Article Text
PDF
Abstract
Objective Examine admissions for bronchiolitis, comparing centres with oxygen saturation thresholds for admission of 90% versus 92%.
Design Prospective multi-centre service evaluation, all admissions for bronchiolitis during 4-week period, November 2018.
Setting Paediatric departments across 12 hospitals in the West Midlands, UK.
Patients 320 patients aged 6 weeks–1 year, diagnosis of bronchiolitis, exclusions: chronic illness or high dependency/intensive care admission.
Main outcome measures Reason for admission, admission saturations and length of stay.
Results Inadequate feeding was the the most common reason for admission (80%). Only 20 patients were admitted solely because of low saturations. Median peripheral oxygen saturation in this group was 88%. Median length of stay in 90% centres was 41 hours, against 59 hours for 92% centres (p=0.0074).
Conclusions Few patients were admitted solely due to low oxygen saturations, only one had a potentially avoidable admission if thresholds were 90%. Length of stay was significantly reduced in the 90% threshold centres.
- bronchiolitis
- paediatrics
Statistics from Altmetric.com
What is already known?
There is variation across centres in the admission threshold for oxygen saturations for patients with bronchiolitis.
While patients with bronchiolitis are most often admitted due to poor feeding, many require oxygen and this impacts on the length of stay.
What this study adds?
There are very few patients who have oxygen saturations of 90%–91% in air, admitted solely because of low saturations.
There would be very few admissions prevented if there was a universal oxygen saturation threshold of 90%.
Reduction of oxygen saturation discharge threshold from 92% to 90% is associated with a deceased length of stay.
Background
The Bronchiolitis of Infancy Discharge Study (BIDS) randomised controlled trial demonstrated that for patients admitted with bronchiolitis, a peripheral oxygen saturation (SpO2) threshold of 90% is safe compared with 94%,1 though admission thresholds were not examined.
Moreover, Schuh et al found that using falsely elevated SpO2 displays led to a significant reduction in admissions for bronchiolitis in their randomised controlled trial.2
Following these studies, many paediatric centres have reduced their SpO2 threshold for both admission and discharge for bronchiolitis from the 92% recommended by the National Institute for Health and Care Excellence (NICE)3 to 90%.
It is not known how this reduction in SpO2 admission threshold has influenced the characteristics of patients admitted with bronchiolitis.
Objective
We aimed to examine patients admitted with bronchiolitis, comparing centres with 90% SpO2 threshold to those with 92% threshold.
Design
Prospective multi-centre service evaluation, using standardised proformas and data collection spreadsheets. Routinely collected data were used, ethical approval was not required for this service evaluation; however, the project was registered with each centre’s audit and clinical governance department.
Setting
12 paediatric departments across the West Midlands region, UK; majority secondary care level, one tertiary children’s hospital.
Patients
All patients admitted under the care of the paediatric team, whether from primary care or emergency department, between 1 November 2018 and 30 November 2018, with admission diagnosis of bronchiolitis, including those admitted only for a short duration of observation, for instance to a paediatric assessment unit.
Exclusions were as per BIDS trial:<6 weeks and >1 year corrected gestational age; prematurity <37/40 with oxygen requirement in last 4 weeks; congenital heart disease with significant cardiovascular compromise; cyanotic congenital heart disease; cystic fibrosis; interstitial lung disease; immunodeficiency; direct admission to high dependency unit (HDU) or paediatric intensive care unit (PICU).
Comparison
Patients from all the 90% SpO2 threshold centres were pooled, and then compared with pooled patients from all the 92% centres.
Analysis was performed using Excel (Microsoft, USA) and Graphpad Quickcalcs (GraphPad Software, USA). χ2 was used for categorical data, Mann-Whitney U test for non-parametric data.
Main outcome measures
Lowest SpO2 in air at time of decision to admit.
Reason(s) for admission: low SpO2 in air, inadequate feeding, social concerns/anxiety, repeat attendance in same illness, clinical instability: tachycardia, tachypnoea, unwell/deteriorating, apnoea.
Length of stay, calculated from admission time/date and discharge time/date.
Results
Data were collected for 394 patients, 320 patients were included in analysis, after 74 exclusions; most common reasons: 34 <6 weeks of age, 10 >1 year of age and 9 HDU admissions. There were 162 patients admitted to the six centres with a 90% SpO2 threshold for admission, which included the tertiary hospital, and 158 admitted to the six centres with a 92% threshold. The overall mean age was 22 weeks (range 42–363 days); 23 weeks in the 90% centres and 21 weeks in the 92% centres (p=0.276).
Individual patients could have multiple reasons given for admission, see table 1. 80% of all patients were admitted because of inadequate feeding, whereas 32% were admitted due to low SpO2. In centres with 90% threshold, 27% were admitted for low SpO2, against 37% in 92% threshold centres (p=0.050).
Reasons given for admission, individual patients could have multiple reasons given, comparing patients admitted to centres with 90% SpO2 threshold for admission with those admitted to 92% threshold centres
The overall median oxygen saturation was 94% (range 66%–100%). The median was 95% in the 90% centres, and 94% in the 92% centres (p=0.085).
