Video Abstract

Video Abstract

Close modal
OBJECTIVES:

To determine if a home oxygen therapy (HOT) management strategy that includes analysis of recorded home oximetry (RHO) data, compared with standard monthly clinic visit assessments, reduces duration of HOT without harm in premature infants.

METHODS:

The RHO trial was an unmasked randomized clinical trial conducted in 9 US medical centers from November 2013 to December 2017, with follow-up to February 2019. Preterm infants with birth gestation ≤37 + 0/7 weeks, discharged on HOT, and attending their first pulmonary visit were enrolled. The intervention was an analysis of transmitted RHO between clinic visits (n = 97); the standard-care group received monthly clinic visits with in-clinic weaning attempts (n = 99). The primary outcomes were the duration of HOT and parent-reported quality of life. There were 2 prespecified secondary safety outcomes: change in weight and adverse events within 6 months of HOT discontinuation.

RESULTS:

Among 196 randomly assigned infants (mean birth gestational age: 26.9 weeks; SD: 2.6 weeks; 37.8% female), 166 (84.7%) completed the trial. In the RHO group, the mean time to discontinue HOT was 78.1 days (SE: 6.4), compared with 100.1 days (SE: 8.0) in the standard-care group (P = .03). The quality-of-life scores improved from baseline to 3 months after discontinuation of HOT in both groups (P = .002), but the degree of improvement did not differ significantly between groups (P = .75).

CONCLUSIONS:

RHO was effective in reducing the duration of HOT in premature infants. Parent quality of life improved after discontinuation. RHO allows physicians to determine which infants can be weaned and which need prolonged oxygen therapy between monthly visits.

What’s Known on This Subject:

Current estimates indicate that up to one-third of premature infants diagnosed with bronchopulmonary dysplasia will require supplemental home oxygen therapy after NICU discharge. Home oxygen is an alternative to prolonged hospitalization, but its use is highly variable.

What This Study Adds:

The recorded home oximetry trial is the first randomized trial used to compare oxygen weaning strategies in premature infants on home oxygen. Use of home oximetry data can decrease duration of home oxygen and decrease adverse events compared with routine clinic assessments alone.

Improvements in neonatology have led to increased numbers of premature infants surviving with bronchopulmonary dysplasia (BPD), including thousands requiring supplemental oxygen (O2) after NICU discharge.1  Up to 28% of early preterm infants (birth gestational age [GA] 23–28 weeks) diagnosed with BPD will require home oxygen therapy (HOT).1,2  When used, HOT is an alternative to prolonged hospitalization, but its use is highly variable.35 

Obstacles to HOT use include lack of evidence-based guidelines for effective management.1,4,5  Commonly, pediatric pulmonologists rely on brief outpatient assessments of oxygenation status during monthly clinic visits, leading to arbitrary timing of weaning and discontinuation.6,7  This may expose patients to potential toxicities of prolonged oxygen exposure. Families have reported that prolonged HOT increases stress, so shortening duration may improve parent quality of life.810 

With availability of better oximeters and portable data storage capability, clinic-based weaning protocols could be enhanced by recorded home oximetry (RHO), allowing data-driven management. Without such data, some infants may be on oxygen longer than necessary, and others may have more episodes of intermittent hypoxemia.6,11 

This trial was designed to test the hypothesis that preterm infants requiring HOT would have HOT discontinued earlier by using RHO compared with infants who did not. The effects of RHO on family quality of life, growth, and respiratory outcomes were also analyzed.8,1214 

The RHO trial was an unblinded randomized trial in 9 US academic medical centers (November 2013 to December 2017, with follow-up through February 2019) (Supplemental Information). By design, randomization assignments were unblinded. Institutional review boards at each center approved the study.

Infants with birth GA ≤37 + 0/7 weeks with an O2 requirement at the time of NICU discharge and attending their first outpatient pulmonary visit were considered eligible. Infants with echocardiography-confirmed pulmonary hypertension (screening at physician discretion), a known diagnosis with a high risk for persistent hypoxia (cardiac disease, etc); infants with an O2 requirement of >1 L/minute; infants with laryngomalacia; infants with tracheomalacia; and infants on caffeine were excluded. Written informed consent was obtained from each infant’s parent or legal guardian.

Groups were assigned by using a computer-generated sequence of randomly permuted blocks (sizes 2 and 4). Separate sequences were generated for each site for a balanced distribution of groups. Assignments were concealed in envelopes and opened in order at the time of enrollment. Multiples were randomly assigned as individual patients.

Both groups were instructed to attend monthly clinic visits, during which the site principal investigator determined whether to increase, maintain, or decrease oxygen flow rates. The decision was based on a structured algorithm identical for infants in both groups that consisted of 20-minute challenges to ensure infants could maintain saturations >93% on the next decrement of O2. Infants who maintained saturations were weaned to an oxygen flow rate that was 50% of the previous rate. Infants who demonstrated a readiness to be weaned from 125 mL/minute were weaned to nocturnal oxygen at 125 cc/minute and room air during awake hours.

