Bolus Versus Continuous Nasogastric Feeds for Infants With Bronchiolitis: A Randomized Trial

Bronchiolitis is a viral lower respiratory tract infection that is the leading cause of hospitalization in infants.^1–5  Around 13 500 children are admitted to hospital with bronchiolitis each year in Australia,¹  most commonly (60% to 70%) due to the respiratory syncytial virus (RSV).^3–9  The management of infants with bronchiolitis is supportive in nature with provision of oxygen supplementation and adequate hydration being the mainstay of international guidelines.^3,5,7,10  Literature suggests that 30% to 60% of infants cannot adequately orally feed and require maintenance of hydration via either nasogastric (NG) feeding or intravenous fluids.^5,11,12  NG feeding is standard practice in Australia and preferred over intravenous fluids. Bolus and continuous NG feeding methods are internationally accepted practices for nutritional support in pediatric patients with disorders of oral feeding, digestion or absorption, increased nutritional or metabolic requirements, and where enteral nutrition is a component of disease treatment.

Infants with moderate to severe respiratory distress are commonly commenced on continuous NG feeds rather than on bolus NG feeds. Diaphragm splinting is assumed to occur when an infant has a full stomach, which is thought to increase respiratory effort and risk of aspiration.^13–15  Despite an extensive search of the literature no published evidence could be found to support this anecdotal practice. After starting on continuous feeds, NG feeds are subsequently transitioned to a bolus regimen before trialing oral feeds as the patient’s clinical condition improves. This transition can take 24 to 72 hours, potentially prolonging hospital admission and significantly adding to the burden on families and the public health system.

Studies comparing NG feeding and intravenous fluids to maintain hydration in bronchiolitis outline NG feeding safety, nutritional benefit, higher success of NG tube insertion, and economic superiority.^{10,12,16–20}  There are no studies evaluating the best method of NG feeding in bronchiolitis. Guidelines are inconsistent as there is no consensus around feeding regimens, with practices being largely based on physician or institution preference.¹³  The 2014 American Academy of Pediatrics Bronchiolitis Clinical Practice Guideline recommended further research into NG feeding for hydration.⁵ 

At this 360-bed free-standing eastern Australian children’s hospital, approximately 500 patients were hospitalized with acute bronchiolitis annually in 2018 and 2019. They had an average length of stay (LOS) of 3.62 days. Literature reveals hospitals in the United States recorded a mean LOS of 2 to 3.5 days for bronchiolitis within the last 10 years.^21,22 

The current study sought to compare continuous versus bolus NG feeding strategies for infants hospitalized with bronchiolitis. The primary hypothesis was that continuous NG feeding would result in an increased LOS when compared with bolus NG feeds. Secondary markers were pulmonary aspiration and admission to the pediatric ICU (PICU).

The study conducted was a randomized, open, parallel-group, superiority clinical trial of bolus versus continuous NG feeding in infants with bronchiolitis. Recruitment occurred during 2 bronchiolitis seasons from May 2018 to October 2019 within the emergency department and general pediatric wards of a children’s hospital providing tertiary and quaternary care in eastern Australia. The free-standing hospital has 360 beds; the emergency department reviewed approximately 72 000 presentations in 2019, and 17 000 of these were treated by pediatricians as inpatients.

The inclusion criteria included (1) infants <12 months corrected age; (2) clinical diagnosis of bronchiolitis; (3) NG feeding (oral intake less than 50% of normal over the previous 12 to 24 hours by observation or parental report); (4) parents or guardian read and understand English. The Australian definition of bronchiolitis includes patients <12 months of age because older children hospitalized for wheezing may have a higher risk of asthma and different pathophysiology and prognosis. We excluded infants who (1) had neurologic disease with impairment of tone, swallowing, or feeding; (2) were already receiving NG or gastrostomy feeding before admission; (3) had any contraindication to insertion of NG tube (eg, choanal atresia); (4) were unable to have a NG tube inserted. Treating doctors identified eligible patients and all were recruited on admission. Patients could be identified in emergency before admission. The doctors provided verbal and written information regarding the purpose of the study and the processes involved before gaining written informed consent from a parent. If consent was not obtained then routine management processes were undertaken.

This study was approved by the Human Research Ethics Committee (HREC/15/QRCH/8) and followed the ethical principles of Declaration of Helsinki.

