Abstract
Introduction:
This retrospective study describes characteristics of serial polysomnograms (PSGs) of BPD patients on home oxygen therapy and describes PSG parameters associated with discontinuation of supplemental oxygen.
Methods:
A single-center study was performed at Children's Hospital Los Angeles, where serial PSGs for 44 patients with BPD infants discharged on home oxygen therapy were extracted for maximum of five PSGs or until oxygen discontinuation. Clinical and polysomnography data was collected. Characteristics of PSG1 were compared amongst the patients who were weaned from oxygen after PSG2 and PSG3.
Results:
Of 44 patients, 68.2% of patients were males with median birth gestational age of 26 weeks (IQR: 24.6–28.1), median birthweight of 777.5 g (IQR: 632.5–1054 g) and 77.3% of the cohort had severe BPD. A total of 138 PSGs were studied between all 44 patients serially. When comparing PSG1 and PSG2 parameters, statistically significant improvement was noted in multiple parameters. Median baseline SpO2, peak RR, and average PETCO2 were found to be potential predictors of prolonged oxygen use. Gestational age and birth weight were not associated with prolonged oxygen use after PSG3. The median age of oxygen discontinuation was calculated to be about 2 years of age.
Conclusions:
The severity of hypoxia and tachypnea on initial infant PSG are associated with prolonged oxygen therapy past 2 years of age. Growth and development of lungs with maturation of control of breathing help improve these parameters over time regardless of BPD severity. The study may inform discussions between providers and parents for patients discharged home on oxygen therapy.
Keywords: bronchopulmonary dysplasia, home oxygen therapy, infant sleep study, polysomnography, prematurity
1 ∣. INTRODUCTION
Bronchopulmonary dysplasia (BPD) has broad multifactorial pathophysiology affecting infants born prematurely.1,2 The aggregate effect of antenatal and postnatal insults has a net effect on the pathogenesis and development of BPD which plays a key role in long-term remodeling and repair at the alveolar and vascular level, impairing lung function and contributing to the severity of BPD.1,3 Abnormalities in airway resistance, forced expiratory flows, reduced diffusing capacity, and functional residual capacity have been documented in BPD neonates into childhood as long-term impairment in lung function.4 This indicates a need for not only ensuring safe discharge from the neonatal intensive care unit (NICU) but also careful long-term surveillance of their respiratory health. At the time of NICU discharge, many patients with BPD have a need for home oxygen therapy. After discharge, this is monitored clinically or via assessments through overnight oximetry studies or polysomnograms (PSGs) based on institution and/or provider practice.5,6 Reliance on clinical outpatient measures such as mean oxygen saturation (SpO2), respiratory rate (RR), and brief clinical evaluations may underestimate underlying intermittent hypoxemia, particularly during sleep.5 A recent study by House et al. showed the presence of persistent nocturnal hypoxemia in patients with BPD who seem otherwise clinically ready for oxygen discontinuation.6 Polysomnography has the potential to be a sensitive tool to personalize care, in assessing gas exchange, titrating, and weaning supplemental oxygen support chronically and identifying other sleep-disordered breathing abnormalities in this fragile BPD patient population.5,7 However, limited literature exists regarding anticipated changes in PSGs from early infancy to childhood in BPD.5,7-10 There is an increased prevalence of OSA in premature children starting from early infancy such that a recent clinical practice guideline from the American Thoracic Society recommended the use of PSG or referral to sleep center if there is ongoing need for oxygen at 2 years of age or symptoms suggestive of sleep-disordered breathing9
At our center, we often perform serial PSGs starting before initial discharge and then on outpatient basis to supplement clinical evaluation for assessing ongoing supplemental oxygen needs and to assess for sleep-disordered breathing in this patient population.
In this study, we aimed to describe serial PSG characteristics over time in BPD patients on an inpatient and outpatient basis. We aimed to compare these parameters in BPD patients with prolonged supplemental oxygen use up to 2 years of age and identify any potential features in the first or initial PSG (PSG1) that might reflect prolonged oxygen requirements.
2 ∣. METHODS
2.1 ∣. Study design and patient population
This study was performed at Children's Hospital Los Angeles. A retrospective study design was used to collect both PSG and clinical data from review of the medical chart. A query of Natus polysomnography database for the year 2016, was conducted to identify patients who had PSG done before the age of 12 months who were born less than 32 weeks of gestation. From this cohort, those who had at least two PSGs done subsequently were included if they were discharged home on oxygen or were seen in outpatient pulmonology clinic on home oxygen therapy. The presence of significant airway abnormality as in craniofacial syndromes, need for tracheostomy placement, cyanotic congenital heart disease, or inability to collect significant clinical data formed the exclusion criteria.
A total of 241 PSGs were extracted from the Natus sleep studies database for the year 2016, based on diagnoses of chronic lung disease/prematurity/BPD/oxygen dependence listed on their history at the time of the study. Of those, 103 were less than 12 months of age at the time of their first PSG. Of those, 47 unique patients had minimum of two PSGs done, were born <32 weeks gestational age and met the criteria for BPD per NIH classification, modified, per Abman et al. paper.3 One patient was excluded from the analysis due to craniofacial abnormalities (Goldenhar Syndrome) and tracheostomy placement. Two additional patients were excluded due to inadequate clinical data being available and/or unable to confirm BPD diagnosis. In total, 44 patients fitting the inclusion criteria were included in the final study cohort. Serial PSG data were collected for a maximum of five PSGs or until oxygen discontinuation or to the available information at the time of data collection. PSGs were ordered by consulting pulmonologists on inpatient and outpatient basis to aid in clinical decision-making to manage home oxygen therapy and assessment of sleep-disordered breathing.
