Article Text


Observational study of two oxygen saturation targets for discharge in bronchiolitis
  1. Steve Cunningham1,
  2. Ann McMurray1
  1. 1Department of Respiratory and Sleep Medicine, Royal Hospital for Sick Children, Edinburgh, UK
  1. Correspondence toDr Steve Cunningham, Department of Respiratory and Sleep Medicine, Royal Hospital for Sick Children, Sciennes Road, Edinburgh EH9 1LF, UK; steve.cunningham{at}


Objectives To assess the potential effect of two guideline discharge oxygen saturation (SpO2) recommendations (≥90% and ≥94%) in recovering bronchiolitis.

Methods and patients Infants aged up to 18 months requiring therapeutic oxygen for SpO2 (≤93%) had SpO2 assessed in air every 2 h. Time from admission to re-establish feeding (>75% normal) and for SpO2 to become stable for at least 4 h at ≥90% and ≥94% were noted.

Results 68 infants, median age 14 weeks, were included. Feeding problems resolved at a median of 11 h (inter-quartile range, IQR 0–47). SpO2 became stable for at least 4 h at 17 h (IQR 0–49) for ≥90% and 63 h (IQR 34–105) for ≥94%. Time for infants to achieve a stable SpO2≥90% and resolve feeding difficulties was a median of 22 h (IQR 7–39 h) sooner than the equivalent for stable SpO2≥94%.

Conclusions Accepting lower SpO2 at discharge could significantly reduce length of stay, but require the clinical and safety effects to be studied.

Statistics from


One in five infants develop symptomatic acute bronchiolitis in the first year of life, and around 3% of all infants require hospital admission, predominantly for help with feeding (nasogastric) and therapeutic oxygen. In the UK in 2007 there were an estimated 28 728 infants under 12 months of age admitted to hospital with bronchiolitis, similar to the number of all children (0–14 years) admitted with acute asthma. The average length of stay for bronchiolitis is 72 h (3 days).1 Oxygen supplementation commenced for an oxygen saturation (SpO2) ≤93% is required by 70% of infants admitted to hospital.1

Two recent evidence-based guidelines considered SpO2 targets for infants discharged home from hospital with acute recovering bronchiolitis. The Scottish Intercollegiate Guidelines Network guideline (number 91)2 concluded that with no evidence available current practice should prevail and infants should be considered ready for discharge once they have regained normoxia (SpO2≥94%). The American Academy of Pediatrics Guideline on Bronchiolitis,3 published in the same year, also accepted that there was no evidence, but the expert opinion of the group considered that infants feeding well with an SpO2≥90% did not require supplemental oxygen (and thus could be considered for discharge). The recommendation of an SpO2 at discharge normally considered hypoxic has however produced controversy.4 A prospective, observational study of 30 Emergency Departments in the USA considered discharge at 94% SpO2 as safe.5 Practice with regard to SpO2 at discharge is highly variable, but in general, practice has been not to discharge as low as 90% SpO2.

The aim of this study was to assess the potential effect of the two recommended SpO2 upper limits on time until fit for discharge from hospital in infants recovering from acute viral bronchiolitis. The study did not address health effects (positive or negative) that may be associated with adopting a lower SpO2 discharge policy.

Methods and patients

This was a prospective, observational study of established practice in an acute medical admissions ward of a tertiary university teaching hospital. Infants up to18 months of age admitted with a clinical diagnosis of acute viral bronchiolitis requiring therapeutic oxygen were studied during the period 1 November 2008 to 31 January 2009. Any infant requiring therapeutic oxygen for hypoxia (SpO2≤93%) in the first 6 h of admission was entered into the observational study. Infants with chronic neurological, cardiac or respiratory conditions were excluded. Infants all received routine symptom-based care. It is not practice to provide corticosteroids (nebulised or oral), bronchodilators or nebulised epinephrine in this institution. Infants are considered ready for discharge once oral feeding is re-established and normoxia has been regained with an SpO2≥94%. Nellcor (Model N-550, Mansfield, Massachusetts, USA) SpO2 monitors (Covidien, USA) were used for study SpO2 observations.

To aid oxygen weaning, it is routine practice in this institution to check SpO2 by removing supplemental oxygen and noting stable SpO2 in air (after 1–2 min). In this study, ward-based nurses caring for the infants were requested to note stable SpO2 in air when therapeutic oxygen was briefly removed every 2 h. Key points of observation were to identify the time when SpO2 had reached ≥90% for three observations (ie, ≥4 h) and subsequently ≥94% for three observations (≥4 h). Patient charts were also reviewed to identify time of admission and time to commence consistent satisfactory oral feeding (defined as ≥75% of that expected for weight and age).

Infants with observation charts that had inadequate observations to identify these key points were excluded from further analysis (N=29). Demographic data, age, sex together with respiratory syncytial virus (RSV) status were also noted. Data were transferred to an Excel spreadsheet; data were presented as mean (SD) or median (inter-quartile range, IQR) as appropriate. Non-parametric tests were used as appropriate (SPSS version 10). Ethical approval was not required for this observational study of current practice.


