Effect of fluctuation of oxygenation and time spent in the target range on retinopathy of prematurity in extremely low birth weight infants

Abstract

BACKGROUND:

The association between saturation of peripheral oxygenation (SpO₂) fluctuation and severity of retinopathy of prematurity (ROP) is well elucidated in extremely low birth weight (ELBW) infants. Time spent in the Target range of SpO₂ is also associated with the severity of ROP.

METHODS:

In a prospective observational study, the SpO2 of all ELBW infants admitted to our unit were monitored for the first four weeks of life, and averaged every minute for analysis. The percent time spent at SpO2 <90%, 90–95%, and >95% and weekly SpO₂ fluctuations [as SpO2 coefficient of variation (CoV)] were calculated.

RESULTS:

During the study period 21 infants had moderate to severe ROP and 35 infants served as controls. Infants with moderate to severe ROP were smaller and younger than their controls [676±124 grams vs. 796±148 grams (p < 0.001); and 24.0±1.0 weeks vs. 25.0±1.7 weeks (p < 0.001) respectively]. There were no significant differences in time spent in the 90–95% range between groups (p = 0.66). However there was a significant increase in weekly SpO₂ CoV in infants with moderate to severe ROP vs. controls (p = 0.007).

CONCLUSION:

In ELBW infants, there was an association between SpO₂ fluctuation during the first four weeks of life and severity of ROP, although, no association was established with time spent in the target range of SpO₂.

Keywords

Retinopathy of prematurity saturation of peripheral oxygen prematurity fluctuation target oxygen saturation

1 Introduction

Preterm birth affects approximately 381,935 infants or 9.63% of all deliveries per year in the United States [1]. Approximately 0.7% of the births are less than 1000 grams in weight, but these infants constitute about 48% of all infant mortality [2]. Extremely low birth weight infants (ELBW) with a birth weight of less than 1000 grams have the highest morbidity which includes respiratory distress syndrome (RDS), bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), intraventricular hemorrhage (IVH) and retinopathy of prematurity (ROP) [3]. The incidence of ROP was 63.7% while severe ROP (stage 3–5 bilaterally) was noted in 12.3% in ELBW infants [4].

Pulse oximetry has become fundamental for non-invasive monitoring of saturation of peripheral oxygen (SpO₂) in premature infants. In a large multicenter randomized controlled study, targeting a high SpO₂ during the first days of life was associated with a higher incidence of ROP while targeting a lower SpO₂ range was associated with a higher mortality; however, in other studies, targeting a lower SpO₂ was not associated with a higher mortality [5 –7]. Therefore, the ultimate targeted range of SpO₂ remains to be determined, especially when it has been very difficult to tightly control the SpO₂ within a defined targeted range [8, 9].

Studies in animal models and premature infants have shown that fluctuation in oxygenation is an important factor in the development of ROP [10 –12]. In this study, we sought to determine if fluctuation in SpO₂ or time spent in the target range, is associated with the severity of ROP, in the same subset of high risk ELBW infants.

2 Methods

In a prospective observational study all ELBW infants who were admitted to our Neonatal Intensive Care Unit (NICU) at Metro Health Medical Center (between December of 2011 and August of 2013) were enrolled.

Our inclusion criteria consisted of all ELBW infants (<1000 grams) who received oxygen during their NICU stay. Our exclusion criteria consisted of infants who died before their retina reached full vascularization on serial ophthalmological examinations, and infants who had significant congenital abnormalities.

All eligible infants were enrolled in the study at birth and their SpO₂ were recorded for the first four weeks of life or until they were weaned off oxygen, whichever came first. The SpO₂ values were directly downloaded from the patients’ bedside monitors via the electronic medical record (EMR; Epic software 2010) interface. SpO₂ values obtained from the patients’ bedside monitors (B850 Care Scape Monitor GE HealthCare) were passed to a Capsule Technologies’ Data Captor application on a swdatacaptor server (software datacaptor server). The Data Captor converted the data it received into an HL7 (Hospital Level 7) format. The HL7 messages from the Data Captor were sent to Orion Health’s Rhapsody Interface Engine. The interface engine filtered the data, allowing only the SpO₂ data to pass to a Visual Basic (VB) program. The VB program inserted the SpO₂ data into a Microsoft SQL database (MSSQLDB). Epic’s Clarity (Oracle) database was extracted daily for all current inpatients in the NICU, and the Epic data were joined to the MSSQLDB SpO₂ data by room, bed and time-frame. The resulting dataset was exported to an Excel sheet from where it was collected. Data gathered from the bedside monitors were averaged over a minute and downloaded every minute for a total of 1440 data-points/day.