Our hospital has quaternary NICU services without an in-house delivery service. Patients who were discharged from CHLA NICU were co-followed while inpatient by a consulting pulmonary attending. PSG was usually obtained to aid in clinical decision-making before discharge. The group of patients who had their initial infant PSG on an outpatient basis, were seen in pulmonary clinic for long-term BPD care after discharge from outside NICUs. Both PSG and clinical data were collected from the medical record chart review. The diagnosis and severity of BPD were assigned and confirmed upon chart review by identifying the level of respiratory support at 36 weeks gestational age.3,11,12 Those with missing clinical information regarding level of respiratory support at 36 weeks of gestation, to determine BPD severity were assigned “Unknown.”
2.2 ∣. Data collection
Demographic and clinical information were collected by chart review of the medical records of the identified patients. Documentation from history and physical examination, progress/procedural notes, diagnostic reports, and laboratory results were reviewed to obtain the following demographic data, including ethnicity, race, gender, and twin status. Clinical information included: hospitalization in our institution's NICU or outpatient referral to pulmonologist, gestational age, postmenstrual age at the time of each PSG, birth weight (BW), history of oligohydramnios, history of in utero substance exposure and maternal pre-eclampsia. From admission documentation and discharge summaries, the following clinical information was extracted: neonatal surfactant use, prenatal steroid use, perinatal antibiotics use, weight at NICU discharge, history of caffeine administration, duration of intubation, duration of noninvasive ventilation (nasal CPAP or nasal intermittent mandatory ventilation), presence of congenital heart disease, presence of PDA, history of PDA surgical ligation, presence of reflux, hemoglobin before PSG1, number of lifetime surgeries, airway evaluation by bronchoscopy, presence of pulmonary hypertension, respiratory medication use at the time of first pulmonary clinic appointment. Number of intubation days was obtained from aggregate dates, if present, documented in the medical chart. The presence of gastroesophageal reflux was inferred if the diagnosis was listed in problem list, documented in admission or discharge notes, or if patient was placed on anti-reflux medications (H2 blocker or proton pump inhibitors). Respiratory medications included albuterol/levalbuterol, budesonide, chlorothiazide, furosemide, and sildenafil.
2.3 ∣. Polysomnography
Equipment used in the sleep lab and at bedside PSGs included Braebon pediatric thermal airflow sensor, Respironics Pro-Tech PTAF2 pressure transducer which was used for airflow and as end-tidal carbon dioxide sensor, BCI/Smith medical capnocheck capnograph 9004 system for monitoring CO2 and SpO2 with averaging set to 2 beats/8 s.
All initial PSGs were oxygen titration studies with minimal and variable diagnostic components, done to aid in clinical management. PSGs for patients in the NICU were performed at bedside. Outpatient PSGs were done in the institution's sleep lab. Polysomnograms were read by board-certified pediatric sleep medicine physicians and scored according to the American Academy of Sleep Medicine Manual for Scoring Sleep and Associated Events 2007 guideline plus 2.3, 2.4, and 2.5 version updates. Scoring updates in various versions were implemented immediately as soon as they were made available to maintain up-to-date information.
PSG parameters measured in the study included: obstructive apnea-hypopnea index (OAHI), central apnea index (CAI), oxygen desaturation index (ODI; the number of oxygen desaturations of 3% or more per hour of sleep), median baseline oxygen saturation (SpO2), oxygen saturation nadir, peak respiratory rate, average respiratory rate, maximum heart rate, obstructive apnea index (OAI), obstructive hypopnea index (OHI), mixed apnea index (MAI), total time in bed, total sleep time, percentage in REM sleep, arousal index, sleep efficiency, presence of hypoventilation, maximum PETCO2, average PETCO2. Though we looked at other parameters like maximum oxygen flow titration, percent time SpO2 < 90% and periodic breathing, as we did not get this data in all studies, these parameters were not included in the data set. Some studies were initiated on supplemental oxygen which was gradually weaned to room air, while others were initiated on room air and oxygen was initiated if more than two desaturation events were less than 89% and titrated to keep SpO2 > 94% as per our hospital guidelines at the time of the study.
2.4 ∣. Statistical methods
Initial descriptive analysis was performed using STATA 17.0. Multivariate Poisson regression with robust standard error estimates13 was used to identify baseline PSG parameters associated with discontinuation of oxygen support after PSG3, around 2 years of age. A variable selection model, as described by Hosmer, Lemeshow, and Sturdivant was used to determine parameters predictive of oxygen discontinuation.14 Graphical (expected vs. the observed number of outcomes) as well as statistical tests, were used to assess goodness-of-fit.15 The rate ratio associated with one unit increase of each variable in the model was estimated, along with 95% confidence intervals. Sensitivity analyses were performed using negative-binomial regression.13,16 In addition, Kaplan–Meier survival curves were constructed to calculate the likelihood of patients on oxygen and to estimate the median age of oxygen discontinuation.
3 ∣. RESULTS
Serial PSGs were studied for 44 patients. All 44 patients had two PSGs. There were 32 patients with three consecutive PSGs, 12 patients with four PSGs, and 6 patients with five PSGs. Before PSG3, 12 patients were weaned off oxygen clinically. Twenty patients were weaned off before PSG4, while 6 were weaned before PSG5. Two patients were lost to follow-up after the fifth sleep study. All initial PSGs (PSG1) were done as oxygen titration studies. We included up to five PSGs as this was the maximum number of sleep studies available at the time of data collection.
In our cohort, 68.2% of patients were males with median birth gestational age of 26 weeks (IQR: 24.6–28.1), median birthweight of 777.5 g (IQR: 632.5–1054 g), and 77.3% of the cohort had severe BPD, of which 40.9% were type 1 severe BPD and 36.4% were type 2 (Tables 1 and 2).
