Feasibility and clinical utility of daytime polysomnography performed in NICU to diagnose sleep disordered breathing in infants

Abstract

Objective

To assess the feasibility and clinical utility of daytime polysomnography (PSG) in infants <3 months of age.

Methods

A prospective observational study of a convenience cohort analysing PSGs that were conducted for clinical purposes in infants less <3 months of age, between 1 May 2021 and 31 May 2024. A comparison was made between results for daytime PSG in the neonatal intensive care unit (NICU) and overnight PSG in the sleep laboratory. The type of PSG performed (daytime vs overnight) was based on the workflow of the sleep laboratory. Primary outcomes were successfully completed PSGs (feasibility) and per cent sleep efficiency (clinical utility). Secondary outcomes compared other sleep parameters between groups. Patient and public feedback directly informed the development of the research question and outcome measures.

Results

Of 60 PSGs, 28 were daytime and 32 were overnight. Daytime studies had a younger age (median 18 vs 55 days, p<0.001) and shorter median recording time (8.2 vs 10.4 hours, p<0.001). All daytime PSGs were successful, indicating feasibility. After adjusting for age at PSG and total recording time, per cent sleep efficiency was equivalent in the two groups (95% CI −12.4 to 5.7; p 0.456), indicating their clinical utility. For secondary outcomes, daytime PSGs had a higher % rapid eye movement (REM) sleep by 9.9% points (95% CI 1.1 to 18.8; p 0.028) compared with overnight PSG. Parameters that were not different included: frequency of spontaneous arousals, REM latency, sleep latency, Apnoea-Hypopnoea Index and Obstructive Apnoea-Hypopnoea Index. A decline in requests for overnight PSGs and a corresponding increase in daytime PSGs over the course of the study were observed.

Conclusion

Daytime PSGs performed in NICU were feasible and provided clinically useful results in infants <3 months of age. Availability of daytime PSGs performed at the infant’s bedside expands resource capacity and has the potential for cost savings.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Polysomnography (PSG) is vital for early diagnosis and treatment of sleep disordered breathing (SDB) in infants; but it is often difficult to obtain.

WHAT THIS STUDY ADDS

• Daytime PSGs conducted in neonatal intensive care unit are feasible and provide clinically useful results in infants <3 months of age, therefore expanding the resource capacity.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE, OR POLICY

• A daytime PSG model may facilitate early diagnosis and management of infants with SDB, expand healthcare capacity in a cost-effective way, but requires further study.

Introduction

Sleep disordered breathing (SDB) is widely prevalent in infants.^{1 2} Infants with conditions such as Robin sequence or other craniofacial syndromes, severe chronic lung disease, airway anomalies, central hypoventilation syndrome or other syndromic conditions causing SDB would need an early diagnosis of their SDB to facilitate management plans. A delayed diagnosis of obstructive sleep apnoea (OSA) or central sleep apnoea in such infants could cause prolonged hospital stay and increase morbidities. In contrast, an early diagnosis and treatment of SDB has the potential to improve feeding, growth and neurodevelopment in affected infants.³,⁶

Attended polysomnography (PSG) is the gold standard test to diagnose SDB.⁷ However, obtaining timely PSGs is limited by resources, long waiting lists for the sleep laboratory and the infants’ clinical instability. The recommended standard for PSGs is for them to be performed overnight in a sleep laboratory. For infants in the neonatal intensive care unit (NICU) in our hospital (Perth Children’s Hospital), a PSG requires a senior nurse to escort the infant from NICU to the sleep laboratory and remain there for the duration of the study. These factors are barriers for physicians to request a PSG and for the PSG to be performed early in the infant’s hospitalisation.

The rationale for PSGs being performed overnight is the nocturnal circadian rhythm of sleep.⁸ However, young infants sleep 15–20 hours per day with an ultradian rhythm, and approximately half of this sleep occurs in the daytime.⁸ Additionally, nearly 60% of their sleep consists of rapid eye movement (REM) sleep, and the majority enter REM sleep soon after sleep onset.^{9 10} The circadian rhythm of sleep begins to emerge around 10–12 weeks and infant sleep becomes increasingly nocturnal thereafter.^{8 11 12} Therefore, infants <3 months of age have significantly different sleep cycles and architecture than older children.¹²,¹⁴

In this study, we sought to assess the feasibility and clinical utility of daytime PSGs conducted in NICU in infants <3 months age.

