Oral Feeding on High-Flow Nasal Cannula in Children Hospitalized With Bronchiolitis Free

We conducted a single-center, retrospective chart review at a university-affiliated, tertiary care, freestanding children’s hospital from March 2017 to May 2020. We included children <24 months old admitted to the pediatric ward on the hospital medicine service with a primary diagnosis of bronchiolitis (International Classification of Diseases, Tenth Edition code J21). If patients had another unrelated admission to the hospital for bronchiolitis during this period (eg, in the subsequent year), only the initial bronchiolitis admission requiring HFNC was included. HFNC was defined as heated, humidified oxygen at flow rates >2 liters per minute (LPM) delivered through a Fisher & Paykel system. As per our HFNC bronchiolitis clinical practice guideline, associated order set, and respiratory therapy protocol, the patient first receives nasal suctioning, and then a decision to start HFNC is determined using the respiratory severity of illness score (Supplemental Table 5). This score includes respiratory rate (RR), oxygen saturation, general appearance, breath sounds, and work of breathing. Patients with a moderate to severe respiratory severity score are started on the maximum HFNC LPM allowed in the pediatric ward and weaned per protocol. During the study period, the maximum HFNC flow rate was increased from 6 LPM to 2 L/kg/min up to a maximum of 12 LPM. Patients are allowed to orally feed ad-lib on HFNC on admission. NG tube placement for feeding is not common practice for patients with bronchiolitis at our institution. Oral feedings may be held at any point during admission for persistent tachypnea, defined as RR consistently >60 breaths per minute, or at the provider’s discretion. Patients may also be made NPO by the provider for severe respiratory distress concerning acute decompensation or escalation to nasal continuous positive airway pressure (nCPAP).

Patients were excluded if they had a chronic medical condition, chronic lung disease, congenital heart disease, bacterial pneumonia, or weight <4 kg. Chronic medical conditions included patients with hypotonia, craniofacial anomalies, genetic disorders, metabolic disease, neuromuscular disease, or gastrostomy tube dependence. We excluded patients with initial respiratory support of nCPAP or who transferred to the PICU before their first feed. Patients were also excluded if their records had incomplete documentation of feeding variables (eg, no feeding route, or no intake recorded in the flowsheets during the entire hospitalization, but nursing notes documented the patient feeding well and the patient was not on intravenous fluids [IVF] and did not have an NPO order). This study was reviewed and determined to be exempt from patient consent by our institutional review board.

Patients who met the above criteria were identified through data query from the electronic medical record (Epic Systems Corporation, Verona, Wisconsin, United States). Outcome variables and covariates were obtained by electronic data query (TR) and manual chart review (SS, BL, ML, EMA). Codes were developed and manualized to ensure consistency during the chart review process. Validation of the electronic data query was performed via chart review. A random sample of charts (15%) from each reviewer was checked biweekly by a second reviewer to ensure the accuracy of the manual chart review. Any identified discrepancies were discussed by all reviewers at a biweekly meeting until a consensus was reached. Review criteria for manually extracted data were regularly reviewed at this meeting to maintain reliability and avoid drift.

The primary outcome measures were the incidence of aspiration pneumonia and the incidence of adverse feeding events among children with bronchiolitis on HFNC. Aspiration pneumonia was defined by clinical documentation of this diagnosis in electronic medical record notes or occupational therapy (OT) assessment, treatment of pneumonia because of aspiration of feeds, or radiographic evidence of aspiration pneumonia. Radiographic diagnosis of aspiration pneumonia was determined through manual chart review on the basis of attending physician documentation and/or radiology report of chest radiograph result.

Adverse feeding events were defined as documented choking or gagging with feeds, aspiration event during feed, or feed(s) held because of worsening respiratory distress while feeding. Adverse feeding events were determined via manual chart review by performing keyword searches of the patient record to identify any documentation of the following terms: choke, choking, gag, gagging, aspiration, feeds, or feeding held. Nursing notes and physician documentation containing any of these keywords were reviewed for additional details and descriptions of any events. The reviewer confirmed whether the event was associated with feeding and should be characterized as an “adverse feeding event”.

Secondary outcome measures included escalation of respiratory support to nCPAP, PICU transfer, NPO status and duration, LOS (in hours), the incidence of 7-day readmission, IVF use, and use of any other route of feeding. These variables were obtained via electronic data query and manual chart review. Patients who were made NPO were screened by manual chart review to ensure the reason was not associated with feeds.

Additional covariates collected by electronic data query included patient demographics (date of birth, race, ethnicity, insurance status), weight, problem list diagnoses, viral testing results, maximum oxygen flow rate, baseline RR, and maximum RR. Additional covariates were also collected manually by chart review by using a structured data collection tool. Manual data extracted included gestational age, smoke exposure, maximum respiratory support in pediatric ward, additional routes of feeding, time to first oral feed in hours (calculated from “time of admission” to “time of first feed”), and reason for readmission.