TABLE 1.
Demographics and clinical characteristics of patient population.
| Demographics | |
|---|---|
| Ethnicity | |
| Hispanic | 59.1% |
| Non-Hispanic | 36.4% |
| Unknown | 4.5% |
| Race | |
| White | 15.9% |
| Black | 13.6% |
| Asian | 2.3% |
| Other/Asian/Pacific Islander | 56.8% |
| Unknown/missing | 11.4% |
| Gender | |
| Male | 68.2% |
| Female | 31.8% |
| Twin status | 18% |
| In utero substance exposure | 9% |
| N = 43a | |
| Antenatal steroids | 85% |
| N = 34a | |
| Surfactant administration | 88% |
| N = 40a | |
| Maternal antibiotics use | 63% |
| N = 33a | |
| Pre-eclampsia | 31% |
| N = 41a | |
| Oligohydramnios | 0.1% |
| N = 41a | |
| Caffeine administration | 88% |
| N = 42a | |
| Acyanotic congenital heart disease (not including PDA, ASD, VSD, PFO) | 11% |
| Patent ductus arteriosus (PDA) | 86% |
| N = 42a | |
| PDA ligation | 46% |
| N = 42a | |
| Reflux | 52% |
| Small for gestational age (SGA) | 17.5% |
| N = 40a | |
| Pulmonary hypertension | 25% |
| Albuterol/levalbuterol | 86% |
| Furosemide | 11% |
| Chlorothiazide | 70% |
| Budesonide | 86% |
| Sildenafil | 9% |
| Days of intubation | 34 (IQR: 9–52) |
| N = 39a | |
| Days on noninvasive ventilation | 32 (IQR: 9–52) |
| N = 35a | |
| Number of lifetime surgeries | 1 (IQR: 1–3) |
| Weight at discharge (g) | 3540 (IQR: 2880–4635) |
| N = 37a | |
| Hemoglobin before PSG1 (g/dL)b | 11.2 (IQR: 10.2–12.1) |
| Age of discontinuation of daytime oxygen | 9 (IQR: 7–12.5) |
Total number of patients (N) is 44, unless specified under each cell, due to data availability limitations. The percentages were calculated from the respective N of each parameter.
Timing of hemoglobin was variable between patients. Data reflect the last hemoglobin level recorded before PSG1.
TABLE 2.
Clinical information and PSG1, 2, 3, 4, and 5 parameters.
| PSG1 median (IQR), N = 44 |
PSG2 median (IQR), N = 44 |
PSG3 median (IQR), N = 32 |
PSG4 median (IQR), N = 12 |
PSG5 median (IQR), N = 6 |
||
|---|---|---|---|---|---|---|
| GA (weeks) | 26 (24.6–28.1) | 26 (24.6–28.1) | 25.9 (24.3–27.5) | 27.1 (25.1–30.1) | 29 (26.7–30.7) | |
| PMA at PSG (weeks) | 43.6 (41.1–47.8) | 70.3 (64.8–96) | 127.8 (103.1–164.5) | 161.8 (112.3–203.7) | 194.3 (135.4–253.1) | |
| Birth weight (g) | 777.5 (632.5–1054) | 777.5 (632.5–1054) | 730 (625–1005) | 927.5 (720–1200) | 954 (815–1502) | |
| Severity of BPD | ||||||
| Mild (N = 1) | 2.3% | 2.3% | 0 | 0 | 0 | |
| Moderate (N = 4) | 9.1% | 9.1% | 6.2% (N = 2) | 0 | 0 | |
| Severe type 1 (N = 18) | 40.9% | 40.9% | 40.6% (N = 13) | 50% (N = 6) | 16.7% (N = 1) | |
| Severe type 2 (N = 16) | 36.4% | 36.4% | 40.6% (N = 13) | 41.7% (N = 5) | 66.7% (N = 4) | |
| Unknown (N = 5) | 11.4% | 11.4% | 12.5% (N = 4) | 8.3% (N = 1) | 16.7% (N = 1) | |
| PSG parameters | Comparison of PSG1 and 2 parameters (p values) | |||||
| OAHI | 4.7 (2–13) | 2.3 (0.95–6.35) | 1.3 (0.4–2.75) | 1.95 (0.9–3.65) | 3.3 (0.6–8.4) | .0195 |
| OAI | 0.49 (0–1.2) | 0 (0–0) | 0 (0–0) | 0 (0–0.15) | 0.08 (0–0.36) | .0001 |
| OHI | 3.19 (1.34–6.79) | 1.8 (0.87–4.48) | 1.21 (0.26–2.43) | 1.26 (0.43–2.86) | 2.7 (0.18–6.76) | .1063 |
| MAI | 0.5 (0–1.54) | 0 (0–0.46) | 0 (0–0.21) | 0 (0–0.31) | 0.29 (0–0.77) | .0059 |
| CAI | 2.5 (1.25–4.15) | 3.5 (1.95–4.65) | 2.35 (1.25–4.1) | 1.1 (0.9–1.8) | 1.9 (1.1–3.8) | .1362 |
| ODI | 23.36 (5.8–71) | 10.64 (6.1–27.75) | 6.15 (1.77–20.1) | 4 (2.05–8.51) | 5.85 (0.5–6) | .1226 |
| Nadir SpO2 | 82 (75–84) | 85 (80.5–89) | 86 (81–90) | 84.5 (81–89.5) | 86 (84–88) | .0065 |
| Median Baseline SpO2a | 93 (89.75–96.5) | 95.5 (93.75–97.25) | 96.5 (94–97.25) | 96 (94.75–96.5) | 96.75 (94.5–97.5) | .0184 |
| Peak RR | 62.5 (53–70) | 33.5 (30–37) | 24 (21–28) | 21 (19–25) | 17.5 (16–22) | .0001 |
| RR | 55 (49–65) | 31 (28–35) | 21 (19–25) | 20.5 (17.5–23) | 16.5 (16–19) | .0001 |
| Max HR | 178 (162–185) | 153 (143–162) | 125.5 (135.5–161.5) | 136 (136–147) | 126.5 (121–187.5) | .0001 |
| TIB | 305.5 (271.75–335.75) | 307.75 (165.75–416.5) | 420.75 (380.25–432) | 423.5 (341.15–449.85) | 433.6 (401–461.1) | .9801 |
| TST | 195.75 (166.5–216.75) | 205.5 (120.5–320.75) | 346 (303.75–374.25) | 336.75 (305–381.5) | 326.25 (302–387.5) | .5647 |
| %REM | 11.8 (4.6–26.9) | 20.4 (14.65–26.65) | 20.4 (14.25–26.3) | 20.4 (12.95–25.05) | 17.4 (12.6–12.1) | .0585 |
