Video Abstract
Bronchiolitis is the leading cause of pediatric hospital admissions. Hospital-at-Home (HAH) delivers hospital-level care at home, relieving pressure on the hospital system.
We aimed to review the feasibility, acceptability, and safety of HAH for bronchiolitis, and assess the cost-impact to hospitals and society.
Ovid Medline, Embase, Pubmed, Cochrane Library, CINAHL, and Web of Science.
Studies (randomized control trials, retrospective audits, prospective observational trials) of infants with bronchiolitis receiving HAH (oxygen, nasogastric feeding, remote monitoring). Studies were limited to English language since 2000.
We reviewed all studies in duplicate for inclusion, data extraction, and risk of bias.
Ten studies met inclusion criteria, all for home oxygen therapy (HOT). One abstract on nasogastric feeding did not meet full inclusion criteria. No studies on remote monitoring were found. HOT appears feasible in terms of uptake (70%–82%) and successful completion, both at altitude and sea-level. Caregiver acceptability was reported in 2 qualitative studies. There were 7 reported adverse events (0.6%) with 0 mortality in 1257 patients. Cost studies showed evidence of savings, although included costs to hospitals only.
Small number of studies with heterogenous study design and quality. No adequately powered randomized control studies.
Evidence exists to support HOT as feasible, acceptable, and safe. Evidence of cost-effectiveness remains limited. Further research is needed to understand the relevant impact of HAH versus alternative interventions to reduce oxygen prescribing. Other models of care looking at nasogastric feeding support and remote monitoring should be explored.
The coronavirus disease 2019 pandemic has disrupted traditional ways of practicing medicine and challenged health care systems across the globe to reconsider ways to best deliver health care. Hospital-at-Home (HAH), where acute hospital-level care is delivered in the patient’s own home, has gained increasing recognition as a safe and high-value alternative to admission to hospital. In November 2020, the Centers for Medicare & Medicaid Services launched a formal funding arrangement leading to increased interest in HAH models. In adults, a small randomized control trial (RCT) of HAH for infective conditions, heart failure, chronic obstructive lung disease, and asthma demonstrated cost benefits to the hospital with higher rates of physical activity and lower rates of readmission rates for patients.1 In children, proposed benefits include both reduced pressure on hospital beds, plus psychological benefits such as less anxiety for the child and less family disruption.2 Pediatric HAH has traditionally been offered for management of chronic conditions (cystic fibrosis, oncology, parenteral nutrition) or semilong-term care (long-term antibiotics, wound care, support with transition to home).3 However, increasingly, institutions are realizing the value of getting children home for even a few days with acute illnesses such as cellulitis, urinary tract infection, and gastroenteritis.4–7 However, although HAH for chronic, semilong-term, and some acute conditions has been demonstrated to be safe and cost-effective in children, it is uncertain whether the same is true for acute respiratory illnesses where there may be a higher readmission rate.8
The most common cause for hospitalization in children is bronchiolitis, a viral infection of the lower respiratory tract.9 In an era of increasing pressure on limited hospital resources, bronchiolitis places substantial stress on hospital beds, particularly during winter months. Surges of coronavirus disease 2019 coinciding with respiratory syncytial surges (the commonest viral agent of bronchiolitis) have seen unprecedented demand being placed on pediatric services over the past 2 years.10
Treatment of bronchiolitis is supportive, with infants with moderate–severe symptoms requiring nasogastric tube (NGT) feeding support and oxygen supplementation.11,12 Rates of morbidity and mortality are low, with ∼5% of affected infants requiring higher levels of respiratory support such as pressure support or ventilation.9 Infants are also frequently admitted for observation and monitoring in the context of high potential for deterioration, typically because of young age or the presence of comorbidities.11
The introduction of pulse oximetry, allowing for noninvasive monitoring of oxygen levels, led to rates of hospitalization for bronchiolitis soaring in the 1980s. Oxygen treatment thresholds vary, with higher thresholds associated with increased admissions and longer length of hospital stay.13–15 Despite this increased hospital resource utilization, no improvements in mortality were seen, raising the question of overdiagnosis and treatment of hypoxia.16 At high altitude, hypoxia occurs at even milder levels of illness, resulting in larger numbers of children meeting thresholds for oxygen therapy.17 The lack of certainty regarding what constitutes a safe oxygen level, combined with a growing recognition that transient desaturations are common and safe in infants with bronchiolitis, has led to the discouragement of continuous oxygen monitoring and reductions in treatment thresholds.16,18 American guidelines now recommend an oxygen saturation threshold of 90% before commencing oxygen therapy with United Kingdom and Australasian guidelines recommending a threshold of 92%.11,12,19 Subsequently, rates of hospitalization in the United States have fallen from 17.9 per 1000 person-years in 2000 to 13.5 in 2016 after the new US guideline was introduced.9
Despite a reduction in hospitalization rates for bronchiolitis per person-years, overall bronchiolitis hospitalizations in the United States have increased from 16% to 18% of total hospitalizations per year.9 Continued efforts to decrease hospitalization, facilitate earlier discharge, and consider alternative models of care are imperative to alleviate the pressure on hospital flow and bed access.
