Retrospective Analysis of Inpatient Polysomnogram Characteristics and Discharge Outcomes in Infants with Bronchopulmonary Dysplasia Requiring Home Oxygen Therapy

Abstract

Rationale

Little is known about the polysomnogram characteristics in infants with BPD, especially severe BPD, who do not need home ventilatory support but are at increased risk for chronic hypoxia and are vulnerable to its effects.

Objective

To assess polysomnogram characteristics and change in discharge outcomes in premature infants with bronchopulmonary dysplasia (BPD) who required oxygen therapy at discharge.

Methods

This is a retrospective chart review of premature infants with BPD admitted to a quaternary newborn and infant intensive care unit from January 1, 2012 to December 31, 2015 and underwent polysomnography prior to discharge

Measurements and Main Results

Data from 127 patients was analyzed. The median gestational age of our patients was 26 weeks and 1 day (IQR 24.71, 28.86). The majority of the patients had moderate to severe BPD. The median obstructive apnea-hypopnea index (OAHI) was 5.3 events per hour (IQR 2.2, 10.1). The median oxygen desaturation index (ODI) was 15.7 events per hour (IQR 4.7, 35). Nadir SpO₂ was 81% (IQR 76–86), Ar/Aw index 21.9 (IQR13.3–30.9). No statistically significant difference was noted between severe and non-severe BPD groups for polysomnogram characteristics. However, average end tidal CO2 was significantly higher in the severe BPD group (p = 0.0438). Infants in the severe BPD group were intubated longer than infants with non-severe BPD (p = 0.0082). Corrected gestational age (CGA) at the time of discharge (CGA-PSG) and polysomnogram (CGA-DC) were higher in severe BPD patients but not statistically different. The majority of premature infants that received a PSG were discharged home with oxygen. 69% required a titration of their level of support based on results from the PSG.

Conclusion

Our results highlight the presence of abnormal polysomnogram characteristics in BPD patients, as early as 43 weeks corrected gestational age. These findings have not been previously described in this patient population prior to initial discharge home. A severe BPD phenotype tends to be associated with higher respiratory morbidity compared to a non-severe BPD phenotype for the comparable CGA. PSG, when available, may be helpful for individualizing and streamlining treatment in preparation for discharge home and mitigating the effects of intermittent hypoxic episodes.

Introduction

Infants admitted to the NICU, whether term or preterm, are at risk for sleep disordered breathing due to a variety of causes (1–3). In particular, infants with BPD are a particularly fragile population that are at risk for sleep disordered breathing, hypoventilation and hypoxemia (4). A study from the National Institute of Child Health and Human Development Neonatal Research Network showed that 68% of infants born at 22–28 weeks and with birthweight from 401–1500 g were diagnosed with BPD, using the severity-based based definition proposed by Jobe et al (5, 6). This classification scheme has been refined by Abman et al. from the Bronchopulmonary Dysplasia Collaborative to distinguish between infants with severe BPD as having “type 1” disease if they are receiving ≥30% O2 or nasal CPAP or high-flow nasal cannula at ≥36 wk post-menstrual age, and “type 2” disease if they are receiving mechanical ventilation at this age (7). Though more of these infants are surviving to childhood and beyond, they remain at risk for life-threatening hypoxia secondary to their impaired lung mechanics (1, 8) and immature respiratory control (4). In addition, chronic or intermittent episodes of hypoxemia have been shown to have adverse effects on cognitive, academic achievement, and behavioral outcomes (9). Providing an adequate regimen of respiratory support is essential to a safe discharge to the home environment, and for optimizing growth, as well as minimizing recurrent lung injury (7).

Clinically unsuspected desaturations can occur frequently in preterm infants with BPD (1). Lung function in these patients remains abnormal for years, and many continue to require monitoring and supplemental oxygen following discharge (10). Though infants may have appropriate oxygen saturations while awake, they can exhibit desaturation events during sleep and feeding, thus prompting clinicians to initiate supplemental oxygen. The Thoracic Society of Australia and New Zealand (TSANZ), the American Thoracic Society (ATS), and the British Thoracic Society (BTS) recommend the use of supplemental oxygen to maintain goal saturations and to facilitate discharge home (10–12). While the TSANZ and ATS acknowledge the potential utility of polysomnography (PSG) to wean infants off oxygen or to determine oxygen need in infants who are not doing well, they do not currently make recommendations for completion of a PSG for all infants with BPD (11, 12). The impact that a PSG, otherwise known as a sleep study, can have on the management of these infants has not been demonstrated to date. It provides data regarding central, mixed, obstructive apnea, hypoventilation information and allows for more precise oxygen titration and provides objective data that can be used to tailor management and optimize respiratory status for the individual patient (2).

