This study attempts to determine whether the risk of pulmonary hypertension in preterm infants with retinopathy of prematurity is increased following treatment with anti-vascular endothelial growth factor therapy as compared with laser treatment.
Key Points
Question
Are intraocular anti-vascular endothelial growth factor (VEGF) agents used to treat retinopathy of prematurity associated with an increased risk of pulmonary hypertension requiring medical therapy?
Findings
In this cohort study of 1577 propensity score–matched patients at 48 tertiary care children’s hospitals, after adjustment for site and year of treatment, infants treated with anti-VEGF therapy had a 2.4% statistically not significant higher absolute risk of receiving pulmonary vasodilators than infants treated with laser.
Meaning
These findings support investigating the possibility of pulmonary hypertension as a potential adverse effect within studies of anti-VEGF therapies in preterm infants.
Abstract
Importance
Anti–vascular endothelial growth factor (VEGF) therapy for retinopathy of prematurity (ROP) has potential ocular and systemic advantages compared with laser, but we believe the systemic risks of anti-VEGF therapy in preterm infants are poorly quantified.
Objective
To determine whether there was an association with increased risk of pulmonary hypertension (PH) in preterm infants with ROP following treatment with anti-VEGF therapy as compared with laser treatment.
Design, Setting, and Participants
This multicenter retrospective cohort study took place at neonatal intensive care units of 48 children’s hospitals in the US in the Pediatric Health Information System database from 2010 to 2020. Participants included preterm infants with gestational age at birth 22 0/7 to 31 6/7 weeks who had ROP treated with anti-VEGF therapy or laser photocoagulation.
Exposures
Anti-VEGF therapy vs laser photocoagulation.
Main Outcomes and Measures
New receipt of pulmonary vasodilators at least 7 days after ROP therapy was compared between exposure groups, matched using propensity scores generated from preexposure variables, and adjusted for birth year and hospital. The odds of receiving an echocardiogram after 30 days of age was also included to adjust for secular trends and interhospital variation in PH screening.
Results
Among 1577 patients (55.9% male) meeting inclusion criteria, 689 received laser photocoagulation and 888 received anti-VEGF treatment (95% bevacizumab, 5% ranibizumab). Patients were first treated for ROP at median 36.4 weeks’ postmenstrual age (IQR, 34.6-38.7). A total of 982 patients (491 in each group) were propensity score matched. Good covariate balance was achieved, as indicated by a model variance ratio of 1.15. More infants who received anti-VEGF therapy were treated for PH, but when adjusted for hospital and year, this was no longer statistically significant (6.7%; 95% CI, 2.6-6.9 vs 4.3% 95% CI, 4.4-10.2; adjusted odds ratio, 1.62; 95% CI, 0.90-2.89; P = .10).
Conclusions and Relevance
Anti-VEGF therapy was not associated with greater use of pulmonary vasodilators after adjustment for hospital and year. Our findings suggest exposure to anti-VEGF may be associated with PH, although we cannot exclude the possibility of residual confounding based on systemic comorbidities or hospital variation in practice. Future studies investigating this possible adverse effect seem warranted.
Introduction
Retinopathy of prematurity (ROP) is a common complication of preterm birth, affecting approximately 60% of very low-birth-weight infants.1 ROP is a leading cause of acquired blindness and appropriate therapy at early stages of disease is important to preserve vision. Traditional treatment is laser photocoagulation, but intravitreal bevacizumab (anti–vascular endothelial growth factor [VEGF] antibody) is an effective therapy, particularly for aggressive ROP.2 Anti-VEGF therapy has several ocular advantages compared with laser, including the potential for increased visual field and lower degrees of refractive error.3,4,5,6 Additionally, it requires less procedural time and anesthesia than laser treatment. However, no anti-VEGF agent is labeled for use by the US Food and Drug Administration for ROP treatment, and there is a paucity of safety data about effects on organ systems beyond the eye. Intravitreal bevacizumab lowers serum VEGF levels5 for up to 8 weeks4 and thus may inhibit vascular development in rapidly developing organ systems, such as the lungs and the brain. Given ongoing pulmonary angiogenesis in preterm infants, this antibody-based ROP therapy could induce or worsen pulmonary hypertension (PH), as has been demonstrated in preclinical models.7 Disrupted pulmonary vascular bed development may be an off-target toxicity of intravitreal anti-VEGF therapy. Our objective was to determine whether there was an association with increased risk of PH in preterm infants with ROP following treatment with anti-VEGF therapy as compared with laser treatment.
Methods
This study was reviewed and deemed exempt by the Children’s Mercy Hospital’s institutional review board (STUDY00001993) under 45 CFR 46.104 (d) category 4 (iii) as secondary research. This study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline8 (eTable 1 in the Supplement).
Data Source
The Pediatric Health Information System (PHIS) is an administrative database containing hospitalization data from 49 children’s hospitals in 27 states, plus Washington, DC, maintained by the Children’s Hospital Association and has been described in other pediatric research.9,10 The database contains detailed data on demographics, diagnosis codes at time of discharge, service locations, procedures, and billing charges. The charges at each site are mapped to a common set of clinical transaction codes (CTCs), which are further categorized into imaging studies, clinical services, laboratory tests, pharmacy, supplies, and room charges, which include date of service. PHIS data are subjected to several validity and reliability checks.
Cohort Selection
All infants admitted to sites participating in PHIS from 2010 to 2020 with a billing charge for neonatal intensive care unit (NICU) services were eligible. One site did not reliably bill for NICU services and was excluded. Infants with documented gestational age (GA) 23 to 32 weeks were included. Infants were excluded if they had major congenital anomalies known to affect lung development or pulmonary blood flow (eTable 2 in the Supplement), if they did not have a diagnosis code for ROP, or if they were not treated for ROP. Additionally, because we assessed for the presence of the primary outcome 7 days after ROP treatment, infants for whom day of therapy could not be determined or who were discharged or transferred fewer than 2 days post–ROP treatment were excluded, as it would not be possible to assess the initiation of pulmonary vasodilators. Infants were only followed up with while inpatient for the hospitalization including ROP treatment. ROP was defined using International Classification of Diseases, Ninth Revision (ICD-9) codes for unspecified ROP, retrolental fibroplasia, and ROP stage 0 through 5 (362.20 through 362.27) and the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) top-level hierarchy (H35.1xx), encompassing the same diagnoses as ICD-9.
