European Journal of Pediatrics

, Volume 171, Issue 12, pp 1839–1844

Impact of home oxygen therapy on hospital stay for infants with acute bronchiolitis


    • Department of Pediatrics, Sainte-Justine University Hospital CenterUniversity of Montreal
    • Department of PediatricsCHU Sainte-Justine
  • Melanie Vincent
    • Department of Pediatrics, Sainte-Justine University Hospital CenterUniversity of Montreal
  • Sylvain Morneau
    • Department of Pediatrics, Sainte-Justine University Hospital CenterUniversity of Montreal
  • Isabelle Chevalier
    • Department of Pediatrics, Sainte-Justine University Hospital CenterUniversity of Montreal
Original Article

DOI: 10.1007/s00431-012-1831-4

Cite this article as:
Gauthier, M., Vincent, M., Morneau, S. et al. Eur J Pediatr (2012) 171: 1839. doi:10.1007/s00431-012-1831-4


Acute bronchiolitis has been associated with an increasing hospitalization rate over the past decades. The aim of this paper was to estimate the impact of home oxygen therapy (HOT) on hospital stay for infants with acute bronchiolitis. A retrospective cohort study was done including all children aged ≤12 months discharged from a pediatric tertiary-care center with a diagnosis of bronchiolitis, between November 2007 and March 2008. Oxygen was administered according to a standardized protocol. We assumed children with the following criteria could have been sent home with O2, instead of being kept in hospital: age ≥2 months, distance between home and hospital <50 km, in-hospital observation ≥24 h, O2 requirement ≤1.0 L/min, stable clinical condition, no enteral tube feeding, and intravenous fluids <50 mL/kg/day. Children with significant underlying disease were excluded. A total of 177 children were included. Median age was 2.0 months (range 0–11), and median length of stay was 3.0 days (range 0–18). Forty-eight percent of patients (85/177) received oxygen during their hospital stay. Criteria for discharge with HOT were met in 7.1 % of patients, a mean of 1.8 days (SD 1.8) prior to real discharge. The number of patient-days of hospitalization which would have been saved had HOT been available was 21, representing 3.0 % of total patient-days of hospitalization for bronchiolitis over the study period (21/701). Conclusions: In this study setting, few children were eligible for an early discharge with HOT. Home oxygen therapy would not significantly decrease the overall burden of hospitalization for bronchiolitis.


BronchiolitisOxygen therapyHome health care



Emergency department


Home oxygen therapy


Pediatric intensive care unit


Acute bronchiolitis is the leading cause of hospital admission in infants and has been associated with an increasing hospitalization rate over the past decades [10, 16, 19]. More than one third of children develop bronchiolitis during the first 2 years of life [1]. Of these, 3 to 5 % [6, 16] will be admitted to hospital, while only 1 % was hospitalized in the first half of the 1980s [16]. The etiology of this increase in hospitalization rate is probably multifactorial. A trend towards enrolment of children at a younger age in large child-care centers and improved survival of prematurely born infants and of children with multisystemic diseases may be partly responsible for this increase [16]. Pulse oxymetry measurements have become common practice over the last two decades in acute respiratory illnesses and may have led to broader criteria for hospitalization [9, 19].

Several approaches may be used to reduce morbidity and costs associated with bronchiolitis in infants. The need for supplemental oxygen (O2) therapy in a previously healthy child, no matter the underlying illness, has long been considered to require hospital admission. However, some variations to this traditional approach have emerged over the years, due in part to economical incentives and the development of home-based care. To date, a few publications have described the use of home O2 in some children with acute bronchiolitis, discharged either from an emergency department (ED) [2, 7], a 24-h observation unit [13] or an inpatient unit [13, 17]. According to these data, home O2 therapy (HOT) is an alternative to traditional hospital O2 therapy in selected children with acute bronchiolitis. However, experience with this approach is still limited, and deserves caution [4, 8].

The objective of this study was to estimate the potential impact of HOT on hospital stay, using pre-defined criteria, for 0–12-month-old children admitted with acute bronchiolitis.

Material and methods

This cohort study was conducted at an urban, academic, tertiary-care children’s hospital located at sea level (57 m) (Sainte-Justine University Hospital Center, Montreal, Canada). This center’s emergency department (ED) has an annual volume of 60,000 patient-visits and includes a 16-bed, 24-h observation unit (children kept in this unit are not considered hospitalized).

