Thus, GERD diagnosis is challenging as both gastroesophageal reflux (GER) events and aerodigestive symptoms can both be physiological and pathological. Distinction between normal and abnormal GERD pathophysiology is now possible using 24-hour pH-impedance (pH-Imp) criteria with symptom correlation. The overall prevalence of GERD diagnosis in premature infants is over 10.3% in the USA, ranging from 2 to 30% by site, which suggests significant practice variation, with a consequence of over $70k cost per diagnosis while prolonging hospital stay by 30 days, all of which escalate health care burden [6].

Clinical diagnosis of GERD among convalescing NICU infants remains a conundrum and often results in unvalidated diagnostic and management strategies. Results from a recent survey of NICUs across the US regarding their current diagnostic and management practices for infant GERD are shown in Fig. 1. Strikingly, 16 (43%) of the centers reported pH-Imp testing capabilities, but only 9 (24%) use pH-Imp to primarily diagnose GERD with 19 centers (51%) utilizing empiric therapies. This clinical management data is concerning as therapies are based on symptoms which may or may not be truly due to GERD (Fig. 1A). When available, pH-Imp testing provides objective pathophysiologic diagnostic mechanisms. Empiric medical and surgical treatments are widely prevalent (Fig. 1B) and likely over-prescribed. Common empiric management strategies include parental reassurance (allowing time for infant maturation while monitoring optimal nutrition and growth), nutritional modifications such as added rice starch (AR) formula, or acid suppressive therapy [1–3, 7]. The indications for use, prescription, and deprescription of these treatment strategies, including effectiveness and safety metrics, remain untested in this population [7, 8]. These persistent gaps in knowledge preclude us from implementing critically needed objective, effective, evidence-guided therapy for GERD in the NICU.

Fig. 1.

Survey results

Response to a 2022 Survey on Neonatal GERD Diagnostic and Management Strategies in US NICUs. As/Bs/Ds: apnea/bradycardia/desaturation events, PPI: proton pump inhibitors. A From 37 responding NICUs, vomiting, arching/irritability, and As/Bs/Ds are frequently assumed to suggest GERD while only vomiting and coughing have been proven to be significantly associated with GERD. B Common GERD therapies used in NICU infants. From 32 responding NICUs, reported therapies included thickened feeds [AR formula (44%), thickening agents added to formula (56%), or human milk (50%)], PPI, and fundoplication

Although the potential for improvement in neonatal GERD may exist, treatment with AR formula may delay gastric emptying and gastrointestinal transit, cause constipation, increase coughing, or exceed safe osmolality thresholds [7–11]. Proton Pump Inhibitor (PPI) use in this population is off-label, commonly over-prescribed, can increase infection risk and has variable effects and side effects [6, 12–14]. A review of hospitalized NICU infants in US children’s hospitals showed that 76% of treated infants remained on PPI at discharge [6]. Chronic use may increase long-term risk of infection, reduced bone mass, fractures, chronic anemia, nutrient malabsorption, and esophageal dysmotility [7, 14–23]. Another treatment often used in the NICU is gastrostomy tube placement with or without fundoplication, which can contribute to further complications along with prolonging NICU discharge. Randomized controlled clinical trials (RCT) using objective metrics are critically needed to address the burden, safety, and efficacy of these current treatment practices in the NICU population; our RCT addresses this critical need.

Prior trials in preterm and full-term infants have not adequately assessed the safety and efficacy of using acid suppression in this population. Use has been shown to add to pharyngo-esophageal dysmotility and therefore potentiate eating problems leading to chronic tube feeding [18]. In all referenced trials, symptom-based treatment durations and measured clinical outcomes varied. Many existing trials were pharmaceutical industry sponsored mainly assessing pharmacological profiles [24–30]. Often, the GERD diagnosis for inclusion is based on subjective symptom scores essentially reflecting the caregivers’ symptom perceptions which may have included normal/non-GERD infants [31–33]. Diagnosis is often limited when the acid reflux index (ARI) severity category (ARI of < 3% as normal, 3–7% as indeterminate and > 7% as severe) is not considered as criteria for inclusion [27, 28, 34]. Other pH-Imp metrics such as distal baseline impedance (DBI) < 900 Ω (impaired esophageal mucosal permeability and/or inflammation marker), symptom association probability (SAP) ≥ 95% (symptom correlation with GER events), and GER > 70 events/day (indicates frequency of causal factor) should also be considered [4, 5, 35, 36].