However, little is known about the PSG characteristics of preterm infants with BPD and the impact the results have on therapy. There are no studies to date that have examined the results of PSG in these infants done prior to discharge or in infants at an early post-menstrual age. Data supporting the routine use of polysomnography to assess the need for supplemental oxygen support in these infants is also lacking (3). Further, there are very few pediatric sleep labs with adequate expertise in infants. The aim of this study is to define polysomnogram (PSG) characteristics in premature infants with bronchopulmonary dysplasia (BPD) and delineate its role in discharge planning.

Methods

Study Design and Patient Population

A retrospective chart review and descriptive study was conducted to identify premature infants born at less than 37 weeks gestational age with BPD who were admitted to the Neonatal Intensive Care Unit (NICU) of a quaternary urban children’s hospital from January 1, 2012 to December 31, 2015 and underwent polysomnography prior to discharge from the hospital. Not every premature infant with diagnosed BPD underwent PSG. PSG was done on premature infants with diagnosed BPD as per the clinical decision-making by the rounding pulmonologist to assist with oxygen titration or if the infant had ongoing intermittent desaturations with feeding, sleeping or while awake despite oxygen supplementation or while on room air. It was done as close to the discharge as possible, when determination of oxygen needs was the last thing the patient needed prior to discharge home. This hospital does not have an in-house delivery service; therefore, all infants are out-born and transferred to the NICU. Patients were identified by querying the Somnostar polysomnography database for infants who had a polysomnography done during this time period. Total of 272 infants were pulled from the database. Of these 133 were preterm infant who underwent sleep study prior to discharge. 4 infants were excluded from the study due to inadequate data being available when they were transferred from the outside hospital and 2 were late preterm infants with complex congenital cardiac abnormalities (hypoplastic left heart and Ebstein’s anomaly) who underwent sleep study for other indications.

Premature infants with BPD who did not have issues with desaturations did not have a PSG done. Patients with type 2 severe BPD with tracheostomy or tracheostomy and ventilator dependence did not undergo PSG and were not included in the study. All PSGs were done at the patient’s bedside. Diagnosis of BPD was made by the neonatologist, confirmed by the pediatric pulmonologist, and based on oxygen dependency at 36 weeks post-menstrual age (5) per the criteria for “new BPD” (Jobe et al. in the 2001 National Institutes of Health and Child Development (NICHD)/National Heart, Lung, and Blood Institute (NHLBI) Workshop and as per Abman et al (7)). During the analysis, a subset of infants was identified for whom documentation from the referring hospital was inadequate for definite classification into type 1 or type 2 severe BPD. These infants were deemed “unclassifiable” per our criteria due to inadequacy of documentation.

Data Collection

We reviewed the electronic medical records of all patients who met the criteria for our study. Demographic data including ethnicity and gender were collected. Gestational age and birthweight were recorded. Maternal history as recorded in the admission history and physical, surfactant and caffeine use, number of days intubated and number of days on non-invasive positive-pressure ventilation (defined as use of nasal cannula intermittent mandatory ventilation or nasal continuous positive airway pressure), use of respiratory medications or diuretics for BPD (defined as albuterol, levalbuterol, budesonide, furosemide, chlorothiazide, and spironolactone) and surgeries were abstracted. Number of days intubated was calculated only for the admission at our institution, unless there was clear documentation from the transfer paperwork of the dates of invasive and non-invasive ventilation. The presence of additional diagnoses including pulmonary hypertension and gastroesophageal reflux, based on inclusion in the diagnosis list of the patient’s medical record, were recorded. Any other significant co-morbidities noted in the diagnosis list in the medical record and any surgeries performed during the admission were also obtained. Infants who did not have continuing desaturations, born at greater than 37 weeks gestation, did not complete a full PSG for any reason, or completed a PSG as an outpatient, and infants who had craniofacial anomalies were excluded from analysis.

Polysomnography

We used following equipment in sleep lab for performing PSG. Braebon thermal airflow sensor (pediatric and adult), airflow and end tidal CO2 canula: PTAF, Pro-tech 2 from Respironics, BCI/Smith Medical Capnocheck, capnograph 9004, system was used for monitoring CO2 and as Pulse oximeter with averaging set to 2 beats/ 8 seconds. Respiratory inductive plethysmography belts used from years 2012–2013 were from AMBU Sleepmate, piezoelectric belts, and from years 2013–2014 we used QDC-Pro from Vyaire Medical/CareFusion.