Infants in the standard-care group underwent an inpatient overnight polysomnography with simultaneous study recordings to determine eligibility to wean off HOT entirely. The physician made the final decision regarding safe discontinuation of HOT using only the polysomnography results. Infants who did not meet the criteria to wean remained on nocturnal oxygen and repeated the test monthly until they passed.

The intervention group received a recording oximeter and specific instructions to use the oximeter continuously as often as possible, but for a minimum of 25 hours per report, and to download or transmit data as often as every 4 to 7 days. Data were sent by families to the data coordinating center, University of Massachusetts Memorial Medical Center, via secure e-mail or via the provided express shipping envelopes. RHO was analyzed between monthly visits by using a structured algorithm, determined by a consensus panel of experts, to determine whether to increase, decrease, or maintain oxygen (Supplemental Information), and recommendations were communicated to the physician within 24 hours of receipt. If the data were deemed to be inadequate, parents were notified, and no change was made until adequate recordings were received.

Primary outcomes were (1) total duration of HOT (NICU discharge to successful HOT discontinuation) and (2) parent quality of life, assessed by differences in postweaning Pediatric Quality of Life Inventory (PedsQL) version 2 scores1519  from baseline. Secondary measures included (1) adverse events (AEs) and (2) change in weight and weight-for-length z score, both measured from enrollment to 6 months after discontinuation.

AEs included, but were not limited to, viral respiratory infection, gastrostomy tube complications, and feeding intolerance requiring intervention or hospital admission. A data and safety monitoring board reviewed all events and determined the appropriate classifications. The growth data were recorded as z scores on the basis of sex-specific World Health Organization growth charts for each clinic visit while on HOT and at the 1- and 6-month postdiscontinuation visits.

The trial was designed to have 97% power at a type I error rate of 5% to detect a 25% intervention effect on duration of HOT, 80% power to detect a 20% absolute difference in change in PedsQL scales after discontinuation of HOT, and 93% power to detect an excess in AE rates beyond the 10% margin of noninferiority.

HOT duration was log-transformed to reduce skew and was compared between groups by using a general linear model accounting for site stratification, covariates, extreme values, and missing data.

As a secondary, nonparametric analysis accounting for right censoring, we constructed Kaplan-Meier time-to-event curves and compared study groups using the log-rank test.

We used multiple regression to model the joint influence of parents’ dose of recording activity (reports per month for RHO: 0 for standard of care) and infants’ oxygen requirements at discharge (4 levels) on log duration of HOT.

In analyses of log duration, adjusted mean log times from the fitted model and their SEs (µ ± ε) were retransformed to natural time units (exp[µ] ± exp[µ] × [exp(ε) − 1]). Group differences (Δ) were converted to percentages (100% × [exp(Δ) − 1]). Similar conversions were made for coefficients of dose (β) and 95% confidence limits.

For PedsQL, the quantity of missing data at 1 or both measurement times led us to assess change by using 2-point repeated measures analysis to make maximal use of available data.

For growth data, we employed a mixed-effects linear model that included time, discrete effects of the study group and period (during or after HOT), and a group × period interaction to allow for different rates of growth before and after weaning.

We compared AE rates between study groups using the Farrington-Manning test20  for noninferiority.

In each parametric analysis, we stratified the model by site, employed robust regression with M-estimation21  to reduce influence of extreme values, and weighted data in inverse proportion to probability of nonmissingness, as estimated by a logistic regression model constructed from those characteristics that differed between participants with missing or nonmissing outcomes.22 

For all analyses, we followed the intention-to-treat principle, analyzing each participant as randomly assigned. SAS software version 9.4 (SAS Institute, Inc, Cary, NC) was used for all computations. A 2-sided P value ≤.05 was interpreted as a statistically significant result.

Two hundred fifty-three infants requiring HOT were identified at the time of NICU discharge or the first pulmonary visit. Of these, 223 families were approached for participation (Fig 1). The remaining 30 infants were excluded, most commonly for pulmonary hypertension. One hundred ninety-six infants were randomly assigned: 99 infants to standard care and 97 infants to RHO. A total of 172 infants were weaned successfully from HOT, and 144 (73%) completed all study procedures through 6 months after discontinuation. Table 1 reveals characteristics of the final cohort. Approximately two-thirds of the infants were male, reflecting the population requiring HOT. Mean birth GA was 26.9 weeks (SD: ±2.6 weeks), and mean birth weight was 934 g (SD: ±440 g). There were 9 pairs of twins who were randomly assigned independently and analyzed without accounting for within-pair correlation. The resulting inflation factor for the SEs of estimate (design effect) was estimated to lie between 1.5% and 4.5% and was judged to be negligible.

