Edited by: Lynette Kay Rogers, The Research Institute at Nationwide Children's Hospital, United States
Reviewed by: Hercília Guimarães, University of Porto, Portugal; Thomas Michael Raffay, Rainbow Babies and Children's Hospital, United States
This article was submitted to Neonatology, a section of the journal Frontiers in Pediatrics
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Globally, ~11.1% of births are premature (
ROP can be classified as type 1 or type 2, based on the Early Treatment for Retinopathy of Prematurity (ETROP) classification (
It has been proposed that ROP may lead to significant visual impairment which consequently portends worse later neurologic outcomes in infants and/or that pathological processes leading to ROP (such as intermittent hypoxia or oxidative stress) could also have detrimental effects elsewhere in the brain due to the similar embryological origins of both the eye and brain (
A retrospective cohort study was performed at the University of California, Los Angeles (UCLA) Mattel Children's Hospital on infants screened for ROP in the neonatal intensive care unit (NICU) between January 1, 2011 and December 31, 2018. The Institutional Review Board at UCLA approved the study protocol and granted waiver of consent.
All neonates screened for ROP while hospitalized in the NICU at the UCLA Mattel Children's Hospital were eligible for the study. Infants eligible for ROP screening were identified by the neonatology team at UCLA. Study inclusion criteria were consistent with American Academy of Pediatrics (AAP) guidelines for ROP screening: infants born at a gestational age ≤30 weeks, birth weight <1,500 g, or gestational age at birth >30 weeks but with an unstable clinical course, such as infants on significant cardiorespiratory support (
Demographic (sex, gestational age, and birth weight), clinical course/outcome information [fetal growth restriction (FGR), small for gestational age (SGA), BPD, and IVH], socioeconomic status information (insurance type), and visual outcomes (myopia, strabismus, amblyopia, and optic atrophy) were collected for each subject via electronic medical review. FGR was determined by the obstetric team through serial prenatal ultrasound. SGA was defined as a birth weight percentile <10% for gestational age and sex (
ROP screening was performed by board-certified pediatric ophthalmologists at the recommended intervals according to the 2013 AAP guidelines (
The primary outcome variables for this study were composite cognition, language, and motor domain scores assessed using the Bayley Scales of Infant and Toddler Development, third edition (Bayley-III) (
For each age group, differences in demographic, clinical and vision outcome data between ROP groups (no ROP, type 1 ROP, and type 2 ROP) were assessed using ANOVA (gestational age and birth weight) and Chi-Square Tests (sex, IVH, BPD, health insurance type, FGR, SGA, and vision outcomes). In univariable analysis, ANOVA was used to assess the association between ROP severity (no ROP, type 1, and type 2 ROP) and Bayley-III composite cognition, language, or motor scores within each age group. In multivariable analysis, the association was assessed between neurodevelopmental scores and infants with any ROP (type 1 or 2) vs. no ROP after controlling for variables found to be highly associated with neurodevelopmental scores. To account for the correlations among Bayley scores from each infant, we used a linear mixed effects model. The model selection steps involved backward eliminations and forward selections during which likelihood tests and Akaike information criteria were used for nested and un-nested model comparisons, respectively. Comparisons of neurodevelopmental scores of three different age cohorts (0–12, 12–24, and 24–36 months) were performed using the final mixed effect models and Tukey–Kramer's multiple comparison adjusted-
Three hundred and sixty infants were screened for ROP exams at the UCLA Mattel Children's Hospital between January 1st, 2011 date and December 31st, 2018 date. One hundred and thirty-two infants did not have available data on neurodevelopmental assessments and were excluded from the study. The remaining 228 infants met study inclusion criteria and were included for analysis.
Our study cohort of 228 infants had a mean gestational age of 28.5 ± 2.8 weeks (range: 22.3–34.6 weeks) and mean birth weight of 1,089 ± 373 g (range: 470–2,370 g). One hundred and six (47%) infants were female, 94 (41%) infants had BPD, 70 (31%) infants had FGR, 38 (16.7%) infants were SGA, and 91 (40%) infants had IVH (grades I–IV). Eighty-eight (42.9%) neonates had public health insurance and 117 (57.1%) neonates had private health insurance. Twenty-three (10%) infants had type 1 ROP, 66 (29%) infants had type 2 ROP, and 139 (61%) infants had no ROP. Thirty-eight infants were treated for ROP; this included infants with type 1 ROP and those with persistent type 2 ROP (
Summary of demographic data and clinical outcomes for infants with no ROP, type 2 ROP, and type 1 ROP.