| Arousal index | 22 (16.7–27.3) | 13.1 (9–20.1) | 12.4 (10.7–15.3) | 10.2 (5.7–13.4) | 10.95 (10.1–17.8) | .0001 |
| Sleep efficiency | 64.85 (55.55–74.9) | 79.1 (62.8–89.05) | 81.85 (72.95–88.65) | 82.25 (75.1–91.85) | 78.15 (74.7–85) | .0083 |
| Peak PETCO2 | 50 (47–53.5) | 47 (44–51) | 49 (44–53) | 50 (49–52) | 48.5 (39–54) | .0129 |
| Av PETCO2 | 38.25 (32.5–0.5) | 38 (34.5–41) | 39 (35.5–43) | 42 (41.5–45.5) | 39.5 (36.5–46.5) | .4352 |
In PSG reports, baseline SpO2 was stated in the form of SpO2 ranges during the study excluding the events; therefore, the average of these two values was taken and assigned Median baseline SpO2.
PSG1—polysomnogram 1, PSG2—polysomnogram 2, OAHI—obstructive apnea-hypopnea index, OAI—obstructive apnea index, OHI—obstructive hypopnea index, MAI—mixed apnea index, CAI—central apnea index, ODI—oxygen desaturation index, RR—respiratory rate, HR—heart rate, REM—rapid eye movement, Peak PET CO2—Peak End Tidal carbon dioxide, Av PETCO2—Average End Tidal carbon dioxide, TST—total sleep time, TIB—total time in bed, GA—gestational age, PMA—postmenstrual age.
64% were hospitalized at our institution's NICU. Others were seen in the pulmonology clinic. 18% were twins and 9% had in utero substance exposure. We found that 85% of the mothers received antenatal steroids and 63% received antenatal antibiotics. 88% of patients received surfactant after birth and 88% were administered caffeine therapy for apnea of prematurity (Table 1). Pre-eclampsia prevalence was 13 out of 41 patients (31%), with missing information for 3 patients (Table 1). Oligohydramnios was present in four patients; two of them were weaned to room air before PSG3 and the remaining two were weaned to room air after PSG3.
11% of patients had acyanotic congenital heart disease. PDA was diagnosed in 86% of patients with 46% needing surgical ligation. About half the patients (52%) were diagnosed with reflux or placed on anti-reflux medications. By the time of first outpatient pulmonary clinic visit, 52% used bronchodilators (albuterol/levalbuterol), more than 2/3rd of the patients used diuretics with 11% being prescribed furosemide, 70% chlorothiazide, 86% inhaled steroid (budesonide), and 9% were on sildenafil (Table 1). The prevalence of pulmonary hypertension beyond 36 weeks was 11 patients out of 44 (25%). Of the 32 patients who underwent PSG3, 8 had pulmonary hypertension. Of 20 patients who needed oxygen after PSG3, 6 (30%) were those who had history of PH.
There were four patients without documented birth weight out of 44 total cohorts. Out of the remaining 40 patients, 7 patients (17.5%) were identified as small for gestational age (SGA). 6 SGA patients (19.3%) out of 31 (BW for one patient was unknown) underwent PSG3. There were 3 SGA patients out of 12 (25%) who underwent PSG4 and 2 out of 6 patients who underwent PSG5 (33%) were SGA. Although there was a trend toward being SGA in patients with prolonged oxygen use, the number of patients was too small for any additional purposeful statistics.
The median days of intubation were 34 (IQR: 9-52) and the median days on noninvasive ventilation were 32 (IQR: 9–52). Median number of lifetime surgeries (by the time of data collection) was 1 (IQR: 1–3) which was chosen as a surrogate of patient complexity. In addition, median weight at discharge was 3540 g (IQR: 2880–4635). Median hemoglobin obtained before PSG1 was 11.2 g/dL (IQR: 10.2–12.1) (Table 1).
39% of patients who underwent PSG1 had escalation in oxygen requirement, while 32% had a decrease immediately post-PSG. The remainder had no change in baseline oxygen therapy. One patient required escalation to nasal CPAP support after PSG1, while the rest of the patients only needed changes in the amount of LPM flow. All 44 patients underwent PSG 2% and 14% of these experienced an increase in the amount of supplemental oxygen while 45% experienced a decrease. Subsequently, after PSG3 (N = 32), 66% of patients had a decrease in oxygen therapy, while none had an increase. After PSG4 (N = 12), 8% of patients had an increase in their oxygen therapy, while 58% had a decrease in therapy. Finally, after PSG5 (N = 5), 60% had decrease in respiratory support, while none had increase in support. Documentation regarding the clinical decision for oxygen therapy after PSG5 was missing for one patient (Table 2).
