The probability of reducing hospitalization rates for bronchiolitis with epinephrine and dexamethasone: A Bayesian analysis

Larry Dong; Terry P Klassen; David W Johnson; Rhonda Correll; Serge Gouin; Maala Bhatt; Hema Patel; Gary Joubert; Karen J L Black; Troy W S Turner; Sandra R Whitehouse; Amy C Plint; Anna Heath

doi:10.1371/journal.pone.0318853

. 2025 May 16;20(5):e0318853. doi: 10.1371/journal.pone.0318853

The probability of reducing hospitalization rates for bronchiolitis with epinephrine and dexamethasone: A Bayesian analysis

Larry Dong ^1,^2,^*, Terry P Klassen ³, David W Johnson ^4,⁵, Rhonda Correll ⁶, Serge Gouin ^7,⁸, Maala Bhatt ^9,¹⁰, Hema Patel ^11,¹², Gary Joubert ¹³, Karen J L Black ¹⁴, Troy W S Turner ^15,¹⁶, Sandra R Whitehouse ¹⁷, Amy C Plint ^10,^18,^☯, Anna Heath ^1,^2,^19,^☯

Editor: Dhammika Leshan Wannigama²⁰

¹Division of Biostatistics, Dalla Lana School of Public Health, University of Toronto, Toronto, Canada

²Child Health Evaluative Sciences, Hospital for Sick Children, Toronto, Canada

³Children’s Hospital Research Institute of Manitoba, Department of Pediatrics and Child Health, Rady Faculty of Health Sciences, University of Manitoba, Winnipeg, Manitoba, Canada

⁴Alberta Children’s Hospital Research Institute, Calgary, Canada

⁵Departments of Pediatrics, Emergency Medicine and Physiology & Pharmacology, Cumming School of Medicine, University of Calgary, Calgary, Canada

⁶Children’s Hospital of Eastern Ontario Research Institute, Ottawa, Canada

⁷Pediatric Emergency Department, Centre Hospitalier Universitaire Sainte-Justine, Montréal, Canada

⁸Department of Pediatrics, Université de Montréal, Montréal, Canada

⁹Department of Pediatrics, University of Ottawa, Ottawa, Canada

¹⁰Children’s Hospital of Eastern Ontario, Ottawa, Canada

¹¹McGill University Health Centre Research Institute, Montréal, Canada

¹²Department of Pediatrics, McGill University, Montreal, Canada

¹³Department of Pediatrics, Children’s Hospital, Western University, London, Canada

¹⁴Division of Pediatric Emergency Medicine, British Columbia Children’s Hospital, University of British Columbia, Vancouver, British Columbia, Canada

¹⁵Pediatric Emergency Department, Stollery Children’s HospitalAlberta, Canada

¹⁶Department of Pediatrics, University of Alberta. Edmonton, Alberta, Canada

¹⁷Department of Pediatrics, University of British Columbia, Vancouver, Canada

¹⁸Department of Pediatrics and Emergency Medicine, University of Ottawa, Ottawa, Canada

¹⁹Department of Statistical Science, University College London, London, United Kingdom

²⁰Yamagata University Faculty of Medicine: Yamagata Daigaku Igakubu Daigakuin Igakukei Kenkyuka, JAPAN

^✉

* E-mail: larry.dong@mail.utoronto.ca

Competing Interests: Amy C. Plint receives in-kind support from Amphastar for a clinical trial in bronchiolitis. The other authors have no relevant conflicts to disclose.

☯ Contributed equally as co-senior authors.

Roles

Larry Dong: Formal analysis, Methodology, Visualization, Writing – original draft

Terry P Klassen: Data curation, Writing – review & editing

David W Johnson: Data curation, Writing – review & editing

Rhonda Correll: Data curation, Writing – review & editing

Serge Gouin: Data curation, Writing – review & editing

Maala Bhatt: Data curation, Writing – review & editing

Hema Patel: Data curation, Writing – review & editing

Gary Joubert: Data curation, Writing – review & editing

Karen J L Black: Data curation, Writing – review & editing

Troy W S Turner: Data curation, Writing – review & editing

Sandra R Whitehouse: Data curation, Writing – review & editing

Amy C Plint: Conceptualization, Data curation, Funding acquisition, Supervision, Writing – original draft

Anna Heath: Conceptualization, Methodology, Supervision, Writing – original draft

Dhammika Leshan Wannigama: Editor

PMCID: PMC12083810 PMID: 40378135

Abstract

Background

Bronchiolitis exerts a high burden on children, their families and the healthcare system. The Canadian Bronchiolitis Epinephrine Steroid Trial (CanBEST) assessed whether administering epinephrine alone, dexamethasone alone, or in combination (EpiDex) could reduce bronchiolitis-related hospitalizations among children less than 12 months of age compared to placebo. CanBEST demonstrated a statistically significant reduction in 7-day hospitalization risk with EpiDex in an unadjusted analysis but not after adjustment.

Objective

To explore the probability that EpiDex results in a reduction in hospitalizations using Bayesian methods.

Study design

Using prior distributions that represent varying levels of preexisting enthusiasm or skepticism, i.e., how confident or doubtful one is that EpiDex may reduce hospitalizations, and information about the treatment effect before data were collected, the posterior distribution of the relative risk of hospitalization compared to placebo was determined. The probability that the treatment effect is less than 1, 0.9, 0.8 and 0.6, indicating increasing reductions in hospitalization risk, are computed alongside 95% credible intervals.

Results

Combining a minimally informative prior distribution with the data from CanBEST provides comparable results to the original analysis. Unless strongly skeptical views about the effectiveness of EpiDex were considered, the 95% credible interval for the treatment effect lies below 1, indicating a reduction in hospitalizations. There is a 90% probability that EpiDex results in a clinically meaningful reduction in hospitalization of 10% even when incorporating skeptical views, with a 67% probability when considering strongly skeptical views.

Conclusion

A Bayesian analysis demonstrates a high chance that EpiDex reduces hospitalization rates for bronchiolitis, although strongly skeptical individuals may require additional evidence to change practice.

Trial registration

Clinical Trial registry name, registration number: Current Controlled Trials number, ISRCTN56745572

Introduction

Bronchiolitis is a respiratory disease that exerts significant burden on the healthcare system [1]. It is the leading cause of infant hospitalization in North America and is associated with substantial healthcare spending during the winter months [2–5]. Few treatments that have conclusively demonstrated a reduction in hospitalization rates for infants with bronchiolitis [6,7]. The Canadian Bronchiolitis Epinephrine Steroid Trial (CanBEST) is one of the largest trials in bronchiolitis and examined the effectiveness of epinephrine, dexamethasone, and their combination in reducing the risk of hospitalization by day 7 in children aged 6 weeks to 12 months of age [8]. CanBEST used a factorial design [9] to randomize participants to one of four treatment categories: a combination of nebulized epinephrine and oral dexamethasone (EpiDex), nebulized epinephrine with oral placebo (Epi), oral dexamethasone with nebulized placebo (Dex) and oral and nebulized placebo (placebo). This design allowed CanBEST to evaluate Epi, Dex and EpiDex to determine whether any of these three intervention arms resulted in a reduction in hospitalization compared to placebo.

CanBEST demonstrated a clinically meaningful 35% reduction in the relative risk (RR) of hospitalization (a 9.3% absolute risk reduction) for EpiDex compared to placebo [8] and used the standard statistical frequentist approach to draw conclusions from the study. The frequentist approach for statistical analysis indirectly evaluates study hypotheses by calculating the chance of observing the available data under the assumption that a null hypothesis is true [10], typically that there is no effect of treatment on the outcome of interest. The frequentist approach assumes that if the chance of observing the data when the null hypothesis is true is small, then the null hypothesis can be rejected in favor of an alternative hypothesis, usually that there is a beneficial treatment effect. Within this framework, answering multiple research questions within the same study, as for the CanBEST study, usually requires an adjustment to maintain appropriate error rates [11]. However, there is a controversy around whether this is required when testing for interactions in a factorial design, such as CanBEST [9,12]. As a result, the CanBEST study presented both an unadjusted and adjusted analysis. The unadjusted analysis resulted in a statistically significant reduction in hospitalization with EpiDex at the 5% level with a p-value of 0.02, while the analysis adjusted for multiple comparisons was not statistically significant with a p-value of 0.07 [8]. The discrepancy has led to challenges in interpreting and translating the results of the CanBEST study to the bedside [13] and currently, national guidelines for bronchiolitis recommend only supportive care for patients with bronchiolitis [6]. However, extensive basic science literature demonstrates that co-administration of beta₂-adrenoceptor agonists and corticosteroids mutually enhance each other’s effectiveness [14–18] and their synergy is also well documented in clinical trials of asthma management [19,20].