Methods

A pilot project conducting daytime PSGs commenced in our centre in May 2021 and expanded after obtaining additional funding in 2023 (figure 1). In this observational study, we prospectively tracked the results of all PSGs (daytime or overnight) among a convenience sample of neonates and infants <3 months of age for whom PSGs were ordered for clinical purposes (inpatient or outpatient setting). The type of PSG (daytime vs overnight) was organised by the sleep laboratory based on available places and workflow of the sleep unit.

Figure 1. The number of daytime and overnight PSGs performed yearly from May 2021 to May 2024. PSG, polysomnography. #Note: PSG numbers in May 2021-2022 were affected by COVID-19 isolation measure.

The feasibility of daytime PSGs was assessed based on the number of successfully conducted studies without the need for repetition (primary outcome 1). The clinical utility of PSGs relies on the quality of sleep captured and other sleep parameters. To enable interpretation of a PSG, the reporting clinician relies on the amount of sleep captured, amount of REM sleep observed, with results enhanced when more consolidated periods of sleep are observed with fewer sleep interruptions due to environmental factors.^{15 16} Since environmental factors such as noise and light levels may differ between day and overnight studies, we chose per cent sleep efficiency as the parameter to assess clinical utility (primary outcome 2). The secondary outcomes were % REM sleep, REM sleep latency, sleep latency, spontaneous arousal index, Obstructive Apnoea Hypopnoea Index (OAHI) and Apnoea Hypopnoea Index (AHI), all defined below.

Data collection

We prospectively collected data on all infants <3 months of age (inpatients and outpatients) who underwent a PSG between 1 May 2021 and 31 May 2024. No exclusion criteria were applied. Data from one PSG were excluded from the final analysis of the overnight PSG group as the study was suboptimal with no PSG data available for analysis. Data were obtained from electronic medical records, PSG request forms, sleep scientist reports and PSG reports. Baseline infant characteristics and results of PSGs were also collected.

Overnight PSG

Overnight PSGs are attended studies conducted in the sleep laboratory at our hospital using the fixed Compumedics Grael PSG (Melbourne, Australia) system with Profusion 4.5 software. Studies were routinely conducted between 18:00 and 06:00 hours. Inpatient infants were mobilised from the NICU to the sleep laboratory and remained accompanied by a senior neonatal nurse throughout. For PSGs done on outpatient infants, one parent was required to accompany the infant for the night.

Daytime PSG

High-risk infants are often admitted to the NICU for monitoring and treatment which provides a safe and conducive environment for conducting a PSG. Daytime PSGs were conducted using Compumedics Grael PSG (Melbourne, Australia) mounted on a mobile trolley in the NICU at the infant bedside (figure 2) with Profusion 4.5 software. Daytime studies were conducted between 07:30 and 16:00 hours to match staff shifts. NICU rooms at our hospital are well-equipped to perform sleep studies in a quiet and dimly lit environment. While daytime PSGs were typically carried out in single bedrooms, two-bedded rooms were used occasionally. Both parents were encouraged to be involved in the infant’s routine care. Like overnight PSGs, daytime PSGs were set up by and attended in person by a sleep scientist or remotely from the sleep laboratory. The routine care and monitoring used in the NICU were continued with preference for conversations to occur outside of the room during the study.

Figure 2. The PSG equipment and setup for the daytime study in NICU. NICU, neonatal intensive care unit; PSG, polysomnography.

PSG channels and reporting

Both daytime and overnight PSGs used standard channels for type 1 PSG studies (online supplemental figure 1) as recommended by the American Association of Sleep Medicine (AASM).¹⁷,¹⁹ All sleep studies were scored by a sleep scientist using the AASM scoring criteria for infants.¹⁹ Standard PSG parameters were analysed and reported (online supplemental table 1).^{18 20} Central SDB was reported and diagnosed considering normative range for AHI based on age⁹ and clinical interpretation of PSG by the reporting clinician. The obstructive SDB was reported based on OAHI cut-off of ≤1/hour to be normal, 1<OAHI≤5 to be mild OSA, 5<OAHI≤10 to be moderate OSA and an OAHI >10/hour as severe OSA.¹⁹

Sample size calculation

This was a pilot (feasibility) observational study involving a prospective convenience sample of neonates and infants under 3 months of age who underwent clinically indicated PSGs (daytime or overnight). The primary aim was to assess the feasibility of identifying eligible participants, conducting PSGs in this age group and collecting interpretable data. As such, a formal sample size or power calculation was not required, in line with the objectives of feasibility studies designed to inform future, adequately powered research.

