The Use of Overnight Oximetry and Polysomnography in Late Preterm to Term Newborns Requiring Home Oxygen at Discharge

ABSTRACT

Background and Objective

Overnight oximetry is routinely used to assess long‐term oxygen therapy (LTOT) in late preterm and term newborns; however, there is a lack of population‐specific evidence‐based guidelines. This single‐center, retrospective cohort study describes the clinical characteristics and management approach for LTOT in this population.

Methods

Newborns ≥ 34 weeks' gestation born between 2016 and 2024 and discharged on LTOT were identified through databases at the Women's and Children's Hospital. Overnight oximetry, completed before and/or after discharge, and polysomnography results were collected, with the difference between first and second overnight oximetry analyzed using a paired t‐test with Wilcoxon matched‐pairs signed‐rank test.

Results

A total of 148 newborns were identified, with 128 meeting eligibility. The median (range) gestational age was 39 (34–42) weeks, and birthweight was 3590 (1590–4790) grams. Overall, newborns had 2 (1–15) overnight oximetry tests and 1 (1–5) polysomnography tests. Airway malacia was the primary diagnosis (48.1%). While desaturation events occurred in all studies, the total number of desaturations (p = 0.0008), % time SpO₂ < 90% (p = 0.002), < 88% (p = 0.01), and desaturation event index (p < 0.0001) improved between initial and subsequent oximetry. Median obstructive apnea‐hypopnoea index was 5.7 (1–46)/hr and central apnea index 5.4 (0–87)/hr. LTOT duration was 229 (38–2347) days.

Conclusion

Late preterm and term newborns showed significant improvement on repeat overnight oximetry. Diagnoses and clinical pathways for this cohort remain heterogeneous, with the decision to undertake repeat oximetry not reflecting initial severity, highlighting the need for standardized population‐specific guidelines.

1. Introduction

Long‐term oxygen therapy (LTOT) is a critical component of post‐discharge care for newborns with ongoing supplemental oxygen requirements, facilitating hospital discharge when all other medical milestones and stability criteria have been met. LTOT is most often prescribed for preterm newborns with bronchopulmonary dysplasia and related chronic lung disease secondary to very preterm birth [1]. Although late preterm to term newborns are less likely to require home oxygen, their higher birth prevalence results in them constituting nearly one‐fifth of all newborns discharged on LTOT [2].

For this population, the indications for LTOT differ from those of more preterm newborns and include respiratory diagnoses other than chronic lung disease, such as sleep‐disordered breathing, central or obstructive apnea, genetic and neuromuscular disorders [2, 3]. While improved oxygenation by LTOT has positive effects on nutrition and growth for newborns diagnosed with intermittent desaturations, with the potential for related chronic hypoxemia [4, 5], the limited evidence and recommendations on when and how LTOT should be initiated and discontinued result in wide practice variations [6].

While polysomnography is the gold standard to objectively assess respiratory stability and guide LTOT use prior to primary hospital discharge [6], it is resource‐intensive, expensive and may not be readily accessible. As a result, overnight oximetry represents a more accessible alternative, particularly when there is no concern of abnormal control of breathing or airway obstruction. However, interpretation of neonatal overnight oximetry data remains largely dependent on clinician experience and consensus guidelines extrapolated from older populations of patients, resulting in an ongoing absence of robust evidence‐based recommendations [7]. This is also the case for the timing and frequency of repeat overnight oximetry +/− polysomnography, which vary widely, and for approaches to eventual weaning of LTOT, resulting in significant practice variation [6, 8]. Here, we review the clinical characteristics of a contemporaneous cohort of late preterm and term newborns discharged on LTOT from a tertiary women's and children's hospital and describe subsequent overnight oximetry and polysomnography findings to provide insight into this patient group.

