High-flow nasal cannula oxygen therapy (HFNC) is increasingly used to treat bronchiolitis. However, HFNC has not reduced time on supplemental oxygen, length of stay (LOS), or ICU admission. Our objective was to reduce HFNC use in children admitted for bronchiolitis from 41% to 20% over 2 years.
Using quality improvement methods, our multidisciplinary team formulated key drivers, including standardization of HFNC use, effective communication, knowledgeable staff, engaged providers and families, data transparency, and high-value care focus. Interventions included: (1) standardized HFNC initiation criteria, (2) staff education, (3) real-time feedback to providers, (4) a script for providers to use with families about expectations during admission, (5) team huddle for patients admitted on HFNC to discuss necessity, and (6) distribution of a bronchiolitis toolkit. We used statistical process control charts to track the percentage of children with bronchiolitis who received HFNC. Data were compared with a comparison institution not actively involved in quality improvement work around HFNC use to ensure improvements were not secondary to the COVID-19 pandemic alone.
Over 10 months of interventions, we saw a decrease in HFNC use for patients admitted with bronchiolitis from 41% to 22%, which was sustained for >12 months. There was no change in HFNC use at the comparison institution. The overall mean LOS for children with bronchiolitis decreased from 60 to 45 hours.
We successfully reduced HFNC use in children with bronchiolitis, improving delivery of high-value and evidence-based care. This reduction was associated with a 25% decrease in LOS.
Bronchiolitis is the most common cause of hospital admission for children under two years of age, leading to substantial health care costs and resource utilization.1 Despite decades of research studying interventions to mitigate the disease process, the standard of care for this disease remains supportive care.2 High-flow nasal cannula (HFNC), a form of respiratory therapy capable of delivering warm, humidified air with titratable oxygen and flow, has been proposed for children with bronchiolitis.3–5 Although initial, small studies demonstrated improvements in physiologic parameters and decreased intubation rates associated with its use, larger, recent studies, including multiple randomized controlled trials (RCTs), have not demonstrated reduced ICU transfer rates or length of stay (LOS).6–10 Even without supporting evidence, its use has significantly increased, potentially contributing to increased intensive care utilization and health care costs in this patient population.11,12
At our institution, HFNC was historically restricted to the emergency department (ED) and ICU. In January 2016, to avoid transferring children to the ICU solely to initiate HFNC therapy, its use was expanded to the general inpatient units. Since that change, HFNC use significantly increased, from 12% of hospitalized children in 2013–2016, to 41% in 2020.13 Although this increase is not surprising, there has not been an associated improvement in patient outcomes.13
The specific patient population most likely to benefit from HFNC remains undefined, which has led to widespread adoption regardless of illness severity and complicates efforts of deimplementation.14 Additionally, bronchiolitis is a condition that is cared for by many providers, including clinicians representing the ED, ICU, and hospital medicine (HM), as well as registered nurses and respiratory therapists, all of whom have their own expertise and face unique challenges.
The increased use of a therapy that did not appear to improve outcomes and contributes to significant health care resources (eg, increased nursing and respiratory therapist assessments, cost of therapy) prompted the formation of a multidisciplinary improvement team tasked with improving the judicious use of HFNC in children hospitalized with bronchiolitis. Our specific aim was to reduce the frequency of HFNC use in children admitted for bronchiolitis from 41% to 20% over 2 years. This goal represented a 50% reduction that was closer to our institution’s HFNC use before its spread to the general inpatient units.
Methods
Context
Our improvement efforts took place at a large, urban, academic medical center in Cincinnati, Ohio, and its satellite campus without an attached ICU. Interventions began in summer 2020 and continued through fall 2021. All patients hospitalized with bronchiolitis were included in this work, regardless of where HFNC was started, and including those patients admitted to the ICU. The COVID-19 pandemic began shortly before our improvement work began and significantly affected the volume of patients hospitalized with bronchiolitis.15 Despite the decreased volume of patients, our improvement team continued to design and implement interventions throughout the pandemic. Following the relaxation of community mask mandates and social distancing, as well as school and daycare reopenings, the number of bronchiolitis hospitalizations increased in summer 2021.
Given the coincident timing of the COVID-19 pandemic and our interventions, and specifically with worries that HFNC may be an aerosol-generating procedure that might cause the reduction of HFNC independent of our improvement interventions, we later collected data from a medium-sized, urban, academic medical center in Rochester, New York. Starting in April 2019, this institution introduced a guideline expanding use of HFNC for bronchiolitis to the general wards in addition to its pediatric ICU, which led to increased HFNC use. Because this institution did not have quality improvement (QI) efforts around reducing HFNC initiation during the pandemic, it served as a comparison to better distinguish if the decrease in HFNC use was driven by improvement interventions or the pandemic. Similar to the primary institution, data were collected on HFNC initiation among all patients hospitalized with bronchiolitis, including those admitted to the ICU.