Gestational age (Weeks) | 117 | 29.9 (2.00) | 55 | 26.8 (2.51) | 19 | 25.5 (2.08) | 61.0 | <0.0001 |
Birth weight (g) | 117 | 1249 (319) | 55 | 872 (296) | 19 | 758 (183) | 42.0 | <0.0001 |
Sex (Female) | 117 | 52 (44.4) | 55 | 29 (52.7) | 19 | 7 (36.8) | 1.76 | 0.42 |
Diagnosis of BPD | 117 | 30 (25.6) | 55 | 36 (65.5) | 19 | 14 (73.7) | 33.1 | <0.0001 |
Diagnosis of IVH | 117 | 36 (30.8) | 55 | 27 (49.1) | 19 | 11 (57.9) | 8.55 | 0.014 |
Diagnosis of FGR | 117 | 37 (31.6) | 55 | 17 (30.9) | 19 | 6 (31.6) | 0.009 | 0.99 |
SGA Status | 117 | 21 (18.0) | 55 | 9 (16.4) | 19 | 3 (15.8) | 0.099 | 0.95 |
Public health insurance | 106 | 31 (29.2) | 52 | 27 (51.9) | 19 | 15 (78.9) | 19.9 | <0.0001 |
Gestational age (Weeks) | 86 | 29.8 (2.09) | 40 | 25.9 (2.53) | 17 | 25.6 (1.97) | 56.1 | <0.0001 |
Birth weight (g) | 86 | 1,277 (362) | 40 | 780 (273) | 17 | 745 (203) | 42.0 | <0.0001 |
Sex (Female) | 86 | 43 (50.0) | 40 | 21 (52.5) | 17 | 7 (41.2) | 0.6 | 0.73 |
Diagnosis of BPD | 86 | 21 (24.4) | 40 | 32 (80.0) | 17 | 11 (64.7) | 37.2 | <0.0001 |
Diagnosis of IVH | 86 | 23 (26.7) | 40 | 24 (60.0) | 17 | 9 (52.9) | 14.2 | 0.001 |
Diagnosis of FGR | 86 | 23 (26.7) | 40 | 12 (30.0) | 17 | 5 (29.4) | 0.2 | 0.92 |
SGA status | 86 | 13 (15.1) | 40 | 7 (17.5) | 17 | 4 (23.5) | 0.7 | 0.69 |
Public health insurance | 71 | 25 (35.2) | 36 | 19 (0.52) | 17 | 13 (76.4) | 10.3 | 0.006 |
Gestational Age (Weeks) | 28 | 29.7 (2.60) | 22 | 25.9 (1.79) | 9 | 24.6 (1.46) | 27.3 | <0.0001 |
Birth Weight (g) | 28 | 1291 (388) | 22 | 771 (239) | 9 | 827 (170) | 19.1 | <0.0001 |
Sex (Female) | 28 | 9 (32.1) | 22 | 12 (54.6) | 9 | 3 (33.3) | 2.8 | 0.25 |
Diagnosis of BPD | 28 | 5 (17.9) | 22 | 16 (72.7) | 9 | 5 (55.6) | 15.6 | <0.0001 |
Diagnosis of IVH | 28 | 6 (21.4) | 22 | 10 (45.5) | 9 | 8 (88.9) | 13.2 | 0.001 |
Diagnosis of FGR | 28 | 7 (25.0) | 22 | 6 (27.3) | 9 | 0 (0.00) | 3.0 | 0.22 |
SGA status | 28 | 4 (14.3) | 22 | 4 (18.2) | 9 | 0 (0.00) | 1.8 | 0.40 |
Public health insurance | 24 | 12 (50.0) | 18 | 14 (77.8) | 9 | 6 (66.7) | 3.5 | 0.18 |
One hundred and thirty-nine infants were seen for follow-up ophthalmology appointments. Out of these 139 children, 27 (19.4%) children had strabismus, 10 (7.2%) children had amblyopia, 8 (5.8%) children had optic nerve atrophy, 1 (0.7%) child had macular dragging, and 8 (5.7%) children had myopia. Rates of myopia, strabismus, and amblyopia were significantly different among infants with no ROP, type 1, and type 2 ROP (
Summary of visual impairment in 139 participants seen for eye examination at the University of California, Los Angeles.