An alternative approach to frequentist statistical analysis [21] is also available and gaining popularity: the Bayesian approach [22,23]. This framework allows you to calculate the probability that an intervention is effective, given the observed data [24,25]. The Bayesian approach also incorporates pre-existing evidence or clinical expertise into the statistical analysis [26] and can thus examine how differences in clinical judgment and experience of an intervention affect the interpretation of results [27]. Finally, as Bayesian analyses are only dependent on the data collected, the proposed model and the prior distributions [28], we circumvent multiple testing issues [9]. Given the extensive health system, patient and family burden of bronchiolitis and the lack of recommended interventions to reduce this burden [1], we undertake an unplanned Bayesian analysis of the data from the CanBEST study. This analysis will calculate the probability that EpiDex reduces hospitalizations for bronchiolitis [27].

Methods

Canadian Bronchiolitis Epinephrine Steroid Trial

CanBEST was a multicenter, double-blinded placebo-controlled clinical trial that assessed the efficacy of epinephrine and dexamethasone, alone and in combination, each compared to placebo, as a treatment for children aged 6 weeks to 12 months who presented at the emergency department with bronchiolitis [8]. All hospitals who participated in CanBEST are members of the national research network, Pediatric Emergency Research Canada (PERC). Children’s Hospital of Eastern Ontario Research Ethics Board gave approval in May 2004 for the CanBEST study with REB Number 02/59E. CanBEST recruited between 1^st December 2004 and 31^st March 2008. Written informed consent was obtained from the parents or guardians of all infants included in the study. The primary outcome was admission to hospital for bronchiolitis within seven days of study enrolment. The complete inclusion and exclusion criteria, outcome definitions and study procedures are provided in the primary publication [8]. Trial participants were randomized equally into one of the four treatment groups: EpiDex, Epi, Dex or placebo, with dosing details provided in the primary publication [8]. The target enrolment was 800 patients. Three participants were lost to follow-up meaning data for 797 participants were available for this post-hoc analysis.

An introduction to Bayesian inference

Bayesian and frequentist methods for statistical analysis differ in their philosophy, leading to differences in their conduct and interpretation [26]. Frequentist analyses reach statistical conclusions by controlling error rates over many analyses conducted in the same manner [29]. When multiple research questions are evaluated within the same study, frequentist reasoning clarifies that the chance of at least one incorrect conclusion is increased and necessitates adjustments to control the error rate of the overall study [11]. In contrast, Bayesian methods aim to make the best conclusions using the data from the specific study [25] meaning that study conclusions depend on the data and assumed model, rather than the analysis method [25].

To perform a Bayesian analysis, a prior distribution is required that represents the available evidence, usually assumed to derive from the literature or relevant clinical experience, about the plausible range of the treatment effect before analyzing the data [30]. This prior distribution is combined with the study data to determine an updated probability distribution for the treatment effect, which then represents our knowledge about the plausible values of the treatment effect. This is known as posterior distribution. From this, we can determine the probability that the intervention is beneficial. This probability is not available from frequentist p-values [24] and allows us to trade-off the chance of experiencing benefit or harm from an intervention [31].

Design of prior distributions: reference priors and data-driven priors

Designing prior distributions is a crucial element of undertaking a Bayesian analysis. Furthermore, it is an inherently subjective process, which has led to criticisms of Bayesian statistics as prior distributions influence the trial analysis [32]. However, we exploit this by selecting a range of prior distributions that explicitly represent different archetypes of beliefs about the efficacy of the interventions and results from previously conducted studies [27]. This allows us to explore how variations in the views on the effectiveness of EpiDex influence the interpretation of the CanBEST study. This is advantageous for two reasons; firstly, given the pre-existing controversy on the efficacy of EpiDex for bronchiolitis, we can gain further insight into the debate around its effectiveness. Secondly, readers can determine which prior best represents their own background assessment of the efficacy of EpiDex, based on their experience and expertise, and interpret the results of the CanBEST study accordingly [27].

We consider two classes of prior distributions in our analysis: reference priors and data-driven priors [27]. Reference priors represent clinical archetypes of beliefs about the treatment effectiveness: strongly enthusiastic, moderately enthusiastic, moderately skeptical, strongly skeptical and no opinion. The “no opinion” option uses a “minimally informative” prior to exert the smallest possible influence on the results and provides a similar numerical result to a frequentist analysis but allowing for a Bayesian interpretation. The other four reference priors are defined to reflect the level of enthusiasm or skepticism about the effect of EpiDex using a normal distribution for the log-RR (Fig 1) [27]. Generally, we assume that skeptics believe there is no treatment effect (corresponding to a treatment effect of 1) while enthusiasts believe the treatment is effective at reducing hospitalization (a treatment effect less than 1). Table 1 presents the five reference priors for treatment effect for EpiDex. As there is no credible evidence that these therapies would increase hospitalization, we did not consider this in our reference priors.

Table 1. The prior distributions used for the CanBEST reanalysis.

Prior Belief	Median relative risk of hospitalization with EpiDex	SD of log relative risk	Equivalent Prior Sample Size	Prior probability that the relative risk is below threshold (%)
Prior Belief	Median relative risk of hospitalization with EpiDex	SD of log relative risk	Equivalent Prior Sample Size	RR < 1	RR < 0.9	RR < 0.8	RR < 0.6
Reference Priors
Minimally Informative	1	10	≈ 0	50	50	49	48
Strongly Enthusiastic	0.6	0.31	134	95	90	82	50
Moderately Enthusiastic	0.8	0.14	668	94	80	50	2
Moderately Skeptical	1	0.31	125	50	37	24	5
Strongly Skeptical	1	0.14	594	50	23	6	≈ 0
Data-Driven Priors
100% Weighting	0.89	0.15	478	77	52	23	≈ 0
50% Weighting	0.89	0.21	272	70	51	30	3
10% Weighting	0.89	0.47	61	59	51	41	20

Open in a new tab

In addition to the reference priors, we used data-driven priors, which were derived by fitting a mixed effects hierarchical model using uninformative priors and data from previously published randomized trials in bronchiolitis [27,33]. Broadly, the results using the data-driven priors can be interpreted as combining data from the CanBEST study with previous studies, like a meta-analysis. Studies deemed sufficiently close to CanBEST were chosen according to the four following criteria: (a) study participants were randomized to either a glucocorticoid steroid, a $β_{2}$ -agonist or their combination; (b) the outcome of interest was related to hospital admission – ideally 7-day cumulative hospitalization, (c) participants were infants less than two years of age and (d) the study was conducted prior to the publication of the initial CanBEST analysis. These inclusion criteria assume that drugs in the same class as dexamethasone and epinephrine will have similar effectiveness. We chose an age range of participants up to 24 months to reflect the variation among clinicians and guidelines in defining the age range of children that may be deemed to have bronchiolitis [7]. Table A.1 in Appendix A in S1 File summarizes the intervention, comparator, population, outcome of interest, relative risk, and sample size for studies used for the data-driven priors. There were differences in the chosen studies, including variation in the choice of drug, its dosing, patient inclusion/exclusion criteria and outcomes. Prior sample sizes were calculated by equating the product of treatment effect variances, prior and posterior, with their corresponding sample sizes [34].

Once the data-driven priors have been specified (Fig 1), we consider scenarios that dilute their impact on the final analysis. These scenarios are based on providing a prior “weight” of 100%, 50% and 10%, which represents the relative contribution of a participant in a previous study compared to the contribution of a participant in the CanBEST study and is controlled by the standard deviation of the prior [34]. This weighting procedure – applied on the variance of treatment effect estimates in the hierarchical mixed effects model – accounts for fundamental differences between the data in CanBEST and the data in the previous studies, such as differences in patient population, interventions, and outcomes of interest. Thus, data-driven priors are rarely developed using a full systematic review and meta-analysis as the down-weighting allows us to “discount” the contribution of studies that do not entirely match the CanBEST study.

Table 1 provides a descriptive summary of all the considered reference and data-driven priors for EpiDex. We report the median RR of hospitalization, the standard deviation (SD) of the log-RR distribution – where smaller SDs imply more certainty about the treatment effect before seeing the data – and the probability of RR being below various thresholds, e.g., P(RR < 0.9) is the probability that the treatment reduces the probability of hospitalization probability by at least 10%. For similar tables pertaining to Epi alone and Dex alone priors, see Tables B.1 and B.2 in the Appendix B in S1 File.