Infants’ demographics and clinical characteristics were collected from the hospital’s EMRs. Birth weight, gestational age, gender, mode of delivery, race, Apgar scores at 1 and 5 minutes and severity of illness [Score for Neonatal Acute Physiology Perinatal Extension (SNAP-PE)] were retrieved from the EMRs. Also, the number of days of mechanical ventilation, days of oxygenation and length of stay were collected. Information regarding prenatal care, prenatal steroid use, number of blood transfusions, and number of erythropoietin doses, blood culture positive sepsis, and presence of symptomatic patent ductus arteriosus (PDA) were retrieved from the EMRs.

The target range of SpO₂ was kept at 90–95% as per policy, during the study period. A single ophthalmologist performed all retinal examinations over the study period. Infants were divided into two groups according to their highest ROP Staging (0–5) after multiple retinal examinations. One group consisted of infants with moderate to severe ROP (with stage 3, 4 or 5 with or without plus disease), and the other group served as control (infants with ROP stage 0, 1 or 2 without any plus disease). The number of retinal examinations, the highest stage of ROP, and number of laser therapies were documented.

Parametric data were expressed as means± standard deviations, non-parametric data were expressed as median and inter-quartile ranges. Categorical data were expressed as percentages. A t-test or a Mann Whitney U-test was used for comparison of parametric and non-parametric continuous data respectively. A chi-square test and a Fisher exact test were used to compare categorical data as appropriate. The weekly SpO₂ fluctuation was measured as SpO₂ coefficient of variation / week. The weekly coefficient of variation for a patient was measured as the standard deviation divided by the mean of SpO₂ values during a week. Analysis of Variance with repeated measures (General Linear Model) was used to compare the percentage of time spent at different SpO₂ ranges between the two groups of patients, and the SpO2 fluctuation over the study period. A co-variance analysis for SpO₂ fluctuation was used to adjust for birth weight and gestational age between the two groups. A p value < 0.05 was considered statistically significant.

The study was approved by the Institution Review Board at Metro Health Medical Center. The requirement for parental consent was waived by the Institutional Review Board of MetroHealth Medical Center.

3 Results

During the study period, Fifty-six infants met our inclusion criteria. Twenty one infants had moderate to severe ROP, and 35 infants served as controls.

Patients’ demographics are illustrated in Table 1. Infants with moderate to severe ROP were younger and smaller than their controls, and had a higher severity of illness than their counterparts.

Table 1
Patients’ demographics

Characteristics Moderate to Severe ROP (N = 21) Control (N = 35) P values

Birth weight (g) 676±124 796±148 <0.001

Gestational Age (wks) 24.0±1.0 25.0±1.7 <0.001

Gender (% male) 10/21 (48%) 17/35 (48%) 0.61

African American (%) 10/21 (48%) 27/35 (77%) 0.33

Caucasian (%) 6/21 (29%) 5/35 (14%) 0.98

Others (%) 5/21 (23%) 3/35 (9%) 0.55

C-section (%) 14/21 (67%) 15/35 (43%) 0.16

Apgar at 1 min 3 (2–5) 4 (3–6) 0.14

Apgar at 5 min 7 (4–8) 7 (6–8) 0.01

SNAP-PE scores 55 (41–61) 35 (29–51) <0.001

Characteristics	Moderate to Severe ROP (N = 21)	Control (N = 35)	P values
Birth weight (g)	676±124	796±148	<0.001
Gestational Age (wks)	24.0±1.0	25.0±1.7	<0.001
Gender (% male)	10/21 (48%)	17/35 (48%)	0.61
African American (%)	10/21 (48%)	27/35 (77%)	0.33
Caucasian (%)	6/21 (29%)	5/35 (14%)	0.98
Others (%)	5/21 (23%)	3/35 (9%)	0.55
C-section (%)	14/21 (67%)	15/35 (43%)	0.16
Apgar at 1 min	3 (2–5)	4 (3–6)	0.14
Apgar at 5 min	7 (4–8)	7 (6–8)	0.01
SNAP-PE scores	55 (41–61)	35 (29–51)	<0.001

Data expressed as means±standard deviations, medians with inter-quartile ranges, and as percentages. C-section: Cesarean section. SNAP-PE: Score for neonatal acute physiology perinatal extension.