PSG was performed when patients were on oxygen flow of 1 liter per minute or less or who had intermittent desaturations who were otherwise medically stable and ready for discharge home. Additionally, infants were off all sedation and narcotic medications for at least 24 hours, unless the infant was to be discharged home with these medications. PSGs were read by a board-certified pediatric sleep medicine attending and scored according to the American Academy of Sleep Medicine Manual for Scoring Sleep 2007 guidelines. As new amendments and versions/updates were available, they were immediately applied to keep scoring guidelines current and up to date, including 2.0 version, then 2.3 versions and updates which happened during study period. The following polysomnogram parameters were evaluated for the study: heart rate, respiratory rate, obstructive apnea hypopnea index (OAHI; number of obstructive apneas and hypopneas per hour of sleep), central apnea index (CI), oxygen desaturation index (ODI; number of oxygen desaturations of 3% or more per hour of sleep), and baseline oxygen saturation as measured by pulse oximetry (SpO₂) and oxygen saturation nadir.. Studies were commenced on room air with the addition of oxygen by nasal cannula for desaturation events or baseline SpO₂ of less than 89%, beginning at 0.25 liters per minute and then oxygen was gradually increased to eliminate hypoxemia with a goal SpO₂ ≥ 94%. Patients who were unable to tolerate room air at the start of the study underwent oxygen titration only.

Statistical Methods

Descriptive statistics are provided to describe the distribution of this study sample characteristics. Continuous variables that are normally distributed are described in mean and standard deviation (SD), whereas those that are not normally distributed are described in median and interquartile range (IQR). Categorical variables are summarized in frequency and percentage. Two-sample t-test and Wilcoxon rank-sum test were used to examine the difference in continuous parameters between non-severe and severe BPD groups. Chi-Square test and Fisher’s Exact test were used for comparing categorical variables between the non-severe and severe groups. Then, generalized linear model based on gamma distribution was used to assess the relationship in days of intubation on OAI, OAHI and MAI univariately. This is to examine the effect of days of intubation on any possible anatomical changes resulting in increase in any obstructive PSG parameters. The association in gestational age and corrected gestation age at the time of sleep study with central apnea index was assessed with a similar statistical model to evaluate role of progressive maturity of central control of respiration in relation to central apneas. Similar model was used to assess the association of reflux on OAI, OAHI, and MAI.

Univariate linear regression model is used to assess the association of hemoglobin (Hb) and pulmonary hypertension (PHTN) on nadir SpO₂. The results for OAI, MAI, OAHI and central apnea index are described as percent change; beta estimate (β) is used to describe the results for SpO₂;. 95% confidence interval and p-value are included in these estimates. Significance level is set at 5% with two-sided test throughout the analyses. All statistical computations were done in Stata/SE 15.1 (StataCorp, College Station, TX).

Results

A total of 127 infants met the criteria for this study. Fifty-seven percent of the patients were male, about forty five percent were Hispanic, and about twenty five percent were Caucasian/white (Table 1). Median gestational age of the patient population was 26 weeks and 1 day (IQR 24.71, 2.86) with a median birthweight of 0.76 kg (IQR 0.64, 1.05). Eighty-three percent of the patients received at least one dose of surfactant. Eighty-one percent of the patients received caffeine during their hospital admission, but this was discontinued by the time of the PSG. Patients were intubated for a median of 41 days (IQR 14, 67) and received non-invasive positive pressure ventilatory support for a median of 25 days (IQR 9, 38). Ninety-six percent of the patients were started on respiratory medications such as albuterol, budesonide, ipratropium and diuretics during their hospital stay and were subsequently discharged home on these medications. Twenty-eight percent of the patients were discharged home with aerosolized medications only. With regards to associated co-morbidities, eighteen percent of infants had pulmonary hypertension diagnosed by echocardiogram at some point during their admission and forty-nine percent had clinical symptoms of gastroesophageal reflux. Almost all the studies performed were titration studies with about 81% of the infants spending median of 27.5 minutes of time on room air (IQR 7.25–63.5 minutes). This was either spend in the beginning of the end of the sleep study as ordered by the physician.