2. Materials and Methods

This retrospective observational study was conducted at the Women's and Children's Hospital (Adelaide, Australia), focusing on newborns ≥ 34 weeks' gestation born between 2016 and 2024 and discharged home on LTOT. Exclusion criteria included newborns > 1 year of age at first oximetry, newborns who did not receive follow‐up at the primary hospital, newborns transferred before discharge or newborns who died before discharge. Deidentified neonatal data were cross‐referenced against corresponding neonatal and pediatric respiratory unit home oxygen databases. Baseline clinical data, including date of birth, gestational age, birthweight, and postnatal age at time of overnight oximetry +/− polysomnography, were recorded for each subject. LTOT duration was estimated from oxygen equipment collection dates, which may not reflect the exact clinical cessation date but are likely to approximate LTOT cessation. This study was approved by the SA Women's and Children's Health Network Human Research Ethics Committee (1359 A). All procedures involving human participants were conducted in accordance with the ethical standards of the Declaration of Helsinki (1964), as revised in 2013. The requirement for informed consent was waived because the study was a retrospective observational study using de‐identified data.

2.1. Data Collection

Overnight oximetry was performed using Nellcor™ oximeters (Medtronic, Macquarie Park, NSW, 2113, Australia) prior to or following discharge, depending on the timing of LTOT assessment. The oximeter had an averaging time of 4 s, with data collected in a standardized way. Data points, including postnatal age at test, duration of oxygen saturations < 90%, < 88%, and < 80%, number of desaturations (≥ 4 points below baseline), mean high, mean low and desaturation event index (number of desaturation events per hour) were generated using the Profox™ software (Profox Associates Inc., Coral Springs, Florida, 33065, USA). Polysomnography data, when available for included subjects, were collected from sleep study reports with SpO₂ nadir, obstructive apnea hypopnea index (OAHI), central apnea index (CAI), transcutaneous carbon dioxide monitoring (TcCO₂) and outcome recorded.

2.2. Statistical Analysis

Descriptive statistics were calculated for the demographic variables, with clinical data for oximetry and polysomnography parameters presented as median (range). Data were analyzed using SPSS version 28 (IBM Corporation, Armonk, New York, USA) and GraphPad Prism version 9 (Boston, Massachusetts, USA). The difference between first and second overnight oximetry was determined by a paired t‐test with Wilcoxon matched‐pairs signed‐rank test. Statistical significance was defined as p < 0.05.

3. Results

Between 2016 and 2024, a total of 148 newborns ≥ 34 weeks’ gestation at birth were discharged from the Women's and Children's neonatal unit on LTOT, comprising 59.2% of all newborns discharged on LTOT. Of these, 17 were excluded due to missing data, an additional 2 were transferred prior to discharge with no follow‐up at the primary hospital, and 1 newborn was excluded due to having their initial oximetry > 1 year of age, resulting in a final cohort of 128 (Figure 1). Overall, 88 (68.8%) of newborns were male, with a median (range) gestational age of 39 (34–42) weeks and birthweight of 3595 (1590–4790) grams. Age at primary hospital discharge was 17 (6–55) days. The underlying diagnosis resulting in the need for LTOT is shown in Table 1. Airway malacia (laryngomalacia, tracheomalacia and bronchomalacia) was the commonest diagnosis confirmed by clinical assessment, followed by mixed apnea, diagnosed by polysomnography.

Figure 1.

PRISMA diagram of eligible records. [Color figure can be viewed at wileyonlinelibrary.com]

Table 1.

Clinical diagnoses of infants discharged on LTOT (long‐term oxygen therapy).

Diagnosis	N	%
Airway malacia	62	48.1
Mixed apnea	33	25.6
Obstructive sleep apnea	20	15.5
Respiratory distress syndrome	12	9.3
Cleft lip and/or Palate	8	6.2
Micro/Retrognathia	7	5.4
Central sleep apnea	6	4.7
Meconium aspiration	6	4.7
Sleep disordered breathing	4	3.1
Hypotonia	3	2.3
Hyaline membrane disease	3	2.3
Pierre Robin Sequence	2	1.6
Hypoxic‐ischemic encephalopathy	1	0.8
Trisomy 21	1	0.8

Note: Categorical data are shown as n, (%).

Overall, newborns underwent 2 (1–15) overnight oximetry tests. The first overnight oximetry was conducted in the neonatal unit for 75.7% of newborns before discharge, with 12.5% having their initial overnight oximetry after discharge, and the remainder not having an overnight oximetry performed before commencing LTOT. The postnatal age at first overnight oximetry was 12 (5–250) days. Repeat overnight oximetry was conducted in 60.9% of newborns, at a postnatal age of 110 (10–470) days.