Analysis

For each of the prior distributions defined in Table 1, we used a Bayesian model to determine the posterior distributions of the treatment effect for Epi, Dex and EpiDex compared to placebo. We used a binomial generalized linear model with a log link to calculate the relative risk of hospitalization for the three interventions, compared to placebo, adjusted for site. The adjustment for participating sites was achieved using a hierarchical model. The Bayesian model was fitted using PyMC version 4.4.0 [35] in Python version 3.9.15 with 20,000 simulations and a burn-in of 10,000 to ensure convergence [36]. Traceplots were examined to check for convergence and autocorrelation [36].

The posterior distributions were summarized using the median relative risk and equi-tailed 95% credible intervals; these quantities are analogues to a frequentist point estimate of effect and confidence interval. Finally, we estimated the probability that the relative risk was below the thresholds 1, 0.9, 0.8 and 0.6 by the proportion of the simulations that were below each of those thresholds. These thresholds represent a reduction in the risk of hospitalization for bronchiolitis of 0%, 10%, 20% and 40%, respectively.

Results

The primary outcome was available for 797 infants, of these 34 who received the EpiDex combination, 47 who received Epi alone, 51 who received Dex alone and 53 who received placebo were admitted to hospital for bronchiolitis within 7 days of study enrolment.

Overall, our Bayesian analysis (Table 2) demonstrated a posterior probability that the use of EpiDex results in a reduction in hospitalization greater than 98%, unless the clinician was strongly skeptical about the effectiveness of EpiDex. This means that there is over a 98% probability that using EpiDex to treat bronchiolitis in infants in the ED makes them less likely to be hospitalized. The complete results from our Bayesian analysis are displayed in Table 2, while the equivalent analyses for the Epi and Dex treatment groups are available in Tables B.1 and B.2 in the Appendix B in S1 File. Posterior distributions for the RR of EpiDex compared to placebo are displayed in Fig 2, with reference priors on the left and data-driven priors on the right.

Table 2. Summary table of group 1 (EpiDex) posterior characteristics: median RR, 95% credible interval and probability of RR smaller than various thresholds.

	Posterior median for the RR of hospitalization (95% Credible Interval)	Posterior probability that the RR of hospitalization is below threshold (%)
		RR < 1	RR < 0.9	RR < 0.8	RR < 0.6
Reference Priors
Minimally informative	0.66 (0.45, 0.96)	99	95	84	30
Strongly enthusiastic	0.63 (0.45, 0.85)	100	99	94	39
Moderately enthusiastic	0.74 (0.60, 0.91)	100	97	77	3
Moderately skeptical	0.75 (0.55, 1.00)	98	89	67	8
Strongly skeptical	0.86 (0.70, 1.06)	92	65	23	≈ 0
Data-Driven Priors
100% weighting	0.77 (0.62, 0.96)	99	92	63	1
50% weighting	0.74 (0.56, 0.96)	99	93	73	7
10% weighting	0.69 (0.48, 0.96)	99	94	82	23

Open in a new tab

Reference priors

All equi-tailed 95% credible intervals exclude a null relative risk value of 1 except when using a strongly skeptical prior. Using minimally informative priors for all treatment effects leads to an estimated posterior median RR of 0.66 for EpiDex and a corresponding 95% equi-tailed credible interval of (0.45, 0.96); this result is similar to the initial CanBEST analysis where the estimated RR and 95% confidence interval were 0.65 and (0.45, 0.96), respectively, in the unadjusted analysis. Comparing estimates for the posterior median, we can see that the RR increases as the prior skepticism increases. Finally, the probability of a reduction in hospitalization rates with EpiDex, compared to placebo (RR < 1) is greater than 98%, unless a strongly skeptical prior is used. Similarly, the probability of a greater than 10% reduction in hospitalization rates is greater than 90%, unless individuals are strongly skeptical.

Data-driven priors

For the data-driven priors, all equi-tailed 95% credible intervals exclude a null relative risk value of 1, indicating that the combined current evidence suggests a reduction in hospitalization rates with EpiDex. Increasing the weighting of the previous studies increases the posterior median RR, from 0.69 to 0.77, indicating that the previous studies demonstrated a smaller treatment effect than the effect observed in CanBEST. We confirm this trend by computing the posterior distributions using data-driven priors with increasing weights between 0% and 100%; this analysis is available in Appendix D in S1 File.

Discussion

Bronchiolitis exerts a huge burden on the healthcare system, patients, and families [1]. Our Bayesian analysis of the results from the pivotal CanBEST trial has demonstrated that there is a greater than 98% probability that EpiDex reduces hospitalizations for bronchiolitis compared to placebo unless clinicians are highly skeptical. Even highly skeptical individuals could be swayed by the data in the CanBEST study as our analysis demonstrates that there is a 90% chance that EpiDex reduces hospitalizations. This finding was also supported when we combined CanBEST with data from previous studies.

Furthermore, our data-driven Bayesian analysis confirmed that CanBEST resulted in a larger reduction in hospitalization rates compared to previous studies as the estimated posterior median RR for EpiDex increases as the weight for the prior studies increases. Overall, we conclude that EpiDex has the potential to reduce admissions to hospital for bronchiolitis and, as a result, the burden of bronchiolitis for infants, their families, and the healthcare system.

The estimated posterior probability of a 10% reduction in the relative risk of hospitalization varies between 65%, for the strongly skeptical prior, to 99%, for the strongly enthusiastic prior and is always greater than 90% for the data-driven priors. This demonstrates not only does EpiDex potentially reduce hospitalizations, but also there is a relatively high chance of a clinically meaningful reduction in hospitalization rates with EpiDex. However, if there is high skepticism about the efficacy of EpiDex, additional evidence may be required before being convinced by the outcome of CanBEST. In contrast, even when moderate skepticism is considered, the CanBEST results demonstrate a probability of clinically meaningful reduction in hospitalization rates of 89%.

Our Bayesian reanalysis has added important nuance to the interpretation of CanBEST. Firstly, we calculated the probability that the interventions are effective at reducing the risk of hospitalization, which is not possible in standard analyses. This can facilitate conversation and aligns with how clinicians make decisions when deciding on patient care [24]. Secondly, by representing a wide spectrum of prior beliefs, we have provided a flexible framework for interpreting the CanBEST results, facilitating discussion between clinical decision makers who may have differing experience and expertise. That being said, it can be argued that, within the range of priors that we have considered, minimally informative and data-driven priors represent impartiality and can be considered most appropriate. Finally, the design of priors and the lack of a strict definition for statistical significance in the Bayesian paradigm encourages an in-depth discussion of the implications of the results from CanBEST and whether EpiDex can be used to alleviate the overwhelming health system impact of bronchiolitis, particularly in face of the recent bronchiolitis surges [37].

There are some limitations to this reanalysis. Firstly, any inherent limitations in CanBEST are not circumvented by this analysis [27]. Bayesian methods provide a different framework for the interpretation and dissemination of results but are unable to overcome challenges in the design of the initial trial. For example, the definition of bronchiolitis varies globally and CanBEST restricted participants to infants less than one year of age who were experiencing wheezing for the first time in the typical “season” for respiratory syncytial virus infection. Other jurisdictions include children up to 24 months of age and do not always restrict the diagnosis to those with a first episode of wheezing [38]. Clinicians may also be concerned about the use of corticosteroids in young children although a recent comprehensive systematic review found no increased risk of short–term adverse effects among children with acute respiratory illnesses treated with corticosteroids compared to placebo [39]. Similarly, no trial of nebulized epinephrine in bronchiolitis has demonstrated serious side effects or clinically important increases in heart rate or blood pressure. A theoretical risk is that children treated with epinephrine and discharged home might clinically worsen as the effect of epinephrine wears off. However, a systematic review of bronchiolitis studies found similar return-to-care rates in children treated with epinephrine as compared with placebo and salbutamol [40]. Strengths of the original trial are also inherent to this analysis, e.g., the CanBEST trial has very limited loss to follow up. A limitation of this analysis is that alternative prior distributions could have been considered and would have changed the results [26]. However, by being explicit about our prior definitions and assumptions and considering a range of previous studies for the data-driven priors, we allow the reader to determine which view and analysis aligns most closely with their beliefs. Lastly, this re-analysis uses data from CanBEST, wherein data was collected between 2004 and 2008. Post-COVID bronchiolitis and its seasonality may have changed, with recent studies highlighting an increased severity of the disease, risk of hospitalization and potential benefit of RSV vaccination as a preventive measure [41,42]. Furthermore, with the advent of RSV monoclonal antibody use among a wider population of infants (e.g., term, healthy infants), the risk of hospitalization from bronchiolitis should be reduced and possibly reduce the need for a change in management [43].