Table 1:

Study Population (n=127)

Gender	% (n)
Male	57% (72)
Female	43% (55)
Ethnicity
Hispanic	44.9% (57)
Non-Hispanic	44.9% (57)
Unknown	10.2% (13)
Race
Asian/Pacific Islander	4.7% (6)
African American/Black	17.3% (22)
White/Caucasian	25.2% (32)
Unknown/other/2 or more race	52.7% (67)
BPD classification
Mild	3% (4)
Moderate	14% (18)
Severe, type 1	37% (47)
Severe, type 2	41% (52)
Unclassifiable severe BPD	5% (6)

The relative proportions of patients with mild, moderate, severe type 1 and severe type 2 BPD are presented in Table 1. Most patients in the study had moderate (14%) to severe (37% with type 1 and 41% with type 2) BPD. Of note, there were six patients with severe BPD who were deemed “unclassifiable” for the reasons detailed above.

Table 2 shows the distribution of sleep study parameters between non-severe and severe BPD groups. 93% of PSG in this study were oxygen titration studies. Median sleep efficiency was 61%. The median OAHI was 5.3 events per hour (IQR 2.2, 10; normal < 1.5 events per hour). The median CI and ODI were 3.4 events per hour (IQR 1.4, 7.2; normal < 5 events per hour) and 15.7 events per hour (IQR 4.5, 35; normal < 1.5 events per hour), respectively. There was no significant difference in these parameters between the non-severe and severe BPD groups. As shown in Table 2, there is a differential distribution in aCO2 (p=0.0438) and days of intubations (p=0.0082) between non-severe and severe BPD groups. REM sleep was detected in 65% of infants during their PSG. The PSG was performed at a median age of 43 weeks and 1 day corrected gestational age and the patients were discharged at a median age of 45 weeks 2 days corrected gestational age. Eight percent of patients had concomitant hypoventilation (defined as more than 25% of total sleep time with end-tidal CO2 more than 50 mm Hg). 72% of mothers had a pregnancy complicated by diagnoses such as preeclampsia, PPROM, diabetes, oligohydramnios, infection, and substance use, in addition to preterm labor.

Table 2.

Polysomnogram and clinical characteristics comparison: severe and non-severe BPD groups

BPD severity
	Total (N=127)	Non-severe (n=22)	Severe (n=105)	P-value
OAHIa	5.3 (2.2, 10)	6.3 (3.6, 11.6)	4.7 (2.1, 9.8)	0.23101
CIa	3.4 (1.4, 7.2)	3.4 (1.4, 7.2)	3.4 (1.3, 7.1)	0.86841
ODIa	15.7 (4.7, 35)	26.5 (9.8, 35)	13.55 (4.1, 33.8)	0.32581
Nadir SpO2b	80 ± 8.46	79.5 ± 9.09	80.10 ± 8.46	0.76392
Hypoventilationc
No	119 (93.70)	21 (95.45)	98 (93.33)	1.0003
Yes	8 (6.30)	1 (4.55)	7 (6.67)
TIB (hour)a	5.18 (3.82, 5.76)	5.49 (4.64, 5.74)	5.16 (3.82, 5.76)	0.52201
TST (hour)a	3.16 (2.07, 3.58)	3.38 (2.2, 3.5)	3.11 (2.04, 3.69)	0.44651
Ar/Awa	22.25 ± 13.88	20.56 ± 14.94	22.60 ± 13.70	0.53222
aCO2b	36.56 ± 6.75	33.86 ± 5.92	37.10 ± 6.80	0.04382
pCO2b	47.94 ± 7.35	45.05 ± 8.11	48.52 ± 7.35	0.05442
MAIa	1.15 (0.47, 2)	1.51 (0.59, 2)	1.06 (0.43, 2.03)	0.58691
OHIa	3.25 (1.41, 6.34)	5.31 (2.02, 7.94)	3.14 (1.35, 5.77)	0.28871
OAIa	1.18 (0.47, 3.34)	1.34 (0.36, 4.04)	1.01 (0.50, 3.34)	0.93341
GA (week)b	27.04 ± 3.25	27.79 ± 3.64	26.88 ± 3.16	0.23692
BW (kg)ad	0.76 (0.64, 1.05)	0.81 (0.67, 1.38)	0.76 (0.63, 1.03)	0.15851
RRb	56.98 ± 12.92	56.45 ± 15.71	57.10 ± 12.35	0.83352
HRa	148 (137, 156)	148 (137, 153)	148 (138, 157)	0.24601
D-Intuba	41 (14, 67)	22 (6, 46)	43 (18, 71)	0.00821
PHTNC
No	104 (81.89)	18 (81.82)	86 (81.90)	1.0003
Yes	23 (18.11)	4 (18.18)	19 (18.10)
Refluxc
No	65 (51.18)	10 (45.45)	55 (52.38)	0.5554
Yes	62 (48.82)	12 (54.55)	50 (47.62)
Hb (g/dL)bc	10.83 ± 1.33	10.40 ± 1.41	10.91 ± 1.30	0.10212
CGA-PSGb	43.60 ± 5.33	41.84 ± 4.84	43.97 ± 5.38	0.08852
CGA-DCa	45.29 (41.43, 50)	43.64 (40.29, 47.14)	45.86 (42.43, 50.57)	0.06191