Before our improvement work, at the time of HFNC spread to the general wards, a clinical practice pathway was developed through group consensus among ED, HM, and ICU leaders to specify which patient population qualified for treatment with HFNC outside of the ICU. The pathway also described management recommendations for patients on HFNC, including feeding practices and weaning methods. Currently, the pathway allows for treatment with 2 L/kg of body weight of HFNC (maximum, 20 L) on the general wards. The pathway does not include any specific clinical criteria to support when to consider HFNC initiation; this decision was at the discretion of the clinical team assessing a patient with bronchiolitis.
Improvement Team
Our multidisciplinary improvement team included physicians and advanced practice providers from HM, the ED, and the ICU, as well as respiratory therapists, registered nurses, and family representatives. The team conducted a failure modes and effects analysis and then identified key drivers (Fig 1), including standardization of HFNC use, effective communication, knowledgeable staff, engaged providers and families, data transparency, and a high-value care focus.
Interventions
Interventions were developed to address key drivers and included: (1) development of standardized HFNC initiation criteria, (2) staff education on appropriate HFNC use, (3) real-time feedback to providers, (4) a script for providers to talk with families about what to expect during hospitalization, (5) implementation of a team huddle for patients admitted on HFNC to discuss necessity of the therapy, and (6) distribution of a bronchiolitis toolkit to staff, containing institutional bronchiolitis protocols and evidence-based care recommendations.
Standardized HFNC Initiation Criteria
To address the lack of clinical criteria supporting HFNC use, our multidisciplinary improvement team first identified evidence-based objective criteria that were used in the RCTs (eg, respiratory rate parameters, standard nasal cannula requirements, specific signs of increased work of breathing) to support HFNC use in patients with bronchiolitis.3,6,7 Our team refined the criteria (Table 1) through group consensus, feedback from frontline providers, and discussions with family representatives. We piloted the criteria using plan-do-study-act cycles.16 We also retrospectively applied the criteria to quantify HFNC overuse at our institution and serve as baseline data to our process improvement. The criteria were distributed to all providers electronically and were posted on nursing units.
Staff Education
Given the evolving evidence around HFNC use, a key intervention was sharing updated education about its mechanism of action and potential for overuse to all providers involved in the care of patients with bronchiolitis. Our team developed an educational module that included updated HFNC facts, including our local frequency of use and cost, details on our improvement project, and our newly developed HFNC initiation criteria. The education was distributed electronically to clinicians from HM, the ED, the ICU, and the Pediatric Residency, as well as bedside nurses and respiratory therapists. Our team also delivered division-level presentations on our work to the stakeholder groups.
Provider Feedback
To provide data transparency to our clinicians, as well as to obtain important qualitative feedback on our project and proposed HFNC initiation criteria, we identified patients who were started on HFNC therapy in real time and contacted providers to both give and elicit feedback on their experience with HFNC. Patients were identified through an electronic health record (EHR) query, and based on where HFNC was initiated (eg, the ED, general inpatient unit), specific team members were asked to follow up with the provider to obtain additional insight into their decision to use HFNC. This intervention allowed our team to gain provider buy-in through 1-on-1 conversations related to specific clinical scenarios. This intervention was limited based on team members’ availability and was ultimately abandoned after ∼6 months, when new feedback was no longer being elicited.
Bronchiolitis Script
Our team developed a bronchiolitis script to be delivered to all families when admitted for bronchiolitis to standardize the admission process and foster family engagement in our work. The script includes an explanation of what bronchiolitis is in family-friendly language, treatment expectations (specifically focusing on the need for frequent nasal suctioning), the overall expected clinical course, contingency plans if the patient is not improving as expected, and an overview of HFNC and how to discontinue the therapy for applicable patients. Copies of the script were kept on the general inpatient units, and charge nurses helped distribute the script to bedside nurses admitting patients with bronchiolitis.
HFNC Huddle
Our team implemented a team huddle for all patients before starting HFNC on the inpatient unit. The huddle includes the involved clinicians, a bedside or charge nurse, and the patient’s respiratory therapist. The purpose of the huddle was to encourage other interventions before starting HFNC, including nasal suctioning, antipyretics, and/or supplemental fluids, as appropriate. The team then would develop a plan for a follow-up evaluation.