Data for newborns who had both first and second oximetry are shown in Table 2. During the first overnight oximetry, all newborns had desaturation events, defined as an absolute drop in SpO₂ of 4% or more, with 142 (5–496) events recorded, and a desaturation event index of 16.9 (0.6–22.2). The mean highest SpO₂ was 98 (85–100) %, and the mean lowest SpO₂ was 91 (79–100)%. Newborns spent a median of 1.7 (0–99.8) % of the time with an SpO₂ < 90% and 0.1 (0–7)% with an SpO₂ < 80% (Table 2). At repeat overnight oximetry, significant improvements were seen in the total number of desaturations (p = 0.003), % of time with an SpO₂ < 90% (p = 0.02) and < 88% (p = 0.05), and desaturation event index (p < 0.0001).

Table 2.

Overnight oximetry data for newborns with both initial and repeat overnight oximetry.

	First Oximetry		Second Oximetry		p
	Median	Range	Median	Range
Age (days)	12	5–250	110	10–470	< 0.0001
% Time with SpO2 < 90	1.7	0–99.8	0.8	0–17.6	0.002
% Time with SpO2 < 80	0.1	0–7	0.1	0–3.2	0.65
% Time with SpO2 < 88	1	0–99.6	0.4	0–6.2	0.01
No. of desaturations	142	5–496	88	3–512	0.0008
Mean high (%)	97.9	85–100	99	95–100	0.12
Mean low (%)	91	79–100	92	85–95	0.28
Desaturation Event Index	16.9	0.60–62.2	9.1	0.30–53.8	< 0.0001

Note: Continuous data are shown as % time or n.

Polysomnography was conducted as a follow‐up to overnight oximetry in 42% of the cohort at a postnatal age of 56 (6–504) days, with 1(1–5) polysomnography tests performed per newborn. For the initial polysomnography, TcCO₂ ranged from 33 to 45 with a SpO₂ nadir of 86 (74–94) %. The OAHI, defined as the total number of obstructive apnea + mixed apnea + hypopnea/total sleep time (hours), was 5.7 (1–46)/hr, with a CAI, defined as the number of central sleep apnea events per hour of sleep, of 5.3 (0–87)/hr. Overall, the duration of LTOT was 229 (38–2347) days, while age at LTOT cessation was 244 (14–2368) days. Infants who subsequently underwent polysomnography experienced more desaturation events and higher desaturation event indices (p < 0.05) on their initial oximetry, while other parameters were similar to those who did not get referred for polysomnography. The duration of LTOT was longer for newborns who had a subsequent polysomnography compared to those who had overnight oximetry alone (313 (74–2347) vs. 206 (45–1635) days, p = 0.01), suggesting greater disease severity in those undergoing polysomnography.

4. Discussion

Here, we review the clinical characteristics of a contemporaneous cohort of late preterm and term newborns discharged on LTOT from a tertiary women's and children's hospital, and subsequent overnight oximetry and polysomnography results. Late preterm and term infants in this cohort typically spent 7.5 months on LTOT, alongside considerable variation in clinical diagnoses and assessment pathways. The literature on the use and interpretation of overnight oximetry to guide LTOT in late preterm and term newborns with persistent oxygen requirements during the neonatal period is limited, contributing to variability in its clinical application. This represents an important knowledge gap, as in the current study, newborns born ≥ 34 weeks' gestation accounted for almost 60% of all newborns discharged home from the neonatal unit on LTOT from a tertiary women's and children's hospital over a contemporaneous 9‐year epoch. These findings highlight the need for evidence‐based, standardized approaches to LTOT initiation, monitoring and weaning in this population.

