Management Practices for Standard-Risk and High-Risk Patients With Bronchiolitis Free

This descriptive, retrospective cohort study was performed in the setting of a 250-bed academic tertiary-care children’s hospital, including a 36-bed PICU. This study was deemed exempt by the institutional review board. An Epic (Epic Systems Corporation 2020©) report was created by using the following parameters: patients admitted to the general pediatric ward, <2 years of age, with an International Classification of Diseases, 10th Revision discharge diagnosis code of bronchiolitis or viral syndrome with evidence of lower respiratory tract involvement (International Classification of Diseases, 10th Revision codes: J21.9, J11.1, J21.0, J42, B97.89, B97.0, J21.1, J21.8, J84.115, B97.4, J12.1, R84.5, R09.02, R06.03) between October 16, 2016 and March 31, 2019. Of the patients in the Epic report, the following exclusion criteria were applied: patients with a diagnosis of sepsis, bacteremia, or empyema, patients with a primary discharge diagnosis of bacterial pneumonia, or patients who were admitted directly to the PICU. Patients were defined as high-risk in the presence of the following conditions on the basis of AAP guidelines and age-based risk factors for severe disease:

1. Prematurity (gestational age <35 weeks)

2. Age <12 weeks

3. Presence of any of the following comorbidities (chronic lung disease [CLD], including those who are oxygen-dependent at baseline, congenital heart disease [CHD], neuromuscular disease, and/or immunodeficiency)¹ 

Demographic characteristics were recorded for standard-risk and high-risk groups, as well as the frequency of diagnostic practices. Risk factor prevalence was recorded; patients could have >1 risk factor counted. Diagnostic practices included viral pathogen testing (respiratory pathogen panel by polymerase chain reaction [PCR] or respiratory viruses by PCR [influenza and respiratory syncytial virus (RSV)]), blood work (complete blood count, blood gas, blood culture, procalcitonin or electrolyte panel), and chest radiography (CXR). Data collected on clinical management included the total number of albuterol, hypertonic saline, or oral corticosteroid doses (classified as either none, 1, 2–3, or >3 doses), the use of percussion and postural drainage (P&PD; defined as chest physiotherapy and prone positioning, respectively), the utilization of high-flow nasal cannula (HFNC; defined as ≥4L/min), noninvasive ventilation (defined as needing continuous positive airway pressure [CPAP] or bilevel positive airway pressure [BiPAP]), and frequency of intubation. HFNC defined as ≥4L/min has been previously used in Cochrane review and in Pediatric Advanced Life Support.^20,21  Admission criteria for the PICU included the need for HFNC flow rates greater than institutional guidelines (institutionally determined age-based cutoffs), the need for noninvasive or invasive mechanical ventilation, clinical concern for further decompensation, or other ICU-level needs. The maximal flow rate of HFNC (L/kg per minute flow) was documented by admission weight. HFNC maximum flow guidelines during the study period were as follows: age <1 month, 1 to 4 L/min; age >1 month to <1 year, 1 to 8 L/min; age >1 year to <8 years, 1 to 12 L/min; age ≥8 years, 5 to 15 L/min.

Primary outcome measures included the need for escalation of care, including the activation of the rapid response team (RRT) and/or transfer to the ICU. To assess the association between risk status, diagnostic testing, and therapy practices with escalation of care, multiple logistic regression was performed. High-risk status was included as an exposure variable for primary outcome measures in multiple logistic regression, in addition to the following covariates: viral pathogen testing, RSV presence, laboratories obtained, CXR obtained, P&PD use, steroid use, hypertonic saline use (by 1 dose, 2–3 doses, or >3 doses), and albuterol use (by 1 dose, 2–3 doses, or >3 doses). HFNC use, noninvasive ventilation use, and intubation were analyzed as descriptive variables only and not used in logistic regression models because of the suspected inherent strong association with RRT activation/ICU transfer, in addition to the generalizability of the results because of unmeasured confounders not accounted for in the study design.

The study population was pulled by electronic data report. Data were collected by a group of trained study team members via a structured chart review. Interrater reliability was assessed for a subset of 10% of total charts. For variables having <90% congruence, data were reanalyzed for all patients. Data identified as needing reanalysis included the presence and type of high-risk conditions and maximum HFNC rate. Variables met 100% congruence after chart review by lead investigators.

Data are reported as n (%), median (interquartile range [IQR]), or mean (standard deviation). For categorical/proportion data, χ-square tests were used. The Student’s t test was used for comparisons of normally distributed variables as determined by the Shapiro-Wilk test, and the Wilcoxon rank test was performed for nonnormally distributed variables. Multiple logistic regression was performed for analysis of RRT activation and transfer to the ICU. Pairwise testing of dosing frequency categories was performed by using χ-squared testing. Bonferroni correction was performed to obtain adjusted P values for pairwise testing and multiple logistic regression. Tests were considered significant if P <.05.