Statistical analysis

Statistical analysis was conducted using Stata V.18 (Statacorp). As the data were not normally distributed, medians and IQRs were reported for continuous data. For unadjusted comparison between groups, Wilcoxon rank-sum (Mann-Whitney) test was used. For categorical data, Fisher’s exact test was used for comparing the two groups. Multiple linear regression analysis was used for model building,²¹ to identify and evaluate confounders and obtain adjusted regression coefficients with 95% CI and p values. Statistical significance was accepted at p<0.05.

Patient and public involvement

The research question and selection of outcome measures were directly informed by the experiences and priorities shared by the families through prior qualitative interviews and routine feedback while developing a standardised pathway for the management of infants with Robin sequence.²² We partnered with the Pierre Robin Sequence (PRS) Australia consumer group to ensure meaningful consumer involvement throughout the project. The outcomes of this work will be shared via the PRS Australia consumer website to facilitate broad community access and engagement.

Results

A total of 60 PSGs were performed in infants <3 months of age during the study period, with 28 daytime PSGs (all inpatient) and 32 overnight PSGs (13 inpatients and 19 outpatients) (table 1). All PSGs were performed at term corrected age or older. The most common indication for requesting a PSG was suspected upper airway obstruction (75.0% vs 65.6%) with OSA being the most common diagnosis on PSG (78.6% vs 68.8%) in both groups.

Table 1. The characteristic of infants <3 months of age who had PSG between 1 May 2021 and 31 May 2024.

PSG parameters	Daytime PSG, number (%)	Overnight PSG, number (%)	P value
Total number	28	32	NA
Female	19 (68)	14 (44)	0.074
Median gestational age, weeks (IQR)*	39.2 (36.6–40.2)	38.5 (36.4–39.5)	0.545
Median birth weight, g (IQR)*	3330 (2780–3745)	3300 (2836–3725)	0.935
Median postnatal age of PSG, days (IQR)*	18 (13–34)	55 (41–73)	<0.001
Primary indication of PSG			0.714
Suspected upper airway obstruction	21 (75.0)	21 (65.6)
Unclassified apnoea or oxygen desaturation	3 (10.7)	4 (12.5)
Suspected hypoventilation	3 (10.7)	3 (9.4)
Suspected central apnoea	1 (3.6)	4 (12.5)
Diagnosis on PSG			0.251
Considered normal for age	1 (3.6)	4 (12.5)
Central SDB	2 (7.1)	0 (0.0)
Obstructive SDB	22 (78.6)	22 (68.8)
Mixed SDB	3 (10.7)	6 (18.8)
Severity of SDB on PSG			0.340
Mild	11 (39.3)	10 (31.2)
Moderate	7 (25)	4 (12.5)
Severe	9 (32.1)	14 (43.7)
Infants requiring respiratory support during PSG	6 (21.4)	1 (3.1)	0.043
Nasopharyngeal airway	4	1
Humidified high flow	1	0
Nasopharyngeal airway and high flow	1	0
Sleep position during PSG			0.002
Supine only	18	28
Prone only	8	0
Half supine and half prone	2	4
Type of referral	28 (100)		NA
Inpatient		13 (40.6)
Inpatient review, outpatient urgent study		8 (25.0)
Outpatient study		11 (34.4)

^*Median and range reported due to skewedness of the data.

PSG, polysomnography; SDB, sleep disordered breathing.