All children less than 12 months of age hospitalized with acute bronchiolitis between November 1, 2007 and March, 31 2008 were eligible for the study. Patients were identified through discharge diagnoses, including the following codes from the ninth revision of the International Classification of Diseases: J21.0 (acute bronchiolitis due to respiratory syncytial virus) and J21.9 (acute bronchiolitis, without causative agent specified). All eligible patients were included, unless they met at least one of the following criteria: (a) cyanotic congenital heart disease, or heart disease requiring digitalisation; (b) chronic pulmonary disease requiring HOT at the time of admission; (c) chronic hematologic disease (e.g., sickle cell anemia); (d) immune deficiency; (e) prior history of cancer; (f) metabolic disease including diabetes mellitus; (g) neuromuscular disease.

Data were obtained through chart review, performed for each patient by one of three independent investigators (MG, MV, SM) using standardized data collection forms. Information was obtained on demographic characteristics, length of stay, clinical course, and treatments received during hospitalization. Information about supplemental O2 use, including duration of administration and O2 flow rates, were abstracted from the nurses’ records in the chart. Over the study period, vital signs, including pulsed O2 saturation (SpO2), and O2 flow rates or concentrations were recorded at least every 4 h for every child admitted with acute respiratory distress. For all children, dates of hospital admission and discharge were defined as recorded in the nurses’ notes.

A clinical protocol outlining the use of supplemental O2 for children without chronic cardiopulmonary disease, hospitalized with an acute respiratory illness, was implemented in 2003 at our hospital. One of its main objectives was to emphasize the importance of O2 weaning to shorten hospital stay. According to this protocol, O2 was administered to maintain a SpO2 ≥ 92 %, except in children with severe respiratory distress (for these patients, a SpO2 ≥ 94 % was the objective). Nurses were instructed to wean O2 in patients for whom SpO2 were above these levels. They were also asked to increase the FiO2 in either of the two following instances: (a) clinically detectable cyanosis; (b) SaO2 <92 % (in this instance, they were instructed to check the reading 5–10 min after the first one, before making the adjustment).

For the purpose of the present study, length of stay was defined as the number of days between admission and discharge. Duration of O2 administration was the difference in days between the day the child began receiving supplemental O2 following hospital admission and the last day the child received any supplemental O2. We considered O2 was administered at a flow rate >1.0 L/min if it was given through sources other than nasal cannula (i.e., endotracheal tube, continuous positive airway pressure, face mask, or hood).

The primary outcome of this study was the proportion of infants hospitalized with acute bronchiolitis who met discharge criteria for HOT. We assumed that children with criteria defined in Table 1 could have been sent home with HOT, instead of being kept in hospital. Intravenous fluid requirement was assessed according to physicians’ prescriptions. The condition of the child was considered as stable over the last 24 h if he/she was described as stable in physicians’ notes, if the on-call team had not been called to his/her bedside and if he/she had not been observed in the pediatric intensive care unit (PICU) over this period of time. Oxygen requirement was not included per se in the definition of “stable condition”.
Table 1

Criteria of potential eligibility for home oxygen therapy

Age ≥2 months

Home within 50 km of the hospital

In-hospital observation ≥24 h

Intravenous fluids <50 mL/kg/day and no enteral tube feeding at the time of potential discharge with oxygen

O2 requirement ≤1.0 L/min and stable clinical condition during the 24 h prior to potential discharge with oxygen

A secondary outcome was the number of patient-days of hospitalization that could have been saved in this cohort, had HOT been available. The potential date of discharge with HOT was defined as the date all criteria for early discharge with HOT were first met. The number of days of hospitalization that could have been saved for an individual patient using HOT was the difference in days between the real date of discharge and the potential date of discharge with HOT. The number of patient-days of hospitalization that could have been saved with HOT in the entire cohort was computed by summing up the number of days potentially saved in individual patients, had HOT been available. This is reported as an absolute number and as a proportion of the total patient-days of hospitalization for bronchiolitis over the study period (calculated by summing up the total number of days of hospitalization for all infants in the cohort).

All analyses were performed using SPSS statistical software, version 17.0.1. Descriptive statistics were calculated for the entire cohort. Odds ratios and corresponding 95 % confidence intervals of meeting criteria for discharge with HOT were calculated through univariate and multivariate logistic regression. Maximum likelihood estimates of regression coefficients were used to estimate crude and adjusted odds ratios for each of the exposure variables. Ninety-five percent confidence intervals (95%CI) were calculated for all estimates reported. The following variables were included in regression models: prematurity, sex, first episode of bronchiolitis, use of endotracheal intubation and mechanical ventilation, use of non-invasive ventilation. This study was approved by Sainte-Justine University Hospital Center’s institutional review board.