There were 20 patients who were admitted solely because of low SpO2. Eight were from 90% threshold centres and 12 from 92% centres. Median saturation in this subgroup was 88% (range 66%–90%) with only one patient having SpO2 of 90%, who was admitted to a 92% threshold centre.
Mean length of stay for all patients was 61 hours (SD 65 hours), median 46 hours (IQR 18–86). In centres with 90% admission, SpO2 threshold median length of stay was significantly shorter at 41 hours (IQR 13–72), against 59 hours (IQR 22–93) for 92% threshold centres (p=0.0037).
One centre which had an admission threshold of 92% but a discharge threshold of 90%, so analysis of length of stay was repeated with this centre included in the 90% threshold group, see table 2. Median length of stay in 90% discharge threshold centres was 41 hours (IQR 13–72) compared with 59 hours (IQR 22–93), (p=0.0074).
Mean and median length of stay, comparing centres with a discharge SpO2 threshold of 90% with those with a threshold of 92%
We examined the data to see whether the most senior decision-maker at the time of admission was a confounding factor. Tier 2 doctors (‘registrar’ equivalent) were the most common decision-makers, 77% of admissions. This was 73% in 90% threshold centres, and 82% in 92% threshold centres. The next most common decision-maker was a consultant at 17% of admissions, 18% in 90% threshold centres and 16% in 92% centres. Tier 1 doctors (‘senior house officer’ equivalent) were decision-makers in only 4% of admissions, 7% and 1% in 90% and 92% threshold centres, respectively.
While there was a significant difference in seniority between 90% and 92% SpO2 centres when examining Tier 1 doctors, Tier 2 doctors and consultants (p=0.020), this was not significant when examining only Tier 2 doctors and consultants (p=0.501).
Conclusions
Our large multi-centre prospective service evaluation of patients admitted with bronchiolitis showed that reducing admission SpO2 threshold affects few patients, though discharge SpO2 threshold does affect length of stay.
We showed that poor feeding was the most common reason for admission, in line with previous evidence.4 Mean SpO2 was not significantly different between the groups, and there were low numbers of patients admitted solely for hypoxia. Only one patient had saturations of 90%, in a 92% threshold centre, suggesting one patient of the 158 admitted to centres with a 92% threshold was a potentially avoidable admission, had the threshold been reduced to 90%. It may be at SpO2 of 90%–92% patients will be struggling to complete feeds, or showing clinical signs, requiring admission regardless.
Although IQRs were high, length of stay was significantly shorter in centres with a discharge SpO2 threshold of 90%, compared with 92%. The reduction in length of stay from reducing SpO2 thresholds was similar to that seen in BIDS,1 and supports previous evidence that oxygen requirement is a key determinant of length of stay in bronchiolitis.4 5
Strengths of this study include it being a large multi-centre service evaluation across a whole region, with a subsequent reasonable size population. In addition, the prospective nature could capture the actual reasons for admission, in real time, from the decision-maker.
However, the limitation of prospective studies is that there is a higher chance of missing data. Moreover, reasons given for admission may be not be fully captured, influenced by subconscious reasoning or subjective interpretation of clinical signs. Though BIDS examined readmission rates, finding no increase with a reduction in SpO2 thresholds, we did not examine this ourselves.
We examined seniority of decision-maker as a potential confounding factor; overall, the centres did not substantially differ. However, there will be other confounding factors not accounted for: socio-economic variation; differing geographical factors across rural and urban areas; and different arrangements in service provision both within hospitals and the community support outside hospitals.
Our study suggests that reducing SpO2 admission thresholds from 92% to 90% does not affect admissions. However, reducing SpO2 discharge thresholds does reduce length of stay, and further studies examining possible benefits in terms of bed capacity and cost would be helpful.
Acknowledgments
This project was a Paediatric Research Across the Midlands (PRAM) collaboration, the authors thank all the local paediatric trainee leads, their supervising consultants, and also thank Dr Amy Henderson and Dr Kevin Morris for support with project planning, and Dr Paul Davis for support with statistical analysis.
Footnotes
Collaborators Dr Bhavna Singham; Dr Andy Taylor; Dr Vic Parsonson; Dr Harriet Swallow; Dr Joanna Nuthall; Dr Sonia Goyal; Dr Hannah Cooper; Dr Katy Francis; Dr Isobel Fullwood; Dr Liz Fairholme; Dr Prasadi Desilva; Dr Deevena Chinthala; Dr Kun Hu; Dr Fiona Halton; Dr Nabil Fassaludhin; Dr Ceri-Louise Chadwick. All affiliated to Paediatric Research Across the Midlands.
Contributors TJvH was involved in project design, statistical analysis and manuscript. BS was involved in project design and liaison with local data collectors. EB was involved in project design. IW was involved with project design. All authors reviewed and approved final manuscript. BS, AT, VP, HS, JN, SG, HC, KF, IF, LF, PD, DC, KH, FH, NF and CC are PRAM affiliated local collaborators, led the project locally at each site: registered with governance, collected and reviewed local data.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.