Given the growing interest in and potential impact of the HAH model, this systematic review seeks to determine whether bronchiolitis requiring hospital-level care is feasible, acceptable to families, and can be safely managed in the home. In addition, we seek to assess the impacts on costs for both the health care system and society.
Methods
This systematic review was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.20 A protocol was registered with the International Prospective Register of Systematic Reviews on August 18, 2021 (registration CRD42021268417).
A systematic search of the Medline, Embase, Pubmed, Cochrane Library, CINAHL, and Web of Science databases was conducted by using Medical Subject Headings and keywords adapted for each database. The full search is available in Supplemental Information, and search terms included: (1) bronchiolitis, (2) ambulatory or outpatient care, and (3) oxygen, enteral feeds, or remote monitoring. We used broad search terms to capture remote monitoring, including telemedicine, pulse oximetry, telemonitoring, apps, m-health, and virtual health care. The search was limited to literature published in English between January 1, 2000, and March 15, 2022.
The population, interventions, comparison, outcomes framework was used to define the inclusion criteria. Inclusion criteria were: (1) majority of participants were infants with bronchiolitis, and (2) management at home with oxygen therapy, and/or NGT feeding, and/or remote monitoring (as defined above). A comparator group of hospital management was not required for the study to be included. We included RCTs, pilot RCTs, observational cohort studies, case series/reports, and retrospective audits. Unpublished data, study protocols, opinion pieces, and commentaries were excluded. The search was checked for completeness by ensuring it captured 3 sentinel papers as determined by the research team.21–23 References of included papers were reviewed for further publications meeting criteria.
Outcomes included feasibility (number of eligible patients, readmission rates), acceptability (parental satisfaction with and uptake of HAH), safety (adverse events), and costs to the health care system and family.
Title and abstract reviews were conducted by 2 independent reviewers (J.L., S.B.) against the inclusion criteria. If any information was not clear in the title or abstract, the full-text article was reviewed. Studies that did not meet the eligibility criteria were excluded. The full text of the remaining papers was critically assessed for inclusion in the review. Any conflicts were resolved by a third reviewer (H.H.).
Data collected including study design, study location and altitude, patient details, sample size, and intervention details, and outcomes were extracted by 2 coauthors (J.L., S.B.) using a standardized data extraction form within the Covidence software (www.covidence.org). The same 2 authors independently assessed risk of bias for all included papers by using the Quality Assessment with Diverse Studies guidelines for assessing quality in studies using different methodologies.24 This tool evaluates methodological quality, evidence quality, and quality of reporting, allowing reflection on areas of strength and weakness across studies of diverse designs.
Results
Study Selection
A total of 1312 unique studies were identified (527 duplicates removed) and 10 studies met full inclusion criteria (Fig 1).
All 10 studies reported on home oxygen therapy (HOT). One abstract reported on NGT feeding at home for infants with bronchiolitis but did not meet inclusion criteria. No reports of remote monitoring were found.
Home Oxygen Therapy (HOT)
Studies dated from 2006 to 2021 across a total of 5 unique study populations (Supplemental Table 2). Seven studies were from 3 high-altitude settings in the United States across 2 states (Colorado, Utah). Studies from Colorado (n = 4) discharged patients directly from the emergency department, whereas studies from Utah (n = 3) used an observational unit (OU)–HOT protocol.21–23,25–28 The OU was a specific ward designed to rapidly discharge infants with or without oxygen.28 Three studies were at sea-level; Perth, Australia (n = 2, discharged from ward admission), and 1 theoretical chart review from Montreal, Canada.29–31 Three of the Colorado papers report on the same population over 3 different time points.21,22,26 In contrast, the Utah and Australian papers report on overlapping cohorts.23,27,28,30,31 Two studies were pilot RCTs, 6 were retrospective chart reviews including 1 theoretical chart review, and 2 were prospective observational studies.