Comparing day versus night studies, gestation at birth (median 39.2 vs 38.5 weeks, p 0.545), birth weight (3330 vs 3300 g, p 0.935) and number of female infants (19 vs 14, p 0.074) were not significantly different in the two groups. Daytime PSGs were performed at a significantly younger age (median 18 vs 55 days, p<0.001), and more infants had respiratory support (6 vs 1, p 0.043) or prone sleep positioning (10 vs 4, p 0.002) at the time of PSG as compared with the overnight PSG group (table 1). Median total recording time was 8.2 hours in daytime PSG group compared with 10.4 hours in overnight PSG group (p<0.001), likely contributing to the lower total sleep time (4.8 vs 7.2 hours, p<0.001) (table 2).

Table 2. Additional sleep and respiratory parameters on PSG assessing the clinical utility of daytime PSG.

PSG parameters	Daytime PSG, median (IQR)	Overnight PSG, median (IQR)	Unadjusted P value
Lights out (time PSG started) (hrs:min)	08:22 (NA)	18:53 (NA)	NA
Lights on (time PSG stopped) (hrs:min)	16:29 (NA)	05:30 (NA)	NA
Total recording time (min)	490 (452–512)	623 (585–647)	<0.001
Total sleep time (min)	290 (249–330)	433 (389–494)	<0.001
Respiratory arousal per hour of sleep time	5.1 (3.1–7.6)	4.4 (2.8–7.5)	0.743
Number of obstructive apnoeas	7 (2–19)	4 (1–11)	0.424
Number of mixed apnoeas	3 (1–12)	4 (2–12)	0.395
Number of central apnoeas	33 (11–56)	56 (27–101)	0.057
Longest apnoea duration (s)	10 (8–14)	10 (9–12)	0.731
Apnoea-Hypopnoea Index (AHI) REM	26.8 (12.7–50.1)	26.6 (15.9–45.2)	0.869
Apnoea-Hypopnoea Index (AHI) NREM	10.5 (6.3–16.3)	9.0 (5.7–20.8)	0.574
Central Apnoea Index (CAI)	8.6 (2.1–12.8)	8.2 (3.6–15.1)	0.551
Obstructive Apnoea Index (OAI)	3.0 (0.9–7.3)	1.4 (0.5–4.2)	0.094
Mean peripheral oxygen saturation	96 (94–97)	96 (94–97)	0.371
Lowest peripheral oxygen saturation	84 (76–86)	82 (77–87)	0.962
Oxygen Desaturation Index 3% (ODI 3%)	64 (29–104)	95 (46–210)	0.161
Oxygen Desaturation Index 4% (ODI 4%)	47 (18–75)	67 (24–166)	0.171
Oxygen Desaturation Index 5% (ODI 5%)	32 (11–53)	43 (15–116)	0.190
Highest CO2 (mm Hg)	51 (49–56)	50 (48–57)	0.429

NREM, non-rapid eye movement; PSG, polysomnography; REM, rapid eye movement.

Feasibility of daytime PSGs

Of the 28 daytime PSGs, 100% were deemed to have acceptable PSG setup and signals and none needed repeating. Of the 32 overnight PSGs, one needed to be repeated due to an unreliable oximeter reading (PSG with reliable data included in the analysis).

Clinical utility of daytime PSGs

In an unadjusted analysis, the per cent sleep efficiency was lower for daytime PSGs (63.4% vs 69.9%, p 0.001) as compared with overnight PSGs. The REM sleep parameters, including per cent REM sleep (median 42.6% vs 36.5%, p 0.095) and REM sleep latency (median 4.8 vs 22.8 min, p 0.105), were not significantly different, despite the REM latency values appearing lower in the daytime PSGs. The number of spontaneous arousals per hour of sleep (median 16.8 vs 15.6, p 0.455) and sleep latency (median 4.5 vs 9.3, p 0.737) were also not different (table 3). Diagnostic markers of central SDB such as AHI (median 18.6 vs 17, p 0.738) and Central Apnoea Index (CAI) (8.6 vs 8.2, p 0.551) were similar (tables2 3). Diagnostic markers for obstructive SDB, such as OAHI (5.5 vs 6.3, p 0.956) and Obstructive Apnoea Index (3.0 vs 1.4, p 0.094), were also not different between the two groups (tables2 3).

Table 3. The primary and secondary outcomes assessing the clinical utility of daytime PSG.