One hundred and ninety-two infants were eligible for the study (Fig. 1). Fifteen patients were excluded because they met at least one exclusion criterion, thus 177 infants were included. Their clinical characteristics are described in Table 2. Median age was 2.0 months (range 0–11) and median length of stay was 3.0 days (range 0–18).
Fig. 1

Flow diagram of patients potentially eligible for home oxygen therapy

Table 2

Clinical characteristics of children hospitalized with acute bronchiolitis (N = 177)

Patient characteristics

 Age at admission—months, median (range)

2.0 (0–11)

 Prematurity (gestational age <37 weeks)—n (%)

19 (10.7)

 Male—n (%)

98 (55.4)

 First episode of bronchiolitis—n (%)

156 (88.1)

 RSV positive—n (% of those tested) a

117 (76.5)

Clinical course in hospital

 Length of stay—days, median (range)

3.0 (0–18)

 ICU admission—n (%)

26 (14.7)

 Treatments received while in hospital—n (%)


  Supplemental oxygen

85 (48)

  Endotracheal intubation and mechanical ventilation

14 (7.9)

  Non-invasive ventilation

15 (8.5)

  Intravenous hydration

115 (65)


84 (47.5)

  Inhaled corticosteroids

4 (2.3)

  Systemic corticosteroids

12 (6.8)

  Inhaled β2-agonist (salbutamol)

45 (25.4)

  Inhaled epinephrine

24 (13.6)

  Enteral tube feeding

22 (12.4)

  Chest physiotherapy

20 (11.3)

aRSV testing results were available for 153 patients

Forty-eight percent of patients (85/177) received supplemental O2 in the course of their hospital stay. Their median length of stay was 5 days (range 1–18). The median duration of O2 administration was 2 days (range 0–14). Almost two thirds (32/52) of infants who received supplemental O2 more than 24 h while in hospital were less than 2 months of age; 22/30 of infants who received O2 more than 48 h were in this age group. Oxygen administration was initiated within 24 h of admission in 88.2 % of children who received supplemental O2. Maximal O2 flow rate given during the entire hospital stay was ≤1.0 L/min in 52.9 % of instances.

Thirteen infants met criteria for discharge with HOT (15.3 % of infants who had received O2 and 7.3 % of all patients), a median of 2 days (range 0–4) prior to real discharge. The number of patient-days of hospitalization which would have been saved had HOT been available was 21, representing 4.2 % of patient-days of hospitalization for children who had received O2 (21/496) and 3.0 % of total patient-days of hospitalization for bronchiolitis over the study period (21/701). If children meeting criteria for HOT were considered eligible for this treatment regardless of the distance between their home and the hospital, 5.4 % of total patient-days of hospitalization for bronchiolitis would have been saved using HOT.

No significant association on crude or multivariate analysis was found between prematurity, sex, first episode of bronchiolitis, or use of non-invasive ventilation and eligibility for HOT, as shown in Table 3. No children who had been intubated and mechanically ventilated were eligible for HOT, therefore no odds ratios were computed for this variable.
Table 3

Crude and adjusted odds ratios (OR) and 95 % confidence intervals (CI) for meeting criteria for discharge with home oxygen therapy


Meeting criteria for discharge with HOTa

Crude OR (95 % CI)

Adjusted ORb (95 % CI)


0.6 (0.1–4.9)

0.6 (0.1–5.5)

Female sex

1.3 (0.4–4.4)

1.4 (0.4–4.8)

First episode of bronchiolitis

1.7 (0.2–15.0)

2.2 (0.2–19.0)

Need for non-invasive ventilation

0.3 (0.0–3.2)