Infants were typically observed in-hospital for a period before being transferred to HAH. This period of observation varied between 4 and 24 hours, with the maximum oxygen level supplied to infants between 0.5 and 1 L per minute. The oxygen saturation threshold to warrant oxygen therapy at home was 92% across all studies at sea-level, whereas studies at higher altitude used lower thresholds of 88% to 90%.
All studies excluded infants with chronic cardiopulmonary conditions or a history of apnea. The lower age range was set between 2 and 3 months in all studies, although Halstead et al report on children as young as 1 month in their retrospective chart review.26 The maximum age range varied from 12 months, which is the age at which bronchiolitis is usually defined in Australia, to >3 years in a study by Zappia et al, which included other lower respiratory tract illnesses complicated by hypoxia.11,31
Outcomes
Supplemental Table 3 provides a summary of the key outcomes.
Feasibility
HOT appears feasible at high altitude, with 25% to 50% of infants with bronchiolitis meeting eligibility criteria and between 70% and 82% of these eligible children being successfully discharged to HAH.21–23,27,30 In the first study by Bajaj et al, 36 of the original 53 patients (68%) assigned to HOT safely completed their treatment at home.21 Of the remaining 17 patients, 10 (18.8%) no longer met criteria (resolved oxygen requirement or worsening clinical status) after an 8-hour period of observation and were either admitted or discharged accordingly. One child returned to hospital (1.8%) and 6 parents withdrew (11.3%). Feasibility at sea-level was demonstrated in the 2 studies from Australia, with a total of 112 patients managed with HOT over a 4-year period (total eligible population unknown).30,31 Twenty-two of these infants participated in an RCT with a reported uptake of 75.8%.30 Reasons for nonuptake were predominantly research logistics, including presenting outside research hours or living out of the catchment area. Three patients (5.2%) refused to partake. The follow-up study by Zappia et al reported 7 readmissions out of 112 patients (6%) for medical reasons and no serious safety events.31 A Canadian retrospective chart review at sea-level reviewed the records of all admitted patients with bronchiolitis and calculated the eligible proportion for HOT to be 7%.29
Readmission rates within 7 days varied between 2.4% and 9.4%, with pooled rates of 6.3% (n = 79 of 1257) among 5 studies.21,22,25,26,31 The findings from Utah report on the impact of both HOT and an observation unit, with similar readmission rates at 2.4 to 5.8%, respectively.23,28 There was no difference in readmission rates between HAH in sea-level versus altitude settings. The shorter observation period (4 hours) in the Colorado study by Flett et al was associated with a higher readmission rate at 9.4%.25
Acceptability
Two studies reported on caregiver satisfaction.21,22 Both reported high rates of caregiver satisfaction and comfort, with 79% to 88% of those receiving HOT preferring HOT to hospitalization (n = 225). In terms of uptake of HAH among eligible patients, this varied between 70% and 82% in the 3 studies that reported this.21,22,30 Reasons for nonuptake were primarily medical (other care needs, escalating respiratory distress, or resolving oxygen requirement); however, 12 of 78 (15%) across 3 studies declined HAH because of caregiver discomfort with the concept.
Safety
Three patients across the reported cohort of 1257 patients required escalation to intensive care. One of these patients required intubation and ventilation. There were no deaths reported and no other adverse events.
Costs
Limited data are provided on costs and impact on hospital length of stay (LOS). The 3 studies that provided financial data provided costs for different outcomes and in different currencies; Halstead et al (Colorado) and Sandweiss et al (Utah) report savings to the health care system of $1262 to $1281 in United States dollars per patient and Zappia et al (Australia) $2100 in Australian dollars (equivalent to $1500 in United States dollars) per day.26,28,31 Ohlsen et al (Utah) report reduction in LOS and cost-savings across long time periods and large patient numbers, with varying results. In the initial abstract, 9887 patients across 17 hospitals between 2007 and 2017 were included.27 All but 1 hospital had declining rates of HOT; therefore, the subsequent article focuses on a single hospital over a longer period of time resulting in a slightly smaller sample of 7116 patients.23 This latter article includes the impact of both the OU and HOT, with only 50% of patients being discharged on HOT. Reductions in LOS are reported as 30.6 hours and cost savings of $4181 United States dollars per episode for the combined impact of an observation unit and HOT protocol.
All studies reporting costs report difficulty in adequately accounting for the costs of HOT and present the cost-saving to the hospital only. No studies report societal (family) costs.