FIGURE 1

Flow diagram of recruitment for the RHO trial. a Infants for whom consent was either withdrawn by parents or the principal investigator or infants who were lost to follow-up but still had data collected for primary aim 2 and secondary outcomes were included in thos outcome’s analysis. Ninety infants in the standard-of-care arm and 80 infants in the RHO arm who completed at least 1 baseline PedsQL assessment were included in the analysis. After weaning, 52 infants in the standard-of-care arm and 56 infants in the RHO arm were included in the comparison analysis. Nonresponding parents did not differ significantly from respondents. We had data for 87 and 72 infants in the standard-of-care arm and RHO arm, respectively, during the O2 weaning process and 80 and 73 infants after O2 discontinuation. The subsamples did not differ significantly from the full sample in sex, race, ethnicity, birth weight, or GA at birth.

FIGURE 1

Flow diagram of recruitment for the RHO trial. a Infants for whom consent was either withdrawn by parents or the principal investigator or infants who were lost to follow-up but still had data collected for primary aim 2 and secondary outcomes were included in thos outcome’s analysis. Ninety infants in the standard-of-care arm and 80 infants in the RHO arm who completed at least 1 baseline PedsQL assessment were included in the analysis. After weaning, 52 infants in the standard-of-care arm and 56 infants in the RHO arm were included in the comparison analysis. Nonresponding parents did not differ significantly from respondents. We had data for 87 and 72 infants in the standard-of-care arm and RHO arm, respectively, during the O2 weaning process and 80 and 73 infants after O2 discontinuation. The subsamples did not differ significantly from the full sample in sex, race, ethnicity, birth weight, or GA at birth.

Close modal
TABLE 1

Baseline Demographic and Clinical Characteristics of the Randomly Assigned Sample for the RHO Trial

Intervention (N = 97)Standard Care (N = 99)
Sex, n (%)   
 Male 64 (66) 58 (59) 
 Female 33 (34) 41 (41) 
Race, n (%)   
 White 59 (60) 57 (57) 
 African American 13 (13) 12 (12) 
 Asian American 4 (4) 1 (1) 
 Other 21 (22) 29 (29) 
 Hispanic 6 (6) 12 (12) 
Multiplicity, n (%)   
 Singleton 85 (88) 93 (94) 
 Twin 12 (12) 6 (6) 
 Very low birth wt (<33 wk, <1500 g) 82 (92)a 84 (90)a 
Respiratory support (36 wk), n (%)   
 Ventilator 6 (6) 3 (3) 
 CPAP or high-flow O2 48 (49) 52 (53) 
 Low-flow nasal cannula 32 (33) 36 (36) 
 Room air 5 (5) 2 (2) 
 Unknown 6 (6) 6 (6) 
Diuretics, n (%)   
 Yes 38 (39) 41 (41) 
 No 51 (53) 52 (53) 
 Unknown 8 (8) 6 (6) 
Initial O2 requirement, n (%)   
 ≤125 mL/min 61 (65)b 54 (56)b 
 250 mL/min 16 (17) 30 (31) 
 ≥500 mL/min 16 (17) 11 (11) 
 Nocturnal 1 (1) 1 (1) 
Birth wt, mean ± SD, g (n929 ± 443 (89) 938 ± 439 (93) 
GA, mean ± SD, wk, (n26.9 ± 2.7 (97) 26.8 ± 2.5 (98) 
Length of stay in neonatal intensive care, mean ± SD, d, (n104 ± 37 (88) 98 ± 33 (94) 
CGA at discharge, mean ± SD, wk, (n42.0 ± 4.8 (92) 40.9 ± 3.2 (95) 
NICU days on ventilator, mean ± SD, (n25 ± 22 (83) 24 ± 25 (83) 
NICU days on CPAP or high-flow O2, mean ± SD, (n39 ± 21 (71) 39 ± 22 (76) 
NICU days on low-flow nasal cannula, mean ± SD, (n28 ± 24 (61) 24 ± 18 (68) 
Intervention (N = 97)Standard Care (N = 99)
Sex, n (%)   
 Male 64 (66) 58 (59) 
 Female 33 (34) 41 (41) 
Race, n (%)   
 White 59 (60) 57 (57) 
 African American 13 (13) 12 (12) 
 Asian American 4 (4) 1 (1) 
 Other 21 (22) 29 (29) 
 Hispanic 6 (6) 12 (12) 
Multiplicity, n (%)   
 Singleton 85 (88) 93 (94) 
 Twin 12 (12) 6 (6) 
 Very low birth wt (<33 wk, <1500 g) 82 (92)a 84 (90)a 
Respiratory support (36 wk), n (%)   
 Ventilator 6 (6) 3 (3) 
 CPAP or high-flow O2 48 (49) 52 (53) 
 Low-flow nasal cannula 32 (33) 36 (36) 
 Room air 5 (5) 2 (2) 
 Unknown 6 (6) 6 (6) 
Diuretics, n (%)   
 Yes 38 (39) 41 (41) 
 No 51 (53) 52 (53) 
 Unknown 8 (8) 6 (6) 
Initial O2 requirement, n (%)   
 ≤125 mL/min 61 (65)b 54 (56)b 
 250 mL/min 16 (17) 30 (31) 
 ≥500 mL/min 16 (17) 11 (11) 
 Nocturnal 1 (1) 1 (1) 
Birth wt, mean ± SD, g (n929 ± 443 (89) 938 ± 439 (93) 
GA, mean ± SD, wk, (n26.9 ± 2.7 (97) 26.8 ± 2.5 (98) 
Length of stay in neonatal intensive care, mean ± SD, d, (n104 ± 37 (88) 98 ± 33 (94) 
CGA at discharge, mean ± SD, wk, (n42.0 ± 4.8 (92) 40.9 ± 3.2 (95) 
NICU days on ventilator, mean ± SD, (n25 ± 22 (83) 24 ± 25 (83) 
NICU days on CPAP or high-flow O2, mean ± SD, (n39 ± 21 (71) 39 ± 22 (76) 
NICU days on low-flow nasal cannula, mean ± SD, (n28 ± 24 (61) 24 ± 18 (68) 
a