Patients’ clinical characteristics are illustrated in Table 2. Infants with moderate to severe ROP had more days of mechanical ventilation, more days of oxygenation and higher length of stay than their controls. Also, infants with moderate to severe ROP had a higher number of eye exams and laser therapies than their controls.

Table 2

Patients’ Clinical Characteristics

	Moderate to Severe ROP (N = 21)	Control (N = 35)	P values
Received Prenatal Care	18/21 (86%)	29/35 (82%)	0.78
Prenatal Steroids	12/21 (57%)	22/35 (62%)	0.67
Days of ventilation	34 (22–41)	8 (2–30)	<0.01
Days of Oxygenation	70 (52–119)	63 (42–82)	0.04
Post Natal Steroids	10/21 (48%)	8/35 (23%)	0.05
Sepsis (Culture positive)	9/21 (43%)	13/35 (37%)	0.67
Symptomatic PDA	15/21 (71%)	20/35 (57%)	0.28
Length of stay	130 (115–185)	101 (83–123)	<0.01
Survival (%)	20/21 (96%)	35/35 (100%)	0.19
Number of eye exams	9 (6–12)	4 (3–6)	<0.01
Laser Therapy	12/21 (57%)	0/35 (0%)	<0.01

Data expressed as percentages and medians with inter-quartile ranges.

SpO₂ data points gathered from the bedside monitors were averaged over a minute and downloaded every minute for a total of 1440 data-points/day. Approximately 2 million data points were analyzed. On average, 57±2 minutes of data/hour/patient were downloaded. Three minutes of data loss every hour was due to the interface automatically discarding SpO₂ data-points that had pulsations which did not correlate with the EKG pulsations for≥15 seconds/minute during the recording time. This algorithm along with data averaging every minute helped to reduce significantly artifacts from loose sensors, excessive subject movements and other mechanical issues. Overall 207 weeks of data collection were recorded, with an average of 3.7±0.63 weeks per infant.

Overall there were no significant differences in the percentage of time spent at three different SpO₂ ranges (<90%, 90–95%, >95%) between the two groups during the first four weeks of life (Fig. 1). The % time spent with a SpO₂ <90% varied between 17.6±7.9% and 25.1±7.8% during the 1st and 4th week of life respectively in infants with moderate to severe ROP, vs. 13.1±7.1% and 20.8±10.9% during the 1st and 4th week of life respectively in the control group (Fig. 1A. GLM; p = 0.10).

The % time spent with a SpO₂ between 90–95% varied between 56.0±11.6% and 41.4±11.4% during the 1st and 4th week of life respectively in infants with moderate to severe ROP, vs. 50.5±16.9% and 41.1±10.2% during the 1st and 4th week of life respectively in the control group (Fig. 1B. GLM; p = 0.66).

Fig.1

The percent time spent at three different saturation of peripheral oxygen (SpO2) ranges (<90%, 90–95%, >95%) in infants with moderate to severe retinopathy of prematurity (ROP) and their controls during the first four weeks of life.

The % time spent with a SpO₂ >95% varied between 26.3±15.7% and 33.3±13.6% during the 1st and 4th week of life respectively in infants with moderate to severe ROP, vs. 36.3±22.6% and 38.1±17.8% during the 1st and 4th week of life respectively in the control group (Fig. 1C. GLM; p = 0.66).

There was a significant difference in SpO₂ fluctuation between infants with moderate to severe ROP and their controls during the first four weeks of life (Fig. 2). The percentage of SpO₂ CoV/week varied between 4.5±0.7% and 6.4±1.1% during the 1st and 4th week of life respectively in infants with moderate to severe ROP, vs. 4.1±0.7% and 5.6±1.4% during the 1st and 4th week of life respectively in the control group (GLM; p = 0.007). The difference in SpO₂ CoV/week between the two groups remained significant after adjusting for BW (GLM; p = 0.021).

Fig.2

Saturation of peripheral oxygen (SpO2) fluctuation expressed as % SpO2 coefficient of variation / week in infants with moderate to severe retinopathy of prematurity (ROP) and their controls during the first four weeks of life.