Trials using AR formula in preterm and full-term infants resulted in decreased regurgitation and perceived improvement in symptom burden with reduced coughing/choking/gagging, but these trials were typically limited to healthy infants [37–41]. In a study of hospitalized infants, coughing increased with the authors concluding that clinicians should avoid thickened feeds in infants with non-regurgitant reflux pending further evaluation [11].

Natural maturation, allowing time for development with supportive optimal nutrition, may be the cornerstone for treating GERD [38, 40, 42–44], but no objective comparison to treatment (PPI or AR formula) exists. Further support for untreated maturation stems from improving nutrition and digestion via feeding strategies which may modify the sensory-motor aspects of motility and peristaltic clearance reflexes over time in premature infants. This suggests better sensory-motor regulation of reflux-clearance-symptom mechanisms with overall growth and maturation [42–52]. However, in high-risk infants with comorbidities, these functions may potentially worsen with time [7, 32].

The objective of this trial is to prospectively compare a 4-week treatment course of three common treatment approaches used for GERD in NICU infants: (1) natural maturation focusing on nutrition, feeding methods, and growth, (2) PPI and (3) AR formula. The two specific aims with sub-aims are as follows:

Aim 1 is to evaluate and compare the efficacy of the randomized interventions. To test the efficacy of the intervention, clinical success or failure is our a priori primary endpoint and is based on a composite metric of oral feeding (based on feeding method at inception) and troublesome symptoms (e.g., vomiting ≥ 2/day or coughing ≥ 18/day) [4, 25]. Primary and secondary endpoints are clarified in Table 1. The working primary hypothesis is that AR formula is superior to (1) maturation alone and (2) PPI. The secondary hypothesis will test if natural maturation is superior to PPI.

Table 1.

Study endpoints

Primary Endpoint (met/not met) This binary outcome is a composite metric of oral feeding ability and absence of troublesome symptom threshold (e.g., vomiting < 2 / day and coughing < 18 / day). To clarify further, primary endpoint determination will occur at discharge or four weeks (whichever is earlier) while still on therapy to measure the efficacy of the intervention based on the infant’s feeding method.  • In infants’ tube-fed at inception: primary endpoint is met if oral feeding volume is increased from baseline or infant attained independent oral feeding -OR- absence of troublesome symptoms.  • In infants oral-fed at inception: primary endpoint is met if infants maintain independent oral feeding (absence of any tube feeding) -AND- the lack of troublesome symptoms. Secondary Endpoints  • Growth metrics, eating, and GI milestones are nutrition and infant physiology dependent and are well-regulated in the NICU setting. Therefore, these values are measured at inception, and a) weekly for four weeks (during therapies), at week 5 (off trial), and at 6- and 12-months corrected age.  • Parent/provider-assessed symptom scores (via I-GERQ-R) are measured at inception, and a) weekly for four weeks (during therapies), at week 5 (off therapy). Although I-GERQ-R has poor sensitivity, specificity, and positive predictive value in NICU infants, we will examine the utility of its negative predictive value to assess recovery in the current trial, as it was high for acid exposure (ARI) and DBI < 900 Ω.  • Adverse effects from GERD or of its therapies will be measured a) weekly for four weeks (during therapies), at week 5 (off trial).  • Economic burden metrics will be measured as the length of hospital stay and # of related readmissions and clinic visits during the 6- and 12-month corrected age.

Sub-aim 1a is to compare the following secondary outcomes: parent/provider-assessed symptom scores, growth, eating and gastrointestinal milestones, safety, adverse effects, and economic burden which is further clarified in Table 1. Sub-aim 1b compares the effects of therapies among ARI severity and feeding methods. Response failures will be referred for follow-up to feeding or gastrointestinal programs.

Aim 2 characterizes the mechanisms of the primary end point (success or failure) using pH-Imp metrics. The working hypothesis is that the pH-Imp mechanisms will differ between assigned therapies as well as clinical success and failure.

Sub-Aim 2a compares the longitudinal effects of assigned interventions by examining the mechanistic and adaptational pH-Imp biomarkers including: ARI, number of GER events, SAP, DBI, and reflux clearance times. Sub-Aim 2b distinguishes pH-Imp mechanistic differences among ARI severity and feeding methods.

Methods

The study protocol described is a RCT comparing three treatments (maturation, PPI, and AR formula) after objectively diagnosed GERD. Note the process flow for screening/eligibility, randomization/allocation, intervention/data collection, and exit/follow up as shown in Fig. 2.