Clinical Characteristics
Collected infant characteristics were GA at birth, birth weight, sex, maternal race and ethnicity, Child Opportunity Index measured at the zip code level, admission source, type of insurance, discharge disposition, length of hospital stay, postmenstrual age (PMA) at first ROP therapy, bronchopulmonary dysplasia (BPD) at 36 weeks, type of respiratory support 2 days prior to ROP therapy, necrotizing enterocolitis (NEC) stage II or III, pneumonia, sepsis, echo risk score, and NICU severity of illness score. The Child Opportunity Index ranks neighborhoods on 29 metrics covering education, health and environment, and social and economic domains.11,12,13 BPD was defined as use of positive pressure respiratory support at 36 weeks’ PMA, both invasive and noninvasive, which corresponds to some grade 2 BPD and all grade 3 BPD using 2019 Neonatal Research Network criteria.14 Inclusion of all infants with grade 2 BPD was not possible as the PHIS data set does not include nasal cannula flow rates. Comorbid conditions were defined by diagnosis codes (eTable 3 in the Supplement). Race in the PHIS data set is reported by site and is not explicitly or consistently defined; some use infant race, while others use maternal race; some were self-reported by caregiver and some were assessed by hospital staff.
Due to increased use of echocardiography over time and the common use of echocardiography to diagnose pulmonary hypertension, we created an echo risk score that estimates the probability of receiving an echocardiogram after 30 days of age as a function of year and site as predictors in a logistic regression model inclusive of all infants with birth weight less than 1000 g in the PHIS database.
Severity of illness was calculated as previously described15 by adapting the Hospitalization Resource Intensity Score for Kids (H-RISK) and restricting the All-Patient Refined Diagnosis Related Groups (APR-DRGs; 3M Health Information Systems) to only those admitted to PHIS NICUs.
Exposure
The primary exposure was type of ROP therapy, laser photocoagulation vs anti-VEGF therapy. Laser photocoagulation was defined using an ICD-9 procedure code (14.24), ICD-10 procedure codes (085E3ZZ and 085F3ZZ), or a CPT code (67229). PHIS CTC and CPT codes for bevacizumab (171603, C9257, and J9035), ranibizumab (188304 and J2778), and aflibercept (188306 and J0178) were used to identify receipt of an anti-VEGF agent. Infants who received both laser photocoagulation and an anti-VEGF therapy during the same hospitalization were excluded. Unfortunately, drug dosing information is not available in the PHIS database and dosing can vary by individual and center. Multiple treatments for ROP was also not available.
Outcome Definitions
The primary outcome was initiation of first pulmonary vasodilator treatment (inhaled nitric oxide, sildenafil, bosentan, milrinone, epoprostenol, or treprostinil) beginning at least 7 days after ROP therapy. Seven days was chosen as a plausible minimum duration needed for systemic VEGF suppression to manifest as pulmonary hypertension. History of treatment with a pulmonary vasodilator prior to ROP therapy was not considered exclusionary and did not influence classification of the primary outcome.
In addition, ICD-9 diagnosis codes 416.0 (primary pulmonary hypertension) and 416.8 (pulmonary hypertension not otherwise specified) and ICD-10 diagnosis codes I27.0 (primary pulmonary hypertension) and I27.2 (other secondary pulmonary hypertension) were used to define a clinical diagnosis of PH in a sensitivity analysis. Additional secondary outcomes included hospital length of stay and NICU length of stay.
Statistical Analyses
Categorical variables were summarized using frequencies and percentages; continuous variables were summarized using medians and interquartile range (IQR). We compared outcomes between the 2 ROP therapy exposure groups using odds ratios (ORs) with 95% CIs. A χ2 test for categorical variables and a Wilcoxon rank sum test for nonnormally distributed continuous variables were used to assess statistical significance.
We used propensity score matching to reduce potential confounding by indication in this observation study. We created a propensity score using multivariable logistic regression to estimate the probability of receiving anti-VEGF therapy. Modeled probabilities were then used to perform a greedy 1-to-1 match (with caliper 0.20) between exposure groups. Balance of covariates was assessed before and after matching using variance ratios and standardized differences.16,17 A standardized difference less than 0.10 indicated good covariate balance between exposure groups.
Variables included in the propensity model were GA in weeks, sex, birth weight group, patient race and ethnicity, Childhood Opportunity Index, inborn status, insurance type, baseline respiratory therapy prior to ROP treatment (eTable 4 in the Supplement), chronologic age and PMA at time of ROP treatment, severity of illness score, stage II or III NEC, and echo risk score. Race and ethnicity were included to provide context and generalizability of our findings to other sites, and because of known disparities in neonatal outcomes by race and ethnicity. BPD was not included in the propensity score match as approximately half of the infants received ROP therapy prior to BPD classification at 36 weeks’ PMA. Instead, we measured and adjusted for the maximal respiratory support present 2 days prior to the exposure. If this information was not available, maximal respiratory support 1 day prior to exposure was used, and if this information was not available, maximal respiratory support on the day of exposure was used.
We did not require matching within sites. Because sites tend to prefer 1 of the compared therapies, the frequency of both exposures within any site was insufficient. Instead, we applied a generalized linear mixed model (GLMM) on the propensity-matched cohort to assess differences in the primary outcome among exposure groups, including random effect for site to account for within-site clustering in practice and year to account to account for changes in practice over time. P values were 2-sided and were not adjusted for multiple analyses. All analyses were performed with SAS version 9.4 (SAS Institute).
Subgroup Analysis
Because PH is associated with BPD, we were interested in the PH rates among patients based on their BPD status. Therefore, we performed a subgroup analysis of PH outcomes in patients with and without BPD and evaluated for any evidence of heterogeneity in treatment effect by tests of statistical interaction.