Of 89 intervention patients and 93 standard-care patients.

b

Of 94 intervention patients and 96 standard-care patients.

Time to discontinue HOT was 22% shorter in infants in the RHO group(retransformed mean ± SE: 78.1 ± 6.4 days) than in infants in the standard-care group (100.1 ± 8.0 days; P = .03) (Table 2). The respective median times were 71 and 90 days. Adjustment for very low birth weight or corrected gestational age (CGA) at NICU discharge did not affect this difference. Adjustment for respiratory support days during the NICU stay (mechanical ventilation, continuous positive airway pressure [CPAP], or high-flow and low-flow nasal cannula O2, covariates indicative of severity of illness) attenuated the difference between study groups, with discontinuation time in infants in the RHO group shorter by 11% to 18% (.08 < P < .40).

TABLE 2

Duration of HOT, as Influenced by Study Group and Initial Oxygen Requirement

InfluenceDuration, Adjusted Mean ± SE, daPb
Study group   
 Intervention 78.1 ± 6.4 .03 
 Standard care 100.1 ± 8.0 — 
Initial O2   
 ≤125 mL/min 73 ± 5 <.0001 
 250 mL/min 98 ± 11 — 
 ≥500 mL/min 149 ± 21 — 
 Nocturnal 80 ± 46 — 
InfluenceDuration, Adjusted Mean ± SE, daPb
Study group   
 Intervention 78.1 ± 6.4 .03 
 Standard care 100.1 ± 8.0 — 
Initial O2   
 ≤125 mL/min 73 ± 5 <.0001 
 250 mL/min 98 ± 11 — 
 ≥500 mL/min 149 ± 21 — 
 Nocturnal 80 ± 46 — 

—, not applicable.

a

Retransformed from log scale.

b

Testing for equal means or for 0 rate of change.

The Kaplan-Meier time-to-event curves (Fig 2) revealed onset of weaning beginning in the intervention group at 15 days after NICU discharge, compared with 37 days in standard care. The Kaplan-Meier estimate of percentage weaned in the intervention group remained above that in standard care up to 90 days, at which curves crossed. The log-rank test indicated that the curves did not differ significantly (χ2 = 0.09; P = .76).

FIGURE 2

Kaplan-Meier time-to-event curves: onset of weaning.

FIGURE 2

Kaplan-Meier time-to-event curves: onset of weaning.

Close modal

Duration of HOT was shorter for those participants making more frequent reports (10.3% shorter duration for each additional report provided per month [Supplemental Figure 4]; 95% confidence interval [CI]: −15.2% to −5.3%; P = .0001) and for infants with lower initial O2 requirements (Table 2).

Parents of 170 infants completed at least 1 preweaning quality-of-life survey (baseline), and among these, 108 (55% of the randomly assigned sample) completed 1 survey after HOT discontinuation. Nonresponding parents did not differ significantly from respondents regarding infants’ demographic and clinical characteristics, except for a 1.3-week lower median birth GA, a 110-g lower birth weight, and a 16-day longer NICU stay. Among respondents, none of the infants’ demographic and clinical characteristics differed between study groups (Supplemental Table 5).

Baseline PedsQL scores did not differ significantly between groups (Table 3). In both groups, scores improved significantly from baseline to 3 months after discontinuation of HOT (mean change ± SE: 4.8 ± 1.3 scale points in intervention infants [P = .0004]; mean change ± SE: 4.2 ± 1.3 scale points in standard-care infants [P = .002]). The degree of improvement did not differ significantly (difference: 0.6 [95% CI: −3.1 to 4.3]; P = .75). The frequency of the parents’ oximetry reports, irrespective of the study group, did not affect these results, nor did weighting data with inverse probabilities on the basis of the characteristics differing between responders and nonresponders (GA, birth weight, and length of stay).