Fig. 2.

Study flow diagram

Setting

The study will take place in the Nationwide Children’s Hospital (NCH) NICU network in Columbus, Ohio which has 306 NICU beds at both our Level 4 main campus and tertiary NICUs. The Innovative Infant Feeding Research Program at NCH is consulted for all GERD evaluations and only treats infants after evidence-based pH-Imp testing revealed potential pathophysiologic mechanisms. Once NICU clinicians seek diagnostic considerations for suspected GERD, diagnostic pH-Imp data is examined regarding the indices for acid exposure, bolus events, symptoms, and mucosal permeability. Depending on the test results and clinical relevance, the three commonly used approaches are: natural maturation and growth, PPI, and AR formula. Infants are monitored in the NICU throughout their treatment course. In the described RCT, these treatments are randomized for those meeting objective, pH-Imp-based diagnostic criteria.

Participants

The population to be studied consists of NICU infants with concern for GERD who are undergoing diagnostic pH-Imp testing. They must have a signed informed parental/legal guardian consent before trial enrollment and evaluation. Inclusion criteria are as follows: (1) infants between 35 and 47 weeks postmenstrual age (PMA) (= birth gestation + chronologic age since birth), (2) a GERD diagnosis using pH-Imp criteria (ARI ≥ 3% plus at least one of the following: # GER events > 70 / day; SAP ≥ 95%, or DBI < 900 Ω), see Table 2 [4, 25, 36], (3) infants must be fully enterally fed, and (4) not on any GERD therapies. Infants will be excluded if they have known lethal chromosomal abnormalities or complex congenital syndromes, severe neurologic pathologies requiring neuroactive medications or neurosurgery, positive airway pressure or oxygen flow > 4 LPM, or upper gastro-intestinal malformations requiring surgery. Complex conditions will be reviewed by the PI for clinical and ethical appropriateness for inclusion/exclusion. Any patient with clinical deterioration during the study period will be discussed among the PI, parents, and clinical team as to determine if the infant should be withdrawn.

Table 2.

Abnormal GERD biomarkers using pH-impedance

Criteria	Definition	Significance
Acid Reflux Index (ARI) ≥ 3% (7, 25) [15, 24]	The percentage of time the esophagus is exposed to acid is calculated as the % of time pH < 4.0 normalized to 24 h. ARI 3–7% is considered indeterminate, and ARI > 7% is considered abnormal.	Prolonged acid exposure may cause epithelial damage and has been correlated with mucosal permeability( low DBI).
# Gastroesophageal Reflux (GER) > 70 events / day [4]	A GER event is defined as > 50% decrease in impedance baseline with retrograde bolus movement reaching as least the two most distal impedance channels.	Abnormal frequency of GER events may indicate frequent transient lower esophageal sphincter (LES) relaxation or hypotonic LES or esophageal dysmotility.
Symptom Association Probability (SAP) ≥ 95% [4, 36]	SAP calculated for each symptom type using Fisher’s Exact Test	Probability that symptoms are likely related to reflux events.
Distal Baseline Impedance (DBI) < 900 Ω [34, 35]	Average DBI calculated over 24 h	A measure of increased mucosal permeability or potential esophagitis resulting in impaired esophageal motility.

Study design

Upon meeting inclusion/exclusion criteria, consented participants will be stratified and randomly assigned into one of three parallel arms using a predefined permuted block randomization design (Fig. 2). This is accessed by the research assistants analyzing the pH-Imp for eligibility as the rest of the study team and the clinical team are blinded to the specific results. Eligible subjects will be randomized 1:1:1 to natural maturation-standard feeding, PPI (omeprazole) off label use, or AR formula group and stratified by ARI (3–7% and > 7%) and feeding method (any tube feeding and full oral feeding). The natural maturation and PPI group will receive standard feeding approaches, whereas the AR formula group will receive the AR formula instead of standard feeding for the 4-week treatment period. Treatment is not blinded as changes in formula, medication, and natural maturation are followed to maintain compliance and need to be communicated to the care team and parents.