Computer generated block randomization was performed, and the treatment allocation was packed into numbered opaque, sealed envelopes. The treating doctor gained consent then used the envelopes to randomly allocate patients to either continuous or bolus NG feeds at a 1:1 ratio. Patients were stratified by age (0–5 months and 6–11 months of age). Group A were commenced on continuous NG feeds (expressed breast milk, or formula) at a rate and volume according to the patient’s weight (two-thirds or full maintenance depending on the clinician’s assessment of hydration; full maintenance determined as per Supplemental Table 4). Group B were commenced on bolus NG feeds at either 2- or 3-hour intervals, and the volume was determined as per Group A. Patient observations (respiratory rate, heart rate, temperature, and oxygen saturation) were obtained at the initiation of allocated NG feeds and again at 2 hours and 6 hours after. The Australasian Bronchiolitis Guideline²  was used to determine severity of bronchiolitis on admission; this coincided with initiation of all patient NG feeds. Respiratory support and duration were documented.

The treating team changed the patient to bolus NG feeds (from continuous NG feeds) or oral feeds (from bolus NG feeds) on morning or afternoon ward rounds in-line with patient improvement as per their usual clinical practice. The treating team, nursing staff, and research team oversaw adherence to the intervention unless a pediatrician or pediatric intensivist deemed it clinically necessary to convert to continuous NG feeds or intravenous therapy. PICU’s common clinical practice is to place patients on NG continuous feeds or intravenous therapy. The patient's NG tube was labeled with the allocated intervention. Enrolled patients who had their feeds altered from bolus to continuous by the treating team, or were admitted to intensive care, exited protocol but remained included in the study with all relevant study data collected for analysis. These patients often had short periods (<12 hours) of intravenous fluid hydration. Families who withdrew consent were requested to allow data collection for analysis.

LOS, measured in hours from admission to discharge time, was the primary outcome measure. A clinically significant difference was considered to be 24 hours. LOS can be affected by many administrative and social factors unrelated to the child’s condition. As such, discharge time for a patient was recorded as the earlier of either (1) actual discharge time in hospital electronic records or (2) time the child was “fit for discharge,” determined by using objective criteria. An infant was considered “fit for discharge” if the infant had not received supplemental oxygen for 12 hours, had stable respiratory status for 4 hours (including slight or nil recession), and was feeding adequately (at least 50% of full quota). Oxygen supplementation is generally applied when an infant’s oxygen saturation as detected by pulse oximeter (Spo₂) was reliably <90% when asleep or <92% when awake.

The secondary outcome measures included the diagnosis of a pulmonary aspiration and whether admission to PICU was required. Pulmonary aspiration was defined as an acute deterioration in respiratory status associated with all 3 of the following (1) clinical concern for aspiration, (2) increase in supplemental oxygen, and (if the first 2 criteria are present) then (3) chest radiograph changes as reported by a radiologist blinded to the child’s allocation. Baseline demographics, clinical characteristics, clinical management, and PICU admissions were recorded by nursing staff, the treating team, or study personnel from electronic hospital records.

In a preceding multisite randomized clinical trial performed by the Perhaps should be ‘Paediatric’ as it is the formal name of their group, not just Australian spelling :) Research in Emergency Departments International Collaborative (PREDICT) group in Australia from 2009 to 2011, the mean (± SD [SD]) LOS in 381 infants with bronchiolitis requiring NG feeding was 85 hours (±58 hours).¹⁰  Based on these data, 93 participants per group were required to demonstrate a difference in LOS of 1 day (24 hours) with power of 80% at a significance level of 5% (by using a 2-sided independent t test). To account for an estimated drop-out rate of 5%, the current study aimed to recruit 100 participants in each arm.

Descriptive statistics were presented for continuous data by using mean and SD or median and interquartile range (IQR) if data were not normally distributed. Normality was assessed by using a Shapiro-Wilk test. Categorical variables were described by using frequencies and percentages. Differences in demographics and baseline clinical characteristics were assessed between groups by using a Mann-Whitney test for nonnormal distributions and χ² test (or Fisher’s exact test, where appropriate) for categorical data.

To assess the efficacy of bolus feeding compared with continuous NG feeding, LOS for each group were compared by using a nonparametric Mann-Whitney test (assumption of normality was not met). Time from admission to discharge (LOS) was analyzed by using the Kaplan-Meier method and compared between groups by using a log-rank test where appropriate. Median LOS were reported along with 95% confidence intervals (CI).