In comparing PSG1 and PSG2 parameters, statistically significant changes were noted in the following: reduction in OAHI, decrease in OAI, decrease in mixed apnea index, improvement in nadir SpO2, improvement in baseline SpO2, reduction in respiratory rate and maximum heart rate, decrease in arousal index, improved sleep efficiency and decrease in peak end-tidal CO2 (Table 2).
The median post-menstrual age at the time of PSG2 was 70.3 weeks (IQR: 64.8–96) and 127.8 weeks (IQR: 103.1–164.5) for PSG3. Tachypnea and lower median baseline SpO2 in PSG1 were found to be a statistically significant finding in patients with prolonged oxygen supplementation after their third sleep study which was around 2 years of age (Table 3). Of the severe BPD group (type 1 and type 2 combined), 38% had oxygen discontinued after PSG3 (around the second year of age).
TABLE 3.
Comparison of GA, PMA, BW, and PSG1 parameters of patients with oxygen discontinued after PSG2 and PSG3.
| Oxygen discontinued after PSG2 | Oxygen discontinued after PSG3 | |||||
|---|---|---|---|---|---|---|
| Room air N = 8 | Non-room air N = 36 | p Value | Room air N = 24 | Non-room air N = 20 | p Value | |
| GA (weeks) | 27.1 (24.9–28.9) | 26 (24.6–27.8) | .3942 | 26 (24.2–28) | 26 (25–28.7) | .7236 |
| PMA (weeks) | 42.1 (40.8–47.1) | 43.7 (41.2–47.8) | .6156 | 44 (41.4–47.9) | 44 (41.4–47.9) | .4721 |
| Birth weight (g) | 902 (715–1400)a | 731 (625–1005)b | .2622 | 797.5 (613–1025)c | 777.5 (672–1069) | .7236 |
| PSG 1 parameters | ||||||
| OAHI | 4.35 (2.2–9.6) | 4.7 (2–13.8) | .6481 | 4.75 (2.65–14.3) | 4.35 (1.25–7.85) | .2158 |
| OAI | 0.83 (0.3–1) | 0.35 (0–1.6) | .7609 | 0.3 (0–1.2) | 0.5 (0–1.4) | .7443 |
| OHI | 2.7 (1.1–8.4) | 3.3 (1.3–6.1) | .7842 | 3.2 (2.1–10.9) | 2.8 (1.2–4.1) | .1439 |
| MAI | 0.39 (0.1–0.7) | 0.5 (0–2.8) | .5032 | 0.44 (0–1.5) | 0.5 (0–1.8) | .7998 |
| CAI | 3.3 (2.6–4.5) | 2 (1.2–4.1) | .2733 | 3 (1.3–4.95) | 1.75 (1.2–3.35) | .1641 |
| ODI | 8.1 (4.2–14.5)a | 32.5 (5.8–75)d | .1554 | 26 (8.1–71)a | 15.35 (2.9–75)d | .4863 |
| Nadir SpO2 | 82.5 (76.5–84.5) | 82 (74–84) | .6264 | 82 (77–84) | 80 (72.5–84.5) | .2660 |
| Median Baseline SpO2 | 94.5 (91–96.7) | 93 (89.2–96.5) | .5328 | 94 (92–97) | 91 (86.5–95.5) | .0387 |
| Peak RR | 64 (52–70)a | 62 (53–70)a | .7613 | 60 (52–64)a | 68 (55–74)a | .0151 |
| RR avg | 62 (52–68)a | 53 (46.5–64)a | .1938 | 52 (46.5–57)a | 62 (49–70)a | .0687 |
| Max HR | 185 (167.5–190) | 176 (160–184) | .1511 | 177 (165.5–187) | 178 (155–185)a | .6594 |
| TIB | 307 (277.5–320) | 305.5 (271.7–339.7) | .7037 | 294 (270.5–320) | 324 (280.7–346.5) | .1116 |
| TST | 203 (153.2–228.7) | 193.5 (167.7–216.2) | .9636 | 193.5 (166.5–231) | 201.5 (161.7–214.5) | .9249 |
| %REM | 11.5 (4.5–23.8) | 12 (4.6–28.2) | .8551 | 16 (6.85–30.5) | 10.4 (2.6–18.3) | .1363 |
| Arousal index | 21.8 (16.7–24.7) | 22.4 (16.7–28.1) | .7150 | 20.9 (15.9–27.9) | 23.8 (17.7–27) | .6458 |
| Sleep efficiency | 65.9 (61–74.3) | 64.1 (54.3–74.9) | .7725 | 65.7 (56.8–74.1) | 62.2 (52.4–74.9) | .8875 |
| Peak PETCO2 | 50.5 (47.5–56) | 50 (47–53) | .7037 | 50 (45–53.5) | 51 (48.5–54) | .7405 |
| Av PETCO2 | 33.2 (31.5–39) | 38.7 (33.5–40.5) | .2938 | 36.2 (32.5–40.7) | 38.5 (31.7–40.5) | .2465 |
Data missing for one patient.
Data missing for three patients.
Data missing for four patients.
Data missing for two patients.
Table 4 depicts the rate ratio (RR, 95% confidence interval) for each half-standard deviation increase for each variable identified by Poisson regression to be predictive of oxygen discontinuation. The final model revealed the following: median baseline SpO2, peak RR, and average PETCO2 were predictors of prolonged oxygen use. Patients with lower median baseline SPO2 on PSG1 are more likely to need oxygen after 2 years of age. Every 3% increase in median baseline SpO2 on PSG1, increases the likelihood of being on room air by 20% at PSG3; every increase of 8 breaths/min increases the likelihood of needing oxygen at PSG3 by 28%. The median age of oxygen discontinuation was calculated to be about 2 years of age (see Kaplan–Meier graph Figure 1).