Conclusion

Bayesian analysis provides an alternative to the commonly used frequentist interpretation of clinical trials. It allows individuals with different prior experience and expertise to contextualize their interpretation of the trial results. For CanBEST, our Bayesian analysis demonstrated a very high probability that the combination of nebulized epinephrine and oral dexamethasone reduces bronchiolitis-related hospital admissions. Thus, use of this combination treatment is likely to reduce the substantial burden of bronchiolitis to both infants and their families and the healthcare system. The use of Bayesian methods circumvents a discussion on whether the analysis should be adjusted for multiple comparisons, which previously complicated the interpretation of CanBEST. Note that due to the uncertain interpretation of CanBEST in the frequentist paradigm, an international randomized trial is currently underway to answer the calls for further evidence on the effectiveness of the combined therapy [44]. Based on our analysis, the results from this international trial will be particularly relevant to clinicians that are highly skeptical about the effectiveness of the combined therapy.

Supporting information

S1 File. Supplementary results from the analysis and description of the data-driven priors.

(DOCX)

pone.0318853.s001.docx^{(82.4KB, docx)}

Acknowledgments

This results from this analysis were presented at the Statistical Society of Canada Conference 2023.

Data Availability

Data cannot be shared publicly because participants of this study did not agree for their data to be shared publicly. Data are available from Amy C. Plint (plint@cheo.on.ca) or the CHEO research institute (researchdatamanagement@cheo.on.ca) for researchers who meet the criteria for access to confidential data.

Funding Statement

This work was supported by Canada Research Chairs in Statistical Trial Design [AH] and Clinical Trials [TPK] (https://www.chairs-chaires.gc.ca/home-accueil-eng.aspx), Tier 1 University of Ottawa Research Chair [ACP] (https://www.uottawa.ca/facultymedicine/research-and-innovation/research-chairs) and an operating grant from Canadian Institutes of Health Research (CIHR; https://cihr-irsc.gc.ca/e/193.html) for the original CanBEST study. The funders plated no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Carroll KN, Gebretsadik T, Griffin MR, Wu P, Dupont WD, Mitchel EF, et al. Increasing burden and risk factors for bronchiolitis-related medical visits in infants enrolled in a state health care insurance plan. Pediatrics. 2008;122(1):58–64. doi: 10.1542/peds.2007-2087 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Shay D. Bronchiolitis-associated hospitalizations among US children, 1980-1996. JAMA. 1999;282(15):1440. [DOI] [PubMed] [Google Scholar]
3.Langley JM, LeBlanc JC, Smith B, Wang EEL. Increasing incidence of hospitalization for bronchiolitis among Canadian children, 1980–2000. J Infect Dis. 2003;188(11):1764–7. [DOI] [PubMed] [Google Scholar]
4.Njoo H, Pelletier L, Spika L. Infectious disease. Respiratory disease in Canada. Ottawa: Canadian Institute for Health Information, Canadian Lung Association, Health Canada, Statistics Canad; 2001. p. 65–84. [Google Scholar]
5.Craig E, Jackson C, Han D, Grimwood K, NZCYES Steering Committee. Monitoring the health of New Zealand children and young people: Indicator handbook. Auckland, New Zealand: Paediatric Society of New Zealand and New Zealand Child and Youth Epidemiology Service. 2007. [Google Scholar]
6.Florin TA, Plint AC, Zorc JJ. Viral bronchiolitis. Lancet. 2017;389(10065):211–24. doi: 10.1016/S0140-6736(16)30951-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Hartling L, Bialy L, Vandermeer B, Tjosvold L, Johnson D, Plint A. Epinephrine for bronchiolitis. Cochrane Database Syst Rev. 2011;6:1–134. [DOI] [PubMed] [Google Scholar]
8.Plint AC, Johnson DW, Patel H, Wiebe N, Correll R, Brant R, et al. Epinephrine and dexamethasone in children with bronchiolitis. N Engl J Med. 2009;360(20):2079–89. doi: 10.1056/NEJMoa0900544 [DOI] [PubMed] [Google Scholar]
9.Gelman A, Hill J, Yajima M. Why we (usually) don’t have to worry about multiple comparisons. J Res Educ Eff. 2012;5(2):189–211. [Google Scholar]
10.Pocock SJ, McMurray JJV, Collier TJ. Making sense of statistics in clinical trial reports. J Am Coll Cardiol. 2015;66(22):2536–49. [DOI] [PubMed] [Google Scholar]
11.Bender R, Lange S. Adjusting for multiple testing--when and how?. J Clin Epidemiol. 2001;54(4):343–9. doi: 10.1016/s0895-4356(00)00314-0 [DOI] [PubMed] [Google Scholar]
12.Sjölander A, Vansteelandt S. Frequentist versus Bayesian approaches to multiple testing. Eur J Epidemiol. 2019;34(9):809–21. doi: 10.1007/s10654-019-00517-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Zorc JJ, Hall CB. Bronchiolitis: recent evidence on diagnosis and management. Pediatrics. 2010;125(2):342–9. doi: 10.1542/peds.2009-2092 [DOI] [PubMed] [Google Scholar]
14.Barnes PJ. Scientific rationale for using a single inhaler for asthma control. Eur Respir J. 2007;29(3):587–95. doi: 10.1183/09031936.00080306 [DOI] [PubMed] [Google Scholar]
15.Giembycz M, Kaur M, Leigh R, Newton R. A holy grail of asthma management: toward understanding how long-acting beta2-adrenoceptor agonists enhance the clinical efficacy of inhaled corticosteroids. Br J Pharmacol. 2008;153(6):1090–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Newton R, Holden NS. Separating transrepression and transactivation: a distressing divorce for the glucocorticoid receptor?. Mol Pharmacol. 2007;72(4):799–809. doi: 10.1124/mol.107.038794 [DOI] [PubMed] [Google Scholar]
17.Kaur M, Chivers JE, Giembycz MA, Newton R. Long-acting beta2-adrenoceptor agonists synergistically enhance glucocorticoid-dependent transcription in human airway epithelial and smooth muscle cells. Mol Pharmacol. 2008;73(1):203–14. doi: 10.1124/mol.107.040121 [DOI] [PubMed] [Google Scholar]
18.Holden N, Bell M, Rider C, King E, Gaunt D, Leigh R, et al. β2-adrenoceptor agonist-induced rgs2 expression is a genomic mechanism of bronchoprotection that is enhanced by glucocorticoids. Proc Natl Acad Sci U S A. 2011;108(49):19713–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Greening AP, Ind PW, Northfield M, Shaw G. Added salmeterol versus higher-dose corticosteroid in asthma patients with symptoms on existing inhaled corticosteroid. Allen & Hanburys Limited UK Study Group. Lancet. 1994;344(8917):219–24. doi: 10.1016/s0140-6736(94)92996-3 [DOI] [PubMed] [Google Scholar]
20.Pauwels RA, Löfdahl CG, Postma DS, Tattersfield AE, O’Byrne P, Barnes PJ. Effect of inhaled formoterol and budesonide on exacerbations of asthma. N Engl J Med. 1997;337(20):1405–11. [DOI] [PubMed] [Google Scholar]
21.Bland JM, Altman DG. Bayesians and frequentists. BMJ. 1998;317(7166):1151–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Tidwell RSS, Peng SA, Chen M, Liu DD, Yuan Y, Lee JJ. Bayesian clinical trials at The University of Texas MD Anderson Cancer Center: An update. Clin Trials. 2019;16(6):645–56. doi: 10.1177/1740774519871471 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Ashby D. Bayesian statistics in medicine: a 25 year review. Stat Med. 2006;25(21):3589–631. doi: 10.1002/sim.2672 [DOI] [PubMed] [Google Scholar]
24.Lee JJ, Chu CT. Bayesian clinical trials in action. Stat Med. 2012;31(25):2955–72. doi: 10.1002/sim.5404 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Goodman SN. Toward evidence-based medical statistics. 1: the p value fallacy. Ann Intern Med. 1999;130(12):995. [DOI] [PubMed] [Google Scholar]
26.Lee P. Bayesian statistics: an introduction. Chichester: Wiley; 1989. [Google Scholar]
27.Goligher EC, Tomlinson G, Hajage D, Wijeysundera DN, Fan E, Jüni P, et al. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome and posterior probability of mortality benefit in a post hoc Bayesian analysis of a randomized clinical trial. JAMA. 2018;320(21):2251. [DOI] [PubMed] [Google Scholar]
28.Berry DA. Interim analysis in clinical trials: the role of the likelihood principle. Am Stat. 1987;41(2):117–22. [Google Scholar]
29.Neyman J. Frequentist probability and frequentist statistics. Synthese. 1977;36(1):97–131. [Google Scholar]
30.Kass R, Wassermann L. The selection of prior distributions by formal rules. J Am Stat Assoc. 1996;91(435):1343–70. [Google Scholar]
31.Parmigiani G, Inoue L. Decision theory: principles and approaches. Chichester: John Wiley & Sons; 2009. [Google Scholar]
32.Gelman A. Objections to Bayesian statistics. Bayesian Anal. 2008;3(3). [Google Scholar]
33.DuMouchel W. Bayesian meta-analysis. In: Berry D, editor. Statistical methodology in the pharmaceutical sciences. Boca Raton, Florida: CRC Press; 1990. p. 509–29. [Google Scholar]
34.Morita S, Thall PF, Müller P. Determining the effective sample size of a parametric prior. Biometrics. 2008;64(2):595–602. doi: 10.1111/j.1541-0420.2007.00888.x [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Salvatier J, Wiecki T, Fonnesbeck C. Probabilistic programming in python using pymc3. PeerJ Comput Sci. 2016;2:e55. [DOI] [PMC free article] [PubMed] [Google Scholar]
36. Gelman A, Carlin J, Stern H, Dunson D, Vehtari A, Rubin DB. Bayesian data analysis. 3rd ed. Chapman and Hall/CRC; 2013. [Google Scholar]
37.Remien KA, Amarin JZ, Horvat CM, Nofziger RA, Page-Goertz CK, Besunder JB, et al. Admissions for Bronchiolitis at Children’s Hospitals Before and During the COVID-19 Pandemic. JAMA Netw Open. 2023;6(10):e2339884. doi: 10.1001/jamanetworkopen.2023.39884 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Dalziel S, Haskell L, O’Brien S, Borland M, Plint A, Babl F. Bronchiolitis. Lancet. 2022;400(10349):392–406. [DOI] [PubMed] [Google Scholar]
39.Fernandes RM, Wingert A, Vandermeer B, Featherstone R, Ali S, Plint AC, et al. Safety of corticosteroids in young children with acute respiratory conditions: a systematic review and meta-analysis. BMJ Open. 2019;9(8):e028511. doi: 10.1136/bmjopen-2018-028511 [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Hartling L, Fernandes R, Bialy L, Milne A, Johnson D, Plint A. Steroids and bronchodilators for acute bronchiolitis in the first two years of life: systematic review and meta-analysis. BMJ. 2011;342(apr06 2):d1714. [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Curatola A, Graglia B, Ferretti S, Covino M, Pansini V, Eftimiadi G, et al. The acute bronchiolitis rebound in children after COVID-19 restrictions: a retrospective, observational analysis. Acta Biomed. 2023;94(1):e2023031. doi: 10.23750/abm.v94i1.13552 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Du Z, Pandey A, Moghadas SM, Bai Y, Wang L, Matrajt L, et al. Impact of RSVpreF vaccination on reducing the burden of respiratory syncytial virus in infants and older adults. Nat Med. 2025;31(2):647–52. doi: 10.1038/s41591-024-03431-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Jones JM, Fleming-Dutra KE, Prill MM, Roper LE, Brooks O, Sánchez PJ, et al. Use of Nirsevimab for the prevention of respiratory syncytial virus disease among infants and young children: recommendations of the advisory committee on immunization practices - United States, 2023. MMWR Morb Mortal Wkly Rep. 2023;72(34):920–5. doi: 10.15585/mmwr.mm7234a4 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Plint A. In clinicaltrials.gov, Reference Number: NCT03567473. 2022. [cited 2024 May 12]. Bronchiolitis in Infants Placebo Versus Epinephrine and Dexamethasone Study (BIPED). Available from: https://clinicaltrials.gov/ct2/show/NCT03567473 [Google Scholar]