Risk of Bias Within Studies
The Quality Assessment with Diverse Studies outcomes for the 10 studies are summarized in Supplemental Table 4. Three studies (Ohlsen et al, Tie et al, Zappia et al) were of low quality because of poorly stated research aims and lack of detail regarding recruitment and analytical methods.27,30,31 Although there were 2 RCTs, neither achieved the required sample size to draw adequately powered conclusions.21,30 Bajaj et al stopped prematurely because HOT became widely adopted. The reason for premature discontinuation was not given by Tie et al. The remaining studies use observational methodology without sample-size calculations.
The highest-quality studies include the first pilot RCT from Bajaj et al (n = 56), a retrospective chart review by Halstead et al (n = 649), and a prospective observational cohort study (n = 225) from Freeman et al, all from the same setting.21,22,26 These studies shared the same protocol (up to 0.5 L per minute oxygen to maintain saturations of 90% after a period of 8 hours observation), although had different inclusion criteria, with Halstead et al including infants as young as aged 1 month, compared with Bajaj et al (2 months) and Freeman et al (3 months). The reason that an older age range was used in the study by Freeman et al is unclear. Sandweiss et al contribute a high-quality article from Utah with their interrupted time series analysis examining the impact of a HOT program.28 Flett et al aimed to describe the outpatient course and burden of care, and leveraged a universal electronic medical record across inpatient and outpatient settings to provide a strong retrospective study.25
Ohlsen et al published 2 studies with large populations, but they report on the combined impact of an OU and HOT making the specific outcomes of HOT difficult to extrapolate.23,27 An interrupted times series design to assess the impact of the OU–HOT protocol provides a strong study design to account for confounding variables and seasonality.
The variation in upper age range introduces a bias to studies that included infants >1 year of age because the potential for other diagnoses, including reactive airway disease and pneumonitis, becomes greater, limiting the specificity of the results. In addition, findings in the high-altitude settings which represent the majority of high-quality studies are poorly generalizable to sea-level settings. Finally, most studies come from the United States where HOT is predominantly provided by an outpatient model rather than the HAH model used in Australia. The health care system and costs differ greatly between these countries, limiting the generalizability between 1 model and the other.
Nasogastric Feeding
One abstract examined parent perspectives on nasogastric feeding at home, but was excluded because the intervention was not performed and therefore the abstract did not meet our inclusion criteria.32
Remote Monitoring
Two further studies reported on bronchiolitis managed in the home compared with hospital: a publication that included bronchiolitis as 1 of the conditions managed through their HAH program and an abstract reporting a nurse-led telephone review/ home-visit service to avoid representations and admissions for respiratory conditions including bronchiolitis.33,34 However, neither fulfilled inclusion criteria in that they lacked information describing their patients, their interventions, or their bronchiolitis-specific outcomes.
Discussion
Despite the immense pressure placed on the pediatric hospital system by infants presenting with bronchiolitis, there is a paucity of evidence regarding the feasibility, acceptability safety, and costs of HAH in bronchiolitis care.
Seven studies from high-altitude United States provide evidence that HOT is feasible and safe among this cohort of patients. Readmission rates measured over many years are <6% and very few infants managed with HOT (<1%) require intensive care. Cost-savings, examined from a hospital perspective only, are suggestive of a positive cost-impact. The degree of impact is hard to ascertain because of differences in methodologies, health care systems, and a failure across all studies to account for the cost of the home therapy. Although therapy at home is less expensive than hospitalization for both hospitals and families, the potential for a slower oxygen wean and therefore an increased LOS means further studies are required to properly measure the cost-impact of HOT.
Further limiting the generalizability of these findings is the varying definition of bronchiolitis, with definitions in the United States including children up to 24 months, which may overlap with other diagnoses such as asthma that respond to alternative therapies such as bronchodilators.
The impact of HOT at sea-level remains uncertain because the evidence comes from 1 Canadian theoretical chart review and 2 low-quality studies from Australia. The Canadian study by Gauthier et al uses a 24 hour observation period before theoretical transfer. This is a longer observation period compared with most real-world studies, which would reduce the pool of eligible infants and therefore underestimate the potential impact.29 The Australian pilot RCT by Tie et al compares 22 children on HOT versus 22 children in hospital.30 This study was underpowered and the outcome measured hospitalization days only without accounting for the costs or LOS of the at-home program. Thus, although HOT appears feasible and safe among small numbers of infants at sea-level, the impacts on parental acceptability and hospital savings remain in question.