^aMedian (Interquartile range);

^bMean ± SD;

^cFrequency (Percentage)

^cHemoglobin data not available for three patients (CBC not done during admission);

^dBW not documented for two patients

¹Mann-Whitney Ranksum Test;

²Two-sample T-test;

³Fisher’s Exact Test;

⁴Chi-Square Test

We used a univariate generalized linear model based on gamma distribution that evaluates the relationship in days of intubations with PSG parameters (OAI, MAI, OAHI) (Table 3). Decrease in all of these parameters for every unit increase in days of intubation was noted, but was not statistically significant. Neither of these PSG parameters are significantly associated with intubation days (all p>0.05). Same model was used to evaluate OAHI, MAI and OAI association with reflux. None were found significantly associated with reflux (Table 3). We also used a univariate linear regression model to evaluate the relationship in Hb and PHTN with nadir SpO₂. Hb is marginally significantly associated with nadir SpO₂ (β=1.12, 95% CI=0.001, 2.24, p=0.050). For patients with and without PHTN, there was no significant difference in nadir SpO₂ (β=0.85, 95% CI=−3.06, 4.76, p=0.668).

Table 3.

Gestational age and corrected gestation age in relation to CIa
	% change	95% CI	P-value
Gestational age (GA)	−3.17	(−9.88, 4.05)	0.380
Corrected gestational age (CGA)	−3.60	(−7.86, 0.85)	0.112
The effect of days of intubation on OAI, OAHI, and MAI univariatelya
	% Change	95% CI	P
OAI	−0.22	(−1.24, 0.81)	0.680
MAI	−0.28	(−0.99, 0.43)	0.440
AHI	−0.26	(−0.69, 0.17)	0.234
The effect of reflux on OAI, OAHI, and MAI univariatelya
	% Change	95% CI	P
OAI	−31.49	(−64.54, 32.33)	0.260
MAI	−27.50	(−54.94, 16.64)	0.185
AHI	−23.96	(−43.62, 2.54)	0.073

^aUnivariate generalized linear model based on gamma distribution

Univariate generalized linear model based on gamma distribution was also used to assesses the association between gestational age and corrected gestational age at the time of sleep study with central apnea index (Table 3). Though there was a 3.17% and 3.60% decrease in central apnea index for every week increase in gestational age and corrected gestational age respectively, neither of them is significantly associated (both p>0.05).

Ninety-eight percent of the patients were discharged home with supplemental oxygen (Table 4). Sixty-nine percent of these patients required a change in their level of supplemental oxygen support prior to discharge; of those infants requiring a change in their oxygen support, 55% required an increase in oxygen support based on the PSG results. Nineteen patients who were previously on room air were found to have abnormal PSG results necessitating oxygen supplementation, whereas only two were found to no longer need oxygen support. It is important to note that this change in level of oxygen support is based solely on a comparison between the baseline level of oxygen the infant was on prior to the PSG and the recommendations for oxygen support based on results from the PSG.

Table 4.

Adjustments to oxygen support based on sleep study results and discharge planning

	% (n)
Required adjustment of supplemental oxygen	69% (95)
- Room air to supplemental oxygen	- 20% (19)
- Supplemental oxygen to room air	- 2% (2)
- Required increase in oxygen support*	- 55% (52)
- Required decrease in oxygen support*	- 45% (43)
Discharge home with oxygen	98% (134)