Twelve infants were treated with Laser therapy (21.4%). There were no differences in SpO₂ fluctuation between the group of patients who received Laser therapy and their counterparts (p = 0.98To stratify patients according to their need for oxygen requirement and or ventilatory support, the NICHD definition for BPD at 36 weeks corrected gestational age was used [13]. 100% (21/21) of infants with moderate to severe ROP had severe BPD (at 36 weeks of corrected gestational age according to the NICHD definition) versus only 14% (5/35) of the control group (p < 0.001). Also 100% (12/12) of all infants who had laser therapy had severe ROP versus only 4.5% (2/44) of infants who did not receive laser therapy (p < 0.001).

4 Discussion

We have shown that there are no significant differences in the % of time spent at different SpO₂ ranges between ELBW infants with moderate to severe ROP and their controls during their first four weeks of life. However, we have shown that there are significant differences in SpO₂ fluctuations between the two groups. ELBW infants with moderate to severe ROP have a higher % of SpO₂ fluctuation than their controls during their first four weeks of life. Several studies have shown the association between SpO₂ ranges and severity of ROP. To our knowledge, our study is the first to demonstrate the association between SpO₂ fluctuation and severity of ROP in ELBW infants.

Maintaining SpO₂ in a targeted range is essential to help prevent the development of ROP in premature infants. In a study of 571 premature infants less than 1,250 grams, Castillo et al., showed that a change in clinical practice by targeting a SpO₂ of 88–93% instead of greater than 93%, and use of better equipment led to a significant reduction in the incidence of severe ROP [14]. In another study, Sears et al. showed that changing the target SpO₂ from 95–100% to lower limits reduced the incidence of ROP from 35% to 13% [15]. Maintaining SpO₂ in a targeted range is also essential to improve survival in premature infants. In an international, randomized, controlled study, there was an increased incidence of mortality at 36 weeks post-menstrual age among premature infants whose SpO₂ was maintained between 85–89% versus 91–95% [16]. The SUPPORT Study Group Trial showed similar findings. In a randomized controlled study, lower SpO₂ was associated with an increased mortality but decreased rate of ROP among premature infants, though the difference was not significant [4]. However in the Canadian Oxygen Trial (COT), a multicenter randomized controlled trial of 1,201 premature infants between 230/7 and 276/7 weeks gestational age, there was no difference in death or severe disability at 18 months of age between the groups of infants who were exposed to a low or high SpO₂ targeted range [5]. In our study, the time spent in the target range was not significantly different between the two groups. This could have been due to the low % time spent in the target range of 90–95% and the difficulty to maintain infants at this range as we have previously shown [17]. Therefore we speculate that the time spent in the target range in our study was low enough to have little impact on the severity of ROP.

The % time spent in the target range of 90% to 95% SpO₂ in our study is consistent with previous studies. The % time spent in the SpO₂ target range (by manual control) has varied between 31% and 69% [18, 19]. In a multicenter study of infants less than 28 weeks GA, maintaining SpO₂ in the intended range varied significantly between centers from 16% to 64% [20]. In another randomized crossover trial, comparing the time spent in the target range by routine nursing care vs. algorithm-based management of oxygen delivery, Clarke et al. found that the % time spent in the target range was 34.6±28.5% during routine nursing care vs. 38.3±29.3% during algorithm-based management of oxygen delivery care [21]. In our study, the 46% time spent in the target range of 90–95% is consistent with previous reports.

In a multicenter randomized controlled, crossover clinical trial of 34 premature infants, Hallenberger et al., have shown that O₂ saturation was kept within the target range more often in infants managed with a closed-loop automatic O₂ control system than manual control (71% vs 61%) [22]. Future studies are needed to determine if SpO₂ fluctuation is also reduced with closed-loop automatic O₂ control systems and if the reduction of SpO₂ fluctuation will have any impact on ROP severity.

DiFiore et al., in their study of 79 preterm infants (gestational age, 24 to 27-6/7 weeks) during the first 8 weeks of life have shown that the incidence of hypoxemic events was higher in infants with ROP requiring laser therapy in comparison to infants with mild or no ROP [10]. Kornacka et al. also found statistically significant influence of drops of SaO₂ <85% and lower mean values of SaO₂ on development of severe retinopathy [23]. In our study, infants with moderate to severe ROP had more SpO₂ fluctuations and drop in their SpO₂ than their counterparts.