TABLE 3

Parents’ Quality-of-Life Scores (PedsQL) During and After Weaning

Preweaning3-mo After WeaningChange
nMean ± SEnMean ± SEEstimate (95% CI)Pa
Intervention 80 72.6 ± 1.9 56 77.4 ± 2.1 4.8 (2.2 to 7.4) .0004 
Standard care 90 72.3 ± 1.8 52 76.5 ± 2.1 4.2 (1.6 to 6.8) .002 
Difference (P— 0.3 ± 2.6 (.91) — 0.9 ± 2.9 (.77) 0.6 (−3.1 to 4.3) .75 
Preweaning3-mo After WeaningChange
nMean ± SEnMean ± SEEstimate (95% CI)Pa
Intervention 80 72.6 ± 1.9 56 77.4 ± 2.1 4.8 (2.2 to 7.4) .0004 
Standard care 90 72.3 ± 1.8 52 76.5 ± 2.1 4.2 (1.6 to 6.8) .002 
Difference (P— 0.3 ± 2.6 (.91) — 0.9 ± 2.9 (.77) 0.6 (−3.1 to 4.3) .75 

—, not applicable.

a

P tests for significant differences between groups or significant change over time.

We obtained growth data for 91 standard-care infants and 81 intervention-group infants during HOT weaning. At postdiscontinuation visits (1 and 6 months combined), we obtained growth data for 93 standard-care infants and 96 intervention-group infants. The subsamples with and without growth data did not differ significantly with respect to demographic and clinical characteristics, except for a 1.3-week lower GA, a 100-g lower birth weight, and a 17-day longer stay in the NICU.

Over the course of HOT, infants randomly assigned to standard care experienced a decrease in the weight z score (estimate ±SE: −0.045 ± 0.018 U per month; P = .02), whereas intervention participants showed a nonsignificant change (estimate ±SE: 0.004 ± 0.020 per month; P = .82) (Fig 3).

FIGURE 3

Weight and weight-for-length growth before and after weaning.

FIGURE 3

Weight and weight-for-length growth before and after weaning.

Close modal

After discontinuation of HOT, intervention participants grew significantly, with an average increase of 0.045 ± 0.017 in weight z score per month (P = .008), whereas the standard-care group showed nonsignificant change (0.015 ± 0.015 per month; P = .34).

The weight-for-length z score revealed significant decreases in both groups during HOT (−0.201 ± 0.037 per month for intervention infants; −0.161 ± 0.033 per month for standard-care infants; P < .0001 for both). This decrease continued after weaning for participants in the standard-care group (−0.067 ± 0.027 per month; P = .01), whereas the weight-for-length z score leveled for intervention participants (−0.021 ± 0.028 per month; P = .46).

The growth results were not substantially affected by weighting the data with inverse probabilities on the basis of the characteristics differing between responders and nonresponders (GA, birth weight, length of stay).

In standard-of-care infants, 30 of 99 (30%) had a serious AE, defined as requiring ICU hospitalization or intubation or as being life-threatening, compared with 19 of 97 RHO participants (20%). No AEs were deemed as being related to the study protocol. One death was reported in the standard-of-care group during the follow-up period. The cause of death was adenovirus and postinfectious bronchiolitis obliterans. There was no significant difference between groups in incidence of total events, emergency department visits, serious AEs, or other categories of AEs (Table 4).

TABLE 4

AEs and Serious AEs From NICU Discharge to 6 Months After Oxygen Discontinuation

Event TypeIntervention (N = 97), n (%)Standard Care (N = 99), n (%)Pa
All events 25 (26) 42 (42) <.0001 
Serious AEs 19 (20) 30 (30) .0005 
Life-threatening, intensive care admission, or intubation (grade 4) 6 (6) 8 (8) .002 
Hospitalizationb 19 (20) 32 (32) .0002 
Intervention without hospitalization drug or nondrug therapyc 25 (26) 40 (40) .0001 
Emergency department visits 17 (18) 19 (19) .02 
October through April (respiratory virus season) 17 (18) 31 (31) <.0001 
May through September 15 (15) 26 (26) .0002 
Respiratory-related events 20 (21) 27 (27) .003 
Nonrespiratory-related events 14 (14) 26 (26) .0001 
Event TypeIntervention (N = 97), n (%)Standard Care (N = 99), n (%)Pa
All events 25 (26) 42 (42) <.0001 
Serious AEs 19 (20) 30 (30) .0005 
Life-threatening, intensive care admission, or intubation (grade 4) 6 (6) 8 (8) .002 
Hospitalizationb 19 (20) 32 (32) .0002 
Intervention without hospitalization drug or nondrug therapyc 25 (26) 40 (40) .0001 
Emergency department visits 17 (18) 19 (19) .02 
October through April (respiratory virus season) 17 (18) 31 (31) <.0001 
May through September 15 (15) 26 (26) .0002 
Respiratory-related events 20 (21) 27 (27) .003 
Nonrespiratory-related events 14 (14) 26 (26) .0001 
a

Small P value from Farrington-Manning noninferiority test indicates intervention group did not incur a significantly worse AE rate than the standard-care group by the prespecified margin of 10%.

b

Hospitalization (grade 3+) includes grade 4 events.

c

Intervention without hospitalization drug or nondrug therapy (grade 2+) includes grade 3.