Results
Cohort Characteristics
A cohort selection flow diagram is shown in Figure 1. Of 477 550 hospitalizations at 48 PHIS hospitals, 1577 infants met all study eligibility criteria (689 laser therapy; 888 anti-VEGF therapy). Most infants were 23 to 26 weeks’ GA at birth (87.2%) with birth weight 400 to 749 g (79.2%), with a slight male predominance (55.9%). Infants were first treated for ROP at a median 36.4 weeks’ PMA (IQR, 34.6-38.7). Most infants (83.1%) remained inpatient at 36 weeks’ PMA. Reasons for transfer to another facility were not available. In the full cohort, infants who received anti-VEGF treatment had significantly younger GA at birth, and lower birth weight and were less likely to be inborn compared with infants receiving laser photocoagulation (Table 1).
Figure 1. Cohort Selection Flow Diagram.
GA indicates gestational age; NICU, neonatal intensive care unit; ROP, retinopathy of prematurity; VEGF, vascular endothelial growth factor.
Table 1. Cohort Demographics and Covariates in Unmatched and Propensity Score Matched (PSM) Groups.
| Characteristic | No. (column %) | Standardized differencea | PSM matched, No. (column %) | Standardized differencea | |||
|---|---|---|---|---|---|---|---|
| Total ROP (N = 1577) | Unmatched | Laser (n = 491) | Anti-VEGF (n = 491) | ||||
| Laser (n = 689) | Anti-VEGF (n = 888) | ||||||
| Gestational age group at birth, wk | |||||||
| 22 | 40 (2.5) | 9 (1.3) | 31 (3.5) | 0.14 | 6 (1.2) | 8 (1.6) | 0.03 |
| 23-24 | 829 (52.6) | 306 (44.4) | 523 (58.9) | 0.29 | 249 (50.7) | 255 (51.9) | 0.02 |
| 25-26 | 545 (34.6) | 276 (40.1) | 269 (30.3) | 0.21 | 185 (37.7) | 180 (36.7) | 0.02 |
| 27-28 | 115 (7.3) | 66 (9.6) | 49 (5.5) | 0.15 | 38 (7.7) | 35 (7.1) | 0.02 |
| 29-30 | 44 (2.8) | 29 (4.2) | 15 (1.7) | 0.15 | 11 (2.2) | 12 (2.4) | 0.01 |
| 31 | 4 (0.3) | 3 (0.4) | 1 (0.1) | 0.06 | 2 (0.4) | 1 (0.2) | 0.04 |
| Birth weight, g | |||||||
| <400 | 108 (6.8) | 25 (3.6) | 83 (9.3) | 0.23 | 21 (4.3) | 22 (4.5) | 0.01 |
| 400-499 | 867 (55.0) | 345 (50.1) | 522 (58.8) | 0.18 | 270 (55.0) | 285 (58.0) | 0.06 |
| 500-749 | 381 (24.2) | 194 (28.2) | 187 (21.1) | 0.17 | 130 (26.5) | 124 (25.3) | 0.03 |
| 750-999 | 48 (3.0) | 25 (3.6) | 23 (2.6) | 0.06 | 18 (3.7) | 14 (2.9) | 0.05 |
| 1000-1249 | 17 (1.1) | 12 (1.7) | 5 (0.6) | 0.11 | 5 (1.0) | 5 (1.0) | 0.00 |
| 1250-1499 | 9 (0.6) | 3 (0.4) | 6 (0.7) | 0.03 | 3 (0.6) | 1 (0.2) | 0.06 |
| >1499 | 5 (0.3) | 2 (0.3) | 3 (0.3) | 0.01 | 2 (0.4) | 2 (0.4) | 0.00 |
| Not reported | 142 (9.0) | 83 (12.0) | 59 (6.6) | 0.19 | 42 (8.6) | 38 (7.7) | 0.03 |
| Sex | |||||||
| Female | 696 (44.1) | 312 (45.3) | 384 (43.2) | 0.04 | 233 (47.5) | 212 (43.2) | 0.09 |
| Male | 881 (55.9) | 377 (54.7) | 504 (56.8) | 0.04 | 258 (52.5) | 279 (56.8) | 0.09 |
| Race and ethnicityb | |||||||
| Asian | 63 (4.0) | 18 (2.6) | 45 (5.1) | 0.13 | 17 (3.5) | 15 (3.1) | 0.02 |
| Black | 304 (19.3) | 141 (20.5) | 163 (18.4) | 0.05 | 97 (19.8) | 90 (18.3) | 0.04 |
| Hispanic | 292 (18.5) | 128 (18.6) | 164 (18.5) | 0.00 | 86 (17.5) | 79 (16.1) | 0.04 |
| Otherc | 208 (13.2) | 104 (15.1) | 104 (11.7) | 0.10 | 70 (14.3) | 70 (14.3) | 0.00 |
| White | 710 (45.0) | 298 (43.3) | 412 (46.4) | 0.06 | 221 (45.0) | 237 (48.3) | 0.07 |
| Child Opportunity Index | |||||||
| Very low | 402 (25.5) | 168 (24.4) | 234 (26.4) | 0.05 | 120 (24.4) | 128 (26.1) | 0.05 |
| Low | 354 (22.4) | 165 (23.9) | 189 (21.3) | 0.06 | 118 (24.0) | 115 (23.4) | 0.01 |
| Moderate | 328 (20.8) | 151 (21.9) | 177 (19.9) | 0.05 | 102 (20.8) | 108 (22.0) | 0.03 |
| High | 250 (15.9) | 112 (16.3) | 138 (15.5) | 0.02 | 80 (16.3) | 71 (14.5) | 0.01 |
| Very high | 242 (15.3) | 92 (13.4) | 150 (16.9) | 0.10 | 71 (14.5) | 69 (14.1) | 0.04 |
| Missing | 1 (0.1) | 1 (0.1) | 0 (0.0) | 0.05 | 0 (0.0) | 0 (0.0) | NA |
| Admission source | |||||||
| Inborn | 177 (11.2) | 59 (8.6) | 118 (13.3) | 0.15 | 51 (10.4) | 53 (10.8) | 0.01 |
| Outborn | 1298 (82.3) | 578 (83.9) | 720 (81.1) | 0.07 | 406 (82.7) | 406 (82.7) | 0.00 |
| Other | 102 (6.5) | 52 (7.5) | 50 (5.6) | 0.08 | 34 (6.9) | 32 (6.5) | 0.02 |
| Insurance type | |||||||
| Commercial | 548 (34.7) | 218 (31.6) | 330 (37.2) | 0.12 | 162 (33.0) | 172 (35.0) | 0.04 |
| Government | 986 (62.5) | 450 (65.3) | 536 (60.4) | 0.10 | 316 (64.4) | 306 (62.3) | 0.04 |
| Self-pay | 10 (0.6) | 5 (0.7) | 5 (0.6) | 0.02 | 4 (0.8) | 3 (0.6) | 0.02 |
