Acceptability of Deimplementing High-Flow Nasal Cannula in Pediatric Bronchiolitis

We conducted a study of health care providers that included quantitative survey data supplemented by a limited number of semistructured interviews. This study was conducted at a large academic children’s hospital in the mid-Atlantic United States with 226 total pediatric beds. Our institution cared for ∼900 children hospitalized with bronchiolitis in 2019. The study included respiratory therapists (RTs), nurses, attending physicians, pediatric resident physicians, and advanced practice providers (APPs). Pediatric resident physicians’ and APPs’ responses were grouped in with the physicians. Participants were recruited via e-mail from the emergency department, general pediatric unit, and PICU. This study took place from July 2021 to September 2021 and was approved by our local institutional review board.

We collected data via electronic surveys and semistructured interviews. Survey questions were developed to collect data on provider characteristics, perceptions of the benefits of HFNC, and acceptability of HFNC deimplementation (Supplemental Fig 2). Survey questions evaluating provider acceptability were informed using the Theoretical Framework for Acceptability (TFA) by Sekohn et al^21,22  TFA is a multifaceted framework consisting of constructs that can be applied to assess prospective acceptability of health care interventions from the perspective of those who will deliver it. Acceptability, in regard to deimplementation, refers to the degree to which the idea of stopping a practice is perceived as agreeable, palatable, or satisfactory. A survey question was designed for 6 of the acceptability domains in the TFA (Supplemental Fig 3). These domains were (1) affective attitude, (2) opportunity costs, (3) ethicality, (4) burden, (5) self-efficacy, and (6) intervention coherence.

Survey responses were anonymous unless participants provided their e-mail address indicating their interest to participate in an interview. Surveys were tested by the authors to ensure questions with logic sequencing were working as intended and that questions were written clearly. The survey was then administered electronically through REDCap 10.0.20 (Vanderbilt University, Nashville, TN), a Health Insurance Portability and Accountability Act-compliant software.

Providers indicated willingness to participate in an interview on the electronic survey. Each of these providers were approached via e-mail to participate in a voluntary interview. We conducted a limited number of semistructured interviews to collect in-depth data on facilitators and barriers to HFNC deimplementation. The interview guide was informed by the Consolidated Framework for Implementation Research. The Consolidated Framework for Implementation Research framework consists of 39 constructs that have been associated with effective implementation²³ ; we used the 7 constructs most likely to influence HFNC deimplementation to create the interview guide and sought to understand those influences more deeply through in-depth discussion (Supplemental Fig 4). When appropriate, interviewees were asked additional questions to better elucidate individual responses from the electronic survey as well. Interviews were conducted and audio recorded by 2 authors (N.G., and A.N.) using an online conference platform. Audio recordings were manually transcribed by 4 authors.

We evaluated 6 outcomes to understand acceptability of HFNC deimplementation. We used affective attitude as a proxy for overall provider acceptability because this outcome refers to how an individual feels about deimplementing an intervention. We used 5 point Likert scales to record all responses. All scales were categorized as negative (numerical value of 1 or 2), neutral (3), or positive (4 or 5) except in regard to familiarity with evidence in which extremely familiar and moderately familiar were positive responses, somewhat familiar and slightly familiar were neutral responses, and not familiar was a negative response. We hypothesized that affective attitudes regarding HFNC deimplementation would differ based on clinical role, primary hospital setting, years of clinical experience, familiarity with HFNC evidence, and presence of a leadership role.

We compared affective attitudes by provider characteristics using Pearson exact χ², Kruskal-Wallis singly ordered exact, and Wilcoxon-Mann-Whitney exact tests. All analyses were performed with SAS software (v9.4, SAS Institute, Cary, NC) and P values less than .05 were considered statistically significant.

Identification of benefits was elicited by responses to the following question on the electronic survey: “Which of the following are benefits to using HFNC? Check all that apply.” Possible benefits included: (1) parent comfort, (2) provider comfort, (3) decrease in length of stay, (4) decrease in length of oxygen supplementation, (5) decrease the risk of intubation, and (6) decrease the risk of PICU transfer. We hypothesized that providers would identify improvements in clinical outcomes (decreased length of stay, length of oxygen supplementation, and risk of PICU transfer) as benefits to using HFNC despite not being evidence-based.

We stratified these responses by clinical role and compared groups using χ² and Wilcoxon-Mann-Whitney tests. When statistically significant differences were found among the clinical role, a secondary analysis using Pearson χ² or Fisher exact testing was performed to individually compare physician responses with RT and nurses separately. All quantitative analyses were performed with SAS software (v9.4, SAS Institute, Cary, NC) and P values less than .05 were considered statistically significant.