Linear mixed models were used to assess the difference in respiratory rate, Spo₂, temperature and heart rate across time (Timepoint 1 [TP1]: initiation of feed vs Timepoint 2 [TP2]: 6 hours after initiation of feed) and between groups. Time, group, and time by group interaction were considered as fixed effects and patient as random effect. Estimates, 95% CI, and P values were outlined for each outcome of interest. The proportion of infants requiring PICU admission in both groups was compared by using a χ² test.

All analyses were performed using the R statistical software.²³  All tests were 2-tailed, and a P value less than 0.05 was considered statistically significant.

A total of 242 patients were assessed for eligibility, and 201 were subsequently randomly assigned. Nine were excluded due to transfer to a private facility midadmission, and 3 families withdrew their consent. The remaining 189 patients were included in our intention-to-treat (ITT) analysis (Fig 1) with 98 allocated to the bolus group and 91 to the continuous group. Patient baseline demographics, clinical characteristics, and clinical management are detailed in Table 1 and 2. There were no significant differences observed between the groups, including the proportion requiring low-flow nasal prong or high-flow nasal cannula oxygen supplementation in each group. When a nasopharyngeal aspirate was performed (59% [111 of 189]), the majority of patients had RSV detected by polymerase chain reaction (60% [67 of 111], Table 1).

There was no significant difference in LOS between the bolus or continuous NG feeding arms (Table 3) in the ITT analysis (median 54.25 hours [IQR 40.25–82] in the bolus group and median 56 hours [IQR 38–78.75] in the continuous group). No significant difference in LOS was found between bolus and continuous feeding in either age group (0–5 months and 6–11 months of age, Table 3). The ITT analysis Kaplan-Meier curves for “time to hospital discharge (LOS)” are presented in Fig 2 and are similar for bolus and continuous NG feeding groups.

After reviewing secondary outcomes, there was a clinically significant higher proportion of admissions to PICU in the continuous NG feeding group (28.57% [26 of 91] vs 11.22% [11 of 98] of the bolus group) in the ITT cohort (Table 3). No infants sustained a pulmonary aspiration. Eight patients allocated to bolus feeds were converted to continuous feeds by their treating team (50% [4 of 8] within PICU, Table 2) after being allocated to bolus NG feeding. These patients were included in the ITT analysis. A notable number of participants had a previous hospital admission for bronchiolitis (27% of the bolus group, 22% of the continuous group, Table 1); however the observed PICU admission was within usual ranges for this population (17% [8 of 46]).

Discharge was delayed beyond the “fit for discharge” criteria for 31% (59 of 189) of patients (Table 2). The delay was primarily (71% [42 of 59]) because of the patient awaiting discharge clearance by the medical team (Table 2). There were no clinically or statistical differences in rates of hospital readmission within 30 days or number of unscheduled medical visits within 7 days postdischarge (Table 3).

The per-protocol analysis found no significant difference in LOS between the bolus or continuous NG feeding arms (median 51.38 hours [IQR 38.12–77.38] in the bolus group and median 44 hours [IQR 35.25–66] in the continuous group, Supplemental Fig 3 and Supplemental Table 5). Supplemental Fig 4 shows the Kaplan-Meier curves for time to hospital discharge for the per-protocol group. Respiratory rate, Spo₂, temperature and heart rate are described by timepoint (TP1: initiation of allocated feed, TP2: 6 hours after initiation of allocated feed) and NG feed group for the per-protocol cohort in Supplemental Table 6. When comparing these vital signs between TP1 and TP2 of all subjects, there was a significant fall in respiratory rate (−4.57 [95%CI −7.44; −1.71]) and heart rate (−10.2 [95% CI −15.3; −5.09]). There were no other clinically or statistically significant differences between timepoints, between bolus and continuous groups, nor between time by group interaction for oxygen saturation and temperature within 6 hours of allocated NG feed initiation (Supplemental Table 6).