PLoS One. 2025 May 16;20(5):e0318853. doi: 10.1371/journal.pone.0318853.r001

Author response to Decision Letter 0

26 Apr 2024

PLoS One. doi: 10.1371/journal.pone.0318853.r002

Decision Letter 0

Helen Howard

10 Feb 2025

PONE-D-24-16700The Probability of Reducing Hospitalization Rates for Bronchiolitis: A Bayesian AnalysisPLOS ONE

Dear Dr. Heath,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The manuscript has been evaluated by two reviewers, and their comments are available below.

The reviewers have raised a number of concerns that need attention. They request improvements to the reporting of methodological and statistical aspects of the study, and improvements to the writing and discussion.

Could you please revise the manuscript to carefully address the concerns raised?

Please submit your revised manuscript by Mar 27 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Helen Howard

Staff Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please note that PLOS ONE has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, we expect all author-generated code to be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/plosone/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse.

3. We note that you have indicated that there are restrictions to data sharing for this study. For studies involving human research participant data or other sensitive data, we encourage authors to share de-identified or anonymized data. However, when data cannot be publicly shared for ethical reasons, we allow authors to make their data sets available upon request. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

Before we proceed with your manuscript, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., a Research Ethics Committee or Institutional Review Board, etc.). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of recommended repositories, please see https://journals.plos.org/plosone/s/recommended-repositories. You also have the option of uploading the data as Supporting Information files, but we would recommend depositing data directly to a data repository if possible.

Please update your Data Availability statement in the submission form accordingly.

4. Your ethics statement should only appear in the Methods section of your manuscript. If your ethics statement is written in any section besides the Methods, please move it to the Methods section and delete it from any other section. Please ensure that your ethics statement is included in your manuscript, as the ethics statement entered into the online submission form will not be published alongside your manuscript.

5. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank you for asking me to review your paper on bronchiolitis. I agree that it will take a further study with strong evidence to convince clinicians to start using epidex for treatment of patients with bronchiolitis. All the evidence to date and guidelines recommends minimal handling and not to use any medications. I feel the paper explains clearly about the Can BEST study and explains the results of the original paper and the reasoning for re-analysis using the Bayesian method. I have to say that I don’t fully understand the statistics and have recommended the paper be reviewed by a statistician. The paper is heavy in statistical methods and discussion so I agree that it will not convince many frontline clinicians to adjust their practise and start using epidex in the management of bronchiolitis. I think it is a valuable debate to have about the methods used for analysis of RCTs but would generally err on supporting simple methods that reflect a pragmatic approach.

It will be interesting to see the results of the BIPED study and then see if the combined data shows that epi and dex truly reduces the hospitalisation of children with bronchiolitis.

Major changes:

I think there needs to be an acknowledgement that the data for this re-analysis is from 2004-2008 when bronchiolitis was different to what we now see. Post covid the bronchiolitis season has dramatically changed and we are only just starting to see a return to normality. It maybe that emerging bronchiolitis studies demonstrate changes in the natural course of the disease.

I would be keen to see if there is any difference when you assess the different risk groups such as those with underlying conditions or born extreme preterm.

It is also important to note that with the introduction of RSV vaccinations the rates of bronchiolitis admissions to hospital should be significantly reduced. This may therefore negate the need for changing in our clinical management of bronchiolitis as it is hoped the burden from bronchiolitis will be significantly reduced.

Minor changes:

I would suggest changing the title to include epinephrine and steroids. Maybe ‘The probability of reducing hospitalisation rates for bronchiolitis with epidex using a Bayesian analysis method.’

In the design of prior distributions section the third criteria for suitable studies close to CanBEST is participants were infants less than two years of age but Can BEST was based on less than 12 months of age. I feel this would lead to significant differences in the cohorts of children being included as the many of the children 1-2 years of age will have a different aetiology to their respiratory illness other than bronchiolitis. This will therefore impact on the results of those studies and so I would assume impact the data driven priors. As stated I am not a statistician and so unsure how you would take into account the difference in ages in the different studies and into the Bayesian analysis model.

In the discussion section it is stated ‘Our Bayesian analysis of the results from the pivotal CanBEST trial has demonstrated that there is a greater than 98% probability that EpiDex reduces hospitalizations for bronchiolitis compared to placebo unless clinicians are highly skeptical.’ which is a repeat of what has been described in the results. I would like to see more detail about what does the greater than 98% probability mean?

In the limitations section the references 38 and 39 are based on evidence from young children or children under 2 years of age. Again I do not feel this is representative of the bronchiolitis group of children who should be defined as less than 12 months of age to avoid confusion with different respiratory illness aetiologies such as episodic wheeze which may be more likely to benefit from steroids or epi. I would also say another limitation of the re-analysis is the unfamiliarity of clinicians with Bayesian analysis and therefore it is unlikely to lead to changes in clinical practise.

Reviewer #2: The aim of this study is to reanalyze data from a previous study that applied frequentist statistics using Bayesian analysis. The study requires a more detailed description of its methodology and results. Another problem is that the present findings lead to uncertain conclusions, depending on the strength of the prior belief. For certain readers, such as policymakers, this uncertainty may not be reassuring. I think there is value in the fact that the focus of the study is on the probability of the treatment effect, however, this should be communicated in an effective manner.