Sixty-eight infants had full data collection during the observation period (42 male, 62%), median age 14 weeks (IQR 6–30). Fifty-nine infants (87%) were RSV positive (eight were RSV negative; for one this information was not available). No significant association was identified between RSV status and time to stable oxygen ≥90%, ≥94% or time to establish oral feeding (p>0.05). Infants in this study, who all required therapeutic oxygen, were discharged from hospital at a median of 86 h (IQR 57–128).

Feeding difficulties resolved at a median of 11 h following admission (IQR 0–47). The time taken from admission to hospital until SpO2 became stable at ≥90% for ≥4 h was 17 h (IQR 0–49). The equivalent time for SpO2 to be stable at ≥94% for ≥4 h was 63 h (IQR 34–105). The median lag between stable oxygen at ≥90% and ≥94% for three observations (≥4 h) was 32 h (IQR 16–59) (figure 1). The range of times for recovery to ≥90% and ≥94% was highly variable. This range correlated poorly with age (r2=−0.16), and had no significant association with sex (p=0.33) or virus type (p=0.21).

Figure 1

Time difference (hours) ≥90% to ≥94% oxygen saturation (SpO2) in air for 4 h.

There were 20 infants (29%) in whom SpO2 stabilised at ≥90% for three observations (≥4 h) before feeding difficulties had resolved. The time for all study infants to achieve a stable SpO2 of ≥90% and resolve feeding difficulties was a median of 22 h (IQR 7–39 h) (figure 2).

Figure 2

Time to resolve feeding, oxygen saturation of ≥90% in air and ≥94% in air in infants with recovering bronchiolitis.

Our previous study1 identified that therapeutic oxygen is required by 70% of infants admitted to hospital with bronchiolitis. This would suggest that adopting a discharge SpO2 of ≥90% in air could facilitate a discharge 22 h earlier in 70% of infants admitted to hospital with bronchiolitis (the authors would not expect discharge times to change in patients not requiring oxygen in whom time to discharge is sooner than those who require oxygen). In the UK, with an estimated 28 728 infants admitted for an average of 3 days (72 h), the effect of earlier discharge once feeding issues have been resolved and oxygen has been stable for ≥4 h at 90% SpO2 would release the equivalent of 18 434 bed days annually (28 728×0.70×22/24), at a cost saving to the National Health Service (NHS) of over £3 million per year (bed cost of £164 per day: NHS Reference Cost 2008).


This study identifies that stopping therapeutic oxygen for acute recovering viral bronchiolitis at stable 90% SpO2 rather than 94% SpO2 could result in a median discharge from hospital 22 h earlier. With an average length of stay of 3 days for an episode of acute bronchiolitis, this difference represents a significant potential gain. The clinical and safety effects of this policy have yet to be assessed.

An important limitation of this study was the number of infants with missing data who were not assessed; this occurred because the study was dependent on busy ward nurses to provide data input and at times the ward was too busy to achieve the level of data collection set for study entry.

The annual, concentrated, winter peak of acute viral bronchiolitis represents a serious logistical challenge to healthcare providers in terms of hospital beds and staffing. Recognising the pressure on secondary healthcare resources, two recent studies6 7 have assessed the impact of the provision of therapeutic oxygen for recovering viral bronchiolitis in the community following discharge from hospital, using primary care resources or home care nursing teams. This study has demonstrated that the lag between establishment of adequate feeding and achieving a stable SpO2≥94% is 32 h (IQR 16–59), and providing home oxygen could sensibly improve logistics for secondary care during peak winter seasons.

The potential clinical implications of short-term borderline hypoxia (around 90% SpO2) are very poorly studied. There are no studies assessing the role of therapeutic oxygen in recovering acute respiratory disease in children (in particular with regard to symptoms). Recurrent variable hypoxia associated with obstructive sleep apnoea in children has been associated with lower levels of cognitive function,8 but acceptance of lower SpO2 targets (91–94%) for management of preterm infants has not been associated with subsequent cognitive defects.9 SpO2 commonly dips to 90% in normal children during sleep. An unpublished study (C O'Callaghan, personal communication) identified that SpO2 in infants going home following bronchiolitis commonly dipped to a percentage in the high 80s/low 90s during sleep with resolution over the subsequent 3–4 nights. The observed and expected improvement in SpO2 in recovering bronchiolitis over 3–4 days would suggest that it may be reasonable to propose that therapeutic oxygen may be stopped once SpO2 is on the plateau of the oxygen dissociation curve (ie, 90%), providing no further respiratory deterioration is expected. Significant respiratory deterioration is not expected once feeding is established.1

Given clinical concerns about the potential for hypoxia to limit recovery, studies are required to gauge the clinical effects of accepting a borderline hypoxic state during recovering acute lung disease. Such a study is planned in bronchiolitis.

View Abstract


  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.