Bronchiolitis Toolkit
As a method of providing continuing education, our team distributed an electronic toolkit to our providers and bedside team for easy access to all updated bronchiolitis materials. The toolkit includes a bronchiolitis suctioning algorithm, facts on HFNC, our HFNC initiation criteria, the bronchiolitis admission script, and the updated HFNC clinical practice guidelines. We shared the toolkit via e-mail and our intranet.
Measures
Our primary outcome measure was the percentage of patients hospitalized with bronchiolitis who were placed on HFNC at any time during their hospitalization. We included patients younger than 2 years of age with International Classification of Diseases Tenth Revision codes for bronchiolitis (J21.0, J21.1, J21.8, J21.9), which we extracted from the EHR. There were no exclusion criteria used, and patients with comorbidities were included. The use of HFNC was extracted from an EHR flowsheet. We shared our eligibility criteria and chart extraction tools with colleagues at the comparison institution. They collected data on our primary outcome measure of HFNC use during the same period when they had no active improvement interventions. Our secondary outcome measure was inpatient LOS for patients hospitalized with bronchiolitis (in hours).
Our primary process measure was the percentage of patients started on HFNC that met at least 1 of our HFNC initiation criteria (Table 1). We extracted data for this measure from the EHR. We chose a goal of 90% for this measure because our team recognized that our criteria were not all inclusive and providers might opt to start HFNC in other scenarios that may be appropriate. To monitor the safety of our interventions and improvement work, we followed the rate of ICU utilization, intubations, and readmission rates for patients with bronchiolitis as balancing measures.
We collected baseline data for all measures from September 2018 to March 2020, representing 2 respiratory viral seasons. We reviewed individual charts of 10% of children admitted for bronchiolitis to validate the accuracy of EHR extraction of important variables, including use of HFNC, LOS, vital signs, ICU utilization, and intubation.
Analysis
We used statistical process control p-charts to analyze our primary and balancing measures.17 We used statistical process control X-bar and S charts to analyze our secondary outcome measure. We identified special causes for centerline shifts using control chart rules.18 Both quantitative and qualitative data informed new interventions. Given the variability of admission volumes over the course of the pandemic, as well as the general seasonality of bronchiolitis cases, we used a fixed subgroup of 25 patients on all our primary charts.
Ethical Considerations
This QI work was reviewed by the institutional review boards at both centers and was deemed not to be human subjects research.
Results
A total of 3225 patients were included, 2100 in the baseline period and 1125 in the intervention phase. The percentage of children hospitalized with bronchiolitis receiving HFNC therapy decreased from 41% to 22% over 10 months (Fig 2). This improvement was then sustained for more than 12 months. The initial interventions of HFNC initiation criteria, staff education, provider feedback, and the bronchiolitis script were associated with the centerline shift. Because there was a significant decline in bronchiolitis admissions during the COVID-19 pandemic, the active intervention timeframe did not include many children, which in turn made it difficult to discern which of these specific interventions were the most impactful.
The percentage of children hospitalized with bronchiolitis receiving HFNC therapy at the comparison institution did not show any significant change during the first 18 months of the pandemic, with ∼48% of patients receiving the therapy (Fig 3).
Average inpatient LOS decreased from 60 hours to 45 hours over 10 months, which has been sustained for more than 12 months (Fig 4). The percentage of patients started on HFNC that met at least 1 of our initiation criteria did not change from our baseline of 65% through the interventions (Fig 5). The balancing measure of ICU utilization decreased from 20% to 8% over the course of our QI work (Fig 6), and rates of intubation remained stable, with approximately 1% of patients requiring intubation. Additionally, readmission rates remained stable at ∼2%.
In fall 2021, our institution participated in the American Academy of Pediatrics Value in Inpatient Pediatrics project on reducing HFNC use. Our site was part of the HFNC Initiation Pause group, which required providers to pause and try other appropriate interventions before initiating HFNC in patients with nonsevere disease. There was no temporal association between our participation in the project and a change in the system.
Discussion
Our improvement interventions were associated with a reduction in HFNC use from 41% to 22% in less than 1 year. The improvements were then sustained for more than 12 months. The involvement of stakeholders from all involved divisions, identification of champions to support the work, standardization of the HFNC initiation process, and repetitive education opportunities contributed to our team’s success.
Our improvement efforts also coincided with a sustained 25% decrease in LOS. By decreasing HFNC use, we eliminated the need to wean the therapy and monitor the patient off HFNC before discharge, potentially reducing the time needed for patients to be deemed medically ready for discharge.19 Though most patients experience a relatively short stay of less than 3 days in the hospital, the impact on families with regard to financial stressors and time away from home is large.