| Other | 33 (2.1) | 16 (2.3) | 17 (1.9) | 0.03 | 9 (1.8) | 10 (2.0) | 0.01 |
| Disposition at discharge | |||||||
| Home | 1301 (82.5) | 569 (82.6) | 732 (82.4) | 0.00 | 401 (81.7) | 405 (82.5) | 0.02 |
| Transfer | 207 (13.1) | 98 (14.2) | 109 (12.3) | 0.06 | 74 (15.1) | 64 (13.0) | 0.06 |
| Mortality | 52 (3.3) | 17 (2.5) | 35 (3.9) | 0.08 | 11 (2.2) | 15 (3.1) | 0.05 |
| Other/unknown | 17 (1.1) | 5 (0.7) | 12 (1.4) | 0.06 | 5 (1.0) | 7 (1.4) | 0.04 |
| Short stays (<7 d LOS) | 181 (11.5) | 123 (17.9) | 58 (6.5) | 0.35 | 57 (11.6) | 41 (8.4) | 0.10 |
| Age at first ROP therapy, median (IQR), d | 82 (71-97) | 87 (75-105) | 78 (69-92) | 0.44 | 84 (73-101) | 83 (73-97) | 0.04 |
| PMA at first ROP therapy, median (IQR), wk | 36.4 (34.6-38.7) | 37.6 (35.7-40.3) | 35.6 (34.0-37.6) | 0.57 | 37.0 (35.3-38.9) | 36.7 (35.3-38.7) | 0.06 |
| In hospital at 36 wk PMA | |||||||
| No | 267 (16.9) | 178 (25.8) | 89 (10.0) | 0.42 | 78 (15.9) | 67 (13.6) | 0.06 |
| Yes | 1310 (83.1) | 511 (74.2) | 799 (90.0) | 0.42 | 413 (84.1) | 424 (86.4) | 0.06 |
| Severe BPD at 36 wk | |||||||
| No | 983 (62.3) | 465 (67.5) | 518 (58.3) | 0.19 | 312 (63.5) | 314 (64.0) | 0.01 |
| Yes | 594 (37.7) | 224 (32.5) | 370 (41.7) | 0.19 | 179 (36.5) | 177 (36.0) | 0.01 |
| NICU SOI score, mean (SE) | 4.65 (0.06) | 4.21 (0.10) | 4.99 (0.07) | 0.33 | 4.62 (0.11) | 4.73 (0.10) | 0.05 |
| Respiratory support up to 2 d prior to ROP Rxa | |||||||
| Invasive | 438 (27.8) | 166 (24.1) | 272 (30.6) | 0.15 | 122 (24.8) | 125 (25.5) | 0.01 |
| Noninvasive | 267 (16.9) | 69 (10.0) | 198 (22.3) | 0.34 | 60 (12.2) | 73 (14.9) | 0.08 |
| Less than positive pressure | 872 (55.3) | 454 (65.9) | 418 (47.1) | 0.39 | 309 (62.9) | 293 (59.7) | 0.07 |
| Other comorbid conditions | |||||||
| Sepsis | 753 (47.7) | 296 (43.0) | 457 (51.5) | 0.17 | 228 (46.4) | 224 (45.6) | 0.02 |
| NEC (stage 2, stage 3) | 213 (13.5) | 69 (10.0) | 144 (16.2) | 0.18 | 59 (12.0) | 57 (11.6) | 0.01 |
| Pneumonia | 191 (12.1) | 64 (9.3) | 127 (14.3) | 0.16 | 53 (10.8) | 65 (13.2) | 0.08 |
| Echo risk score [min = 0/max = 1], median (IQR) | 0.388 (0.308-0.466) | 0.361 (0.282-0.438) | 0.407 (0.340-0.484) | 0.43 | 0.376 (0.289-0.456) | 0.381 (0.309-0.454) | 0.07 |
Abbreviations: BPD, bronchopulmonary dysplasia; LOS, length of stay; NICU SOI, neonatal intensive care unit severity of illness; NEC, necrotizing enterocolitis; PMA, postmenstrual age; ROP, retinopathy of prematurity; Rx, prescription; VEGF, vascular endothelial growth factor.
Standardized difference less than 0.10 indicates imbalance between laser and anti-VEGF.
Race in the Pediatric Health Information System data set is reported by site and is not explicitly or consistently defined; some use infant race, while others use maternal race; some were self-reported by caregiver, and some were assessed by hospital staff.
This category includes American Indian, Native Hawaiian/Pacific Islander, missing, and other.
For the propensity score analysis, 491 infants from each treatment group were matched, yielding a cohort of 982 infants. Overall, good covariate balance was achieved as indicated by a model variance ratio of 1.15 (reference, 0.50-2.00). None of the assessed demographic or neonatal criteria differed significantly between the anti-VEGF and laser photocoagulation groups (Table 1) after propensity score matching.
ROP Treatment Over Time
Significantly increasing proportions of infants were treated with anti-VEGF therapy compared with laser over the study period, with anti-VEGF therapy becoming the predominant therapy beginning in 2015 (Figure 2). However, increases in anti-VEGF therapy use were not uniform among all sites. Four sites provided only anti-VEGF therapy and 1 provided only laser photocoagulation. Among the 43 sites that provided both treatments, 11 used anti-VEGF therapy 75% or more of the time and 4 used laser photocoagulation 75% or more of the time. Of note, infants in the propensity-matched cohort represented all sites. A total of 95% of infants treated with anti-VEGF therapy received bevacizumab and 5% received ranibizumab; none received aflibercept.
Figure 2. Primary Results: Use of Pulmonary Vasodilators Following Retinopathy of Prematurity (ROP) Treatments.
Adjusted analyses include year of treatment and site. PH indicates pulmonary hypertension; VEGF, vascular endothelial growth factor.