TABLE 4.
Poisson regression model for oxygen discontinuation.
| PSG parameter at baseline (PSG1) |
Rate ratio (RR) |
95% Confidence interval |
p Value |
|---|---|---|---|
| Median baseline SpO2 (%) | 0.93 | (0.90–0.96) | <.001 |
| Peak RR (breaths/min) | 1.03 | (1.02–1.05) | <.001 |
| Av PETCO2 (mmHg) | 1.05 | (1.00–1.09) | .038 |
FIGURE 1.
Kaplan–Meier graph. The median age at discontinuation as calculated to be about 2 years of age.
The median age of discontinuation of daytime oxygen or using oxygen when asleep only, was 9 months of age (IQR: 7–12.5).
There were patients for whom oxygen was discontinued later and the oxygen available at home for as needed (PRN) use. We had 35% of the patients in our cohort with documentation of PRN oxygen use. Data was available for 40 patients. Four patients had limited information documented.
Gestational age at birth (p value .7333) and birth weight (p value .8602) were not associated with prolonged oxygen use after PSG3 (Table 3).
Nine of the 20 patients who continued to need oxygen after PSG3 (roughly corresponding to 2-year chronological age), had one or more combined risk factors for severe BPD, including pulmonary hypertension, SGA, and oligohydramnios. In the entire cohort, we had only six patients who underwent airway evaluation by bronchoscopy. Of these, only two needed prolonged oxygen therapy after PSG3, of whom one had normal airway evaluation and the other had mild tracheomalacia and needed supraglottoplasty for laryngomalacia. Of the four patients who did not need prolonged oxygen, two had normal evaluation, one had grade 1 subglottic stenosis, and another patient also had mild tracheomalacia and the need for supraglottoplasty. None of the patients needed any tonsillectomy or adenoidectomy before 2 years of age.
4 ∣. DISCUSSION
Our study is the largest study that is known to us that reviews serial PSG characteristics in BPD patients discharged on home oxygen therapy. It described changes in PSG parameters with age and discontinuation of supplemental oxygen. There is scarcity of polysomnographic data on BPD population during infancy and even lesser data assessing change in the PSG variables over time. There was consistent improvement in these parameters over time which is expected with lung growth, improved pulmonary compliance, and maturation of control of breathing with age. We compared PSG1 data between patients who needed oxygen beyond the second and third PSGs, as these times corresponded to roughly first and second years of chronological age to identify any PSG parameters that can reflect possible long-term oxygen use in these patients. In our study population, those with lower median baseline SpO2, higher RR, and higher average PETCO2 on PSG1 were more likely to have prolonged oxygen use which is reflective of their pulmonary status.
Daftary et al. worked to define normative values of parameters in diagnostic polysomnography in newborns and described high indices of obstruction and lower nadir in their cohort of healthy newborns.17 Fenlon et al. described indices of obstruction and hypoxemia in titration studies done in their BPD cohort before discharge.8 Our cohort also underwent primary oxygen titration studies and had similar abnormal PSG parameters at the time of discharge or PSG1.
The use of polysomnography as a form of disease severity assessment was explored in other studies such as correlation with computed tomography (CT) scan abnormalities in severe BPD patients in a study by Mastrigt et al. The study showed that 74% of the infants had an elevated ODI > 5 in addition to architectural distortion on chest imaging.18 An increase in gestational age of 1 week was associated with lower ODI; in addition, increased weight gain between 2 and 6 months of corrected age was associated with less severe desaturations during sleep.18
Improved sleep quality was observed by a reduction in arousal Index and improved sleep efficiency over time. We speculate that this is due to maturation of sleep architecture and decreased need for care interventions after graduation from the NICU. The CAI decreased over time indicating overall maturity of respiratory control mechanisms. Overall, all PSG indices showed improving trend with ODI decreasing from median of 23.36 to 4 events/h by PSG4, with statistically significant improvement in nadir SpO2, Median baseline SpO2, respiratory rate, peak PETCO2, which reflects lung growth and improvement in gas exchange surface area with time. The median age of discontinuation of supplemental oxygen was around 2 years of age. Gestational age between patients weaned off oxygen by PSG2 and PSG3 was similar (Table 3). These findings suggest the role of postnatal factors contributing to long-term respiratory morbidity (reflected by prolonged oxygen use).19
We acknowledge that prolonged oxygen need, as noted in our study, may have wider causal base with multiple factors contributing, such as variability in clinical practice for oxygen continuation when interpreting PSG reports timely getting PSG—which in turn is affected by many factors in itself (appointment availability, patient sickness related delay as studies are avoided when the patient is actively sick with respiratory illness, missed appointments), wait time until study is read and followed up by clinician. In our cohort, there was median delay of 1 month (IQR 1–5) from the time of PSG to the date of follow-up when oxygen was discontinued. There were also instances when patient's families self-discontinued oxygen before clinic visits, but the order for discontinuation was placed later. Hence, the oxygen discontinuation date reported in the medical record would be different from the actual time of parent discontinuation of oxygen. In addition, the physician variability in the management of oxygen therapy could have also contributed to seemingly prolonged oxygen requirement in our cohort.
In a few instances, sleep studies were performed for the concerns for sleep-disordered breathing and oxygen was resumed.
In our cohort, about 35% of the patients were on as needed oxygen supplementation before removal of oxygen from home. Factors contributing were varied from provider preference or parental worry about discontinuing or parental perception of need for oxygen only when the child is sick to the patient's intolerance of nasal cannula.
Despite this, we still had 20 patients in our cohort who displayed the need for ongoing oxygen supplementation post-PSG3. Of these 20 patients, 4 had PH alone as risk factor, 2 had both PH and SGA, 2 had SGA alone, and 1 with oligohydramnios; thus, overall, 9 of the 20 carried some multiple risk factors. These numbers are small for any statistical calculation but reflect the need for oxygen contributed by these underlying factors.