Data will be collected to analyze primary outcomes at the endpoints stated and follow-up with be undertaken (Table 1). The pH-Imp at time-1 is a diagnostic study and is the standard of care at NCH. A repeat pH-Imp study will be completed prior to discontinuing therapy at four weeks or prior to discharge, whichever is sooner. The pH-Imp at time-2 is for research purposes to define therapeutic efficacy and validate the mechanisms of clinical success or failure. The timeline for data collection and procedures is stated in Table 3. Data will be collected electronically in the password protected REDCap database and will be reviewed by two research coordinators for accuracy and completeness. The informatics manager will receive only coded information that is entered into the database under study-specific identification numbers. Electronic communication with outside collaborators will involve only deidentified information. Research coordinators will follow the participants closely even after discharge to continue to collect all data and ensure compliance. Long term follow-up data will be collected from electronic medical records.

Table 3.

Tests, Monitoring, and Timelines in each subject

Characteristic	Pre-Study	Week 1	Week 2	Week 3	Week 4	Week 5	6 months	1 year
Symptom Scores/I-GERQ-R	X	X	X	X	X1	X
pH-Impedance Testing	X				X2
Treatments Applications		X	X	X	X
Outcomes Measurements	X	X	X	X	X1	X	X	X

Weeks 1–4 on therapy and week-5 off therapy

¹ Measurement for primary and secondary aims will be taken at four weeks or at discharge

² Will be done at four weeks or at discharge, whichever is the earliest. Overall, we expect 294 evaluable subjects over a 4-year recruitment period. The last year is dedicated to data/statistical analyses and completing the protocol if recruitment is needed

Statistical analysis

Under Aim 1, efficacy will be assessed by testing the hypothesis that AR formula provided over four weeks is superior to both the natural maturation (standard feeding group) and the PPI group with standard feeding. We will have two primary superiority tests: AR vs. maturation, and AR vs. PPI. The PPI and natural maturation group will only be compared if at least one of the primary tests is significant. The overall type I error will be controlled at 0.05. With this sequential testing strategy, PPI and maturation groups will be tested at the one-sided alpha of 0.025 or 0.05, depending on the significance of the primary tests. If at least one of the primary tests is significant, PPI and maturation will be tested for superiority first (maturation is superior to PPI). If the superiority test fails, we will switch to the non-inferiority test with a 10% non-inferiority margin at the same significance level. The success rate (primary endpoint) will be summarized with a 95% confidence interval for each group and compared between groups. Logistic regression models will study the success rate among the treatment groups and ARI severity and feeding method categories and their interactions, while adjusting for measured, potential confounding factors. Linear mixed models or Generalized Estimating Equation models (depending on the nature of the outcome variable) will study the associations of interest within and between study groups across repeated outcome measures. A linear regression model and Negative Binomial model will assess associating economic metrics (the length of hospital stay, # readmissions, and clinic visits) with the treatment groups. Bonferroni correction will adjust for multiple comparisons.

Under Aim 2, mechanistic outcomes will be examined by testing the hypotheses that the pH-Imp mechanisms are distinct between (a) assigned therapies and (b) clinical success and failure. We will assess the mechanistic and adaptational pH-Imp biomarkers (i.e., ARI, #GER events, SAP, DBI, bolus clearance metrics) at time-1 and time-2. Sub-Aim 2a will address the longitudinal effects. In sub-Aim 2b, we will distinguish between ARI severity and feeding method categories. Linear mixed models or Generalized Estimating Equations models (depending on the nature of the outcome variable) will test the pH-Imp mechanisms between therapies among ARI severity and feeding method categories. Bonferroni corrections will control Type I error due to multiple comparisons. Potential confounders, and interaction between groups, will be included in the models. We have utilized similar models previously [43, 45, 49–51, 53, 54].

Sample size

The primary endpoint is clinical success or failure measured at follow-up (week 4 while on therapy or prior to discharge). In fully oral-fed infants, success means maintaining oral feeding and the absence of troublesome symptoms. Failure occurs if the infant continues to have or develops troublesome symptoms or requires tube feeding at four weeks. In infants with any tube feeds at inception, success means improving oral feeds or attaining full oral feeding or the absence of troublesome symptoms. Using pilot data for calculations shows with a sample of 294 evaluable patients in total (98 per group, 1:1:1 allocation ratio) stratified by ARI category (3–7%; >7%) and feeding method (Tube; Oral), we will achieve 76% power to detect a 15% increase in the defined success in the AR formula group compared to the maturation group, assuming the success rate is 75% in the maturation group and using a one-sided alpha of 0.025. We will also have 94% power to detect a 20% increase in the defined success rate of 90% in the AR formula group compared to the success rate of 70% in the PPI group, using a one-sided alpha of 0.025. The overall type I error will be controlled at 0.05 using the Bonferroni method adjusted for these two comparisons. The sample size calculations were performed using East 6.5 software, assuming a group sequential design would be implemented with two interim analyses for both futility and efficacy and one final analysis for each comparison. Considering a dropout rate of 20%, we aim to enroll 369 patients for this study for 294 evaluable subjects. The trigger of the interim analyses will be at 33% and 66% recruitment (i.e., 98 and 196 evaluable patients in total); this approach gives the team an opportunity to reassess sample size as there is no other similar study data available to precisely estimate sample size in this high-risk NICU population. Lan and DeMets version of the O’Brien-Fleming spending function will be utilized for the futility and efficacy stopping boundaries. This will be evaluated by the study team and the data safety monitoring board.