Studies have shown that broad use of HFNC in bronchiolitis has not been associated with improved clinical outcomes and may in fact be a contributor to increased ICU utilization in this cohort.6,7,11 A remaining question around HFNC use is whether there is a specific patient population or disease severity level that would benefit from the therapy, and there are no published clinical criteria specifically used to support its initiation in bronchiolitis. The development of HFNC initiation criteria offered guidance to providers, who were previously relying solely on clinical judgment when deciding whether to start HFNC or not. Our criteria appear to be safe when used because we did not observe an increase in intubations after implementation and, in fact, found decreased utilization of ICU-level care. Despite being 1 of our goals, we did not increase the percentage of patients started on HFNC who met at least 1 initiation criteria, suggesting adherence to the criteria alone may not have been responsible for the large overall decrease in HFNC use. Qualitative feedback from providers has supported the impact that standardizing the initiation process has had on bringing awareness to our improvement work, similar to findings seen in other studies.20–22 Additional effort to standardize the use of the criteria more reliably is warranted.
As the evidence around the limitations of HFNC therapy in bronchiolitis grows, the urgent need for rapid deimplementation efforts is clear. Previous studies have shown the feasibility of using QI efforts to decrease time on HFNC therapy and LOS,23,24 and combined with our work, demonstrate the usefulness of QI in addressing deimplementation of HFNC in bronchiolitis. Deimplementation work is inherently challenging because it asks providers to pause and do less, which goes against our desire to intervene and hence improve care.25,26 A specific framework within deimplementation work is behavior substitution, which provides an alternative, more appropriate intervention to replace the one being deimplemented.27 In the case of bronchiolitis, supportive care remains the standard of care, thus reinforcing a previous nasal suctioning algorithm and encouraging other interventions, including standard nasal cannula when appropriate, may have made our deimplementation efforts more acceptable to providers. Directly related to deimplementation work is high-value care, or care that improves health, avoids harm, and eliminates wasteful practices.28 By educating providers on the costs associated with HFNC, including that it is 16 times more expensive than standard nasal cannula,3 we provided additional rationale for clinicians to prioritize patient outcomes while minimizing cost.
Another factor significantly affecting our deimplementation work was the COVID-19 pandemic. Bronchiolitis admissions nearly disappeared during the first year of the pandemic because social distancing, mask mandates, and daycare closures were common. This decline in admissions forced our team to implement small tests of change, which allowed deep learning from each patient as we adapted our various interventions. Additionally, the lack of bronchiolitis cases provided an opportunity for additional education around evidence-based bronchiolitis care, particularly for our learners because they had seen few cases of bronchiolitis in their training thus far and were therefore open to the idea of reserving HFNC use for patients with severe disease.
Our work has several limitations. First, this work occurred at a large academic institution with readily available QI resources, potentially limiting the generalizability of our results. We intentionally included our satellite campus, where there is no onsite ICU, in the hope that our results will be more applicable to hospitals in the community setting. Another limitation is our reliance on data from the EHR, although we conducted a manual chart review to validate all important variables, making this less of a concern. Finally, given the initial concern that HFNC represented an aerosol-generating procedure, the COVID-19 pandemic may have accelerated our initial success. If that were the primary driver of our improvement, however, we would not have expected sustained success as seen in our system. The lack of change in HFNC use at the comparison institution further supports that the improvement seen at our institution was not from effects of the pandemic.
Conclusions
We successfully used QI interventions that were associated with reduced HFNC use in patients admitted with bronchiolitis, leading to improved delivery of high-value and evidence-based care. This improvement was associated with a 25% reduction in average inpatient LOS for this cohort. The next steps include evaluating the impact of our work on additional patient outcomes, including frequency of missed feeds and hospital costs.
Drs Treasure, Brady, and Statile drafted the initial manuscript, conceptualized and designed the study, interpreted the data, and reviewed and revised the manuscript for important intellectual content; Drs Lipshaw, and Dean, Mr Paff, Ms Arnsperger, Mr Meyer, Drs Gillen and Segev, Ms Woeste, Ms Mullaney, Mr O’Neill, and Drs Fallon and Gildner conceptualized and designed the study or led in the acquisition, analysis, or interpretation of data for the work and reviewed and revised the manuscript for important intellectual content; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: This work was supported under a novice research grant from the Gerber Foundation. The funders had no role in the collection, analysis, or interpretation of the data, or in the decision to publish or the writing of the manuscript.
CONFLICT OF INTEREST DISCLOSURES: Dr Treasure discloses grant funding from the Gerber Foundation. The other authors have indicated they have no potential conflicts of interest to disclose.