Parental attitudes toward management at home were only captured in 2 studies, both of which were at high altitude. Although logistical concerns with obtaining oxygen at home were commonly stated, caregiver acceptability of managing acute bronchiolitis in the home environment is encouraging, with 88% of caregivers in Freeman’s study (n = 188) reporting they would prefer home oxygen again if on offer.22
From these studies, it is possible to conclude that HOT may play a role in high-altitude settings where hypoxia occurs at a lower level of acuity and acts as a barrier to discharge in a larger proportion of infants.35 However, the role of HOT at sea-level remains underresearched. Alternative methods to reduce LOS among infants at sea-level, such as reducing the use of continuous pulse oximetry, reducing the oxygen threshold for treatment, and interpreting oxygen saturations within the context of the child and their illness, should also be explored. A child with fleeting desaturations who is otherwise recovering and feeding well may benefit more from discharge and cessation of monitoring than potential overtreatment with oxygen through a HAH program.
At sea-level, more children with bronchiolitis are admitted to hospital for feeding support than oxygen therapy.36 NGT feeding is already provided in the home environment for infants and children with long-term feeding needs and has long been established as a safe and feasible practice. Converting this to the acute HAH setting would require parental training and evaluation of training time required versus bed-days saved to better understand whether the training investment for an illness with such a short duration would be worthwhile. The study by Wasala et al suggests that parental appetite for such a model exists and further investigation is therefore warranted.32
One cohort of potential interest for home management is composed of infants who are admitted for observation only. No studies have assessed the role of observation in the home as a safety net for moderately unwell patients who need neither HOT nor NGT support. Admitting an infant who does not require any hospital level intervention simply because of risk of deterioration, when this condition overall has a low morbidity and mortality, raises the opportunity for alternative models of care. These may be able to address both provider and caregiver anxiety about these infants. Simple discharge is 1 option but may lead to recurrent representation to emergency departments. With today’s access to virtual monitoring, further research at sea-level should include consideration of alternative models of care for infants with bronchiolitis, such as remote monitoring and telehealth review for the infant at risk for deterioration. Given the risk of overtreatment through continuous monitoring, intermittent oximetry measurements interpreted in the context of a clinical review via teleconsultation would likely provide a cost-effective alternative to observation in hospital.18,37
Strengths and Limitations
The strengths of this review are its comprehensiveness: a systematic search of the literature across 5 databases using a broad search strategy make it unlikely that relevant publications have been missed. However, there was a paucity of data from sea-level and it is possible that there is a risk of publication bias, with most studies coming from high altitude.
Conclusions are therefore limited by a small number of published studies among a limited number of unique settings. Cost studies found evidence of cost savings; however, they included hospital costs only, omitting costs to families and the costs of the home program, thereby limiting the ability to draw conclusions about the cost-effectiveness of HAH.
Conclusions
There is evidence to support the safety and feasibility of managing oxygen requirements for bronchiolitis in the home, especially in areas of high altitude. Further studies at sea-level are required to better understand the role of HAH in bronchiolitis management. There is a need for more research to investigate the cost-effectiveness by including costs to society of oxygen therapy and exploring health care systems outside of the United States to investigate alternate strategies to reduce inpatient durations (such as lowering oxygen treatment thresholds further) and to examine the safety and feasibility of models of care to support other care requirements such as NGT feeding and observation of those at risk for deterioration.
Without further research into alternative ways to care for children with moderately severe bronchiolitis out of hospital, this illness will continue to place heavy demand on pediatric health care systems.
Acknowledgment
This research was undertaken as a part of a doctor of public health degree at the University of New South Wales.
Dr Lawrence conceptualized and designed the review, screened articles, extracted data, assessed for quality, and drafted, reviewed, and revised the manuscript; Dr Boyce screened articles, extracted data, assessed for quality, and reviewed the manuscript; Drs Walpola and Hiscock conceptualized and supervised the design of the review, advised on data extraction processes, and critically reviewed the manuscript; Dr Sharma conceptualized the review and critically reviewed the manuscript; Dr Bryant critically reviewed the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
This study is registered at PROSPERO, CRD42021268417, http://www.crd.york.ac.uk/prospero/display_record.asp?ID=CRD42021268417.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2022-058042.
FUNDING: No external funding.
CONFLICT OF INTEREST DISCLAIMER: The authors have indicated they have no conflicts of interest relevant to this article to disclose.