^*from baseline prior to sleep study, based on sleep study recommendations

Discussion

This study provides data on PSG characteristics on the largest cohort of premature infants with moderate to severe BPD who had a PSG performed as an inpatient prior to discharge home, a practice that is unique to very few medical centers. Our cohort had predominantly severe BPD patients (82%) born around 26 weeks of gestation. All of these infants had abnormal polysomnogram characteristics with lower sleep efficiency of around 61%, elevated median OAHI (5.3 events/hour), ODI (15.7 event/hour), lower nadir SpO₂ (80%), elevated Ar/Aw index (22.25), and elevated OHI (3.25 events/hour). Considering that almost all these studies were titration studies done for clinical purposes, they are not truly comparable to indices obtained and reported from diagnostic studies. It is thence likely that indices such as OAHI, OHI, MAI, nadir SpO₂, end tidal CO2 etc. can potentially be much more severe than obtained. End tidal CO2 can be inaccurate due to faster respiratory rate and washing of CO2 with supplemental oxygen. Supplemental oxygen in addition prevents significant desaturation events thus leading to underscoring of the respiratory events. These are significant limitations of performing sleep studies, especially titration studies in this age group. Future designed prospective studies could consider diagnostic studies only rather titration studies and could potentially add transcutaneous CO2 monitoring for more accurate measurements. Regardless, these indices were abnormal not only when compared to children at one year of age and older, but some were also abnormal when compared to those reported by Daftary et al in their study of 30 normal neonates born at 37–42 weeks of gestation (14). These results identify a population of infants that are at increased risk for hypoxemia and its consequences and represent a potential target for improving neonatal outcomes though provision of supplemental oxygen tailored to the individual infant’s needs. The median nadir oxygen saturation was 80%. Further, we found that 87 infants (69%) who had a PSG required adjustment of their supplemental oxygen as compared to the clinical assessment of oxygen needs based on bedside monitoring, with 50 of those patients requiring an increase in oxygen support, and the rest requiring a decrease in support. Ensuring that infants with BPD are discharged home safely with adequate oxygenation is critical. Frequent episodes of hypoxia in infants with BPD may affect growth, cardiac functions and long term neurodevelopmental outcomes (15). Further, infants with BPD are vulnerable to episodes of hypoxemia and hypoventilation, which may be clinically inapparent, particularly during sleep (1, 16).

TSANZ, ATS, and BTS have made recommendations regarding target saturation, supplemental oxygen need, determination of oxygen need, and discharge home with oxygen (10–12). All three groups recommend use of supplemental oxygen to maintain saturations within goal and to facilitate discharge home and affirm the adequacy of continuous pulse oximetry to assess the need for and level of supplemental oxygen support. The ATS makes the additional recommendation in its practice guideline on home oxygen therapy for children that oximetry measurements should include the infant’s awake, sleeping and feeding states to adequately capture desaturation episodes (13). The TSANZ and ATS acknowledge the potential utility of PSG to wean infants off oxygen or to determine oxygen need in infants who are not doing well, but do not make recommendations for completion of a PSG for all infants with BPD (11, 12). At our institution, we use pulse oximetry as outlined in the above society guidelines to monitor saturations but use PSG when an infant who is otherwise approaching discharge home is having ongoing oxygen desaturations. We use the information from the PSG for oxygen titration and to determine the presence of apnea, its type (central, mixed or obstructive) and its severity. This allows us to titrate the level of oxygen support the infant needs to ensure a safe discharge home or pursue investigations like brain imaging, bronchoscopy or optimizing respiratory support if hypoventilation is present.

Completion of an inpatient infant PSG when possible should be considered in premature infants with severe BPD for several reasons. In its 2018 Clinical Practice Guideline regarding Home Oxygen Therapy for Children, the ATS strongly recommends that home oxygen therapy be prescribed for infants with BPD complicated by chronic hypoxemia, with chronic hypoxemia defined as ≥ 5% of the recording time spent with SpO₂ ≤ 93% if measurements are obtained continuously, or three separate readings of SpO₂ ≤ 93% if measurements are obtained intermittently (13). They note that assessment of oxygen saturation using pulse oximetry is enough to diagnose chronic hypoxemia in infants. However, hospital cardiorespiratory monitors can detect some, but not all, episodes of hypoxia in infants. The pulse oximetry output on hospital monitors represents the average of the patient’s oxygen saturation over 10–15 seconds, whereas the pulse oximetry probe used during a PSG averages a patient’s oxygen saturation every three seconds, thereby providing much greater resolution and detail to the information it provides. Additionally, oxygen saturations while the infant is awake may not be predictive of desaturations during sleep, and more effective management of oxygen treatment may be done with more continuous monitoring during sleep (17).