Other studies have demonstrated the utility of overnight oximetry studies as a modality of detecting the presence of hypoxemia in otherwise seemingly ready-to-be weaned off the oxygen.6 The RHO trial conducted by Rhein et al. regarding the use of oximetry studies in premature patients had earlier oxygen discontinuation compared to our cohort.20 As seen in the RHO trial, we agree regimented follow up and earlier diagnostic testing can eliminate prolonged oxygen use. Inherent differences exist between our study and RHO trial. RHO was done with a randomized control study design, while our study is a retrospective study which would have limitations in terms of adequate data collection and the ability to have PSG or a sleep study done in timely manner with quick follow-up. In addition, our study cohort had predominantly severe BPD patients (about 76%) compared to the RHO trial which had about 55% of their patients with severe BPD. Pulmonary hypertension patients were excluded from that study, while our cohort included 25% of BPD patients with pulmonary hypertension in addition to 31% of patients with maternal history of pre-eclampsia and 17.5% of patients were SGA. Such differences in patient characteristics and study design would contribute to variability in oxygen use duration seen between the two cohorts.
In terms of sleep studies performed in our study, end-tidal carbon dioxide was used instead of transcutaneous CO2 or capillary blood gas. This might underestimate hypercapnia due to known tachypnea in these patients due to CO2 washout.
In addition, our study not only aimed to assess the duration of oxygen requirement but also to look at the natural history of sleep-disordered breathing in this patient population. We do acknowledge that an accurate natural disease course would be a prospective evaluation in room air of patients with BPD using polysomnography in a controlled manner. However, given the retrospective design and if the infant is already hypoxemic and on oxygen, it would be clinically and ethically challenging to perform a sleep study strictly on room air in these circumstances.
Despite this, Table 2 depicts changes in PSG parameters with subsequent studies, that can be used to inform families' progression of these parameters with time. We noticed a statistically significant difference just after the second PSG for many parameters. Notably, OAHI, OAI, MAI, nadir SpO2, median baseline SpO2, peak RR, RR, max HR, improvement in arousal index, sleep efficiency, and peak Pet CO2. These and other parameters we collected continue to improve in subsequent PSGs. This reflects some degree of natural course despite study limitations.
There were several limitations in our study as seen with retrospective study design with limited documentation and missing data particularly in patients who were discharged from outside NICUs. All PSGs were titration sleep studies, which would underestimate the frequency and severity of respiratory events, such as the scoring of apneas and hypopneas. Another limitation is the selection bias inherent to the patient population managed at CHLA, as most patients were referred for higher level of acute care or surgical management to our NICU, thus conferring higher complexity and disease severity. We also selected our cohort based on the presence of at least two PSGs. As not all infants with BPD would necessarily receive PSGs and often may not get a second one, we may have selected group of high-risk patients, thus increasing the duration of oxygen use when compared to other studies.
RHO study did not discuss in detail what parameters were considered by experts to wean child from oxygen. Group of experts deciding versus different providers making their own decision is different and is bound to have limitations as such.
It would be useful to have future studies looking into provider surveys as to how they interpret overnight oximetry study results and sleep study reports for clinical decision-making. Eventually, guidelines detailing HOT when considering overnight oximetry studies and PSGs would be helpful.
Lack of evidence-based optimal SpO2 guidelines in this population leads to variability in practice with providers choosing >92% or >93% (or >94% like in our institution) as adequate and is taken into consideration when deciding supplemental oxygen use.21 Though it is very welcomed information to family when oxygen is discontinued, it is important to ponder that we are making decisions with limited understanding of not just late neurodevelopmental outcomes but also what effects it may bear on cardiopulmonary dysanapsis into their adulthood.22,23
Chronic oxygen therapy in BPD patients is a potential indicator of healthcare utilization need and frequent rehospitalizations, longer admission durations, and overall higher healthcare costs.24 More studies are needed to understand and assess this delicate group of patients to institute early counseling and possible preventative measures to optimize long-term outcomes and reduce healthcare burden.
Taking our limitations into consideration, through this study, we attempted to understand the natural disease course of sleep-disordered breathing and oxygen dependency in BPD patients. To our knowledge, this is one of the few larger cohort studies describing multiple serial PSGs in prolonged oxygen use with an attempt to identify early PSG markers that may help guide discussions with the families about the estimated length of oxygen therapy.
5 ∣. CONCLUSIONS
BPD patients have abnormal PSG parameters. The severity of hypoxia and tachypnea on initial infant PSG is associated with prolonged oxygen therapy past 2 years of age. Growth and development of lungs with the maturation of control of breathing help improve these parameters over time regardless of BPD severity and gestation age. Providers may reference our findings when counseling families regarding estimates of the length of time that supplemental oxygen therapy will be needed for their infant with severe BPD. Future studies could be designed as prospective studies with diagnostic PSGs and may be at standardized intervals to follow changes over time to allow for better association with the severity of BPD, long-term outcomes, and duration of oxygen need.
ACKNOWLEDGMENTS
This work was supported by grants UL1TR001855 and UL1TR000130 from the National Center for Advancing Translational Science (NCATS) of the U.S. National Institutes of Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.
DATA AVAILABILITY STATEMENT
Research data are not shared.