A total of 376 patients were included in the Epic report. After exclusion criteria were applied, 265 (70.5%) patients were left with 122 (46.0%) of those classified as standard-risk and 143 (54.0%) high-risk. Table 1 describes the characteristics of included patients. The most common high-risk conditions were age <12 weeks (n = 100, 69.9%) followed by gestational age <35 weeks (n = 40, 28.0%). Eighteen patients in the high-risk group had 2 or more risk factors present; 11 patients had a history of prematurity and CLD; 2 patients had both CLD and CHD; 1 patient had CHD and neuromuscular weakness; only 4 patients <12 weeks of age had a concomitant history of prematurity but none of these patients had CLD. The patients in the high-risk group were overall younger than in the standard-risk group (median age 2.0 months [IQR: 1.0–4.0] vs 12.0 months [IQR: 7.0–18.0 months], P <.001) and weighed less (median weight 5.2 kg [IQR: 4.1–6.5 kg] vs 9.4 kg [IQR 8.3–11.3 kg], P <.001). There was no statistical difference in diagnostic workup (viral pathogen testing, laboratory work, and CXR) between the standard and high-risk groups.

Albuterol use was frequently observed among standard-risk and high-risk patients, with more albuterol use in the standard-risk group (any albuterol dosing, standard-risk 65.6% vs high-risk 44.1%, P = .003; median albuterol dosing, standard-risk 1.0 doses [IQR: 0.0–6.0] vs high-risk 0.0 doses [IQR: 0.0–3.0], P = .007). Albuterol use was significantly higher for the standard-risk group when comparing the administration of a single dose with none at all between the standard- and high-risk groups (P = .033). The use of multiple doses (>3 doses) was frequently observed (standard-risk 28.7% and high-risk 23.8%; Fig 1). For patients ultimately needing ICU admission, albuterol use was consistent across both groups (standard-risk 100% and high-risk 85%, P = .11). Multiple logistic regression analysis revealed more frequent dosing of albuterol (eg, 2–3 or >3 doses) to be associated with increased RRT activation, but not ICU transfer (Tables 2 and 3). There was no difference in the frequency of hypertonic saline use between groups, and the majority of patients received >3 doses (standard-risk 12.3% and high-risk 16.1%), which was also associated with RRT activation and ICU transfer. Increased steroid use was observed in the standard-risk group (any steroid dosing, standard-risk 19.7% vs high-risk 14.7%, P = .018), with a single administration of steroid dosing more commonly observed for the standard-risk group (P = .033) but 2 to 3 doses more commonly observed in the high-risk group (P = .039). The practice of P&PD was commonly used in both standard and high-risk groups (44.3% vs 40.6%, P = .629, respectively; Fig 1) and was associated with RRT activation and ICU transfer (Tables 2 and 3).

No differences were observed for frequencies of need for HFNC (standard-risk 60.7% vs high-risk 53.8%, P = .321), noninvasive ventilation (standard-risk 5.7% vs high-risk 8.4%, P = .551), or intubation (standard-risk 0.0% vs high-risk 2.1%, P = .305; Fig 2). The high-risk group required a higher average max flow based on weight (standard-risk mean 0.8 L/kg/min [95% confidence interval (CI) 0.6–1.0] vs high-risk mean 1.3 L/kg/min [95% CI 0.9–1.7], P = .001).

There was no difference in the frequency of RRT activation or ICU transfer between standard- and high-risk groups using χ-square analysis (Fig 2). Logistic regression models did not reveal high-risk status to be more associated with the need for RRT activation or ICU transfer (odds ratio RRT: 2.72 [95% CI: 1.06–7.03], P = .497; odds ratio ICU transfer: 3.78 [95% CI: 1.32–10.86], P = .177; Tables 2 and 3).

With this retrospective cohort study, we aimed to describe and compare current care in standard-risk versus high-risk patients with bronchiolitis admitted to the general pediatric ward in a single-center academic tertiary care hospital. Our findings reveal that many non-evidenced-based diagnostic tests and therapies continue to be used and that they are often used regardless of risk status. This poses the clinical question of why providers chose to use these therapies, even for patients without an obvious clinical indication based on high-risk status (eg, young age, history of prematurity, comorbidities).

We found that nonevidence-based diagnostic testing and therapies for bronchiolitis (eg, albuterol, hypertonic saline, P&PD, and corticosteroids) were frequently used among both risk groups. This is consistent with other studies that have examined the use of or aimed to reduce the use of therapies known to be ineffective in bronchiolitis.^22–28  This finding, at a minimum, reveals an institutional deviation from best-practice recommendations. The use of these therapies may reflect variations of care practices in different departments but also remains inconsistent with trends observed nationally since the introduction of a revised bronchiolitis guideline.⁴  Surprisingly, our data indicate there was significantly more albuterol and oral corticosteroid use in the standard-risk group compared with the high-risk group at our institution, although there was no difference in rates of HFNC use. It is unclear whether these therapies benefit the high-risk group, but the observed increased frequencies, at least in the standard-risk group, may reflect a resistance toward the de-implementation of ineffective medical therapies,²⁹  which may take time and continued multidisciplinary collaboration to achieve. Additional focus on whether de-implementation strategies for the high-risk group are needed should also be considered.