This has important implications particularly for infants with BPD who seem stable without oxygen support even on conventional unit monitoring, but on further study are found to require oxygen supplementation to maintain adequate saturations. There were eighteen of such infants in our study. PSG provides a wide range of objective and physiologic data that can direct individualized medical management in both the short-term, while the infant is admitted and monitored, and long-term, when the infant is at home (18). The results can also complement findings from direct visualization of the airway via bronchoscopy and thereby provide a full picture of the pathophysiologic processes contributing to sleep-disordered breathing in preterm infants (18). Our practice is to complete an inpatient PSG in a select group of patients with moderate to severe BPD to streamline clinical management and facilitate safe discharge home. The infant’s healthcare team, both neonatologists and pulmonologists, interpret the PSG findings in the context of the infant’s clinical picture to ensure that adequate oxygenation, and therefore safety, is optimized prior to discharge. We have found that using a portable PSG system and completing the full technician-attended PSG at the bedside to be safe and minimally disruptive to the care of the infant.

Ensuring that an infant is discharged home with adequate and optimal oxygen saturation is an important continuation of the practice of keeping premature infants at relatively higher oxygen saturation targets, as data from the BOOST-II and SUPPORT trials suggested increased mortality among infants in the lower oxygen saturation target group (19, 20). Maintaining oxygen saturations between 92–96% can improve growth, pulmonary hypertension, neurodevelopmental and behavioral outcomes, and decrease risk of sudden death (21–23). Infants with BPD are at increased risk for adverse neurodevelopmental and motor outcomes, and this risk appears to be positively correlated with the severity of BPD (15, 24).

Given the incidence of BPD in premature infants, we recognize that performing a PSG in all infants with BPD is resource-intensive and not feasible. The resources to perform inpatient infant PSG is not readily available in most centers but should be considered a helpful adjunct in individualizing therapy in particularly vulnerable patients such as those with severe BPD. Available literature supports pulse oximeter is sufficient to determine oxygen need in infants with BPD. Statements and guidelines from the Thoracic Society of Australia and New Zealand, the American Thoracic Society (ATS), and the British Thoracic Society all affirm the adequacy of continuous pulse oximetry to assess the need for and level of supplemental oxygen support (10–12). An alternative approach to performing an inpatient infant PSG is to obtain overnight pulse oximetry data download to assess the requirement for oxygen or to determine if the level oxygen being provided is optimal. This strategy has its own challenges, in that oxygen titration may not be done as accurately while the study is in process and no other data can be accurately obtained as in a PSG. If it is not possible to obtain any concrete data as an inpatient, the patient can be discharged on supplemental oxygen and complete the PSG study in the outpatient setting based on need and the judgement of the treating pulmonologist. Thus, a close follow-up with a pediatric primary care provider or pulmonologist is an important component to a safe discharge home, regardless of whether a PSG is performed or not, in order to monitor oxygen saturations, titrate the level of support, and provide further management. At our institution, we use PSG to ensure adequacy of respiratory support to this vulnerable patient population. Further, patients who continue to be followed at this institution undergo serial PSG as per their outpatient pulmonologist. The findings allow for individualized titration of oxygen therapy as well as a better understanding of degree of obstructive, mixed and central apnea in this vulnerable population who is at risk of adverse neurodevelopmental outcomes as well. This is additional information beyond what pulse oximetry can provide and may be helpful in the management of preterm infants with severe BPD.

A large multi-center cohort study found that tracheobronchomalacia is a common finding in neonates with BPD who undergo bronchoscopy (36% of infants in the study) and was associated with longer admissions and more complicated hospitalizations (25). Infants with tracheobronchomalacia often require CPAP to minimize airway collapse and obstruction (26). We did not routinely perform or examine bronchoscopy findings (if done) in our patient population, but it is plausible that a portion of our population likely has tracheobronchomalacia, which may contribute to the degree of OSA that we observe here. Additionally, studies have found evidence of upper airway remodeling in preterm infants that may result in OSA and contribute to an increased OAHI (27, 28). While infants with BPD may have a component of OSA, this elevated OAHI may not be solely due to airways obstruction secondary to altered anatomy. Of note, as discussed earlier OAHI can very well be under-estimated as almost all of the studies were titration studies. Since the index is related to desaturations of 3% or more from baseline that occur during both central and obstructive apneas, the score may be also an indication of the degree to which the lungs remain immature and injured, and not entirely reflect airways obstruction but may be a consequence of poor pulmonary reserve.