PH
In the full cohort, 107 infants (6.8%) were diagnosed with PH by receipt of pulmonary vasodilators (Table 2) and met the primary outcome definition. An additional 83 infants (5.2%) were diagnosed by ICD-9/10 diagnosis code and were included in the secondary pulmonary hypertension outcome. Sildenafil and iNO were the most commonly prescribed vasodilators, and median time between ROP therapy and day of first PH medication was 24 (IQR, 7-62) days. In the unmatched analysis, patients receiving anti-VEGF therapy were more likely to receive pulmonary vasodilators following ROP therapy compared with patients receiving laser photocoagulation (8.3% vs 4.8%; OR, 1.81; 95% CI, 1.19-2.79; P = .01) (Table 2) (Figure 3).
Table 2. Primary Outcome of Pulmonary Vasodilators by Retinopathy of Prematurity (ROP) Treatment Group.
| Characteristic | Total ROP (N = 1577) | Unmatched | Effect estimate (95% CI)a | P value | PSM matched | Effect estimate (95% CI)a | P value | ||
|---|---|---|---|---|---|---|---|---|---|
| Laser (n = 689) | Anti-VEGF (n = 888) | Laser (n = 491) | Anti-VEGF (n = 491) | ||||||
| Adjusted probability PH, % (95% CI)b | 5.5 (4.0-7.6) | 4.2 (2.7-6.4) | 7.3 (5.3-10.0) | 1.80 (1.14-2.85) | .01 | 4.3 (2.6-6.9) | 6.7 (4.4-10.2) | 1.62 (0.90-2.89) | .10 |
| PH (meds), No. (column %) | .01 | .05 | |||||||
| No | 1470 (93.2) | 656 (95.2) | 814 (91.7) | 1 [Reference] | NA | 468 (95.3) | 453 (92.3) | 1 [Reference] | NA |
| Yes | 107 (6.8) | 33 (4.8) | 74 (8.3) | 1.81 (1.19-2.79) | NA | 23 (4.7) | 38 (7.7) | 1.71 (1.00-2.95) | NA |
| Length of stay, median (IQR) | 116 (83 145) | 106 (51 138) | 123 (97 152) | 17.0 (11.7-22.3) | <.001 | 111 (79 140) | 118 (92 141) | 7.0 (1.5-12.5) | .05 |
| NICU days, median (IQR) | 105 (62 138) | 95 (32 126) | 114 (78 144) | 19.0 (12.3-25.7) | <.001 | 103 (53 131) | 105 (70 134) | 2.0 (−4.6-8.6) | .06 |
Abbreviations: GLMM, generalized linear mixed model; IQR, interquartile range; NA, not applicable; NICU, neonatal intensive care unit; PH, pulmonary hypertension; PSM, propensity score matched; VEGF, vascular endothelial growth factor.
Effect estimates for PH = odds ratio; effect estimate for length of stay and NICU days = difference.
Adjusted for clustering within site and year using GLMM.
Figure 3. Trends of Retinopathy of Prematurity Therapy Modality Over Time.
Number of infants treated with each modality over study period. P < .001 for trend. VEGF indicates vascular endothelial growth factor.
In the unadjusted propensity-score matched cohort, more infants who received anti-VEGF therapy were treated for PH (7.7 vs 4.7%; OR, 1.71; 95% CI, 1.00-2.95; P = .05) (Table 2) (Figure 2). This difference was attenuated after adjustment for site and year (6.7 vs 4.3%; OR, 1.62; 95% CI, 0.90-2.89; P = .10) and was the primary outcome of this study.
Sensitivity Analysis
To ascertain whether PH was diagnosed after ROP therapy, regardless of treatment, we evaluated PH by diagnosis code. There was no significant difference in PH between groups when PH was defined by ICD code and/or any treatment (11.8% vs 9.2%). Forty-four of 107 infants receiving pulmonary vasodilators had a diagnosis code for PH, while 44 of 127 infants with a PH diagnosis code received pulmonary vasodilators. Infant age at PH diagnosis and at time of pulmonary vasodilator initiation could not be ascertained.
Length of Stay
In unadjusted analyses of the full cohort, length of NICU stay was 114 (IQR, 78-144) days among infants treated with anti-VEGF therapy compared with 95 (IQR, 32-126) days among infants treated with laser (P < .001), and length of hospital stay was 123 (IQR, 97-152) days among infants treated with anti-VEGF therapy compared with 114 (IQR, 51-138) days among infants treated with laser (P < .001) (Table 2).
In the propensity-score matched cohort, length of NICU stay was 105 (IQR, 70-134) days among infants treated with anti-VEGF therapy compared with 102 (IQR, 53-131) days among infants treated with laser (P = .06), and length of hospital stay was 118 (IQR, 92-141) days among infants treated with anti-VEGF therapy compared with 111 (IQR, 79-140) days among infants treated with laser (P = .05) (Table 2).
Subgroup Analysis Among Infants With and Without BPD
In the entire cohort, grade 2 to 3 BPD at 36 weeks’ PMA was diagnosed in 594 infants (37.7%). Among infants without BPD, there was no difference in pulmonary vasodilator use who received anti-VEGF therapy compared with those who received laser photocoagulation (5.3%; 95% CI, 2.9-9.5 vs 2.4%; 95% CI, 1.1-5.3; P = .06). Among patients with BPD, there was no difference in pulmonary vasodilator ruse between infants who received anti-VEGF therapy compared with those who received laser photocoagulation (7.4%; 95% CI, 3.4-15.1 vs 6.8% 95% CI, 3.3-13.5; P = .85). There was not a significant interaction between type of ROP treatment and BPD.
Discussion
In this retrospective cohort study of preterm infants admitted to US children’s hospitals over a 10-year period, anti-VEGF therapy for ROP was associated with a 2.4% statistically not significant increase in the risk of medical therapy for pulmonary hypertension in a propensity-matched cohort. Although there are ocular advantages of anti-VEGF over laser therapy, there may be off-target organ toxicity; for example, off-target impact in the brain may affect neurodevelopment.18,19,20 Findings from the current study raise concern about similar effects in the developing lung vasculature. Future studies evaluating the safety of anti-VEGF therapy should include pulmonary hypertension as a key clinical outcome. This is especially important as anti-VEGF treatment usage is increasing over time, which underscores the importance of the adjusted, rather than unadjusted, analysis.