Previous studies have addressed the effect of oxygen fluctuation on ROP. Using frequent transcutaneous oxygen pressure (TcPO₂) recording in 18 extremely premature infants, Saito et al. showed that infants with wide fluctuation of their arterial oxygen tension (PO₂) had a greater risk of developing progressive ROP [11]. In rat pups, Penn et al., showed that 24 hour cycling of environmental oxygen of 10% to 50% had a greater effect on the severity of ROP than cycling of environmental oxygen of 40% to 80%. The group of animals exposed to cycles of 10% to 50% O₂ had a greater retardation of retinal blood vessel development than the group of animals exposed to 40% to 80% O₂, and their incidence of pre-retinal neovascularization was 97% and 72% respectively after 4 days of post-exposure to room air [12]. Wang et al., showed in their experiment on neonatal mice exposed to variable oxygen concentrations that an abrupt discontinuation of O₂ exposure lead to retinal changes consistent with ROP in comparison to the normal retina of a control group exposed only to room air. They also demonstrated that mice exposed to alternating oxygen and room air had significantly greater changes of retinopathy compared to the control group [24]. In our study we have shown a similar trend, whereas the fluctuation of SpO₂ is associated with an increased severity of ROP.

York et al., showed that VLBW infants experiencing fluctuating PaO₂ are at higher risk of threshold ROP. They studied 231 VLBW infants retrospectively and found that the Odds ratio of developing threshold ROP versus pre-threshold was 1.82 in the first 10 days and 1.68 from day 11–30 [25]. Cunningham et al. demonstrated that infants with severe ROP showed an increased variability of transcutaneous oxygen in week 1 and 2 but not week 3. They showed that variability of transcutaneous oxygen in the first 2 weeks of life is a significant predictor of severe ROP [26]. Poets et al. showed that in premature infants who survived till 36 weeks post menstrual age, prolonged hypoxemic episodes during the first 2 to 3 months after birth were associated with adverse 18-month outcomes [27]. The findings of our study is similar with the addition that given the same group of premature infants, fluctuations may affect the severity of ROP more than the time spent in above or below the target range of oxygenation.

Our study has limitations. Patients who had moderate to severe ROP were sicker at birth with a higher SNAP-PE scores, they were ventilated longer and had a higher rate of severe BPD than their counterparts. Therefore, the severity of their ROP illness could have been related to the underlying severity of their respiratory illness. An increase in oxygen and ventilation requirements along with SpO₂ fluctuations could have been responsible for the severity of their ROP. Our study was underpowered to adjust for multiple covariables. Future larger studies are needed to address potential confounders.

5 Conclusion

In ELBW infants, an increase in SpO₂ fluctuation is associated with an increase in the severity of ROP. For the same group of infants, SpO₂ fluctuation was significantly associated with the severity of ROP while time spent in SpO₂ ranges other than the target range was not. Future studies are needed to determine if SpO₂ fluctuation is a marker or a cause of the ROP severity of illness.

Conflict of interest

None documented.

Financial support

None.

References

National Vital Statistics Report, 2013;62(9).

Lau

, Ambalavanan

, Chakraborty

, Wingate

, Carlo

. Extremely low birthweight and infant mortality rates in the United States. Pediatrics. 2013;131(5):855–60.

Stoll

, Hansen

, Bell

, et al. Neonatal outcomes of extremely preterm infants from the NICHD Neonatal Research Network. Pediatrics. 2010;126(3):443–56.

Glass

, Costarino

, Stayer

, Brett

, Cladis

, Davis

. Outcomes for extremely premature infants. Anesth Analg. 2015;120(6):1337–51.

Saugstad

, Aune

. Optimal oxygenation of extremely low birth weight infants: A meta-analysis and systematic review of the oxygen saturation target studies. Neonatology. 2014;105(1):55–63.

SUPPORT Study Group of the Eunice Kennedy Shriver NICHD Neonatal Research Network, Carlo

, Finer

, et al. Target ranges of oxygen saturation in extremely preterm infants. N Engl J Med. 2010;362(21):1959–69.

Schmidt

, Whyte

, Asztalos

, et al. Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: A randomized clinical trial. JAMA. 2013;309(20):2111–20.

Sola

, Golombek

, Montes Bueno

, et al.

Safe oxygen saturation targeting and monitoring in preterm infants: Can we avoid hypoxia and hyperoxia?

Acta Paediatr. 2014;103(10):1009–18.