Descriptive analysis was conducted by using means ± standard deviation (SD) or medians (interquartile range [IQR]). We used t tests or Mann-Whitney tests to compare differences between patients who were made NPO versus patients who were not made NPO. All statistical analysis was performed by using SAS 9.4.

Between March 2017 and May 2020, 876 children <24 months of age were admitted to the pediatric ward for bronchiolitis and received HFNC. Two hundred patients did not meet the inclusion criteria, leaving 676 patients for analysis (Fig 1). The average age was 9.4 (SD ± 6.3) months with a predominance of males (62.9%), full-term (82.3%), and Hispanic (48.1%) and white (27.7%) ethnic/racial background (Table 1). The average initial RR was 53 ± 12 with a maximum RR of 62 ± 10. Most patients’ initial respiratory support in the pediatric ward was HFNC (85%); the remainder were escalated to HFNC from room air or standard nasal cannula. The average flow rate was 7.9 ± 2.4 LPM (weight-adjusted 1.0 ± 0.4 L/kg/min) with a median of 8 (IQR 4). Viral testing was performed in 183 patients with 52.9% positive for respiratory syncytial virus and 28.4% for rhinovirus/enterovirus. The median length of stay was 56.8 hours (IQR 40.2 hours).

No patients were diagnosed with aspiration pneumonia clinically or radiographically during admission (Table 2). All patients were fed orally (breast, bottle, solid food, or combination) with a median time to the first oral feed of 2 hours (IQR 3 hours). No NG tubes were placed in any children. Most patients also received some IVF (82.5%). Adverse feeding events were rare with only 11 total patients (1.6%) having a documented event: 3 patients (0.4%) had concern of possible aspiration or microaspiration events, 4 patients (0.6%) had clinical signs of coughing, choking, or gagging with feeds, and 5 patients (0.7%) had worsening respiratory distress with feeding (Table 2 and Table 3). The flow rate at the time of the event varied between 0.5 to 1.9 L/kg/min. Of these 11 patients, 4 were made NPO and 2 were escalated to nCPAP; none were transferred to the PICU (Supplemental Fig 2). Of the 4 patients made NPO after the event, 1 choked with a feed, 2 had worsened respiratory distress with feeding, and 1 had concern for possible aspiration with worsening distress after feeding. None were ultimately diagnosed with or treated for aspiration pneumonia, and no patients were readmitted for aspiration pneumonia. Of the 2 patients with worsening respiratory distress during feeding who were made NPO, both required escalation of respiratory support to nCPAP after the episode. The remaining 9 patients with adverse feeding events improved after modifications were made to feeding, such as pacing, positioning, smaller volume feeds, and OT consultation.

In total, 55 children (8%) were made NPO by a physician while on HFNC, including the 4 patients noted above who had an adverse feeding event. The remainder were made NPO because of concerns for worsening respiratory status, not associated with feeds. Patients remaining on HFNC were made NPO for a median duration of 4 hours (IQR 11); most continued to feed orally throughout the rest of their hospitalization. Among patients on HFNC, those who were made NPO were younger (6.4 months vs 9.7 months, P <.001), had higher maximum RR (67 vs 62, P <.001), and had a longer LOS (104 hours vs 61 hours, P <.001) than those who were not made NPO (Table 4). All patients returned to full oral feeding by discharge.

In this retrospective study of 676 previously healthy children admitted to the pediatric ward with viral bronchiolitis, oral feeding on HFNC of up to 2 L/kg/min with an overall maximum of 12 LPM was well-tolerated. All patients were able to feed orally without anyone being diagnosed with aspiration pneumonia. Patients tolerated feeds well with a low incidence of adverse feeding events, such as choking on feeds or increased respiratory distress (1.6%). To our knowledge, this is the largest study of patients with bronchiolitis treated with HFNC in the pediatric ward who were fed orally from admission.

HFNC feeding practices vary across institutions, and, often, oral feeds are delayed until clinical improvement, weaning of HFNC, or after the approval of the physician, OT, respiratory therapist, or nurse.^24–26  Although more recent studies performed in PICU settings suggest that feeding is well-tolerated on HFNC, the initiation of feeds was determined by a physician when support was weaned and the patient was improving.^21–23  Our study extends the results beyond PICU settings and suggests that oral feedings started on admission to the pediatric ward while patients are on HFNC can be well tolerated.