Few studies have investigated the PSG characteristics of preterm infants with BPD. To our knowledge, there are no studies to date that have studied the results of inpatient PSGs in preterm infants with moderate to severe BPD as young as our population. McGrath-Morrow et al. performed a retrospective study to determine the utility of overnight polysomnography to assess pulmonary reserve in infants with BPD. The mean age at which the PSG was done in this group of patients was ten months, they found that the respiratory disturbance index (RDI), defined as the number of hypopneas, obstructive, mixed and central apneas per hour of total sleep time, oxygen saturation nadir did not correlate with outpatient clinic measures of respiratory rate or oxygen saturation. Therefore, in order to help prevent some of the long-term complications of BPD in these infants, it may provide extra benefit to the patient to complete an overnight PSG and allow for a full assessment of oxygen needs prior to discharge home. Underdevelopment of the lungs in premature infants leads to impaired respiratory mechanics, which should improve over time as these infants continue to grow and develop (29). Our median nadir SpO₂ was 80%, lower than the mean nadir SpO₂ of 86% in the study by McGrath-Morrow. Our patient population was younger at the time of their PSG (43 weeks corrected gestation age or approximately 4 months of age in our population versus ten months of age), and therefore their lungs are likely not at the same level of development and growth as in McGrath-Morrow’s study.

Our study is the earliest assessment to date that examines the characteristics of inpatient PSG done prior to the initial hospital discharge of infants with BPD. It is also the largest study to date in terms of the number of infants that were eligible for our study. Our institution is unique in our practice of completing a relatively large number of inpatient infant PSG prior to discharge home, which provides us with a robust number of patients from which to collect data.

There are important limitations to note in our study. This is a retrospective study, so we are unable to draw any conclusions of causality from our results. There is inadvertent selection bias that exists due to retrospective nature of the study. Titration studies have limitations and can underestimate many PSG indices as discussed above. The adequacy of documentation can also impact the quality of the analysis, as evidenced by the group of “unclassifiable” patients. This is also a descriptive study, and we did not include premature infants without BPD as controls with which to compare our study group to. Due to the nature of being a tertiary referral center for the region and being situated in an urban setting, our patient population may be at higher risk for respiratory morbidity and have more severe BPD than other populations. In addition, not every infant diagnosed with BPD was followed by a pulmonologist, which may further limit our population to patients with more severe disease. Due to the variability in documentation practices and in level of detail in the records received from the referring hospitals, data on exact number of days on invasive or non-invasive ventilation was not available for some of our patients. We did not record whether infants received post-natal steroids for prevention of BPD. We also did not collect data on Apgar scores, oxygen dependence at twenty-four hours of age or administration of antenatal steroids, all of which are factors that are associated with the development of BPD but may or may not affect the severity of sleep disordered breathing. Given that our study focused on a subset of premature infants with moderate to severe BPD, our results may not be representative or generalizable to all infants with BPD, and outcome data to validate our approach is needed.

Further studies are needed to address the questions that remain from this current study. While we have provided demographic information, general clinical descriptors, and co-morbidities of our study population, further analysis can be done to examine for any relationships between these parameters and PSG findings. A study that includes a diagnostic PSG component, larger number of BPD patients and if possible control group should be considered. An analysis of airway exams in this population of infants along with correlation with PSG results may yield further insights. Long-term outcomes such as neurodevelopmental impairment, degree of respiratory morbidities, and pulmonary function, should also be further explored.

Conclusion

Results from the largest cohort to date of inpatient PSGs in infants with BPD show several abnormal PSG characteristics. Most infants required an alteration in the oxygen therapy as compared to their baseline oxygen support prior to the PSG. Performing a PSG during the index hospital admission and prior to discharge may be helpful for individualizing and optimizing inpatient treatment and planning for a safe discharge home. Further studies into the use of PSG in this vulnerable population are needed. It is possible that with more available data, a set of guidelines pertaining to use of PSG in infants with BPD may be developed.

Abbreviations:

PSG

polysomnogram

BPD

bronchopulmonary dysplasia

NICU

neonatal intensive care unit

OAHI

obstructive apnea-hypopnea index

ODI

oxygen desaturation index

OSA

obstructive sleep apnea

CI

central apnea index

SpO₂

oxygen saturation measured by pulse oximeter

TIB

Total time in bed

TST

Total sleep time

Ar/Aw

Arousal/Awakening index

ACO2

Average carbon dioxide level (in torr)

PCO2

Peak carbon dioxide level (in torr)

MAI

Mixed apnea index

OAI

Obstructive apnea index

OHI

Obstructive hypopnea index

REM

Rapid Eye movement

PB

Periodic breathing