Interview data were analyzed using a theory-informed inductive approach to thematic content analysis.²⁴  After the first 7 interviews, 4 authors met to discuss emerging concepts and themes. Following 2 more interviews, the researchers confirmed no new themes were emerging in the data and that they had reached thematic saturation; data collection then ceased. After completing the 9 interviews, the same 4 authors met weekly to develop a codebook and used this to code each transcript. After group-coding while meeting via an online conference platform, the team reviewed the coded data for discrepancies, which were addressed through negotiated consensus. The 3 most common themes for facilitators and barriers were identified and described with exemplary quotes.

We distributed the electronic survey to 387 providers; 152 responded, for an overall 39% response rate. Of the 387 providers approached, 52 (13%) were physicians, 225 (58%) were nurses, 59 (15%) were RTs, 39 (10%) were pediatric residents, and 12 (3%) were APPs. Of the 152 participants who completed the survey, 44 (29%) were physicians/APPs, 78 (51%) were nurses, and 30 (20%) were RTs. Thirty providers (20%) identified their primary setting as the emergency department, 63 (42%) as the general pediatric unit, and 59 (39%) as the PICU. Participants were diverse in terms of clinical role, primary hospital setting, years of clinical experience, familiarity with HFNC evidence, and presence of a leadership role.

Overall, 83 (55%) providers reported feeling positively about deimplementing HFNC, 51 (34%) reported feeling neutrally and 18 (12%) reported feeling negatively. Physicians/APPs (34; 77%) were significantly more likely to report feeling positively about reducing HFNC use compared with nurses (33; 42%) and RTs (16; 53%) (P ≤ .001 and .03, respectively). Providers with leadership roles were more likely to report feeling more positively about reducing HFNC (P = .04). There was no statistically significant difference in reporting feeling positively based on primary hospital setting (P = .98) or years of clinical experience (P = .55). Table 1 describes provider characteristics and compares providers’ affective attitudes in regard to HFNC deimplementation.

Providers who reported being at least moderately familiar with HFNC evidence (33; 72%) were more likely to report feeling positively about deimplementing HFNC compared with those who reported being less familiar with HFNC evidence (P = .04). Figure 1 illustrates the acceptability of HFNC deimplementation based on each individual outcome.

Providers attributed nonevidence-based clinical benefits to HFNC: 51 (34%) reported decreased length of stay, 57 (38%) reported decreased length of oxygen supplementation, and 79 (52%) reported decreased risk of PICU transfer. Overall, we found that 100 (66%) participants believe HFNC improves at least 1 of these clinical outcomes despite being nonevidence-based. RTs (27; 90%) were significantly more likely to report a nonevidence-based clinical benefit of HFNC than physicians/APPs (14; 32%) (P ≤ .001). Among all providers, 97 (64%) reported decreased risk of intubation as a benefit of HFNC, a benefit that aligns with current evidence.

Of all the participants, 58 (38%) reported provider comfort and 70 (46%) reported parent comfort as a benefit of HFNC. Conversely, physicians/APPs were significantly more likely to report provider and parent comfort as a benefit of HFNC compared with RTs (P = .004 for both comparisons) but not nurses (P = .05 and P = .08, respectively). Table 2 illustrates perceived benefits of HFNC by clinical role.

A total of 35 providers indicated willingness to participate in an interview in the electronic survey; 19 were physicians/APPs, 12 were nurses, and 4 were RTs. We conducted 9 semistructured interviews: 1 RT, 4 nurses, and 4 physicians.

We identified several barriers and facilitators to HFNC deimplementation. The 3 most common barriers included the following: (1) discomfort with not intervening, (2) perception that HFNC helps, and (3) variation in risk tolerance and clinical experience. The 3 most common facilitators included (1) staff education, (2) a culture of safely doing less, and (3) enhanced interdisciplinary communication. Each facilitator naturally addressed each barrier. These are described in Table 3 with exemplary quotes and suggested interventions that were drawn from the interviews.

Reducing the use of HFNC in children with bronchiolitis is an important area of pediatric research. However, to the best of our knowledge, no previous studies have assessed the acceptability of deimplementing HFNC among providers. Overall, providers found deimplementation acceptable with most reporting feeling positively about reducing HFNC use in children with mild to moderate bronchiolitis and very few feeling negatively. We also found that most participants believe HFNC improves at least 1 clinical outcome that is not evidence-based. Finally, we identified several barriers to deimplementation: discomfort with not intervening, perception that HFNC helps, and variation in risk tolerance and clinical experience. Providers also described facilitators such as staff education, a culture of safely doing less, and enhanced multidisciplinary communication. Hospital leaders interested in improving the care of children hospitalized with bronchiolitis can leverage these findings to inform future HFNC deimplementation efforts.