The RHO trial represents a novel approach to HOT weaning in infants with BPD, revealing the effectiveness of these 2 distinct HOT management protocols. The use of RHO shortened HOT duration in a dose-dependent manner; data transmission frequency was inversely correlated with decreased HOT duration. It is likely that in the absence of RHO to inform HOT management, pediatric pulmonologists err on the side of extending HOT to minimize risk, but infants are often ready to be weaned between visits. RHO allowed for earlier identification of this readiness. The use of RHO did not reduce family quality of life, and earlier HOT discontinuation did not result in adverse growth for RHO infants.

The authors of previous retrospective studies have described HOT durations varying up to 18 months’ CGA.1,23  Infants in this trial demonstrated significantly shorter durations of HOT.13,24  For standard-care infants, the average duration was ∼3 months from NICU discharge; for RHO infants, the duration was only 2.5 months. The health status of the RHO trial’s cohort is similar to that of previous cohorts and is unlikely to have impacted results. The exclusion of infants with persistent pulmonary hypertension of the newborn, laryngomalacia, and tracheomalacia may have played a role because these diagnoses likely increase duration of HOT and predispose infants to more respiratory exacerbations.

This cohort, one of the largest of infants on HOT, represented multiple centers and is generalizable for infants with BPD on HOT. Previous studies were purely descriptive, and the authors of those studies did not use standardized protocols to manage HOT.1,6  Simply using a standardized protocol, regardless of specific O2 threshold parameters, may allow for earlier discontinuation of HOT. Either of the management strategies described seem to allow for shorter HOT duration.

No standardized guideline for determining eligibility for discharge on HOT was followed. Infants in the RHO group had a slightly longer length of stay compared with the standard-care infants. With safe and effective outpatient management strategies, using RHO, more neonatologists may choose to discharge infants sooner on HOT if an O2 requirement is the final discharge-delaying factor.

In this trial, we used consensus guidelines from pediatric pulmonologists, primary care physicians, and neonatologists to determine when to wean, increase, or maintain HOT. These guidelines are the first to set standard oxygen threshold targets for weaning low-flow nasal cannula. The specific O2 saturation targets used to adjust flow rates may have allowed for shorter duration of HOT. The optimal saturation targets remain controversial in neonatology, but the guidelines used in this protocol are consistent with the published guidelines.11,2532  Infants in both groups had relatively fewer AEs compared with those in previous reports, suggesting that the thresholds used were not unsafe.4,23,24  Previous trials have revealed that higher oxygen targets, leading to prolonged oxygen use, may contribute to pulmonary toxicity.33  This may explain the slightly higher AE rate in the standard-care cohort, whose HOT duration was longer. The guidelines used are easily generalizable and could potentially be used in both inpatient and outpatient settings to manage low-flow nasal cannula.

Family-centered outcomes were also examined. There was no significant difference in parent-reported quality of life between groups. Having a child on HOT creates substantial restrictions on daily household activities and personal interactions with family and friends.4,810  Additional research into parental preference for HOT is necessary, but our findings suggest that the process of transmitting data is not more stressful.

Authors of some studies have suggested that the use of HOT can improve growth; as a corollary, early discontinuation of HOT may result in poor growth.12,23,34  Our results were reassuring, revealing that RHO infants were able to maintain weight while being weaned and were able to grow appropriately and significantly better than standard-care infants. Decreases in weight z scores during HOT in standard-care infants may have been due to clinically inapparent hypoxemia. Alternatively, despite adequate O2 levels, increased pulmonary effort may have a greater effect on growth. However, the extended duration of HOT in these patients suggests that they were on HOT longer than needed. Alternatively, families of RHO infants may have been more actively engaged in their HOT management and, therefore, more attentive to their infant’s nutritional needs. The lack of weight loss in both groups after HOT discontinuation is reassuring that none were weaned too early. Alternatively, weight and oxygen status may not be as tightly correlated as previously suggested.

The use of HOT in either of the structured protocols described are safe. Evidence of the safety of both protocols may allow earlier NICU discharge for infants unable to be weaned to room air. Infants randomly assigned to RHO experienced fewer AEs and better growth. The shorter duration of HOT may have led to decreased AEs because of less oxygen toxicity because previous studies have revealed that exposure to higher levels of O2 may cause respiratory exacerbations.33  More likely, RHO parents may have been more engaged in their infant’s care and able to address issues before they progressed to AEs.

This study had several potential limitations. First, in the absence of verified O2 saturation guidelines, a consensus guideline was used to adjust flow rates. Different parameters may have led to different results. Initial feedback on the algorithm suggested that thresholds were conservative and may prolong HOT. Instead, this algorithm resulted in shorter HOT durations.