The increased use of albuterol and corticosteroids in the standard-risk group compared with the high-risk group was unexpected. Although the location of albuterol use was not determined (ie, whether given in the emergency department vs general care ward vs ICU), the frequent use of >3 doses of albuterol indicates a tendency toward use beyond the simple trial of therapy. Although a trial of albuterol in select patients is an understandable practice, the number of patients in both groups receiving >3 doses of albuterol points toward deviation from recommendations. This may have been due to the perceived benefit of albuterol use for some patients; however, this finding likely indicates continued overuse by clinicians. It was unknown if individual patient factors such as the presence of wheezing or a family history of atopy/asthma impacted the use of various therapies in these 2 groups. However, we suspect that few patients in the study would have routinely used bronchodilators at baseline given the overall young age of included patients (standard-risk median age 12 months, high-risk median age 2 months); the estimated prevalence of childhood asthma being only ∼2.6% for children aged 0 to 4 years³⁰  and the small number with underlying CLD or neuromuscular weakness (10.5% and 2.1% of the high-risk group, respectively). Our data are consistent with studies describing high early use of albuterol with an older age demographic for hospital admissions for bronchiolitis,³¹  especially for the standard-risk group.

The finding of increased therapy usage in the standard-risk group raises the question of what factors providers rely on for the escalation of therapy. Providers may not be considering underlying risk status in their clinical assessments, instead reverting to an inclination to trial and continue a therapy in the hope of clinical improvement or using other clinical factors, such as laboratory values, respiratory scores, etc. to determine care plans. Although multiple studies have revealed that albuterol does not lead to disease resolution, decrease hospitalization, decrease the length of stay,^32,33  or mitigate the need for either noninvasive or invasive ventilation,⁷  it is still frequently used. Our finding of albuterol, hypertonic saline, and P&PD use being associated with RRT activation supports the idea that providers trial therapies in the setting of worsening respiratory status regardless of underlying risk status and evidence for use. Future studies should investigate provider rationale for the trial of therapy, with particular regard to risk status. Studies for those patients who are classified as high-risk should also investigate which therapies are beneficial in mitigating the severity of clinical illness, including the timing and frequency of use.

Our study has several limitations. First, this study describes practice at a single institution and likely does not reflect observed adherence to best-practice recommendations for care at other institutions. The patients included in the study were admitted to the general pediatric floor, and although transfers from the general floor to the ICU were analyzed, patients admitted directly to the ICU were not included. This may have affected data because patients presenting with severe respiratory distress or deemed to be at high-risk for decompensation would not have been included. The increased association of various therapies with RRT activation may reflect confounding by indication because sicker patients will inherently draw increased attention and management strategies to mitigate the severity of illness. This substantially limits the interpretation of this finding and is a fundamental challenge posed by the study design. In addition, our institution’s guidelines for HFNC flow rates were not based on patient weight during the study period. This may account for the observed difference in higher average flow rate for high-risk patients, whereas the same rate was initiated despite patients being younger and weighing less. It also may have influenced the number of patients who required escalation of care because they exceeded the accepted threshold of HFNC on the general care ward. Additionally, the frequency of patients with a family history of atopy/asthma or previous wheezing was unknown for patients who received albuterol.

Another key limitation is the varied definition of high-risk status. Although we defined the high-risk group on the basis of AAP guidelines for bronchiolitis management, we acknowledge that other conditions, such as trisomy 21 and complex chronic conditions, place children at increased risk for worse outcomes. Similarly, the utilization of specific interventions was not sub-analyzed on the basis of specific high-risk conditions (such as CLD or neuromuscular weakness), which limits the interpretation of observed differences between the 2 groups. Our definition of high-risk, as well as the small sample size, may also have impacted our finding of high-risk status no longer being associated with the need for RRT activation or transfer to the ICU after adjusting the logistic regression model using Bonferroni correction. Other measures of illness severity, such as respiratory score, were similarly not measured or corrected for, which may confound results. Finally, because many studies have revealed increased rates of noninvasive ventilation and intubation in those with high-risk features,^6,7,9,10,34  our study was not powered to detect differences in these overall infrequently observed events (posthoc power analysis revealed power of 18% and 44% respectively).

Our study revealed a varied approach to the management of bronchiolitis in both standard-risk and high-risk children at our institution. Some therapies, such as bronchodilators and steroids, continue to be frequently used despite best practice recommendations and regardless of risk status. Future research is needed to determine interventions, such as rapid response checklists, to reduce the overuse of therapies in all risk groups along with specific focus on interventions and outcomes for those in the high-risk group.