To successfully deimplement nonevidence-based interventions, we must understand providers’ attitudes toward acceptability.^25,26  This is supported by Tyler et al,’s findings that provider attitudes influence adoption of best practices, and that providers at high-using hospitals worry about the potential harms of doing less, whereas those at low-using hospitals focus on the potential harms of unnecessary treatments.¹⁹  This phenomenon was similarly described by Charvat et al., who encountered uncertainty and discomfort among RTs and nurses regarding weaning HFNC as the authors strove to reduce HFNC duration.¹⁷  Our study elicited similar attitudes; however, overall we noted that most providers found it acceptable to deimplement HFNC. Our study adds to the literature by describing the differences in the acceptability by provider characteristics: acceptability varied by clinical role, but not by years of clinical experience or primary hospital setting. Our study also adds that providers who were more familiar with the evidence were more likely to find HFNC deimplementation acceptable. Thus, access to literature illustrating the lack of significant improvements in clinical outcomes with HFNC may increase acceptability and support HFNC deimplementation.

Although a level of acceptability is necessary, it is not sufficient. There are well-established factors that can affect the adoption of interventions that have been applied toward deimplementation. However, there are distinct factors associated with the discontinuation of established practices that are not as well understood. Staff education on the evidence, communication, and permission to change are facilitators congruent with those described by McDaniel et al.’s work exploring physician experience with deimplementation.²⁰  Additional facilitators described by this study include stakeholder engagement, normalization, and preempting expectations of action. In addition, both studies describe provider fear or discomfort with doing less as barriers to deimplementation. Our study additionally described the variation in risk tolerance and clinical experience as a barrier to deimplementation. Although McDaniel et al,’s study was limited to physicians, our study identified similar findings among nurses and RTs.

Providers identified the importance of developing a culture of safely doing less as a key facilitator of HFNC deimplementation. A study by Haskell et al described theory-based strategies to reduce the use of low-value interventions, such as antibiotics and steroids, in children with bronchiolitis.²⁷  Although this study did not specifically address HFNC, the interventions described are similar to those elicited in our interviews. Participants suggested the following as interventions: use of departmental champions to motivate individuals, creation of clinical pathways that prompt recommended behaviors and allow providers to do less in a formal way, dissemination of education, cultivation of multidisciplinary relationships, and real-time results on the impact of deimplementation efforts to alleviate provider fear of potential negative consequences. These strategies can be leveraged by hospital leaders to create a culture for safely doing less. Given the striking impact of local culture on deimplementation efforts^19,20  and the differences in acceptability of deimplementation across provider roles, investigating specific barriers and facilitators across roles and the impact of mitigation strategies by role is an important area for future research.

Limitations of this research should be considered for the generalizability of our findings. First, the study was conducted at a single academic site that may have meaningful differences from other institutions. Second, only 39% of providers completed the survey and a limited number completed interviews. We noted that those finding HFNC deimplementation more acceptable were more willing to participate in the survey and interviews, suggesting a nonresponse or participation bias. Of those who were interviewed, all reported feeling neutral or positive regarding HFNC deimplementation on the electronic survey. Although we identified 3 providers willing to participate in an interview who reported feeling negatively, these providers were ultimately not available to be interviewed. Furthermore, although acceptability among providers in their first few years of practicing clinically may differ from those with more years of experience, we were unable to analyze acceptability among providers with fewer than 6 years of experience. Last, although we reached thematic saturation among those who felt neutrally or positively, obtaining a better understanding of those who felt negatively may inform different barriers and facilitators, and thus, is an important area for future research.

Our study provides insight into the acceptability of HFNC deimplementation in children hospitalized with bronchiolitis. Providers commonly attributed improvements in clinical outcomes as a benefit for using HFNC despite not being evidence-based. Nevertheless, we found that HFNC deimplementation is acceptable among most providers in different roles and hospital settings. Hospital leaders should incorporate staff education, a culture for safely doing less, and enhanced multidisciplinary communication to facilitate HFNC deimplementation.

The authors acknowledge Mike Sheridan and all staff members at our institution who made this study possible. Deidentified individual participant data will not be made available.