Myopia | 83 | 0 (0.0) | 38 | 4 (10.8) | 18 | 4 (22.2) | <0.0001 |
Strabismus | 83 | 10 (12.0) | 38 | 10 (26.3) | 18 | 7 (38.9) | 0.015 |
Amblyopia | 83 | 0 (0.0) | 38 | 5 (13.2) | 18 | 5 (27.8) | <0.0001 |
Optic nerve atrophy | 83 | 2 (2.4) | 38 | 3 (7.9) | 18 | 3 (16.7) | 0.050 |
Macular dragging | 83 | 0 (0.0) | 38 | 1 (2.63) | 18 | 0 (0.0) | 0.262 |
Any visual impairment | 83 | 10 (12.0) | 38 | 13 (34.2) | 18 | 9 (50.0) | <0.0001 |
Given that neurodevelopmental assessment data was available over 0–36 months, neurodevelopmental outcomes were grouped and assessed separately at three different time points. One hundred and niney-one infants completed neurodevelopmental assessments at 0–12 months corrected age, 142 infants completed neurodevelopmental assessments at 12–24 months corrected age, and 59 infants completed neurodevelopmental assessments at 24–36 months corrected age. Neurodevelopmental information for infants in each age group are represented in
Summary of Bayley-III neurodevelopmental scores for infants assessed at 0–12, 12–24, and 24–36 months.
Cognition | 117 | 101.0 (17.1) | 55 | 93.8 (19.6) | 19 | 85.0 (17.1) | 8.05 | 0.0004 |
Language | 117 | 94.6 (12.9) | 55 | 88.8 (13.3) | 19 | 89.9 (16.7) | 3.96 | 0.021 |
Motor | 117 | 95.2 (19.6) | 55 | 87.9 (19.5) | 19 | 80.2 (18.0) | 6.29 | 0.002 |
Cognition | 85 | 97.7 (15.6) | 40 | 91.3 (19.8) | 17 | 84.1 (17.8) | 5.30 | 0.006 |
Language | 85 | 90.1 (16.3) | 40 | 84.2 (20.7) | 17 | 78.4 (18.0) | 3.78 | 0.025 |
Motor | 85 | 91.3 (18.1) | 40 | 82.1 (19.0) | 17 | 79.0 (20.7) | 5.17 | 0.007 |
Cognition | 28 | 94.5 (18.9) | 22 | 86.6 (18.0) | 9 | 79.4 (15.3) | 2.73 | 0.074 |
Language | 26 | 88.0 (17.3) | 21 | 77.6 (18.3) | 9 | 75.6 (10.7) | 3.00 | 0.058 |
Motor | 28 | 84.7 (19.2) | 22 | 80.1 (17.2) | 9 | 72.0 (16.8) | 1.71 | 0.191 |
Boxplots of Bayley-III Neurodevelopmental Composite Scores, by univariable analysis (uncorrected for covariates). Boxplot lines and rectangles indicate the median and 1.5 × interquartile range below the first quartile or above the third quartile. The navy circles represent scores falling outside of that range, though all scores were included in analyses. Asterisks represent significant differences in Bayley-III neurodevelopmental scores between infants without ROP and infants with type 1 or type 2 ROP as measured by
Fifty-nine children were seen for neurodevelopmental assessment at 24–36 months, which was less than the number of infants assessed at 0–12 and 12–24 months. Children identified as high-risk because of continued significant medical and developmental concerns which necessitate a higher level of care coordination continue to receive neurodevelopmental assessments after the age of 24 months at the UCLA High Risk Infant Follow-up Clinic, whereas children who are categorized as lower risk because of reassuring improvements in their medical conditions and neurodevelopmental testing scores are “graduated” from the clinic around 24 months of age. In our cohort, children seen for neurodevelopmental assessment at 24–36 months had higher rates of type 1 ROP and type 2 ROP compared to children not assessed at 24–36 months (
Gestational age, birth weight, diagnosis of IVH, and diagnosis of BPD were significantly different between infants with type 1 ROP, type 2 ROP, or no ROP in all three age groups and insurance type was significantly different between infants with type 1 ROP, type 2 ROP, or no ROP at ages 0–24 months (0–12 months: gestational age
The relationships between ROP severity and Bayley-III composite cognition, language, and motor scores for each age group were assessed using ANOVA. ROP severity was related to worse Bayley-III cognition, language, and motor scores at ages 0–12 months (cognition:
When comparing Bayley-III composite scores between infants with any ROP (type 1 or 2) to infants without ROP, infants with ROP were significantly more likely to have lower cognition, language, and motor scores at 0–12 months (cognition:
In order to assess if ROP was independently related to neurodevelopmental outcomes after co-varying for risks associated with poor neurodevelopmental outcomes in premature infants, we performed multivariable analysis using linear mixed effects models (for continuous Bayley scores) and generalized linear mixed effect models (for dichotomized Bayley scores- moderate to severe impairment compared to no or mild impairment). Lower birth weight, higher IVH grade, male sex and public insurance were identified as independently associated with worse Bayley scores. Moreover, given that there were differences in Bayley scores based upon the age at testing, this variable was also included in the model (
To provide clinically interpretable results, a Glimmix model for the dichotomized outcome of moderate to severe impairment vs. no or mild impairment was performed using the variables identified as significant in the multivariable mixed effect model (
Summary of odds ratios and 95% confidence intervals for variables considered in the generalized linear mixed effect (Glimmix) model for each Bayley domain.