Second, several families in both groups self-weaned without physician guidance, but the distribution of these families was similar between groups. BPD status could not be determined in 6% of infants because of missing or incomplete medical records from the birth hospital.

The study included patients with a range of illness severity, the persistent effects of which may account for the attenuated RHO effect when adjusted for duration of ventilatory support during the NICU stay. We excluded patients with confounding causes of prolonged hypoxemia, besides prematurity, that might affect HOT duration. This large cohort of premature infants on HOT included multiple centers and infants with diverse illness severity, so the findings should be generalizable for patients with BPD on HOT.

RHO is an effective protocol to manage HOT in premature infants. Increased reporting of RHO further shortened the duration. RHO use is associated with fewer AEs and did not influence parent-reported quality of life.

Dr Rhein conceptualized and designed the study, drafted the initial manuscript, and reviewed and revised the manuscript; Ms White coordinated and supervised data collection, monitored protocol initiation and compliance, drafted the initial manuscript, and reviewed and revised the final manuscript; Drs Simoneau, Traeger, Lahiri, Kremer, Sheils, Meyer, Rosenkrantz, Krishnan, Hartman, and Abu Jawdeh coordinated and supervised data collection at their clinical sites and critically reviewed the manuscript; Dr Feldman conducted the data analysis and drafted and critically revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

This trial has been registered at www.clinicaltrials.gov (identifier NCT01994954).

FUNDING: Funded by the Patient-Centered Outcomes Research Institute (CER-1402-10292) and by the Boston Children’s Hospital Provider-Payor Quality Initiative.