The practice at our institution has been to allow patients to feed orally on HFNC from admission with nursing parameters in place to hold feeds for persistent tachypnea (RR >60 breaths per minute) or if concern for acutely worsening clinical status. If a patient is told to stop feeding, the patient was made NPO. Most patients were allowed to feed orally for the majority of their hospitalization, and few patients were made NPO. Patients tolerated all routes of oral feeding, including solid food, which is typically more concerning than a liquid diet for aspiration risk. None of our patients had NG tubes placed. Because most of our patients feed orally for the majority of the hospitalization, we believe placing an NG tube is likely an unnecessary intervention. Most patients received IVF, likely because of dehydration in the setting of increased insensible losses and inability to meet hydration needs during acute respiratory illness. Nevertheless, all patients were able to meet hydration needs by oral intake and wean off IVF by the time of discharge.

Patients who were made NPO because of concerns for worsening clinical status were more likely to be younger, have higher respiratory rates, and have longer LOS, possibly indicating they were a more vulnerable population within the study sample. One-half of these patients remained on HFNC and were still able to feed orally during most of their hospitalization with an average NPO time of only 4 hours. The decision to make a patient NPO was at the clinical discretion of the treating provider. This decision is often subjective and varies between different providers. Adverse feeding events may have been prevented by making these patients NPO; however, it is possible that some of these patients may have still been able to feed safely.

There are currently no best practice guidelines to help clinicians decide when to feed children with respiratory distress.²  HFNC practices between institutions vary, including whether HFNC can be used in the pediatric ward or whether it should only be used in the PICU. In sicker children who may be in the ICU, their severe respiratory status may be a higher priority than feeding and may lead to more patients being NPO than necessary. Traditionally, reasons to withhold oral nutrition from patients with moderate to severe bronchiolitis on HFNC include concern for perceived swallowing dysfunction and risk of aspiration, increased secretions, upper airway edema, and iatrogenic nasal prong airway obstruction.^6,13–17 

In our sample, most patients in the adverse feeding group did not require any additional support and improved with small modifications to their feeding. We do not know the baseline rate of “feeding events” in healthy infants, and it is possible these patients had “adverse feeding events” at baseline because of feeding practices that were also used at home and were unrelated to their acute illness. We had 4 patients who were evaluated by OT for concern for swallowing dysfunction. No patient in our study met clinical concern to prompt a formal dysphagia study. Nevertheless, OT assistance with pacing or adjusting nipple size seemed to resolve feeding issues and allowed a continuation of feeding, as was seen in 3 of our 4 patients.

In addition, pediatricians are often trained to be wary of aspiration events when feeding during any respiratory difficulty. Individual provider concerns about aspiration events likely prevent providers from allowing oral feeding when patients are on HFNC. However, our study of primarily healthy infants over a 3-year period without chronic medical conditions did not find any aspiration events. Providers may also have higher concerns for aspiration among premature infants, but in our patient population of premature infants who were >4 kg and without comorbidities, there was no increased morbidity. We specifically did not include any patients with underlying hypotonia or craniofacial abnormalities because those issues could be confounders and increase the risk of baseline aspiration.

An unintended negative association of withholding nutrition in patients admitted for bronchiolitis is an increased LOS.^9,11  However, we were unable to compare children who were fed with those not fed orally from admission. Therefore, we were not able to determine if feeding impacts LOS. The authors of future studies should prospectively compare children on HFNC who are allowed to feed orally from admission with those who were kept NPO until there was respiratory improvement to see if this impacts LOS, ICU transfers, or adverse feeding events.

Our study has several limitations to note. The study was a retrospective review at a single tertiary care institution and the assessment of adverse feeding events was limited by documentation in the EHR. If nurses or parents stopped feedings briefly and this was not documented in the EHR, we could not capture this information. However, patients should have NPO orders if parents were told to stop feeding their child. Therefore, we feel we have captured the majority of what clinically occurred. Our hospital has a robust pediatric ward with an established clinical practice guideline and respiratory therapy protocol, which may not be generalizable to institutions with limited support for higher acuity patients. The results may also not be generalizable to higher flow rates >12 LPM, as the average maximum flow rate was 1 L/kg/min; however, a subset of patients received 2 L/kg/min and did not experience adverse outcomes. In addition, we only evaluated readmission data at our own institution. Nevertheless, we care for >90% of the pediatric inpatient population in our large catchment area and children with more severe illnesses are typically transferred to our facility for care. Finally, we aimed to study otherwise healthy children with bronchiolitis, so we cannot draw conclusions regarding children with bronchiolitis with comorbid chronic conditions, such as hypotonia or neuromuscular disease, that may affect their ability to safely feed orally. We also did not include children with severe illness who initially required PICU level of care or nCPAP.

Oral feeding on admission while on HFNC in otherwise healthy children hospitalized with bronchiolitis on a pediatric ward was well-tolerated. Policies that restrict oral feeding on HFNC in patients with bronchiolitis should be reconsidered. This study, however, is limited by its retrospective, single-center, descriptive design, and future prospective studies with a control group are needed.