It is challenging to evaluate the comparative safety of ROP treatments. Randomized trials are ideal to answer this question but are expensive to perform and lack statistical power to detect differences in important, yet rare outcomes. The largest randomized trial of anti-VEGF treatment, BEAT-ROP,2 reported mortality from respiratory decompensation in 4 infants (of 75) in the anti-VEGF group, compared with 1 infant (of 75) in the laser therapy group. More recently, the RAINBOW study randomized infants to ranibizumab vs laser, but found no differences in mortality, with 4 deaths in each arm.21 Nonrandomized studies have reported higher rates of death in anti-VEGF–treated infants6,20,22,23 but are difficult to interpret due to small numbers and the likelihood of confounding by indication, since anti-VEGF is increasingly preferred in the most premature infants who are more likely to have aggressive posterior ROP and to develop pulmonary disease. Favoring safety of anti-VEGF treatment, there is less anesthesia required at the time of treatment and, in contrast to the present study, a recent comparative study found higher respiratory support 1 week post treatment with laser compared with anti-VEGF.24
Intravitreal anti-VEGF use may have clinically relevant extraocular effects because intravitreal anti-VEGF reduces serum VEGF levels following treatment.5,25 Most studies have evaluated bevacizumab, with several showing persistent reduction in serum VEGF for 8 to 12 weeks.25,26,27 Even with variable dosing regimens, doses as low as 0.6% of the original BEAT-ROP dose were effective, yet still showed similar reductions in serum VEGF levels.28 As lower-dose bevacizumab becomes more commonly used, these studies need to be repeated to better understand the association between bevacizumab dose, serum VEGF levels, and clinical sequelae. Unfortunately, dosing information was not available in the PHIS database. There are several small studies with aflibercept, which have also shown persistent reductions in serum VEGF.29,30 These effects might impact the developing pulmonary vascular bed, resulting in pulmonary hypertension.
One potential alternative approach that may result in fewer off-target effects is use of different anti-VEGF agents, such as ranibizumab which has been approved in Europe for treatment of ROP. This agent appears to affect serum VEGF levels much more transiently, presumably due to differences in the underlying molecular structure,27 resulting in ranibizumab being cleared from the retina more quickly with less drug released into the systemic circulation.31,32 Importantly, low-dose ranibizumab does not appear to suppress systemic VEGF levels.21,33 Thus, while anti-VEGF use in general has the potential to affect serum VEGF levels, and therefore cause off-target toxicity, this may vary based on the specific dose and medication used. Subgroup analysis was not performed in this study, as only 5% of our cohort received ranibizumab. In addition, although we did not include infants who received both anti-VEGF therapy and laser photocoagulation, the latter is capable of breaking down the blood-retinal barrier, theoretically allowing anti-VEGF agents to escape the eye and potentially increase risks associated with systemic exposure.
The multicenter design of the current study allowed for a relatively large sample size, which is essential for detection of rare events like PH. Inclusion of multiple sites with different practices increases the generalizability of our findings. Propensity score matching helps address bias due to confounding by indication, although residual confounding from unmeasured covariates is possible. We also developed an echocardiogram risk score to adjust for possible detection bias due to echocardiographic screening and secular trends in PH monitoring, which has become more common since a 2015 recommendation.34 A strength of the PHIS data set is reliable ascertainment of billing, which should accurately reflect medication administration, so we are confident in assessing both anti-VEGF agent administration and prescription of PH medications. Additionally, the PHIS data set allows robust adjustment for social disadvantage using the Child Opportunity Index.
Limitations
Use of administrative data has inherent limitations, particularly in classification of medical diagnoses as compared with medications or procedures.35,36 We chose a primary outcome defined by PH treatment administration instead of ICD billing codes based on a presumption that drug treatment billing would more accurately reflect clinically important PH. We did not exclude infants who received pulmonary vasodilators before ROP treatment (eg, in the first week of life) but included only infants who received a first postprocedure dose of pulmonary vasodilator within the 7 days of ROP treatment, thus excluding any infants who were receiving pulmonary vasodilators at the time of ROP treatment. Due to some sites exclusively using 1 form of ROP treatment, we were unable to perform propensity score matching between sites. Individual preference for use of laser photocoagulation or anti-VEGF therapy is likely not random, and it is possible that the latter could be chosen by individual ophthalmologists and neonatologists based on variables not included in our PS model. Similarly, preference for treatment or nontreatment of PH could also lead to residual confounding. Severity of ROP is associated with an increased probability of receiving bevacizumab and correlates with a diagnosis of PH.37 Unfortunately, ROP severity was not available from all sites, so was unable to be included in this study, but infants treated with anti-VEGF agents may have already been at a higher risk of being diagnosed with PH. This would diminish the importance of our findings, but awareness of the possibility of an additional diagnosis (PH) is essential to appropriate management. PS matching is limited to balance on measured confounders, unlike randomization. In addition, data on the duration of PH treatment, long-term ocular outcomes, or attributed adverse effects from PH medications or anti-VEGF agents were not available. It should also be understood that it is difficult to discriminate between infants developing cardiopulmonary disorders secondary to prematurity or due to the use of anti-VEGF agents. Additionally, we relied on the need for PH therapy or ICD diagnosis, rather than objective echocardiographic or catheterization data to diagnose PH. The decision to initiate therapy for PH is somewhat subjective, though diagnostic and treatment recommendations have become more standardized since the recent guidelines.34,38
Conclusions
In summary, anti-VEGF therapy for ROP was associated with a statistically not significant, but potentially clinically important increase in the rate of treatment for pulmonary hypertension. To our knowledge, this is the largest cohort to date evaluated for this safety outcome. Our findings suggest that exposure to anti-VEGF may be associated with incident PH, although we cannot exclude the possibility of residual confounding based on systemic comorbidities, an unmeasured association between probability of receiving anti-VEGF therapy and developing PH, or hospital variation in ROP or PH treatment practices. Alternatives to bevacizumab, such as ranibizumab, may have a smaller impact. Because of the rarity of the outcome, randomized trials to evaluate PH are unlikely to be definitive due to inadequate power and families may be unwilling to randomize their children to a trial with harm as an end point. Pharmacovigilance studies, such as ours, may be the strongest available evidence. Neonatal clinicians should continue to partner with families to evaluate possible adverse effects of anti-VEGF therapy, evaluate the relative systemic risks of different anti-VEGF agents, and consider the potential harms of these medications alongside their potential benefits.
eTable 1. STROBE checklist.
eTable 2. Major congenital anomalies.
eTable 3. Comorbid Conditions.
eTable 4. Baseline levels of respiratory support and timing.