REFERENCES
- 1.Kalikkot Thekkeveedu R, Guaman MC, Shivanna B. Bronchopulmonary dysplasia: A review of pathogenesis and pathophysiology. Respir Med. 2017;132:170–177. doi: 10.1016/j.rmed.2017.10.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Allen J, Zwerdling R, Ehrenkranz R, et al. Statement on the care of the child with chronic lung disease of infancy and childhood. Am J Respir Crit Care Med. 2003;168(3):356–396. doi: 10.1164/rccm.168.3.356 [DOI] [PubMed] [Google Scholar]
- 3.Abman SH, Collaco JM, Shepherd EG, et al. Interdisciplinary care of children with severe bronchopulmonary dysplasia. J Pediatr. 2017;181:12–28.e1. doi: 10.1016/j.jpeds.2016.10.082 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Baraldi E, Filippone M, Trevisanuto D, Zanardo V, Zacchello F. Pulmonary function until two years of life in infants with bronchopulmonary dysplasia. Am J Respir Crit Care Med. 1997;155(1):149–155. [DOI] [PubMed] [Google Scholar]
- 5.McGrath-Morrow SA, Ryan T, McGinley BM, Okelo SO, Sterni LM, Collaco JM. Polysomnography in preterm infants and children with chronic lung disease. Pediatr Pulmonol. 2012;47(2):172–179. doi: 10.1002/ppul.21522 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.House M, Klein S, Parham D, Hysinger EB, Brady JM. Frequent hypoxemia found in infants with bronchopulmonary dysplasia after weaning home oxygen. Pediatr Pulmonol. 2022;57(11):2638–2645. doi: 10.1002/ppul.26076 [DOI] [PubMed] [Google Scholar]
- 7.Ortiz LE, McGrath-Morrow SA, Sterni LM, Collaco JM. Sleep disordered breathing in bronchopulmonary dysplasia. Pediatr Pulmonol. 2017;52(12):1583–1591. doi: 10.1002/ppul.23769 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Flores-Fenlon N, Wright N, Lew C, et al. Retrospective analysis of inpatient polysomnogram characteristics and discharge outcomes in infants with bronchopulmonary dysplasia requiring home oxygen therapy. Pediatr Pulmonol. 2021;56(1):88–96. doi: 10.1002/ppul.25129 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Cristea AI, Ren CL, Amin R, et al. Outpatient respiratory management of infants, children, and adolescents with post-prematurity respiratory disease an official American Thoracic Society Clinical Practice guideline. Am J Respir Crit Care Med. 2021;204(12):e115–e133. doi: 10.1164/rccm.202110-2269ST [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Collaco JM, McGrath-Morrow SA. Respiratory phenotypes for preterm infants, children, and adults: bronchopulmonary dysplasia and more. Ann Am Thorac Soc. 2018;15(5):530–538. doi: 10.1513/AnnalsATS.201709-756FR [DOI] [PubMed] [Google Scholar]
- 11.Jobe AH. Background and Reason for Workshop. 2001. www.atsjournals.org [Google Scholar]
- 12.Higgins RD, Jobe AH, Koso-Thomas M, et al. Bronchopulmonary dysplasia: executive summary of a workshop. J Pediatr. 2018;197(1):300–308. doi: 10.1016/j.jpeds.2018.01.043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Zou G. A modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159(7):702–706. doi: 10.1093/aje/kwh090 [DOI] [PubMed] [Google Scholar]
- 14.Hosmer DW, Lemeshow S, Sturdivant RX. Applied Logistic Regression. 3rd ed. John Wiley & Sons; 2013. [Google Scholar]
- 15.Manjón M, Martínez O. The Chi-squared Goodness-of-fit Test for Count-data Models. 2014. [Google Scholar]
- 16.Chen W, Qian L, Shi J, Franklin M. Comparing performance between log-binomial and robust Poisson regression models for estimating risk ratios under model misspecification. BMC Med Res Methodol. 2018;18:63. doi: 10.1186/s12874-018-0519-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Daftary AS, Jalou HE, Shively L, Slaven JE, Davis SD. Polysomnography reference values in healthy newborns. J Clin Sleep Med. 2019;15(3):437–443. doi: 10.5664/jcsm.7670 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.van Mastrigt E, Kakar E, Ciet P, et al. Structural and functional ventilatory impairment in infants with severe bronchopulmonary dysplasia. Pediatr Pulmonol. 2017;52(8):1029–1037. doi: 10.1002/ppul.23696 [DOI] [PubMed] [Google Scholar]
- 19.Jensen EA, Schmidt B. Epidemiology of bronchopulmonary dysplasia. Birth Defects Res A Clin Mol Teratol. 2014;100(3):145–157. doi: 10.1002/bdra.23235 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Rhein L, White H, Simoneau T, et al. Transmitted home oximetry and duration of home oxygen in premature infants. Pediatrics. 2020;146(2):e20200079. http://publications.aap.org/pediatrics/article-pdf/146/2/e20200079/1079642/peds_20200079.pdf [DOI] [PubMed] [Google Scholar]
- 21.Hayes D, Wilson KC, Krivchenia K, et al. Home oxygen therapy for children. An official American Thoracic Society Clinical Practice guideline. Am J Respir Crit Care Med. 2019;199(3):e5–e23. doi: 10.1164/rccm.201812-2276ST [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.DeMauro SB. Recorded continuous oximetry improves post-discharge management of bronchopulmonary dysplasia. Pediatrics. 2020;146(2):e2020002774. doi: 10.1542/peds.2020-002774 [DOI] [PubMed] [Google Scholar]
- 23.Duke JW, Lewandowski AJ, Abman SH, Lovering AT. Physiological aspects of cardiopulmonary dysanapsis on exercise in adults born preterm. J Physiol. 2022;600(3):463–482. doi: 10.1113/JP281848 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Greenough A, Alexander J, Burgess S, et al. Home oxygen status and rehospitalisation and primary care requirements of infants with chronic lung disease. Arch Dis Child. 2002;86(1):40–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Research data are not shared.