Insurance ( |
0.320 (0.050–2.042) | 0.246 (0.067–0.898) |
0.124 (0.027–0.558) |
Age at Assessment ( |
1.172 (0.557–2.466) | 2.795 (1.501–5.205) |
1.070 (0.604–1.894) |
BW ( |
0.998 (0.995–1.001) | 0.998 (0.996–1.000) |
0.998 (0.996–1.000) |
IVH ( |
7.961 (1.147–55.244) |
1.927 (0.593–6.263) | 4.755 (1.266–17.859) |
Sex ( |
14.2 (1.788-113.247) |
6.358 (1.692-23.888) |
8.663 (1.921-39.073) |
Receiver operator curves (ROC) from the generalized linear mixed effect model for having moderate to severe impairment vs. no to mild impairment in cognitive
In summary, this study found that having ROP is not associated with worse neurodevelopmental outcomes as assessed by Bayley developmental testing after adjusting for important factors associated with prematurity-related poor neurodevelopmental outcomes (birthweight, IVH, male sex, and insurance status). These results confirmed our hypothesis that poorer neurodevelopmental outcomes in preterm neonates are most likely related to co-morbidities related to younger gestational age at birth and socioeconomic determinants of premature birth, not severity of ROP itself.
Counseling parents and caregivers of preterm neonates on the long-lasting effects of ROP can be challenging (
Interestingly, studies showing significant relationships between ROP and worse neurodevelopmental outcomes often evaluated neurodevelopmental performance earlier in childhood than those that did not demonstrate an association (
In our study population, ROP severity did not relate to neurodevelopmental outcomes at 0–36 months corrected age after co-varying for birth weight, IVH, male sex, and insurance type. A previous study by Glass et al. had a similar study design and found that infants with severe ROP had poorer Bayley-III cognition and motor scores at 18 months corrected age, after controlling for gestational age. Differences in categorizing ROP severity may explain the conflicting findings. Glass et al. (
The limitations of this study include that this is a single center retrospective study, and the small sample size and potential selection bias of the infants assessed at 24–36 months (
Currently, one of the most challenging aspects for clinicians working in the NICU is counseling parents on the likelihood of neurodevelopmental impairment in preterm neonates diagnosed with ROP. Our results emphasize that ROP is not associated with worse neurodevelopment performance at 0–36 months corrected age after adjusting for co-variates known to be associated with worse neurodevelopmental outcomes in preterm infants and despite infants with ROP having more visual impairments. Our study supports the overarching theme that the more premature and lower birth weight a neonate is/has, the higher risk they are for medical co-morbidities, including ROP, as well as worse neurodevelopmental outcomes. However, the co-morbidity of ROP itself does not appear to contribute to neurodevelopmental impairment.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
The studies involving human participants were reviewed and approved by the Institutional Review Board at UCLA. The UCLA IRB granted waiver of consent.
IT, AC, and MS created study design and statistical methods. AK and MA collected data. MA, BG, and MS performed statistical analyses. MA wrote the manuscript. MA, IT, AC, and MS contributed to data interpretation and critical revision of the manuscript. All authors contributed to the article and approved the submitted version.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Our team would like to give special thanks to Yasmeen Dhindsa, B.A. at the David Geffen School of Medicine at UCLA for her contribution to this manuscript and she does not have conflicts of interest to report.
The Supplementary Material for this article can be found online at:
Average Bayley scores in infants assessed at 0–12, 12–24, and 24–36 months. Bars represent least square means with standard errors estimated from the mixed effect model adjusting for insurance, sex, BW, and IVH.