Historically, research has predominantly focused on very preterm infants, where LTOT is most commonly indicated for chronic lung disease secondary to bronchopulmonary dysplasia. In the largest retrospective American cohort assessing LTOT use, late preterm to term infants comprised 19% (246 of 1286) of infants discharged on LTOT [2]. Gestational age and mechanical ventilation emerged as the key predictors of LTOT use; however, 62% of the late preterm to term infants had never required ventilation. This highlights the greater clinical heterogeneity observed in the late preterm and term population. While mixed apnea was the commonest underlying cause confirmed by polysomnography, we observed a high prevalence of co‐existing clinical diagnoses. Approximately three‐quarters of newborns in this group were diagnosed with more than one contributing condition, with airway malacia (laryngomalacia, tracheomalacia and/or bronchomalacia) the most prevalent co‐diagnosis present in the cohort.

Most newborns discharged on LTOT underwent overnight oximetry prior to discharge, with the initial overnight oximetry performed on average at 25 days old. While a practical and accessible alternative to polysomnography, overnight oximetry is limited in its utility [9]. An Australian survey found significant variability in the interpretation of overnight oximetry between pediatric clinicians and respiratory specialists [7]. Despite respiratory specialists performing better on an oximetry quiz, there was notable variability between their answers (κ = 0.45). The authors concluded that inadequate evidence defining clinically significant intermittent hypoxemia, whether in terms of desaturation duration, frequency or nadir, was the most likely underlying reason for this variation in clinical approach. This highlights the pressing need for the development of standardized guidelines specific to this population.

In Australia, the Australasian Sleep Association (ASA) pediatric guidelines are currently used to define abnormal newborn overnight oximetry, even though they are only validated for older pediatric patients and designed specifically for patients with suspected obstructive sleep apnea [10]. As a result, we have a limited understanding of what constitutes a normal overnight oximetry test in the healthy term newborn [11, 12]. Based on a systematic search of 31 studies, the American Thoracic Society Guideline states mean (±SD) normative oxygen saturation values for children < 1 year of age while asleep of 96.3% ± 1.3, with a desaturation nadir of 86% ± 1.5 [13]. These values are comparable to our findings, where the mean high was 98% ± 1.74, while the mean low was 91% ±2.29 during initial overnight oximetry. Newborns in our sample spent a median of 1.7 (0–99.8) % of time with < 90% SpO₂ during the initial overnight oximetry. While this median is consistent with the 2.7 (0.12–7.94) % reported by Flint and colleagues in healthy term newborns who underwent overnight oximetry earlier in the first postnatal week, our cohort exhibited a considerably wider range [11]. To better define normal ranges and clinically relevant abnormalities, future research should establish normative reference values for overnight oximetry and polysomnography in late preterm and term populations.

A critical component of the current clinical guidelines is the number of desaturations below a pre‐specified percentage of baseline SpO₂, the incidence of falls in SpO₂ below 90%, 85%, and 80%, and the number of clusters of desaturations (≥ 5 desaturations within a 10–30 min period of sleep) [10]. Despite cluster analysis defining desaturation events as a ≥ 4% fall in peripheral cutaneous oxygen saturation from baseline, the ASA currently recommends use of a 3% desaturation index for overnight oximetry [10], an approach also adopted by the American Academy of Sleep Medicine in 2012 [14]. Little is known about the frequency of these episodes in young infants. Indeed, the first report of both 4% and 3% desaturation indices in healthy term newborns undergoing overnight pulse oximetry at home at 1 and 3–4 months of age was only published in 2018, with a sample size of 45 babies studied at 1 month, 38 babies at 3–4 months, and paired studies of both for 32 [15]. For our cohort, a ≥ 4% fall in peripheral cutaneous oxygen saturation from baseline was used. For the initial overnight oximetry, a median of 142 (5–496) events were recorded with a desaturation event index of 16.9 (0.6–22.2), a finding consistent with reported values in healthy term newborns not being investigated for ongoing need for oxygen [11, 15].

Repeat overnight oximetry was performed in just over half of our cohort of late preterm and term newborns. In line with Evans and colleagues, in healthy term newborns not requiring LTOT [15], significant improvements were seen in the total number of desaturations, % of time with SpO₂ < 90% and < 88%, and the desaturation event index. Age‐related improvements in overnight oximetry may occur in healthy infants due to maturation of respiratory control centers and changes in sleep patterns [15], with additional improvements in infants with airway malacia due to progressive airway maturation during the first 12–24 months of age [16]. Therefore, in a large majority of late preterm and term newborns discharged on LTOT who follow this clinical pathway, overnight oximetry alone may be sufficient to determine weaning from LTOT, without the need for polysomnography. However, it is important to note that factors beyond the natural improvement of airway malacia may be responsible for improved oximetry parameters. Infants who underwent repeat overnight oximetry had similar initial overnight oximetry parameters, but significantly longer LTOT duration than those who did not. However, baseline oxygen requirement was not available and may have also influenced the decision to perform repeat testing. Repeat overnight oximetry was not associated with greater initial oximetry severity in this cohort. Further work is needed to establish which newborns may benefit from repeat testing and the optimal timing of assessment.