0.3 (0.1–3.0)

aNo children who had been intubated and mechanically ventilated were eligible for HOT

bAdjusted OR are adjusted for all variables included in the model


Our study shows that in the setting of a pediatric tertiary-care hospital, using pre-defined criteria, HOT would not significantly decrease the overall burden of hospitalization for 0–12-month-old children hospitalized with bronchiolitis. More than half of children who received O2 during their hospital stay were less than 2 months of age, and the majority of patients who required supplemental O2 for more than 24 or 48 h were in this age group. These children were not considered eligible for HOT in the study design. Young age was thus a major obstacle to HOT in our study. Supplemental O2, when it was used, was given for a short duration (median of 2 days); a longer need for O2 administration would have increased the potential usefulness of HOT. The 21 patient-days of hospitalization potentially saved with HOT among the 13 eligible patients described in Fig. 1 could be considered as clinically relevant—indeed, for a few patients, HOT could have saved 1–2 days of hospitalization—but they only represented 4 % of patient-days of hospitalization for children who had received O2 and 3.0 % of total patient-days of hospitalization for bronchiolitis over the study period. This could be considered significant in other settings where demand for hospital beds is not being met, and inpatient costs are high. However, other ways of limiting duration of hospitalization for infants with bronchiolitis, namely reducing SpO2 threshold to 90 % in stable children instead of 92–93 % or instituting procedures to wean O2 more efficiently, may have more effect than HOT, without its potential risks and the need for logistical organization.

Fifteen percent of our patients were admitted to the PICU during their hospital stay, 8 % were intubated, and 8.5 % required non-invasive ventilation. These numbers indicate similar, if not greater, seriousness of disease compared with another series describing children hospitalized for bronchiolitis at the Children’s Medical Center in Dallas, Texas, between 2002 and 2007, where requirement for PICU and ventilatory support were 10 and 5 %, respectively [6]. In this same series, the percentage of children requiring supplemental O2 administration was 53 %, and the mean duration of O2 administration was 2 days; these numbers are comparable to ours. In a recent cohort study including infants aged up to 6 months admitted to hospital with bronchiolitis, oxygen was administered in 61 % (201/328) of patients, but only 3 % (11/328) required artificial ventilation [18]. Thus, the low potential impact of HOT on hospital days found in our study cannot be explained by the fact that bronchiolitis was mild in our patients.

In this cohort, all infants were followed on a daily basis by full time hospital-based pediatricians, and a clinical protocol for O2 administration on the pediatric wards had been implemented for several years at the time of the study. Results may have been different in another context, particularly in a setting allowing a more liberal use of O2.

So far, four publications have reported the use of HOT in children with bronchiolitis. Two studies were prospective randomized trials, comparing discharge with HOT to either hospitalization [2] or prolongation of hospitalization [17] in children aged 2–24 months. The third was a prospective observational study describing a group of 20 children of unspecified age discharged with HOT either from a 24-h observation unit or after inpatient admission [13]. In these three studies, a total of 79 patients were treated with HOT and only two required readmission to hospital; there were no significant complications related to O2 therapy at home. Neither of these three studies evaluated the impact of HOT on hospital stay for the entire group of children treated for bronchiolitis at their institution, be it in the ED or on a hospital ward. In two of these studies, the authors used a convenience sample [2, 13], and it is unclear what fraction of all children with bronchiolitis these children represented. Halstead et al. performed a retrospective chart review of children who presented at the Children’s Hospital of Denver (Colorado) ED with bronchiolitis during a 5-year period [7]. The objective of their study was to evaluate the impact of a home O2 clinical care protocol on admission rates. In this study, inclusion criteria for HOT were met in 4,194 instances overall, and in 15 % of these (649/4,194), patients were discharged with home O2. The overall admission rate for bronchiolitis dropped from 40 to 31 % during the study period, and there were no PICU admissions in patients discharged on HOT. So far, published data are thus in favor of HOT in selected groups of children with bronchiolitis. However, three of these studies come from US cities located at moderate altitude, Denver (1,600 m) [2, 7] and Salt Lake City (1,300 m) [13]. At moderate altitudes, “normal” SpO2 values are somewhat lower [5]. It is therefore possible that children with SpO2 <90–92 % at higher altitude have less severe disease on average compared to those with similar SpO2 at sea levels [7]. Moreover, at higher elevations, the need for oxygen may often be the only reason for hospital admission, as suggested by Bajal et al. [2], whereas, in our own experience, young children are rarely admitted to hospital only for the administration of O2. All these factors limit the generalizability of studies coming from hospitals at higher altitude to areas at sea level, as mentioned by Halstead et al. themselves [7], and may explain the fact that our results do not indicate a significant impact of HOT on hospital days.