AE

adverse event

BPD

bronchopulmonary dysplasia

CGA

corrected gestational age

CI

confidence interval

CPAP

continuous positive airway pressure

GA

gestational age

HOT

home oxygen therapy

O2

supplemental oxygen

PedsQL

Pediatric Quality of Life Inventory

RHO

recorded home oximetry

1
Lagatta
JM
,
Clark
RH
,
Brousseau
DC
,
Hoffmann
RG
,
Spitzer
AR
.
Varying patterns of home oxygen use in infants at 23-43 weeks’ gestation discharged from United States neonatal intensive care units
.
J Pediatr
.
2013
;
163
(
4
):
976
982.e2
2
Yeh
J
,
McGrath-Morrow
SA
,
Collaco
JM
.
Oxygen weaning after hospital discharge in children with bronchopulmonary dysplasia
.
Pediatr Pulmonol
.
2016
;
51
(
11
):
1206
1211
3
MacLean
JE
,
Fitzgerald
DA
.
A rational approach to home oxygen use in infants and children
.
Paediatr Respir Rev
.
2006
;
7
(
3
):
215
222
4
Lagatta
J
,
Clark
R
,
Spitzer
A
.
Clinical predictors and institutional variation in home oxygen use in preterm infants
.
J Pediatr
.
2012
;
160
(
2
):
232
238
5
Halbower
AC
,
McGrath
SA
.
Home oxygen therapy: the jury is still in session
.
J Perinatol
.
2004
;
24
(
2
):
59
61
6
Palm
K
,
Simoneau
T
,
Sawicki
G
,
Rhein
L
.
Assessment of current strategies for weaning premature infants from supplemental oxygen in the outpatient setting
.
Adv Neonatal Care
.
2011
;
11
(
5
):
349
356
7
Simoes
EA
,
Rosenberg
AA
,
King
SJ
,
Groothuis
JR
.
Room air challenge: prediction for successful weaning of oxygen-dependent infants
.
J Perinatol
.
1997
;
17
(
2
):
125
129
8
Zanardo
V
,
Freato
F
.
Home oxygen therapy in infants with bronchopulmonary dysplasia: assessment of parental anxiety
.
Early Hum Dev
.
2001
;
65
(
1
):
39
46
9
Manns
SV
.
Life after the NNU: the long term effects on mothers’ lives, managing a child at home with broncho-pulmonary dysplasia and on home oxygen
.
Neuro Endocrinol Lett
.
2004
;
25
(
suppl 1
):
127
132
10
McLean
A
,
Townsend
A
,
Clark
J
, et al
.
Quality of life of mothers and families caring for preterm infants requiring home oxygen therapy: a brief report
.
J Paediatr Child Health
.
2000
;
36
(
5
):
440
444
11
Balfour-Lynn
IM
,
Field
DJ
,
Gringras
P
, et al;
Paediatric Section of the Home Oxygen Guideline Development Group of the BTS Standards of Care Committee
.
BTS guidelines for home oxygen in children
.
Thorax
.
2009
;
64
(
suppl 2
):
ii1
-
ii26
12
Groothuis
JR
,
Rosenberg
AA
.
Home oxygen promotes weight gain in infants with bronchopulmonary dysplasia
.
Am J Dis Child
.
1987
;
141
(
9
):
992
995
13
Procaskey
A
,
White
H
,
Simoneau
T
, et al
.
The optimization of home oxygen weaning in premature infants trial: design, rationale, methods, and lessons learned
.
Contemp Clin Trials
.
2018
;
75
:
72
77
14
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
15
Stein
R
,
Jessop
DJ
.
Tables Documenting the Psychometric Properties of a Measure of the Impact of Chronic Illness on a Family
.
Bronx, NY
:
Albert Einstein College of Medicine of Yeshiva University
;
1985
16
Stein
RE
,
Jessop
DJ
.
The impact on family scale revisited: further psychometric data
.
J Dev Behav Pediatr
.
2003
;
24
(
1
):
9
16
17
Stein
RE
,
Riessman
CK
.
The development of an impact-on-family scale: preliminary findings
.
Med Care
.
1980
;
18
(
4
):
465
472
18
Varni
JW
,
Sherman
SA
,
Burwinkle
TM
,
Dickinson
PE
,
Dixon
P
.
The PedsQL family impact module: preliminary reliability and validity
.
Health Qual Life Outcomes
.
2004
;
2
:
55
19
Varni
JW
,
Beaujean
AA
,
Limbers
CA
.
Factorial invariance of pediatric patient self-reported fatigue across age and gender: a multigroup confirmatory factor analysis approach utilizing the PedsQLTM multidimensional fatigue scale
.
Qual Life Res
.
2013
;
22
(
9
):
2581
2594
20
Farrington
CP
,
Manning
G
.
Test statistics and sample size formulae for comparative binomial trials with null hypothesis of non-zero risk difference or non-unity relative risk
.
Stat Med
.
1990
;
9
(
12
):
1447
1454
21
Holland
PW
,
Welsch
RE
.
Robust regression using iteratively reweighted least-squares
.
Commun Stat Theory Methods
.
1977
;
6
(
9
):
813
827
22
Seaman
SR
,
White
IR
.
Review of inverse probability weighting for dealing with missing data
.
Stat Methods Med Res
.
2013
;
22
(
3
):
278
295
23
DeMauro
SB
,
Jensen
EA
,
Bann
CM
, et al
.
Home oxygen and 2-year outcomes of preterm infants with bronchopulmonary dysplasia
.
Pediatrics
.
2019
;
143
(
5
):
e20182956
24
Pinney
MA
,
Cotton
EK
.
Home management of bronchopulmonary dysplasia
.
Pediatrics
.
1976
;
58
(
6
):
856
859
25
Silva
DT
,
Hagan
R
,
Sly
PD
.
Home oxygen management of neonatal chronic lung disease in Western Australia
.
J Paediatr Child Health
.
1995
;
31
(
3
):
185
188
26
Rhein
L
,
Simoneau
T
,
Davis
J
, et al
.
Reference values of nocturnal oxygenation for use in outpatient oxygen weaning protocols in premature infants
.
Pediatr Pulmonol
.
2012
;
47
(
5
):
453
459
27
Poets
CF
,
Stebbens
VA
,
Southall
DP
.
Arterial oxygen saturation and breathing movements during the first year of life
.
J Dev Physiol
.
1991
;
15
(
6
):
341
345
28
Urschitz
MS
,
Wolff
J
,
Von Einem
V
,
Urschitz-Duprat
PM
,
Schlaud
M
,
Poets
CF
.
Reference values for nocturnal home pulse oximetry during sleep in primary school children
.
Chest
.
2003
;
123
(
1
):
96
101
29
Hayes
D
 Jr.
,
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
30
Fitzgerald
DA
,
Massie
RJ
,
Nixon
GM
, et al;
Thoracic Society of Australia and New Zealand
.
Infants with chronic neonatal lung disease: recommendations for the use of home oxygen therapy
.
Med J Aust
.
2008
;
189
(
10
):
578
582
31
Flynn
JT
,
Bancalari
E
.
On “supplemental therapeutic oxygen for prethreshold retinopathy of prematurity (STOP-ROP), a randomized, controlled trial. I: primary outcomes”
.
JAAPOS
.
2000
;
4
(
2
):
65
66
32
Brockmann
PE
,
Poets
A
,
Poets
CF
.
Reference values for respiratory events in overnight polygraphy from infants aged 1 and 3 months
.
Sleep Med
.
2013
;
14
(
12
):
1323
1327
33
Askie
LM
,
Henderson-Smart
DJ
,
Irwig
L
,
Simpson
JM
.
Oxygen-saturation targets and outcomes in extremely preterm infants
.
N Engl J Med
.
2003
;
349
(
10
):
959
967
34
Moyer-Mileur
LJ
,
Nielson
DW
,
Pfeffer
KD
,
Witte
MK
,
Chapman
DL
.
Eliminating sleep-associated hypoxemia improves growth in infants with bronchopulmonary dysplasia
.
Pediatrics
.
1996
;
98
(
4 pt 1
):
779
783

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data