PSG parameters	Daytime PSG, median (IQR)	Overnight PSG, median (IQR)	Unadjusted P value	Adjusted coefficient* (95% CI)	Adjusted P value*	Adjusted coefficient† (95% CI)	Adjusted P value†
Primary outcomes
Per cent sleep efficiency	63.4 (51.2–68.2)	69.9 (65.5–76.1)	0.001	−5.6 (−11.7 to 0.5)	0.072	−3.39 (−12.4 to 5.7)	0.456
Secondary outcomes
Per cent REM sleep	42.6 (34.4–46.9)	36.5 (30.7–41)	0.095	4.7 (−1.4 to 10.8)	0.129	9.9 (1.1 to 18.8)	0.028
Spontaneous arousal per hour of sleep time (number)	16.8 (15.4–20.7)	15.6 (12.6–24.2)	0.455	−3.8 (−7.6 to 0.12)	0.058	−4.0 (−9.8 to 1.7)	0.165
Sleep REM latency (min)	4.8 (0.3–31.8)	22.8 (7–32.7)	0.105	2.1 (−16.1 to 20.1)	0.822	12.1 (−14.5 to 38.8)	0.366
Sleep latency (min)	4.5 (2–15.3)	9.3 (0–24.5)	0.737	−11.2 (−23.0 to 0.6)	0.064	−9.6 (−27.2 to 8.0)	0.281
Apnoea-Hypopnoea Index (AHI)	18.6 (11.5–33.5)	17 (10.3–30.4)	0.738	−2.8 (−14.5 to 8.9)	0.635	−0.04 (−17.5 to 17.4)	0.996
Obstructive Apnoea-Hypopnoea Index (OAHI)	5.5 (3.6–11.2)	6.3 (1.5–14.3)	0.956	−2.9 (−10.1 to 4.3)	0.417	−0.50 (−11.1 to 10.2)	0.925

^*Adjusted for covariate, age at PSG.

^†Adjusted for covariates, age at PSG and total recording time.

PSG, polysomnography; REM, rapid eye movement.

Multiple linear regression models were built with type of PSG (daytime vs overnight) as the primary predictor and age at PSG and total recording time as covariates, based on the observed associations (table 3). Expectedly, the total sleep time was highly correlated with total recording time (correlation coefficient of 1.0); therefore, total recording time (modifiable factor) was used in the regression models.

After adjusting only for age at PSG, both the per cent sleep efficiency (95% CI −11.7 to 0.5; p 0.072) and per cent REM sleep were similar (95% CI −1.4 to 10.8; p 0.129) (table 3). Other sleep parameters such as REM latency, sleep latency, number of spontaneous arousals per hour of sleep time, OAHI and AHI remained similar on adjusted analysis (table 3).

After adjusting for age at PSG and total recording time, the per cent sleep efficiency was similar in the two groups (95% CI −12.4 to 5.7; p 0.456), but per cent REM sleep was 9.9% points higher in daytime PSGs (95% CI 1.1 to 18.8; p 0.028) as compared with overnight PSGs. The number of spontaneous arousals per hour of sleep (95% CI −9.8 to 1.7; p 0.165), REM latency (95% CI −14.5 to 38.8; p 0.366), sleep latency (95% CI −27.2 to 8.0; p 0.281), AHI (95% CI −17.5 to 17.4; p 0.996) and OAHI (95% CI −11.1 to 10.2; p 0.925) were not statistically different in the two groups (table 3).

Additional PSG parameters

Additional PSG parameters depicting gas exchange, such as highest carbon dioxide level (median 51 vs 50 mm Hg, p 0.429), mean peripheral oxygen saturation (median 96% vs 96%, p 0.371), minimum peripheral oxygen saturation (median 84% vs 82%, p 0.962) and oxygen desaturation index (3%, 4% or 5%), were similar in the two groups in the unadjusted analysis (table 3).

Impact on sleep laboratory’s capacity to conduct PSG

There was a trend, over the course of this study, for a decrease in overnight laboratory infant PSGs and an increase in daytime infant NICU PSGs. In 2022–2023, of 25 PSGs performed in infants <3 months of age, 88% were overnight PSGs (inpatient or outpatient). But in 2023–2024 after additional funding permitted expansion of the daytime PSG programme, of 24 PSGs performed in infants <3 months of age, 87.5% were daytime PSGs (inpatient) (figure 1).