References
- 1.Zin A, Gole GA. Retinopathy of prematurity-incidence today. Clin Perinatol. 2013;40(2):185-200. doi: 10.1016/j.clp.2013.02.001 [DOI] [PubMed] [Google Scholar]
- 2.Mintz-Hittner HA, Kennedy KA, Chuang AZ; BEAT-ROP Cooperative Group . Efficacy of intravitreal bevacizumab for stage 3+ retinopathy of prematurity. N Engl J Med. 2011;364(7):603-615. doi: 10.1056/NEJMoa1007374 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Barry GP, Yu Y, Ying GS, et al. ; G-ROP Study Group . Retinal detachment after treatment of retinopathy of prematurity with laser versus intravitreal anti-vascular endothelial growth factor. Ophthalmology. 2021;128(8):1188-1196. doi: 10.1016/j.ophtha.2020.12.028 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.VanderVeen DK, Melia M, Yang MB, Hutchinson AK, Wilson LB, Lambert SR. Anti-vascular endothelial growth factor therapy for primary treatment of type 1 retinopathy of prematurity: a report by the American Academy of Ophthalmology. Ophthalmology. 2017;124(5):619-633. doi: 10.1016/j.ophtha.2016.12.025 [DOI] [PubMed] [Google Scholar]
- 5.Pertl L, Steinwender G, Mayer C, et al. A systematic review and meta-analysis on the safety of vascular endothelial growth factor (VEGF) inhibitors for the treatment of retinopathy of prematurity. PLoS One. 2015;10(6):e0129383. doi: 10.1371/journal.pone.0129383 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Sankar MJ, Sankar J, Chandra P. Anti-vascular endothelial growth factor (VEGF) drugs for treatment of retinopathy of prematurity. Cochrane Database Syst Rev. 2018;1:CD009734. doi: 10.1002/14651858.CD009734.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Demir C, Karaman M, Uçan ES, et al. Effects of bevacizumab administration on the hypoxia-induced pulmonary hypertension rat model. Turk J Med Sci. 2021;51(5):2752-2762. doi: 10.3906/sag-2101-76 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med. 2007;4(10):e296. doi: 10.1371/journal.pmed.0040296 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Mahant S, Richardson T, Keren R, Srivastava R, Meier J, Pediatric Research in Inpatient Setting (PRIS) Network . Variation in tonsillectomy cost and revisit rates: analysis of administrative and billing data from US children’s hospitals. BMJ Qual Saf. 2021;30(5):388-396. doi: 10.1136/bmjqs-2019-010730 [DOI] [PubMed] [Google Scholar]
- 10.Keren R, Luan X, Localio R, et al. ; Pediatric Research in Inpatient Settings (PRIS) Network . Prioritization of comparative effectiveness research topics in hospital pediatrics. Arch Pediatr Adolesc Med. 2012;166(12):1155-1164. doi: 10.1001/archpediatrics.2012.1266 [DOI] [PubMed] [Google Scholar]
- 11.Brandeis University, The Heller School for Social Policy and Management . Child Opportunity Index 2.0 ZIP code estimates. Accessed September 2, 2022. https://www.diversitydatakids.org/
- 12.Noelke C, McArdle N, Baek M, et al. ; Diversity Data Kids . How we built it. Accessed September 2, 2022. http://new.diversitydatakids.org/research-library/research-brief/how-we-built-it
- 13.Acevedo-Garcia D, Noelke C, McArdle N, et al. Racial and ethnic inequities in children’s neighborhoods: evidence from the New Child Opportunity Index 2.0. Health Aff (Millwood). 2020;39(10):1693-1701. doi: 10.1377/hlthaff.2020.00735 [DOI] [PubMed] [Google Scholar]
- 14.Jensen EA, Dysart K, Gantz MG, et al. The diagnosis of bronchopulmonary dysplasia in very preterm infants. an evidence-based approach. Am J Respir Crit Care Med. 2019;200(6):751-759. doi: 10.1164/rccm.201812-2348OC [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.King BC, Richardson T, Patel RM, et al. Prioritization framework for improving the value of care for very low birth weight and very preterm infants. J Perinatol. 2021;41(10):2463-2473. doi: 10.1038/s41372-021-01114-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Rubin DB. Using propensity scores to help design observational studies: application to the tobacco litigation. Health Serv Outcomes Res Methodol. 2001;2(3):169-188. doi: 10.1023/A:1020363010465 [DOI] [Google Scholar]
- 17.Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25(1):1-21. doi: 10.1214/09-STS313 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Tan H, Blasco P, Lewis T, Ostmo S, Chiang MF, Campbell JP. Neurodevelopmental outcomes in preterm infants with retinopathy of prematurity. Surv Ophthalmol. 2021;66(5):877-891. doi: 10.1016/j.survophthal.2021.02.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Morin J, Luu TM, Superstein R, et al. ; Canadian Neonatal Network and the Canadian Neonatal Follow-Up Network Investigators . Neurodevelopmental outcomes following bevacizumab injections for retinopathy of prematurity. Pediatrics. 2016;137(4):e20153218. doi: 10.1542/peds.2015-3218 [DOI] [PubMed] [Google Scholar]
- 20.Natarajan G, Shankaran S, Nolen TL, et al. Neurodevelopmental outcomes of preterm infants with retinopathy of prematurity by treatment. Pediatrics. 2019;144(2):e20183537. doi: 10.1542/peds.2018-3537 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Stahl A, Lepore D, Fielder A, et al. Ranibizumab versus laser therapy for the treatment of very low birthweight infants with retinopathy of prematurity (RAINBOW): an open-label randomised controlled trial. Lancet. 2019;394(10208):1551-1559. doi: 10.1016/S0140-6736(19)31344-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hillier RJ, Connor AJ, Shafiq AE. Ultra-low-dose intravitreal bevacizumab for the treatment of retinopathy of prematurity: a case series. Br J Ophthalmol. 2018;102(2):260-264. doi: 10.1136/bjophthalmol-2017-310408 [DOI] [PubMed] [Google Scholar]