First, there is a mismatch between the presented results in Table 2 and Figure 2. The color of strongly skeptical shows a posterior distribution with highest density for a value around 0,75. The color of skeptical distribution shows the highest density for a value around 0,86. The moderately enthusiastic distribution peaks at something around 0,6. The strongly enthusiastic distribution peaks at around 0,75. Finally, the minimally informative peaks at around 0,66. This is inconsistent with Table 2.

Furthermore, inconsistencies also happen for the results using data-driven priors in Figure 2. The 100% weight appears to have the smallest median value. The 50% weight has the largest median value. The 10% weight has an intermediate median value.

Second, the use of data-driven prior distributions is an interesting approach for the reason pointed out in the manuscript that the analysis considers the data of the present study with prior knowledge of the previous studies. There are two problems. One is the methodology is still not clear on how the weights are applied. The other problem is that the upper bound of the posterior interval remains unchanged regardless of how informative the priors are. For instance, it is difficult to accept no change in the upper bound for the 10% weight distribution.

Third, for a reader that is not used with Bayesian analysis, results that vary depending on the views might be hard to accept. In the absence of previous data, I would suggest the minimally informative prior. IN this case, according to Table 2, the probability of RR below threshold is high. Since there is previous knowledge, the data-driven priors are an interesting approach. However, it is not clear how this was applied to be able to fully judge the results.

Although I see some value in using different degrees of views (skeptical, enthusiastic etc.), I would say that a manuscript with minimally informative, and data-driven only priors would better communicate results. This means all other results not incorporated in the manuscript.

Additional comments below.

Abstract

The term Bayesian distribution should be replaced with posterior distribution for accuracy.

Correction: The probability that the treatment effect is less than 1, 0.9, 0.8 and 0.6

For a reader of the abstract, the meaning of skeptical views is not clear. It may be helpful to introduce this concept in the background section. The same applies to strongly skeptical individuals. After re-reading the manuscript, I understand what is meant, but the abstract should be accessible to a general audience.

Introduction

Statement about 9.3% absolute risk reduction (line 16): is this from ref 8?

Methods

I do not think the description of frequentist methods in line 64 is fair and should be revised.

"Frequentist analyses reach statistical conclusions by controlling error rates over many analyses conducted in the same manner (29). When multiple research questions are evaluated within the same study, the chance of making at least one incorrect conclusion is increased and necessitates adjustments to control the error rate of the overall study."

The phrase "treatment effect after seeing the data" (line 76) seems very conversational and should be revised.

About the statement:"Secondly, readers can determine which prior best represents their own background assessment of the efficacy of EpiDex, based on their experience and expertise, and interpret the CanBEST study results accordingly." This is a strong claim, implying that the results are subjective, which I do not agree with. It should be reworded to avoid this implication.

Regarding Table 1: Inform which distributions are being used.

It is not well understood how the weighting is applied to data-driven priors. Is it applied to the standard deviation to make the prior less informative? Further clarification on this point would be helpful. Formulas can be useful as well.

Results

For some unknown reason, when I download Figure 1 (JPEG), only Figure 2 is downloaded. I wanted to view Figure 1 in higher resolution.

Figures 1 and 2: The tables use the term moderately skeptical. I recommend maintaining this term in the figures for consistency.

When a probability is found to be 0, it is best to report it as less than a negligible number, such as 0.001.

The previous comments about Figure 2 on the mismatch of the distributions.

How is the equivalent prior sample size calculated? This was not explained in the methodology.

Figure D1: The distribution of RR seems bounded on top by 1. What explains this behavior?

Discussion

Last sentence: should be revised.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Martin Edwards

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2025 May 16;20(5):e0318853. doi: 10.1371/journal.pone.0318853.r003

Author response to Decision Letter 0

2 Apr 2025

It will be interesting to see the results of the BIPED study and then see if the combined data shows that epi and dex truly reduces the hospitalisation of children with bronchiolitis.

Dear Reviewer #1,

Firstly, thank you for taking the time to review our paper and propose changes with the goal to improve it.

Major changes:

RESPONSE: While we agree that the seasonality of bronchiolitis changed during the pandemic and is now returning to a more typical seasonality, we are not aware of evidence suggesting changes in the natural history of the disease per se related to COVID-19. However, the use of RSV monoclonal antibody among a wider population of children may influence severity and risk of hospitalization and we have included a statement to that effect in our limitations. As such, we have edited the last sentence of the limitations to read: “Lastly, this re-analysis uses data from CanBEST, wherein data was collected between 2004 and 2008. Post-COVID bronchiolitis and its seasonality may have changed, with recent studies highlighting increased severity and risk of hospitalization, potentially decreasing the relevance of this analysis. Furthermore, with the advent of RSV monoclonal antibody use among a wider population of infants (e.g. term, healthy infants), the risk of hospitalization from bronchiolitis should be reduced and possibly reduce the need for a change in management.”

I think the limitations of this re-analysis needs to further develop the potential risks of using dex and epi in children with bronchiolitis. The importance of safe prescribing can not be ignored and we know that the current management of infants with bronchiolitis with minimal intervention is safe and causing little harm. I would be keen to see if there is any difference when you assess the different risk groups such as those with underlying conditions or born extreme preterm.

RESPONSE: Unfortunately, we do have the data available to investigate these subgroup analyses. Overall only five percent of children in the original CanBEST study had an underlying condition, and children at high risk of severe bronchiolitis (those with underlying significant heart disease, chronic lung disease, and immunodeficiency) were not enrolled in the trial. Only 10% of children in CanBEST were born preterm (defined as less than 37 weeks in the trial). However, we agree that this point would be worth exploring in future research.

RESPONSE: We agree that the introduction of RSV monoclonal antibody use for otherwise healthy infants will/has reduced the burden of bronchiolitis and has lessened the risk of hospitalization. We have added this to our limitations as outlined above

Minor changes:

I would suggest changing the title to include epinephrine and steroids. Maybe ‘The probability of reducing hospitalisation rates for bronchiolitis with epidex using a Bayesian analysis method.’

RESPONSE: We have made these changes.

RESPONSE: We chose to include studies with children up to two years of age to reflect the variation in national guidelines and clinician practice in defining the age range of children that may be deemed to have bronchiolitis. However, we acknowledge that the selection of studies for inclusion in the prior is not easy, which is why we decided to use the weighing approach to reduce the impact of the prior on the results. This aims to account for differences between the study populations. We have also added the following sentence when discussing the design of the prior distributions: “These inclusion criteria assume [...] that outcomes in participants between ages one and two are comparable to those in CanBEST.”

RESPONSE: We clarify using the following passage: “this means that there is over a 98% probability that using EpiDex to treat bronchiolitis in infants makes them less likely to be hospitalized”.

RESPONSE: As highlighted above, we have included these children to reflect differences in guidelines.

I would also say another limitation of the re-analysis is the unfamiliarity of clinicians with Bayesian analysis and therefore it is unlikely to lead to changes in clinical practise.

RESPONSE: While this may be true, we do not believe that it is an inherent limitation of the study we have performed. In this study, we have aimed to introduce the concept of Bayesian analysis and use a range of priors to explore how the CanBEST analysis is influenced by the choice of prior. We have clarified that the goal of this analysis is to support discussion: “by representing a wide spectrum of prior beliefs, we have provided a flexible framework for interpreting the CanBEST results, facilitating discussion between clinical decision makers who may have differing experience and expertise.”

Reviewer #2:

Dear Reviewer #2, we thank you for your detailed comments on our manuscript. We address each of them below.

The aim of this study is to reanalyze data from a previous study that applied frequentist statistics using Bayesian analysis. The study requires a more detailed description of its methodology and results. Another problem is that the present findings lead to uncertain conclusions, depending on the strength of the prior belief. For certain readers, such as policymakers, this uncertainty may not be reassuring. I think there is value in the fact that the focus of the study is on the probability of the treatment effect, however, this should be communicated in an effective manner.

RESPONSE: Thank you for the comments on our manuscript, we’ve made changes throughout based on these reviewer comments to improve clarity.

RESPONSE: In response to this comment, we discovered an error in labelling our plots; these have been updated. We highlight the changes in color labelling below as “prior strength: color of posterior in previous manuscript version -> color in the revised manuscript”:

Minimally informative: blue -> blue

Strongly enthusiastic: green -> orange

Moderately enthusiastic: orange -> green

Moderately skeptical: purple -> red

Strongly skeptical: red -> purple

In short, we kept the legend in the plots the same and ensured that the labelling is accurate.