Discussion

This prospective study reports that standard PSGs can be performed during the daytime and at the bedside (NICU) of infants <3 months of age and can provide reliable results. Despite being performed outside the sleep laboratory, daytime PSGs were universally successful (feasible) and captured similar per cent sleep efficiency (clinically useful) with 9.9% higher per cent of REM sleep. Daytime studies showed equivalent results for the number of spontaneous arousals, sleep latency and REM latency and the obstructive and central apnoea hypopnoea indices to the overnight PSG group. Notably, daytime PSGs were performed at a younger age, indicating reduced wait times.

Sleep plays a critical role in the healthy development of infants and the well-being of their caregivers. Disrupted sleep in infancy can impact neurodevelopment, emotional regulation and growth, while also contributing to caregiver stress, reduced quality of life and increased healthcare use.^{23 24} It is important to identify and assess SDB early in infants due to its impact on growth and the developing brain.²⁵,²⁷ While screening tools such as home ambulatory PSG, nap PSG, polygraphy, sleep oximetry or oxycapnography are used in older children,²⁸,³¹ they are not validated to diagnose SDB in infants <3 months of age.^{17 29 32}

Given the challenges in obtaining standard PSGs, there is a need to explore options to improve access to PSGs, especially for neonates and young infants. One approach could be to conduct overnight PSGs in NICU, but the limiting factor with such an approach is the availability of sleep scientists as they are needed for regular PSGs in the sleep laboratory. A second approach could be to conduct PSGs in NICU during the daytime (our approach).

There are few reports of inpatient PSG studies in the NICU. Meerkov et al reported on NICU PSGs conducted in 42 infants at risk of seizures (excluding infants at risk of SDB).³³ The PSG was described as standard attended PSG planned for 8–12 hours of recording time capturing mean total sleep time of 9.5 hours. In the cohort study by Shelhaas et al in infants with myelomeningocele, standard attended PSG was conducted for 12 hours in the NICU.³⁴ It was unclear whether in the above two studies PSGs were conducted during the day or night. Moreover, they had a different patient cohort than ours; hence, it is difficult to compare their results with ours.

Kim et al reported on 31 infants with suspected SDB who underwent standard PSG performed in NICU; again, it was unclear whether they were conducted during the day or night.⁵ They reported a mean total sleep time of 5.4 hours and sleep efficiency of 67% which is similar to the findings in our daytime PSG group. A significantly higher arousal index (per hour of sleep time) of 52.2 was observed in their study as compared with 16.8 in our daytime PSG group, which supports the testing reliability of daytime PSGs.

Our proposed daytime PSG model eliminates the need for additional NICU senior staff, brings the service to the bedside of the infant and expands the sleep laboratory’s capacity to conduct additional studies. Conducting PSGs during the day is not possible in older children due to an established circadian rhythm, which results in difficulty falling, staying asleep or capturing a reasonable amount of REM sleep during the daytime, which is crucial to assess SDB.^{15 16} Daytime nap PSGs previously attempted to assess SDB in older children were shorter in duration (1 hour of sleep) and rarely captured REM sleep, causing negative or inconclusive results.^{15 16} In contrast, infants <3 months spend a high proportion of their daytime hours in sleep, with the majority being REM sleep and physiologically a shorter REM latency.¹²,¹⁴ This provides a unique opportunity to perform daytime PSGs in infants. In our daytime cohort, we captured almost 5 hours of sleep, of which 43% was REM sleep (normal for age).

The use of daytime PSGs in infants is sparingly described, with only one study identified. Singh et al reported on 100 infants <6 months of age who underwent a daytime PSG in the NICU.³⁵ Like ours, their study confirmed the equivalence of sleep and respiratory parameters between daytime and nighttime PSGs. However, their service aimed for a minimum of 6 hours of recording time as compared with 8 hours of recording time in our model, resulting in a lower recorded sleep time than ours (approx. 4 hours vs 5 hours).³⁵ The average age at which the daytime PSG was conducted in their study was 8.3 weeks as compared to 2.5 weeks in our daytime PSG group.