In Australia, NG feeding is the preferred method of fluid supplementation for infants who cannot feed adequately orally and are hospitalized with bronchiolitis. There are no previous studies comparing bolus and continuous NG feeding, therefore, international practices vary according to institution or clinician preference. A Cochrane Review of preterm infants <1500g found that babies receiving continuous feeding may reach full enteral feeding later than intermittent feeding. However, the evidence was of low certainty. The clinical risks and benefits of each regimen could not be reliably discerned.²⁴ 

In this study, no significant difference in LOS was found between bolus versus continuous NG feeding regimens in infants hospitalized with bronchiolitis. Continuous nasogastric feeding was hypothesized to prolong LOS; however, it was unlikely to alter LOS by 24 hours. Additional patient recruitment to detect a 12-hour difference in LOS would be clinically significant.

A higher percentage of the continuous NG feeding group were admitted to PICU. Patients in PICU received high-flow nasal cannula oxygen at an elevated rate (3L/kg/min, Table 2). Each cohort had similar bronchiolitis severity ratings and high-flow nasal cannula requirements. As the current study was an open label study, a number of factors may have contributed to this observed difference. There could be an element of subjectivity in clinical decision-making for PICU admissions; patients on continuous NG feeds could have been perceived as being sicker, or there could be another unaccounted for reason leading to escalation in clinical care. Further exploration of this in future research would be beneficial.

The clinical practice of using continuous NG feeding in moderate to severe bronchiolitis remains unsupported. Using a bolus NG feeding regimen may best mimic usual infant oral feeding and physiology. Active infants can usually manage easier with less time connected to a mechanical device. The bolus NG feeding group had equivalent vital signs and oxygen supplementation profiles. In future, international clinical practice guidelines can support infants hospitalized with bronchiolitis to be initiated on either bolus or continuous NG feeding regimens without impacting their hospital length of stay (Supplemental Tables 5 and 6). Clinicians can make a considered, patient-centered decision.

The LOS study data were reliably available from hospital electronic records and included a definition of “fit for discharge” criteria. This assisted in preventing bias because of social or practical issues affecting discharge time. There is a potential that patients could have received other escalations of clinical care such as an increased nurse-to-patient ratio, or a PICU clinician review that was not accounted for within the study. Approximately 24% of patients had a previous admission with bronchiolitis; this aligns with a study of 296 618 infants finding that 21.3% had more than 1 bronchiolitis admission during their first year of life.²⁵  The 20% PICU admission rate for trial participants aligned with the 22% PICU admission rate for this institution throughout 2018 and 2019. Comparatively, 24.5% of infants hospitalized with bronchiolitis in 2019 were admitted to intensive care in the United States.²⁶ 

The results are generalizable to a wide population of infants with bronchiolitis; however, the study did have limitations. The interventions could not be masked after allocation for pragmatic reasons. Staff within the emergency department, general pediatric wards, and PICU were educated about the trial. This may have sensitized the clinicians to transition earlier in both groups, thereby potentially reducing the absolute difference between the 2 groups.

Younger infants are at risk for more severe bronchiolitis so patients were stratified into 0 to 5 months and 6 to 11 months of age to account for their potentially longer LOS.⁵  No statistical difference in LOS was found between bolus and continuous feeding in either age group. There were clinically significant differences; however, the study was not powered to detect subgroup differences in the LOS between infants less than or more than 6 months of age.

The study did not aim to compare the differences in the cost-effectiveness of 2 methods of NG feeding. Parental satisfaction was not assessed in this study; however, it is important to note that several families within the continuous NG feed group raised concern for child hunger to the research team. This interesting anecdotal finding related to clinical care within our study suggests further studies should include this aspect within their outcomes.

There was no significant difference in LOS between bolus and continuous NG feeding in infants hospitalized with bronchiolitis. There were no clinically significant aspiration events; however, a higher proportion of PICU admissions was noted in the continuous feeding group. The current study suggests there is clinical equivalence of bolus versus continuous NG feeding regimens for fluid supplementation in infants hospitalized with bronchiolitis. Further research within the critical care setting that is powered for infants <6 months of age would be of interest.

We sincerely thank Shreya Bhushan, Shaun Cousen, Katherine Harlen, Veenaa Pakeerathan, William Saunders, Calvin Skews, Suzen Siew, Timothy Teo, Mounavi Vemula, Noemi Lie Man Wong, and Julia Zhu from QCH for their significant contributions of data collection. We also thank the QCH general pediatric department, general pediatric wards, emergency department (including the Paediatric Emergency Research Unit and in particular Natalie Phillips), and PICU (including the Paediatric Critical Care Research Group) teams for championing and facilitating the study.