Discussion

As there are currently no objective guidelines for GERD diagnosis and treatment in NICU infants, this prospective RCT has the potential to establish safe and effective guidelines for three commonly used treatments. Keeping the specifics of the pH-Imp results blinded and doing the follow-up study while receiving the randomized treatment are important metrics to prevent bias and establish efficacy. Improvement in the primary outcome, oral feeding, and symptoms, will provide pragmatic evidence supportive of real-world clinical improvement with the randomized therapy.

The data generated on primary, secondary, mechanistic, and economic metrics from this RCT will instantly inform clinical practice and definitive diagnostic and therapeutic criteria for future multi-center RCTs. Wherein the dose and therapy durations will need further examination depending on the rates of success or failure to meet the primary endpoint. Algorithms can be generated upon completing this study to develop a framework for GERD evaluation, diagnostic considerations, potential new therapies, and refractory cases which will require specialty GI follow-up.

Importantly, diagnosis of infant GERD can be further refined based on level 1 screening using the Infant Gastroesophageal Reflux Questionnaire-Revised (I-GERQ-R) score and determining objective diagnostic grades for classification. Then, levels of therapies can be determined using the lowest risk and most effective type and duration of therapy. Thus, prescription and de-prescription strategies can be developed to prevent long-term consequences of unmonitored therapies. This study can provide evidence for anticipatory guidance, which will help the pediatricians and follow-up programs in counselling parents of NICU graduates with history of or concern for GERD. This RCT challenges the dogma of symptom-based diagnosis and empiric management of GERD in convalescing NICU infants by providing a scientific basis for managing these vulnerable infants.

Abbreviations

GERD

Gastroesophageal reflux disease

NICU

Neonatal intensive care unit

GER

Gastroesophageal reflux

pH-Imp

pH-impedance

AR

Added rice starch

PPI

Proton pump inhibitor

RCT

Randomized controlled trial

ARI

Acid reflux index

DBI

Distal baseline impedance

SAP

Symptom association probability

I-GERQ-R

Infant gastroesophageal reflux questionnaire revised

Author’ contributions

SJ, EO, JS, LW and CD made substantial contributions to the conception and design of the study. SJ, EO, ZS, RH, JBS, and JS substantially contributed to the acquisition of data. LW conducted the statistical analysis for the study proposal. SJ, EO and JS, drafted the manuscript, and all authors read, revised and approved the submitted version.

Funding

Research described in this publication is funded by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK, healthinfo@niddk.nih.gov) of the National Institute of Health (NIH) under award number 1R01DK136762, “Pathophysiological Evidence Driven Management of GERD in Neonatal ICU Infants: Randomized Controlled Trial”.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This RCT has been reviewed by the Nationwide Children’s Hospital Institutional Review Board (IRB@nationwidechildrens.org, reference #STUDY00003300) and was approved on 7/20/23. Protocol modifications (now version 5) were reviewed and approved by the IRB on 8/18/2025; changes will be updated on clinicaltrials.gov (NCT06114836). Modifications were to expand the screening and age criteria as well as to include a parent recruitment video and flyer. Informed consent from a parent or guardian for all participants (as under 16 years old) will be obtained for all study participants according to the IRB approved protocol and consent form. There is an IRB approved Data Safety Monitoring Board comprised of pediatricians/neonatologists, ethical and quality improvement experts, nurses, a nutritionist, a pharmacist and a biostatistician which who meets quarterly to review enrollment, compliance, protocol adherence, side effects, complications, and follow-up. Recruitment began on 1/31/2024 with the first enrollment on 3/13/2024.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.