RESPONSE: This has also been fixed:

100% weight: pink -> brown

50% weight: cyan -> pink

10% weight: brown -> cyan

We apologize for the confusion.

RESPONSE: We have added a description and citation to highlight how the weights are applied: “These scenarios are based on providing a prior “weight” of 100%, 50% and 10%, which represents the relative contribution of a participant in a previous study compared to the contribution of a participant in the CanBEST study and is controlled by the standard deviation of the prior (A). This weighting procedure – applied on , which increases the variance of treatment effect estimates in the hierarchical mixed effects model – accounts for fundamental differences between the data in CanBEST and the data in the previous studies, such as differences in patient population, interventions, and outcomes of interest.”

We have also verified our results and confirmed that the upper bound of the CI does not change as we are rounding to 2 decimal places.

RESPONSE: We have clarified that it is the interpretation of the results that change depending on prior beliefs: “Secondly, readers can determine which prior best represents their own background assessment of the efficacy of EpiDex, based on their experience and expertise, and interpret the results of the CanBEST study results accordingly” We have also clarified how the data-driven priors are incorporated into the analysis, particularly regarding the down-weighting procedure which is “applied on the variance of treatment effect estimates in the hierarchical mixed effects model”.

RESPONSE: The results of the CANBEST analysis have already been published and, as such, this analysis is not aiming to provide a neutral analysis of the CANBEST results. Clinicians often have views on the efficacy of an intervention based on their occasional use in practice or use in other disease areas. As such, we believe that the priors representing different “degrees of belief” can help support debate around the interpretation of the CANBEST analysis and have decided to keep all analyses. This way the reader can reflect on their prior beliefs and find their own unique interpretation. However, we have added a sentence to clarify that the minimally informative and data driven priors demonstrate impartiality: That being said, it can be argued that, within the range of priors that we have considered, minimally informative and data-driven priors represent impartiality and can be considered most appropriate.

Additional comments below.

Abstract

The term Bayesian distribution should be replaced with posterior distribution for accuracy.

RESPONSE: We made this change.

Correction: The probability that the treatment effect is less than 1, 0.9, 0.8 and 0.6

RESPONSE: We made this change.

RESPONSE: We have added the following explanation into the abstract: “Using prior distributions that represent varying levels of preexisting enthusiasm or skepticism, i.e. how confident or doubtful one is that EpiDex may reduce hospitalizations,”

Introduction

Statement about 9.3% absolute risk reduction (line 16): is this from ref 8?

RESPONSE: Yes.

Methods

I do not think the description of frequentist methods in line 64 is fair and should be revised.

RESPONSE: We have revised this to read: "Frequentist analyses reach statistical conclusions by controlling error rates over many analyses conducted in the same manner (29). When multiple research questions are evaluated within the same study, frequentist reasoning clarifies that the chance of at least one incorrect conclusion is increased and necessitates adjustments to control the error rate of the overall study."

The phrase "treatment effect after seeing the data" (line 76) seems very conversational and should be revised.

RESPONSE: We have removed this phrasing and the sentence finishes at “treatment effect.”

About the statement: "Secondly, readers can determine which prior best represents their own background assessment of the efficacy of EpiDex, based on their experience and expertise, and interpret the CanBEST study results accordingly." This is a strong claim, implying that the results are subjective, which I do not agree with. It should be reworded to avoid this implication.

RESPONSE: The choice of a prior distribution is a personal one as we have clarified in the abstract, it represents how confident the individual is about the efficacy of EpiDex before seeing the data. As the amount of data increases, the influence of

Attachment

Submitted filename: Response to Reveiwers.docx

pone.0318853.s003.docx^{(29.4KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0318853.r004

Decision Letter 1

Dhammika Leshan Wannigama

16 Apr 2025

The Probability of Reducing Hospitalization Rates for Bronchiolitis with Epinephrine and Dexamethasone: A Bayesian Analysis

PONE-D-24-16700R1

Dear Dr. Heath,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. If you have any questions relating to publication charges, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Dhammika Leshan Wannigama, MD PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

All the questions raised have been adequately addressed, and the study provides critical new insights that could shift how we treat bronchiolitis with steroids.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #2: Yes

**********

6. Review Comments to the Author

Reviewer #2: All comments have been addressed by the authors.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

**********

PLoS One. doi: 10.1371/journal.pone.0318853.r005

Acceptance letter

Dhammika Leshan Wannigama

PONE-D-24-16700R1

PLOS ONE

Dear Dr. Heath,

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now being handed over to our production team.

At this stage, our production department will prepare your paper for publication. This includes ensuring the following:

* All references, tables, and figures are properly cited

* All relevant supporting information is included in the manuscript submission,

* There are no issues that prevent the paper from being properly typeset

You will receive further instructions from the production team, including instructions on how to review your proof when it is ready. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few days to review your paper and let you know the next and final steps.

Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

If we can help with anything else, please email us at customercare@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Dhammika Leshan Wannigama

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 File. Supplementary results from the analysis and description of the data-driven priors.

(DOCX)

pone.0318853.s001.docx^{(82.4KB, docx)}

Attachment

Submitted filename: Response to Reveiwers.docx

pone.0318853.s003.docx^{(29.4KB, docx)}

Data Availability Statement

[pone.0318853.ref001] 1.Carroll KN, Gebretsadik T, Griffin MR, Wu P, Dupont WD, Mitchel EF, et al. Increasing burden and risk factors for bronchiolitis-related medical visits in infants enrolled in a state health care insurance plan. Pediatrics. 2008;122(1):58–64. doi: 10.1542/peds.2007-2087 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref002] 2.Shay D. Bronchiolitis-associated hospitalizations among US children, 1980-1996. JAMA. 1999;282(15):1440. [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref003] 3.Langley JM, LeBlanc JC, Smith B, Wang EEL. Increasing incidence of hospitalization for bronchiolitis among Canadian children, 1980–2000. J Infect Dis. 2003;188(11):1764–7. [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref004] 4.Njoo H, Pelletier L, Spika L. Infectious disease. Respiratory disease in Canada. Ottawa: Canadian Institute for Health Information, Canadian Lung Association, Health Canada, Statistics Canad; 2001. p. 65–84. [Google Scholar]

[pone.0318853.ref005] 5.Craig E, Jackson C, Han D, Grimwood K, NZCYES Steering Committee. Monitoring the health of New Zealand children and young people: Indicator handbook. Auckland, New Zealand: Paediatric Society of New Zealand and New Zealand Child and Youth Epidemiology Service. 2007. [Google Scholar]

[pone.0318853.ref006] 6.Florin TA, Plint AC, Zorc JJ. Viral bronchiolitis. Lancet. 2017;389(10065):211–24. doi: 10.1016/S0140-6736(16)30951-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref007] 7.Hartling L, Bialy L, Vandermeer B, Tjosvold L, Johnson D, Plint A. Epinephrine for bronchiolitis. Cochrane Database Syst Rev. 2011;6:1–134. [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref008] 8.Plint AC, Johnson DW, Patel H, Wiebe N, Correll R, Brant R, et al. Epinephrine and dexamethasone in children with bronchiolitis. N Engl J Med. 2009;360(20):2079–89. doi: 10.1056/NEJMoa0900544 [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref009] 9.Gelman A, Hill J, Yajima M. Why we (usually) don’t have to worry about multiple comparisons. J Res Educ Eff. 2012;5(2):189–211. [Google Scholar]