It is necessary to establish discharge criteria before sending young children home with O2. Globally, our discharge criteria did not differ significantly from those already reported for the use of HOT in bronchiolitis [2, 7, 13, 17]. Some of these criteria appear unquestionable, namely clinical stability as defined above, and no need for artificial hydration. We selected 2 months as the minimal age for HOT, as Bajaj et al. did [2]; a limit of 3 months has also been proposed [7, 13, 17]. Other selected discharge criteria (e.g., desired levels of SpO2, O2 flow rate limits, distance from home to hospital) could be more debatable.

The ideal threshold of SpO2 in the context of acute bronchiolitis is a matter of much debate [1, 14, 15]. The American Academy of Pediatrics recommends the use of supplemental O2 in otherwise healthy infants with bronchiolitis to maintain SpO2 at or above 90 % at sea level [1]. Thresholds of 88–92 % to define hypoxia and administer O2 have been described in the abovementioned publications on HOT [2, 13, 17], and may vary according to altitude [12, 13]. Our objective was to maintain SpO2 ≥92 % in stabilized children, and despite this “comfortable” level, the impact of HOT was low. In our study, an O2 flow rate ≤1 L/min was required during the 24 h before potential discharge. An acceptable SpO2 on an O2 flow rate ≤1 L/min was also an inclusion criterion for Tie et al. [17] and Bajaj et al. [2], while others were more conservative, selecting ≤0.5 L/min as a threshold for discharge [7, 11].

Home oxygen therapy requires a well-organized structure, including equipment readily available at home to administer O2, and close clinical follow-up to implement O2 weaning. The setup to ensure follow-up may involve primary care practitioners [2, 7, 13] or a trained nurse [17]; daily re-assessment of the child at a day hospital could also be used. In our setting, appropriate follow-up via local primary care practitioners or nurses would be difficult to organize in the short term. Patients discharged with HOT would have to be reassessed at our own institution. For practical reasons, we thus determined that infants living >50 km from the hospital would not be eligible for HOT, given that frequent transportation to and from the hospital would be too cumbersome for their parents. Other authors have also taken some practical aspects of surveillance in consideration when defining criteria for HOT eligibility, namely distance between home and health care facility [2], availability of “hospital in the home” nurses to do home visits at least twice daily [17], and availability of primary care physicians to perform a follow-up visit [13]. Had we considered that all infants in the study were eligible for HOT regardless of the distance between their home and the hospital, the proportion of patient-days of hospitalizations saved would have been only marginally increased (5 vs 4 %). In other words, the distance criterion did not have a significant impact on our results, most likely because to be eligible for HOT, several criteria had to be met at the same time, distance between home and hospital being only one of them.

There were some limitations to our study. First, it was held at a single pediatric tertiary-care center and patients were included over a period of only one winter season. Second, it was limited to children aged 0 to 12 months. We chose not to include the 12–24-month-old group because of the heterogeneity of diagnoses in this population. Children older than 12 months who are admitted for respiratory distress and wheezing after a viral upper respiratory prodrome may be diagnosed as having bronchiolitis, asthma, or bronchial hyperresponsiveness, depending on clinicians. Third, retrospective assessment through chart review may have led, in some cases, to under- or overestimation of clinical stability. Fourth, some unnecessary increments of the FiO2 could have happened despite our instructions. As oxygen requirement ≤1.0 L/min was one criterion for potential discharge with HOT, it is therefore possible that some patients were considered not eligible for HOT because of unnecessary upward adjustments of the FiO2. Fifth, there are several potential barriers to discharging children hospitalized with bronchiolitis, such as the need for suctioning, and parental and treating physician’s discomfort [13]; an acceptable SpO2 on room air can also be required [2, 3]. These elements were not evaluated in our study; had they been, the impact of HOT would possibly have been even lower.


At a single pediatric tertiary-care center located at sea level, HOT would be minimally effective at reducing the number of days of hospitalization in 0–12-month-old children with bronchiolitis. In a state-of-the-art on HOT in children, Balfour-Lynn mentioned that acute HOT can be considered for children who have acute bronchiolitis, after a period of hospital observation [3]. HOT is described as a “novel approach” for bronchiolitis by Zorc and Breese Hall [19]. Clearly, more data are needed before this option becomes “routine care”, including effectiveness, and cost analysis of the redistribution of care costs from hospital to home [4].

Conflict of interest

The authors declare that they have no conflict of interest. They did not have any affiliation, financial agreement, or other involvement with any company for this study.

Copyright information

© Springer-Verlag 2012