The American Thoracic Society continues to recommend that overnight oximetry be performed before weaning from LTOT [17]. The median duration of LTOT use in our cohort was approximately 229 days, almost double that reported in infants < 32 weeks’ gestation by Teoh and colleagues [18]. Oxygen was ceased at a median age of approximately 8 months, with only one newborn still requiring ongoing LTOT at the time of the current retrospective analysis. In an Australian sample of 93 preterm infants, polysomnography was conducted in 97% of infants before ceasing oxygen [19]. Conversely, less than half of our cohort underwent polysomnography, with those who did more likely to receive LTOT for longer. Given the marked improvements observed between the first and second overnight oximetry assessments, an argument could be made that routine polysomnography may not be necessary in all newborn cases. However, polysomnography was critical in the diagnosis of obstructive, central, or mixed apnea observed in almost half of our cohort. Polysomnography may clarify the underlying etiology of persistent oxygen requirement, facilitating more informed counseling of parents and/or caregivers, and assisting in clinical decision‐making regarding the need for further intervention. Despite this, we do not know how many newborns in the sample required intervention as a result (ENT surgery or continuous positive airway pressure). As a result, an important role remains for polysomnography, but its use could be reserved for the subset of infants who fail to demonstrate improvements on follow‐up oximetry or who have persistent clinical concerns. Considering the limited accessibility and costs associated with polysomnography, such an approach may better inform treatment pathways for this group.

Despite being a single‐center retrospective study, the current data do have several strengths. The 9‐year epoch studied means the sample size is large in comparison to other recent reports focusing on the use of overnight oximetry in the newborn population. The data were also contemporaneous with approaches to management of this patient group, consistent across the period studied. Finally, the Women's and Children's Hospital was South Australia's only tertiary women's and children's hospital with established integration of neonatal and pediatric respiratory care across the 9‐year study epoch. However, the study does have some limitations. The sample only represents infants who were discharged home from the neonatal unit on LTOT. As a result, a small portion of infants who were transferred from the neonatal unit to pediatric wards or down‐transferred to regional facilities closer to home for prolonged hospitalization while still receiving oxygen therapy were not included in the analysis. Some infants may also have had follow‐up overnight oximetry and/or polysomnography with other care providers out of the public health system. Finally, in the absence of a control group or information regarding disease severity, these findings describe patterns in overnight oximetry and polysomnography results but cannot establish their impact on clinical decision‐making or outcomes.

The current study describes our model of care after discharge in late preterm and term infants requiring LTOT. Here, we have shown that most late preterm and term newborns with ongoing oxygen requirements improve over time, as indicated by improved overnight oximetry parameters. There is, however, substantial variability in assessment pathways, driven by the complexity of additional co‐diagnoses and underlying etiology. This underscores the need for standardized reporting guidelines for overnight oximetry in late preterm to term newborns to improve clinical judgment and optimize LTOT management in this population.

Author Contributions

Amelia N. Noone: conceptualization, methodology, investigation, formal analysis, project administration, writing – original draft, writing – review and editing, visualization. Tom Goddard: conceptualization, supervision, writing – review and editing, resources. Tara M. Crawford: supervision, conceptualization, writing – review and editing. Chad C. Andersen: conceptualization, methodology, supervision, writing – review and editing. Michael J. Stark: conceptualization, methodology, supervision, writing – review and editing.

Funding

The authors have nothing to report.

Ethics Statement

This study was approved by the SA Women's and Children's Health Network Human Research Ethics Committee (1359 A). The requirement for informed consent was waived because this was a retrospective observational study using de‐identified data.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.