Although not the primary objective of the study, we found that daytime PSGs were performed at a younger age compared with overnight PSGs. Also, the availability of daytime PSGs was found to be associated with a decline in the number of overnight PSGs for infants <3 months of age, potentially vacating additional PSG spots for older infants and children. To gain further understanding, we reviewed the age at which overnight PSGs were performed in our sleep laboratory prior to the availability of daytime PSGs. In the prior 6 years (January 2014 to December 2020), of the 92 PSGs performed in infants <3 months of age, the median age at PSG was 47 days (range 4–90), close to the overnight PSG group (May 2021 to May 2024), as compared with a median age of 18 days in our daytime PSG group.

An additional observed benefit of daytime NICU PSGs could be the perceived increase in patient safety due to the ongoing monitoring and support provided in the intensive care unit during the study, particularly for infants requiring respiratory support or prone sleep positioning (no need to move the infant), therefore explaining the presence of more infants on respiratory support during PSG in our daytime PSG group. The daytime PSG may also support a conducive environment for mother-baby bonding and parental reassurance. Given that there is advocacy for making PSGs more ‘child and family friendly’,^{36 37} future studies are required to confirm the impact on families of a daytime PSG service.

While we did not undertake formal economic analysis, the potential cost-savings with a daytime PSG model could be substantial in our hospital setting. No additional infrastructure was required after the once-off procurement of the mobile sleep study equipment. There was a 50% reduction in cost of staff when comparing 10 hours of sleep scientist support (including additional time for study setup) to 12 hours of senior nursing staff (including additional time for study setup and transport). The earlier access to PSG could also plausibly reduce the time spent in the NICU by days to weeks. While it is possible that the availability of daytime PSG could lead to an increase in the number of PSGs requested, we found that the availability of daytime PSGs led to a decline in outpatient PSG requests with no significant rise in the overall PSG numbers conducted in infants <3 months of age (figure 1). Future studies undertaking cost analysis are required.

For practical reasons, we planned for a recording time of 8.2 hours for daytime PSGs as compared with 10.4 hours for overnight PSGs, which explains the lesser total sleep time on daytime PSGs. Despite the lower total sleep time, the distribution of sleep stages was reassuring. The higher per cent of REM sleep observed in daytime PSG group is likely reflective of the shorter sleep REM latency and relatively higher proportion of REM sleep seen at younger age when these daytime PSGs were undertaken as compared with overnight PSG.^{8 9} The equivalence of our sleep and respiratory outcomes provides reassurance that the slightly shorter recording time in our daytime PSG model is acceptable.

Strengths

This is the first prospective study to objectively test and compare the feasibility and clinical utility of conducting daytime PSGs in the NICU to overnight PSGs. Our model of providing daytime PSGs brings the service to the bedside of the infant. Providing comparative data against our current standard of care model enabled detailed analysis of the various testing outcomes and comparison of PSG reports conducted by the same department, mitigating reporting and technological bias.

Limitations

Our study has a relatively small number of infants (n=60) which is reflective of the challenges of obtaining sleep testing for a younger age group. The two groups differ in the age at which the PSG was conducted. This could be partly due to the challenges of obtaining an inpatient overnight PSG due to sleep laboratory resource burden, which sometimes meant the infant was discharged before the PSG was scheduled. This could also explain the increased proportion of outpatient studies in the overnight PSG group. There is some concern that performing PSGs at an earlier age could result in inadvertently identifying a higher number of central apnoeas which may be normal for age.⁹ Despite the infant population being younger in the daytime PSG group, the number of central apnoeas and CAI were comparable and within the described normative data for age,⁹ reassuring that no overestimation of central SDB occurred. However, it is possible that confounders such as the sleep position or respiratory support at the time of the PSG could have affected the estimation of obstructive SDB. Moreover, due to resource issues, we could not follow a before and after design where the same infant undergoes a daytime and overnight PSG.

Conclusion

Daytime PSGs performed in NICU were feasible and provided clinically useful results in infants <3 months of age. The daytime NICU PSG model may facilitate earlier access to PSG, has the potential to improve patient safety, possibly expand the service delivery capacity of the sleep laboratory, and may even have cost-saving implications for the healthcare system. Further studies are required to evaluate the impact of daytime NICU PSGs on the outcomes of infants with SDB, including length of hospital stay, growth and neurodevelopment, and cost savings.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.