- 23.Kennedy KA, Mintz-Hittner HA; BEAT-ROP Cooperative Group . Medical and developmental outcomes of bevacizumab versus laser for retinopathy of prematurity. J AAPOS. 2018;22(1):61-65.e1. doi: 10.1016/j.jaapos.2017.10.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Arthur N, Byrne E, Kehinde F, Bhandari V, Quinones Cardona V. Respiratory complications in infants with retinopathy of prematurity (ROP) requiring laser photocoagulation. Am J Perinatol. 2021. Published online December 10, 2021. doi: 10.1055/s-0041-1740347 [DOI] [PubMed] [Google Scholar]
- 25.Kong L, Bhatt AR, Demny AB, et al. Pharmacokinetics of bevacizumab and its effects on serum VEGF and IGF-1 in infants with retinopathy of prematurity. Invest Ophthalmol Vis Sci. 2015;56(2):956-961. doi: 10.1167/iovs.14-15842 [DOI] [PubMed] [Google Scholar]
- 26.Wu WC, Lien R, Liao PJ, et al. Serum levels of vascular endothelial growth factor and related factors after intravitreous bevacizumab injection for retinopathy of prematurity. JAMA Ophthalmol. 2015;133(4):391-397. doi: 10.1001/jamaophthalmol.2014.5373 [DOI] [PubMed] [Google Scholar]
- 27.Wu WC, Shih CP, Lien R, et al. Serum vascular endothelial growth factor after bevacizumab or ranibizumab treatment for retinopathy of prematurity. Retina. 2017;37(4):694-701. doi: 10.1097/IAE.0000000000001209 [DOI] [PubMed] [Google Scholar]
- 28.Wallace DK, Kraker RT, Freedman SF, et al. ; Pediatric Eye Disease Investigator Group (PEDIG) . Short-term outcomes after very low-dose intravitreous bevacizumab for retinopathy of prematurity. JAMA Ophthalmol. 2020;138(6):698-701. doi: 10.1001/jamaophthalmol.2020.0334 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Huang CY, Lien R, Wang NK, et al. Changes in systemic vascular endothelial growth factor levels after intravitreal injection of aflibercept in infants with retinopathy of prematurity. Graefes Arch Clin Exp Ophthalmol. 2018;256(3):479-487. doi: 10.1007/S00417-017-3878-4 [DOI] [PubMed] [Google Scholar]
- 30.Furuncuoglu U, Vural A, Kural A, Onur IU, Yigit FU. Serum vascular endothelial growth factor, insulin-like growth factor-1 and aflibercept levels in retinopathy of prematurity. Jpn J Ophthalmol. 2022;66(2):151-158. doi: 10.1007/s10384-021-00895-9 [DOI] [PubMed] [Google Scholar]
- 31.Bakri SJ, Snyder MR, Reid JM, Pulido JS, Ezzat MK, Singh RJ. Pharmacokinetics of intravitreal ranibizumab (Lucentis). Ophthalmology. 2007;114(12):2179-2182. doi: 10.1016/j.ophtha.2007.09.012 [DOI] [PubMed] [Google Scholar]
- 32.Mordenti J, Cuthbertson RA, Ferrara N, et al. Comparisons of the intraocular tissue distribution, pharmacokinetics, and safety of 125I-labeled full-length and Fab antibodies in rhesus monkeys following intravitreal administration. Toxicol Pathol. 1999;27(5):536-544. doi: 10.1177/019262339902700507 [DOI] [PubMed] [Google Scholar]
- 33.Fidler M, Fleck BW, Stahl A, et al. ; RAINBOW study group† . Ranibizumab population pharmacokinetics and free VEGF pharmacodynamics in preterm infants with retinopathy of prematurity in the RAINBOW trial. Transl Vis Sci Technol. 2020;9(8):43. doi: 10.1167/tvst.9.8.43 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Abman SH, Hansmann G, Archer SL, et al. ; American Heart Association Council on Cardiopulmonary, Critical Care, Perioperative and Resuscitation; Council on Clinical Cardiology; Council on Cardiovascular Disease in the Young; Council on Cardiovascular Radiology and Intervention; Council on Cardiovascular Surgery and Anesthesia; and the American Thoracic Society . Pediatric pulmonary hypertension: guidelines from the American Heart Association and American Thoracic Society. Circulation. 2015;132(21):2037-2099. doi: 10.1161/CIR.0000000000000329 [DOI] [PubMed] [Google Scholar]
- 35.Beam KS, Lee M, Hirst K, Beam A, Parad RB. Specificity of International Classification of Diseases codes for bronchopulmonary dysplasia: an investigation using electronic health record data and a large insurance database. J Perinatol. 2021;41(4):764-771. doi: 10.1038/s41372-021-00965-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Muir KW, Gupta C, Gill P, Stein JD. Accuracy of international classification of diseases, ninth revision, clinical modification billing codes for common ophthalmic conditions. JAMA Ophthalmol. 2013;131(1):119-120. doi: 10.1001/jamaophthalmol.2013.577 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Bhatti F, Yu Y, Ying GS, Tomlinson LA, Binenbaum G; G-ROP Study Group . Association of cardiovascular disease with retinopathy of prematurity. Ophthalmic Epidemiol. Published online February 9, 2022. doi: 10.1080/09286586.2022.2036766 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Krishnan U, Feinstein JA, Adatia I, et al. ; Pediatric Pulmonary Hypertension Network (PPHNet) . Evaluation and management of pulmonary hypertension in children with bronchopulmonary dysplasia. J Pediatr. 2017;188:24-34.e1. doi: 10.1016/j.jpeds.2017.05.029 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. STROBE checklist.
eTable 2. Major congenital anomalies.
eTable 3. Comorbid Conditions.
eTable 4. Baseline levels of respiratory support and timing.