Lakshminrusimha

, Manja

, Mathew

, Suresh

. Oxygen targeting in preterm infants: A physiological interpretation. J Perinatol. 2015;35(1):8–15.

10.

Di Fiore

, Bloom

, Orge

, et al. A higher incidence of intermittent hypoxemic episodes is associated with severe retinopathy of prematurity. J Pediatr. 2010;157(1):69–73.

11.

Saito

, Omoto

, Cho

, Hatsukawa

, Fujimura

, Takeuchi

. The progression of retinopathy of prematurity and fluctuation in blood gas tension. Graefes Arch Clin Exp Ophthalmol. 1993;231(3):151–6.

12.

Penn

, Henry

, Tolman

. Exposure to alternating hypoxia and hyperoxia causes severe proliferative retinopathy in the newborn rat. Pediatr Res. 1994;36(6):724–31. Erratum in: Pediatr Res. 1995;37(3):353.

13.

Jobe

, Bancalari

. Bronchopulmonary dysplasia. Am J Respir Crit Care Med. 2001;163(7):1723–9.

14.

Castillo

, Deulofeut

, Critz

, Sola

. Prevention of retinopathy of prematurity in preterm infants through changes in clinical practice and SpO₂ technology. Acta Paediatr. 2011;100(2):188–92.

15.

Sears

, Pietz

, Sonnie

, Dolcini

, Hoppe

. A change in oxygen supplementation can decrease the incidence of retinopathy of prematurity. Ophthalmology. 2009;116(3):513–8.

16.

BOOST II United Kingdom Collaborative Group; BOOST II Australia Collaborative Group; BOOST II New Zealand Collaborative Group, et al. Oxygen saturation and outcomes in preterm infants. N Engl J Med. 2013;368(22):2094–104.

17.

Das

, Mhanna

, Teleron-Khorshad

, et al. A comparison of manual versus automated saturation of peripheral oxygenation in the neonatal intensive care unit. J Matern Fetal Neonatal Med. 2016;29(10):1631–5.

18.

Lim

, Wheeler

, Gale

, et al. Oxygen saturation targeting in preterm infants receiving continuous positive airway pressure. J Pediatr. 2014;164(4):730–736.e.

19.

Waitz

, Schmid

, Fuchs

, Mendler

, Dreyhaupt

, Hummler

. Effects of automated adjustment of the inspired oxygen on fluctuations of arterial and regional cerebral tissue oxygenation in preterm infants with frequent desaturations. J Pediatr. 2015;166(2):240–244.e.

20.

Hagadorn

, Furey

, Nghiem

, et al. Achieved versus intended pulse oximeter saturation in infants born less than 28 weeks’ gestation: The AVIOx study. Pediatrics. 2006;118(4):1574–82.

21.

Clarke

, Yeomans

, Elsayed

, et al. A randomised crossover trial of clinical algorithm for oxygen saturation targeting in preterm infants with frequent desaturation episodes. Neonatology. 2015;107(2):130–6.

22.

Hallenberger

, Poets

, Horn

, et al. Closed-loop automatic oxygen control (CLAC) in preterm infants: A randomized controlled trial. Pediatrics. 2014;133(2):e379–85.

23.

Kornacka

, Tupieka

, Czajka

, Gajewska

, Gołebiewska

. [Tissue and blood oxygenation in newborns <1250g and incidence of ROP]. Ginekol Pol. 2003;74(7):538–44.

24.

Wang

, Chen

, Shi

, Xiao

, Tong

, Zhou

. [Effects of different oxygen inhalation modes on retinal vessels development in neonatal mice]. Zhongguo Dang Dai Er Ke Za Zhi. 2006;8(2):129–32.

25.

York

, Landers

, Kirby

, Arbogast

, Penn

. Arterial oxygen fluctuation and retinopathy of prematurity in very-low-birth-weight infants. J Perinatol. 2004;24(2):82–7.

26.

Cunningham

, Fleck

, Elton

, McIntosh

. Transcutaneous oxygen levels in retinopathy of prematurity. Lancet. 1995;346(8988):1464–5.

27.

Poets

, Roberts

, Schmidt

, et al. Canadian Oxygen Trial Investigators. Association between Intermittent Hypoxemia or Bradycardia and Late Death or Disability in Extremely Preterm Infants. JAMA. 2015;314(6):595–603.