[pone.0318853.ref010] 10.Pocock SJ, McMurray JJV, Collier TJ. Making sense of statistics in clinical trial reports. J Am Coll Cardiol. 2015;66(22):2536–49. [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref011] 11.Bender R, Lange S. Adjusting for multiple testing--when and how?. J Clin Epidemiol. 2001;54(4):343–9. doi: 10.1016/s0895-4356(00)00314-0 [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref012] 12.Sjölander A, Vansteelandt S. Frequentist versus Bayesian approaches to multiple testing. Eur J Epidemiol. 2019;34(9):809–21. doi: 10.1007/s10654-019-00517-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref013] 13.Zorc JJ, Hall CB. Bronchiolitis: recent evidence on diagnosis and management. Pediatrics. 2010;125(2):342–9. doi: 10.1542/peds.2009-2092 [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref014] 14.Barnes PJ. Scientific rationale for using a single inhaler for asthma control. Eur Respir J. 2007;29(3):587–95. doi: 10.1183/09031936.00080306 [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref015] 15.Giembycz M, Kaur M, Leigh R, Newton R. A holy grail of asthma management: toward understanding how long-acting beta2-adrenoceptor agonists enhance the clinical efficacy of inhaled corticosteroids. Br J Pharmacol. 2008;153(6):1090–104. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref016] 16.Newton R, Holden NS. Separating transrepression and transactivation: a distressing divorce for the glucocorticoid receptor?. Mol Pharmacol. 2007;72(4):799–809. doi: 10.1124/mol.107.038794 [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref017] 17.Kaur M, Chivers JE, Giembycz MA, Newton R. Long-acting beta2-adrenoceptor agonists synergistically enhance glucocorticoid-dependent transcription in human airway epithelial and smooth muscle cells. Mol Pharmacol. 2008;73(1):203–14. doi: 10.1124/mol.107.040121 [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref018] 18.Holden N, Bell M, Rider C, King E, Gaunt D, Leigh R, et al. β2-adrenoceptor agonist-induced rgs2 expression is a genomic mechanism of bronchoprotection that is enhanced by glucocorticoids. Proc Natl Acad Sci U S A. 2011;108(49):19713–8. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref019] 19.Greening AP, Ind PW, Northfield M, Shaw G. Added salmeterol versus higher-dose corticosteroid in asthma patients with symptoms on existing inhaled corticosteroid. Allen & Hanburys Limited UK Study Group. Lancet. 1994;344(8917):219–24. doi: 10.1016/s0140-6736(94)92996-3 [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref020] 20.Pauwels RA, Löfdahl CG, Postma DS, Tattersfield AE, O’Byrne P, Barnes PJ. Effect of inhaled formoterol and budesonide on exacerbations of asthma. N Engl J Med. 1997;337(20):1405–11. [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref021] 21.Bland JM, Altman DG. Bayesians and frequentists. BMJ. 1998;317(7166):1151–60. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref022] 22.Tidwell RSS, Peng SA, Chen M, Liu DD, Yuan Y, Lee JJ. Bayesian clinical trials at The University of Texas MD Anderson Cancer Center: An update. Clin Trials. 2019;16(6):645–56. doi: 10.1177/1740774519871471 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref023] 23.Ashby D. Bayesian statistics in medicine: a 25 year review. Stat Med. 2006;25(21):3589–631. doi: 10.1002/sim.2672 [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref024] 24.Lee JJ, Chu CT. Bayesian clinical trials in action. Stat Med. 2012;31(25):2955–72. doi: 10.1002/sim.5404 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref025] 25.Goodman SN. Toward evidence-based medical statistics. 1: the p value fallacy. Ann Intern Med. 1999;130(12):995. [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref026] 26.Lee P. Bayesian statistics: an introduction. Chichester: Wiley; 1989. [Google Scholar]

[pone.0318853.ref027] 27.Goligher EC, Tomlinson G, Hajage D, Wijeysundera DN, Fan E, Jüni P, et al. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome and posterior probability of mortality benefit in a post hoc Bayesian analysis of a randomized clinical trial. JAMA. 2018;320(21):2251. [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref028] 28.Berry DA. Interim analysis in clinical trials: the role of the likelihood principle. Am Stat. 1987;41(2):117–22. [Google Scholar]

[pone.0318853.ref029] 29.Neyman J. Frequentist probability and frequentist statistics. Synthese. 1977;36(1):97–131. [Google Scholar]

[pone.0318853.ref030] 30.Kass R, Wassermann L. The selection of prior distributions by formal rules. J Am Stat Assoc. 1996;91(435):1343–70. [Google Scholar]

[pone.0318853.ref031] 31.Parmigiani G, Inoue L. Decision theory: principles and approaches. Chichester: John Wiley & Sons; 2009. [Google Scholar]

[pone.0318853.ref032] 32.Gelman A. Objections to Bayesian statistics. Bayesian Anal. 2008;3(3). [Google Scholar]

[pone.0318853.ref033] 33.DuMouchel W. Bayesian meta-analysis. In: Berry D, editor. Statistical methodology in the pharmaceutical sciences. Boca Raton, Florida: CRC Press; 1990. p. 509–29. [Google Scholar]

[pone.0318853.ref034] 34.Morita S, Thall PF, Müller P. Determining the effective sample size of a parametric prior. Biometrics. 2008;64(2):595–602. doi: 10.1111/j.1541-0420.2007.00888.x [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref035] 35.Salvatier J, Wiecki T, Fonnesbeck C. Probabilistic programming in python using pymc3. PeerJ Comput Sci. 2016;2:e55. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref036] 36. Gelman A, Carlin J, Stern H, Dunson D, Vehtari A, Rubin DB. Bayesian data analysis. 3rd ed. Chapman and Hall/CRC; 2013. [Google Scholar]

[pone.0318853.ref037] 37.Remien KA, Amarin JZ, Horvat CM, Nofziger RA, Page-Goertz CK, Besunder JB, et al. Admissions for Bronchiolitis at Children’s Hospitals Before and During the COVID-19 Pandemic. JAMA Netw Open. 2023;6(10):e2339884. doi: 10.1001/jamanetworkopen.2023.39884 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref038] 38.Dalziel S, Haskell L, O’Brien S, Borland M, Plint A, Babl F. Bronchiolitis. Lancet. 2022;400(10349):392–406. [DOI] [PubMed] [Google Scholar]

[pone.0318853.ref039] 39.Fernandes RM, Wingert A, Vandermeer B, Featherstone R, Ali S, Plint AC, et al. Safety of corticosteroids in young children with acute respiratory conditions: a systematic review and meta-analysis. BMJ Open. 2019;9(8):e028511. doi: 10.1136/bmjopen-2018-028511 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref040] 40.Hartling L, Fernandes R, Bialy L, Milne A, Johnson D, Plint A. Steroids and bronchodilators for acute bronchiolitis in the first two years of life: systematic review and meta-analysis. BMJ. 2011;342(apr06 2):d1714. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref041] 41.Curatola A, Graglia B, Ferretti S, Covino M, Pansini V, Eftimiadi G, et al. The acute bronchiolitis rebound in children after COVID-19 restrictions: a retrospective, observational analysis. Acta Biomed. 2023;94(1):e2023031. doi: 10.23750/abm.v94i1.13552 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref042] 42.Du Z, Pandey A, Moghadas SM, Bai Y, Wang L, Matrajt L, et al. Impact of RSVpreF vaccination on reducing the burden of respiratory syncytial virus in infants and older adults. Nat Med. 2025;31(2):647–52. doi: 10.1038/s41591-024-03431-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref043] 43.Jones JM, Fleming-Dutra KE, Prill MM, Roper LE, Brooks O, Sánchez PJ, et al. Use of Nirsevimab for the prevention of respiratory syncytial virus disease among infants and young children: recommendations of the advisory committee on immunization practices - United States, 2023. MMWR Morb Mortal Wkly Rep. 2023;72(34):920–5. doi: 10.15585/mmwr.mm7234a4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0318853.ref044] 44.Plint A. In clinicaltrials.gov, Reference Number: NCT03567473. 2022. [cited 2024 May 12]. Bronchiolitis in Infants Placebo Versus Epinephrine and Dexamethasone Study (BIPED). Available from: https://clinicaltrials.gov/ct2/show/NCT03567473 [Google Scholar]

PERMALINK

The probability of reducing hospitalization rates for bronchiolitis with epinephrine and dexamethasone: A Bayesian analysis

Larry Dong

Terry P Klassen

David W Johnson

Rhonda Correll

Serge Gouin

Maala Bhatt

Hema Patel

Gary Joubert

Karen J L Black

Troy W S Turner

Sandra R Whitehouse

Amy C Plint

Anna Heath

Roles

Abstract

Background

Objective

Study design

Results

Conclusion

Trial registration

Introduction

Methods

Canadian Bronchiolitis Epinephrine Steroid Trial

An introduction to Bayesian inference

Design of prior distributions: reference priors and data-driven priors

Fig 1. Left: Reference priors for the relative risk of hospitalization within 7 days after treatment administration. Right: Data-driven priors using different weights to control the influence from previous studies.

Table 1. The prior distributions used for the CanBEST reanalysis.

Analysis

Results

Table 2. Summary table of group 1 (EpiDex) posterior characteristics: median RR, 95% credible interval and probability of RR smaller than various thresholds.

Fig 2. Posterior distributions for the relative risk of hospitalization within 7 days after treatment under different reference priors (left) and data-driven priors (right).

Reference priors

Data-driven priors

Discussion

Conclusion

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Author response to Decision Letter 0

Decision Letter 0

Helen Howard

Roles

Author response to Decision Letter 0

Decision Letter 1

Dhammika Leshan Wannigama

Roles

Acceptance letter

Dhammika Leshan Wannigama

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases