This cohort study investigates the association between socioeconomic and clinical factors and neurodevelopmental outcomes in a cohort of infants screened for retinopathy of prematurity.
Key Points
Question
What is the association between socioeconomic and clinical factors and neurodevelopmental outcomes in a relatively large, diverse cohort of infants screened for retinopathy of prematurity?
Findings
In this cohort study including 706 infants, public health insurance was associated with an increased risk of neurodevelopmental impairment in the cognitive, language, and motor domains. Infants treated with either laser or anti–vascular endothelial growth factor (VEGF) therapy had worse neurodevelopmental outcomes in unadjusted analysis, and in a multivariable model, treatment type was no longer associated with worse neurodevelopmental outcomes.
Meaning
This study highlights the potential role of socioeconomic factors in the neurodevelopment of high-risk infants and supports the early neurodevelopmental safety of anti-VEGF treatment in infants diagnosed with retinopathy of prematurity.
Abstract
Importance
Preterm infants screened for retinopathy of prematurity (ROP) are at risk for heterogenous neurodevelopment outcomes that are difficult to predict.
Objective
To characterize the potential association between socioeconomic and clinical risk factors and neurodevelopmental outcomes in a diverse, multicenter cohort of premature neonates screened for ROP.
Design, Setting, and Participants
This was a retrospective cohort study using electronic medical records and US Census Bureau income data. This study was performed at academic (University of California, Los Angeles [UCLA] Mattel Children’s Hospital and UCLA Santa Monica Hospital), community (Cedars-Sinai Medical Center), and LA county (Harbor-UCLA Medical Center) neonatal intensive care units. Participants included infants who met American Academy of Pediatrics guidelines for ROP screening and had records from at least 1 Bayley Scales of Infant and Toddler Development (BSID) neurodevelopment assessment between 0 and 36 months of adjusted age. Data analyses were conducted from January 1, 2011, to September 1, 2022.
Exposures
Demographic and clinical information, proxy household income, and health insurance type were collected as risk factors.
Main Outcomes and Measures
Neurodevelopmental outcomes in the cognitive, language, and motor domains measured via BSID were the primary outcomes.
Results
A total of 706 infants (mean [SD] age, 28.6 [2.4] weeks; 375 male [53.1%]) met inclusion criteria. In a multivariable model, which included adjustments for birth weight, sex, insurance type, intraventricular hemorrhage (IVH), and age at assessment, public health insurance was associated with a 4-fold increased risk of moderate to severe neurodevelopmental impairment (NDI) in cognitive and language domains (cognitive, odds ratio [OR], 3.65; 95% CI, 2.28-5.86; P = 8.1 × 10−8; language, OR, 3.96; 95% CI, 2.61-6.02; P = 1.0 × 10−10) and a 3-fold increased risk in the motor domain (motor, OR, 2.60; 95% CI, 1.59-4.24; P = 1.4 × 10−4). In this adjusted model, clinical factors that were associated with an increased risk of moderate to severe NDI included lower birth weight, diagnosis of IVH, male sex, and older age at time of Bayley assessment. In unadjusted analyses, infants who received either laser or anti-VEGF treatment, compared with infants without treatment-requiring ROP, had lower BSID scores in multiple domains at 0 to 12 months, 12 to 24 months, and 24 to 36 months (DATA). In the multivariable model, treatment type was no longer associated with worse neurodevelopmental outcomes in any domain.
Conclusions and Relevance
Study results suggest an association between public insurance type and NDI in a diverse population screened for ROP, indicating the complexities of neurodevelopment. This study also supports the early neurodevelopmental safety of anti-VEGF treatment, as anti-VEGF therapy was not found to be independently associated with worse NDI in any domain.
Introduction
Neurodevelopmental impairment (NDI) in preterm infants ranges from mild motor and cognitive delay to severe cerebral palsy, with approximately 50% of infants born before 28 weeks of gestational age (GA) being affected to some degree.1 The heterogeneity of long-term outcomes makes it challenging for health care practitioners to counsel families. Influencing factors for NDI include intraventricular hemorrhage (IVH), birth weight (BW), as well as enrollment in early intervention programs.2,3,4
It has been proposed that neonates diagnosed with severe retinopathy of prematurity (ROP) are at risk of worsened neurologic outcomes, including NDI.5,6 Significant early and late visual impairment in infants with ROP has been proposed as a potential mechanism underlying increased NDI in this population.7 In addition, it has been proposed that the association between NDI and ROP may be independent of visual outcomes and rather associated with the interruption of common pathways of both retinal and brain development in premature neonates.8 A recent single-site study9 by our group demonstrated that severe ROP was not independently associated with worse neurodevelopmental outcomes; rather, associations between severe NDI and ROP were due to shared mediators or confounders such as low BW.
This study aimed to better understand the associations between socioeconomic and clinical factors and neurodevelopmental outcomes in a diverse cohort of infants screened for ROP in Los Angeles County. We hypothesized that socioeconomic factors, including household income and insurance type, could be associated with adverse neurodevelopmental outcomes in a multicenter cohort of infants screened for ROP.
Methods
The University of California, Los Angeles (UCLA), institutional review board approved the study protocol and granted a waiver of consent owing to the use of deidentified patient data. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
Study Participants
This was a multicenter, retrospective, cohort study performed at UCLA Mattel Children’s Hospital, UCLA Santa Monica Medical Center, Cedars Sinai Medical Center, and Harbor-UCLA Medical Center. Included in the study were all neonates screened for ROP while hospitalized at these institutions between January 1, 2011, and September 1, 2022. We identified a cohort of infants who qualified for ROP screening based on the American Academy of Pediatrics (AAP) guidelines, which included infants born at a GA of 30 weeks or less, BW of 1500 g or less, or a GA at birth greater than 30 weeks but with an unstable clinical course deemed to confer higher risk for ROP.10 Infants who met AAP guidelines for ROP screening and had records from at least 1 Bayley Scales of Infant and Toddler Development (BSID) neurodevelopment assessment between 0 and 36 months of adjusted age were included in the study.
Clinical and Demographic Data
Data were collected from electronic medical records. Demographic data collected included sex, race, ethnicity, primary language, zip code, and health insurance type. Clinical data included GA, BW, Bayley scores, and medical comorbidities including bronchopulmonary dysplasia (BPD), IVH, necrotizing enterocolitis (NEC), patent ductus arteriosus (PDA), and sepsis. Participant race and ethnicity information was included in this study to better characterize potential social factors that may impact neonatal and neurodevelopmental outcomes. Race and ethnicity were self-reported. The neonate’s parent or other guardian indicated if they identified as Asian, Black, White, or unknown or other (included patients who self-reported as Native American or Pacific Islander), and indicated if they identified as Hispanic or non-Hispanic. Health insurance type at time of admission to the neonatal intensive care unit (NICU) was designated as public insurance (Medi-Cal or Medicaid) or private insurance. Household income was determined by proxy using the neonate’s reported zip code, which was matched with data from the US Census Bureau to obtain median household income by zip code.11
ROP Screening
Neonates were screened for ROP in accordance with AAP guidelines.10 Neonates were grouped into 3 cohorts based on ROP outcome: (1) neonates without treatment-requiring ROP (ie, no ROP or low-grade ROP), (2) neonates who received laser monotherapy for type 1 ROP, and (3) neonates who received anti–vascular endothelial growth factor (VEGF) treatment (bevacizumab or ranibizumab) for type 1 ROP (with or without laser therapy). Treatment type was determined by ROP severity, with neonates with posterior zone II and zone I disease treated with anti-VEGF therapy, and neonates with less severe disease receiving laser monotherapy. No neonates in this cohort required primary vitrectomy for ROP treatment.
Neurodevelopment Outcomes
The BSID, third edition (BSID-III), assesses cognitive, language, and motor domains.12 Composite scores in these 3 domains are used to evaluate neurodevelopmental progression. BSID-III assessments were performed at NICU follow-up clinics across the 4 hospitals. Across the cohort, infants received a variable number of BSID-III assessments at variable time intervals. To account for this variability, neurodevelopmental assessments were categorized into 3 corrected age groups: 0 to 12, 12 to 24, and 24 to 36 months. If an infant had more than 1 assessment completed within a single age group, the more recent assessment was used. All univariable analyses were performed independently within each age group.
Statistical Analysis
The primary outcome measures in this study were individual composite scores within each Bayley domain (cognitive, language, motor) and the risk of moderate to severe NDI in each domain, defined as a Bayley composite score of less than 85 or more than 1 SD below the mean.
For each age group, differences in demographic and clinical outcome data between ROP outcome groups were assessed using analysis of variance and χ2 tests, for continuous and categorical variables, respectively. In univariable analyses, analysis of variance was used to assess the association between ROP outcome group (no treatment, laser treatment, and anti-VEGF treatment) and Bayley composite cognition, language, or motor scores within each age group. Post hoc comparisons were performed using the Bonferroni correction to determine adjusted significance.
In multivariable analyses, we sought to further evaluate the association between ROP outcome and neurodevelopmental impairment after controlling for potential confounders for poor neurodevelopmental outcomes. Linear mixed-effects modeling (for the continuous outcome measure of composite Bayley scores among all age groups) was used to identify variables independently associated with worse neurodevelopmental scores. Correlations among Bayley scores from each infant were accounted for as a random effect. Model selection steps involved backward elimination based on likelihood ratios. Variables that were identified as associated with Bayley scores in this model were then included in a mixed-effects logistic regression model along with the variable of interest, ROP outcome. Odds ratios (ORs) and 95% CIs were calculated for the binary outcome of moderate to severe NDI in each Bayley domain (cognitive, language, motor). All P values were 2-sided, and P values were not adjusted for multiple analyses. The cutoff for statistical significance was P < .05. All statistical analyses were performed using R software, version 4.1.1 (R Project for Statistical Computing).
Results
A total of 1366 neonates were screened for ROP at the 4 hospital systems studied between January 1, 2011, and September 1, 2022. Of these, 660 infants were excluded due to lack of available documentation of a Bayley neurodevelopment assessment, leaving 706 remaining infants (375 male [53.1%]; 331 female [46.9%]) meeting inclusion criteria (eFigure 1 in Supplement 1).
Demographic and Clinical Data
Our study cohort of 706 infants had a mean (SD) GA of 28.6 (2.4) weeks (range, 22.3-38.9 weeks) and a mean (SD) BW of 1123 (348) g (range, 360-3501 g). Proxy mean (SD) annual household income of the cohort was $80 044 ($30 633; range, $25 800-$209 531). Parents identified the following race and ethnicity categories for their infants: 87 non-Hispanic Asian (12.3%), 85 non-Hispanic Black (12.0%), 175 Hispanic (24.8%), 308 non-Hispanic White (43.8%), and 50 non-Hispanic unknown or other race (7.1%). A total of 208 infants (29.5%) had public health insurance, 219 (31.0%) had BPD, 166 (23.5%) had IVH, 54 (9.1%) had NEC, 236 (33.4%) had PDA, and 122 (17.3%) had sepsis.
Neurodevelopmental assessments were completed at 0 to 12 months in 565 infants, at 12 to 24 months in 379 infants, and at 24 to 36 months in 238 infants. In all age groups, GA and BW varied by ROP outcome (Table 1, Table 2, and Table 3). Proxy median household income, race and ethnicity, and insurance type also varied by ROP outcome in all age groups (Table 1, Table 2, and Table 3). Specifically, neonates who received laser or anti-VEGF treatment for type 1 ROP were born at earlier GAs, had lower BWs, had lower median household incomes, and had higher rates of public health insurance. Sex did not vary by ROP outcome in any age group. Medical comorbidities including BPD, IVH, NEC, and PDA varied by ROP outcome in all 3 age groups (Table 1, Table 2, and Table 3). Sepsis varied by ROP outcome in the groups 0 to 12 months and 12 to 24 months but did not vary in the group 24 to 36 months (Table 1, Table 2, and Table 3).
Table 1. Demographic and Clinical Characteristics for Infants Receiving Bayley Assessments at 0 to 12 Months.
| Variable | No treatment (n = 520) | Laser monotherapy for type 1 ROP (n = 20) | Anti-VEGF treatment for type 1 ROP (n = 25) | Test statistic (F or χ2)a | P valuea |
|---|---|---|---|---|---|
| Bayley age: 0-12 mo | |||||
| Gestational age, mean (SD), wk | 28.9 (2.2) | 26.0 (2.3) | 24.8 (1.2) | 54.89 | <.001 |
| Birth weight, mean (SD), g | 1159 (333) | 771 (193) | 660 (111) | 40.86 | <.001 |
| Household income, mean (SD), thousands of dollars/y | 81.92 (30.76) | 81.59 (34.19) | 65.81 (23.22) | 3.30 | .04 |
| Sex, No. (%)b | |||||
| Female | 237 (45.6) | 12 (60.0) | 11 (44.0) | 1.66 | .44 |
| Male | 283 (54.4) | 8 (40.0) | 14 (56.0) | ||
| Infant race and ethnicity, No. (%)b | |||||
| Asian | 69 (13.3) | 3 (15.0) | 3 (12.0) | 16.64 | .03 |
| Black | 67 (12.9) | 1 (5.0) | 3 (12.0) | ||
| Hispanic | 97 (18.7) | 9 (45.0) | 9 (36.0) | ||
| Non-Hispanic White | 249 (47.9) | 4 (20.0) | 8 (32.0) | ||
| Otherc | 38 (7.3) | 3 (15.0) | 2 (8.0) | ||
| Insurance type, No. (%) | |||||
| Medicaid/Medi-Cal | 124 (23.8) | 11 (55.0) | 10 (40.0) | 12.62 | .002 |
| Private | 396 (76.2) | 9 (45.0) | 15 (60.0) | ||
| Medical comorbidities, No. (%) | |||||
| BPD | 135 (26.0) | 11 (55.0) | 17 (68.0) | 27.44 | <.001 |
| IVH | 113 (21.7) | 7 (35.0) | 14 (56.0) | 16.94 | <.001 |
| NEC | 37 (7.1) | 3 (15.0) | 7 (28.0) | 14.86 | <.001 |
| PDA | 157 (30.2) | 14 (70.0) | 21 (84.0) | 42.77 | <.001 |
| Sepsis | 75 (14.4) | 5 (25.0) | 10 (40.0) | 12.93 | .002 |
Abbreviations: BPD, bronchopulmonary dysplasia; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis; PDA, patent ductus arteriosis; ROP, retinopathy of prematurity; VEGF, vascular endothelial growth factor.
For continuous variables, refers to F statistic derived from univariate analysis of variance tests. For categorical variables, refers to χ2 statistic derived from χ2 tests. P values correspond to adjacent F and χ2 values.
Sex was reported as assigned at birth. Race and ethnicity were self-reported by each neonate’s guardian.
Other race and ethnicity included Native American or Pacific Islander.
Table 2. Demographic and Clinical Characteristics for Infants Receiving Bayley Assessments at 12 to 24 Months.
| Variable | No treatment (n = 338) | Laser monotherapy for type 1 ROP (n = 22) | Anti-VEGF treatment for type 1 ROP (n = 19) | Test statistic (F or χ2)a | P valuea |
|---|---|---|---|---|---|
| Bayley age: 12-24 mo | |||||
| Gestational age, mean (SD), wk | 28.6 (2.4) | 25.5 (1.7) | 25.1 (1.4) | 39.30 | <.001 |
| Birth weight, mean (SD), g | 1135 (364) | 745 (193) | 657 (158) | 28.20 | <.001 |
| Household income, mean (SD), thousands of dollars/y | 80.32 (29.92) | 71.34 (28.94) | 58.70 (22.94) | 5.53 | .004 |
| Sex, No. (%)b | |||||
| Female | 150 (44.4) | 15 (68.2) | 11 (57.9) | 5.76 | .06 |
| Male | 188 (55.6) | 7 (31.8) | 8 (42.1) | ||
| Infant race and ethnicity, No. (%)b | |||||
| Asian | 36 (10.7) | 2 (9.1) | 2 (10.5) | 19.26 | .01 |
| Black | 44 (13.0) | 0 (0) | 3 (15.8) | ||
| Hispanic | 84 (24.9) | 13 (59.1) | 9 (47.4) | ||
| Non-Hispanic White | 152 (45.0) | 5 (22.7) | 4 (21.1) | ||
| Other race and ethnicityc | 22 (6.5) | 2 (9.1) | 1 (5.3) | ||
| Insurance type, No. (%) | |||||
| Medicaid/Medi-Cal | 88 (26.0) | 14 (63.6) | 11 (57.9) | 21.49 | <.001 |
| Private | 250 (74.0) | 8 (36.4) | 8 (42.1) | ||
| Medical comorbidities, No. (%) | |||||
| BPD | 106 (31.4) | 14 (63.6) | 13 (68.4) | 19.2 | <.001 |
| IVH | 76 (22.5) | 12 (54.5) | 14 (73.7) | 33.04 | <.001 |
| NEC | 28 (8.3) | 6 (27.3) | 5 (26.3) | 13.63 | .001 |
| PDA | 105 (31.1) | 16 (72.7) | 15 (78.9) | 31.71 | <.001 |
| Sepsis | 48 (14.2) | 7 (31.8) | 8 (42.1) | 14.00 | <.001 |
Abbreviations: BPD, bronchopulmonary dysplasia; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis; PDA, patent ductus arteriosis; ROP, retinopathy of prematurity; VEGF, vascular endothelial growth factor.
For continuous variables, refers to F statistic derived from univariate analysis of variance tests. For categorical variables, refers to χ2 statistic derived from χ2 tests. P values correspond to adjacent F and χ2 values.
Sex was reported as assigned at birth. Race and ethnicity were self-reported by each neonate’s guardian.
Other race and ethnicity included Native American or Pacific Islander.
Table 3. Demographic and Clinical Characteristics for Infants Receiving Bayley Assessments at 24 to 36 Months.
| Variable | No treatment (n = 211) | Laser monotherapy for type 1 ROP (n = 15) | Anti-VEGF treatment for type 1 ROP (n = 12) | Test statistic (F or χ2)a | P valuea |
|---|---|---|---|---|---|
| Bayley age: 24-36 mo | |||||
| Gestational age, mean (SD), wk | 28.5 (2.3) | 25.3 (0.9) | 24.8 (1.5) | 29.12 | <.001 |
| Birth weight, mean (SD), g | 1111 (330) | 750 (107) | 707 (107) | 17.39 | <.001 |
| Household income, mean (SD), thousands of dollars/y | 79.66 (34.49) | 56.43 (19.70) | 62.86 (21.98) | 4.60 | .01 |
| Sex, No. (%)b | |||||
| Female | 93 (44.1) | 9 (60.0) | 6 (50.0) | 1.54 | .46 |
| Male | 118 (55.9) | 6 (40.0) | 6 (50.0) | ||
| Infant race and ethnicity, No. (%)b | |||||
| Asian | 25 (11.8) | 1 (6.7) | 2 (16.7) | 17.05 | .03 |
| Black | 25 (11.8) | 1 (6.7) | 0 (0) | ||
| Hispanic | 62 (29.4) | 9 (60.0) | 8 (66.7) | ||
| Non-Hispanic White | 87 (41.2) | 2 (13.3) | 2 (16.7) | ||
| Other race and ethnicityc | 12 (5.7) | 2 (13.3) | 0 (0) | ||
| Insurance type, No. (%) | |||||
| Medicaid/Medi-Cal | 75 (35.5) | 11 (73.3) | 5 (41.7) | 8.53 | .01 |
| Private | 136 (64.5) | 4 (26.7) | 7 (58.3) | ||
| Medical comorbidities, No. (%) | |||||
| BPD | 65 (30.8) | 11 (73.3) | 9 (75.0) | 19.53 | <.001 |
| IVH | 50 (23.7) | 6 (40.0) | 8 (66.7) | 12.06 | .002 |
| NEC | 14 (6.6) | 4 (26.7) | 3 (25.0) | 11.10 | .004 |
| PDA | 58 (27.5) | 12 (80.0) | 12 (100) | 41.14 | <.001 |
| Sepsis | 39 (18.5) | 5 (33.3) | 4 (33.3) | 3.28 | .19 |
Abbreviations: BPD, bronchopulmonary dysplasia; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis; PDA, patent ductus arteriosis; ROP, retinopathy of prematurity; VEGF, vascular endothelial growth factor.
For continuous variables, refers to F statistic derived from univariate analysis of variance tests. For categorical variables, refers to χ2 statistic derived from χ2 tests. P values correspond to adjacent F and χ2 statistics.
Sex was reported as assigned at birth. Race and ethnicity were self-reported by each neonate’s guardian.
Other race and ethnicity included Native American or Pacific Islander.
Univariable Analyses
The association between ROP outcome (no treatment, laser treatment, and anti-VEGF treatment) and neurodevelopmental examination scores in BSID-III domains (cognitive, language, and motor) were analyzed in univariate analyses within each corrected age cohort. In the cohort of infants with a neurodevelopmental assessment between 0 to 12 months, Bayley scores in all 3 domains varied by ROP outcome (Table 4; eFigure 2A-C in Supplement 1). Post hoc analyses in this cohort revealed that infants who received laser treatment had lower cognitive (mean [SD] score with no treatment = 99.6 [14.3]; mean [SD] score with laser monotherapy = 86.0 [19.2]; P < .001) and motor (mean [SD] score with no treatment = 94.4 [17.7]; mean [SD] score with laser monotherapy = 78.4 [19.8]; P < .001) scores compared with infants without treatment-requiring ROP and that infants who received anti-VEGF treatment had lower cognitive (mean [SD] score with no treatment = 99.6 [14.3]; mean [SD] score with anti-VEGF therapy = 91.4 [17.8]; P = .02), language (mean [SD] score with no therapy = 93.8 [12.9]; mean [SD] score with anti-VEGF therapy = 87.0 [16.8]; P = .03), and motor (mean [SD] score with no treatment = 94.4 [17.7]; mean [SD] score with anti-VEGF therapy = 84.3 [19.7]; P = .02) scores compared with infants without treatment-requiring ROP. Post hoc comparisons of infants treated with laser vs anti-VEGF revealed no differences in cognitive, language, or motor scores.
Table 4. Bayley Neurodevelopmental Scores for Infants Assessed at 0 to 12, 12 to 24, and 24 to 36 Months of Age.
| Test | No treatment | Laser monotherapy for type 1 ROP | Anti-VEGF treatment for type 1 ROP | F | P valuea | |||
|---|---|---|---|---|---|---|---|---|
| No. | Mean (SD) | No. | Mean (SD) | No. | Mean (SD) | |||
| Bayley age: 0-12 mo | ||||||||
| Cognitive | 520 | 99.6 (14.3) | 20 | 86.0 (19.2) | 25 | 91.4 (17.8) | 11.46 | <.001 |
| Language | 520 | 93.8 (12.9) | 20 | 89.6 (13.5) | 25 | 87.0 (16.8) | 4.10 | .02 |
| Motor | 520 | 94.4 (17.7) | 20 | 78.4 (19.8) | 25 | 84.3 (19.7) | 11.11 | <.001 |
| Bayley age: 12-24 mo | ||||||||
| Cognitive | 338 | 95.2 (16.2) | 22 | 85.0 (15.7) | 19 | 82.6 (13.2) | 9.15 | <.001 |
| Language | 338 | 89.2 (17.0) | 22 | 80.0 (15.4) | 19 | 77.3 (15.2) | 7.20 | <.001 |
| Motor | 338 | 90.9 (17.3) | 22 | 79.1 (18.9) | 19 | 81.3 (17.7) | 7.04 | <.001 |
| Bayley age: 24-36 mo | ||||||||
| Cognitive | 211 | 95.8 (16.3) | 15 | 85.0 (13.5) | 12 | 81.8 (21.4) | 6.84 | .001 |
| Language | 211 | 86.6 (16.7) | 15 | 77.9 (13.5) | 12 | 79.4 (11.1) | 2.96 | .05 |
| Motor | 211 | 89.2 (14.9) | 15 | 77.2 (14.2) | 12 | 77.3 (20.7) | 7.45 | <.001 |
Abbreviations: ROP, retinopathy of prematurity; VEGF, vascular endothelial growth factor.
P values correspond to adjacent F statistics, which are derived from univariate analysis of variance models comparing neurodevelopmental scores between the 3 ROP treatment groups (no treatment, laser treatment, and anti-VEGF treatment).
In the cohort of infants with a neurodevelopmental assessment between 12 and 24 months, Bayley scores in all domains varied by ROP outcome group (Table 4; eFigure 2A-C in the Supplement). Post hoc analyses in this cohort revealed that infants who received laser treatment had lower cognitive (mean [SD] score with no treatment = 95.2 [16.2]; mean [SD] score with laser monotherapy = 85.0 [15.7]; P = .01), language (mean [SD] score with no treatment = 89.2 [17.0]; mean [SD] score with laser monotherapy = 80.0 [15.4]; P = .04), and motor (mean [SD] score with no treatment = 90.9 [17.3]; mean [SD] score with laser monotherapy = 79.1 [18.9]; P = .007) scores compared with infants without treatment-requiring ROP and that infants who received anti-VEGF treatment had lower cognitive (mean [SD] score with no treatment = 95.2 [16.2]; mean [SD] score with anti-VEGF therapy = 82.6 [13.2]; P = .003) and language (mean [SD] score with no treatment = 89.2 [17.0]; mean [SD] score with anti-VEGF therapy = 77.3 [15.2]; P = .008) scores compared with infants without treatment-requiring ROP. Post hoc comparisons of infants treated with laser vs anti-VEGF therapy revealed no differences in cognitive, language, or motor scores.
In the 24- to 36-month cohort, cognitive and motor scores varied by ROP outcome. Language scores did not vary by ROP outcome (Table 4; eFigure 2A-C in Supplement 1). Post hoc analyses revealed that infants who received laser treatment had lower cognitive (mean [SD] score with no treatment = 95.8 [16.3]; mean [SD] score with laser monotherapy = 85.0 [13.5]; P = .04) and motor (mean [SD] score with no treatment = 89.2 [14.9]; mean [SD] score with laser monotherapy = 77.2 [14.2]; P = .01) scores compared with infants without treatment-requiring ROP, and infants who received anti-VEGF treatment had lower cognitive (mean [SD] score with no treatment = 95.8 [16.3]; mean [SD] score with anti-VEGF therapy = 81.8 [21.4]; P = .01) and motor (mean [SD] score with no treatment = 89.2 [14.9]; mean [SD] score with anti-VEGF therapy = 77.3 [20.7]; P = .02) scores compared with infants without treatment-requiring ROP. Post hoc comparisons of infants treated with laser vs anti-VEGF revealed no differences in cognitive, language, or motor scores in this age group.
Multivariable Analysis
To further explore the association between ROP outcomes and neurodevelopmental outcomes, mixed-effects logistic regression was used to model the risk of moderate-severe NDI in each Bayley domain. In multivariable analysis, lower BW, male sex, public health insurance, IVH, and older age at neurodevelopment assessment were independently associated with worse neurodevelopmental outcomes (Table 5).
Table 5. Multivariable Analysis Results.
| Variable | OR (95% CI)a | ||
|---|---|---|---|
| Cognitive | Language | Motor | |
| Insurance (public vs private) | 3.65 (2.28-5.86) | 3.96 (2.61-6.02) | 2.60 (1.59-4.24) |
| Birth weight (per 100-g increase) | 0.91 (0.84-0.97) | 0.95 (0.90-1.00) | 0.91 (0.85-0.98) |
| IVH | 2.61 (1.61-4.24) | 1.85 (1.21-2.83) | 3.11 (1.85-5.24) |
| Sex (female vs male) | 0.36 (0.22-0.57) | 0.52 (0.36-0.76) | 0.40 (0.26-0.64) |
| Age at Bayley assessment (with each increment of 0-12, 12-24, 24-36 mo) | 1.34 (1.05-1.71) | 1.73 (1.41-2.11) | 1.18 (0.94-1.47) |
| Laser treatment (vs no treatment) | 1.82 (0.74-4.46) | 1.33 (0.56-3.13) | 3.28 (1.17-9.19) |
| Anti-VEGF treatment (vs no treatment) | 2.01 (0.81-4.99) | 1.91 (0.80-4.58) | 1.43 (0.50-4.08) |
Abbreviations: IVH, intraventricular hemorrhage; OR, odds ratio; VEGF, vascular endothelial growth factor.
ORs and 95% CIs for predictors included in the mixed-effect logistic regression models, where the outcome is the risk of moderate to severe neurodevelopmental impairment within each Bayley domain (defined as a score more than 1 SD below the mean).
When evaluating socioeconomic factors that may impact neurodevelopmental outcomes, public insurance status was found to be associated with a 4-fold higher risk of moderate to severe neurodevelopmental impairment in the cognitive and language Bayley assessment domains (cognitive, OR, 3.65; 95% CI, 2.28-5.86; P = 8.1 × 10−8; language, OR, 3.96; 95% CI, 2.61-6.02; P = 1.0 × 10−10) and a 3-fold increased risk in the and motor Bayley assessment domain (motor, OR, 2.60; 95% CI, 1.59-4.24; P = 1.4 × 10−4) (Table 5). Illustration of Bayley composite score in the cognitive, language, and motor domains by insurance type can be found in eFigure 3A to C in Supplement 1. Race and ethnicity and proxy median household income were not independently associated with lower Bayley scores and were thus not included in the final adjusted model.
When analyzing clinical factors, a 100-g increase in BW was associated with a 0.09, 0.05, and 0.09 reduction, respectively, in likelihood of moderate to severe NDI in the cognitive, language, and motor domains (cognitive, OR, 0.91; 95% CI, 0.84-0.97; P =.007; language, OR, 0.95; 95% CI, 0.90-1.00; P = .07; motor, OR, 0.91; 95% CI, 0.85-0.98; P = .01). Infants with IVH were approximately 2.6, 1.9, and 3.1 times more likely, respectively, to have moderate to severe NDI in the cognitive, language, and motor domains (cognitive, OR, 2.61; 95% CI, 1.61-4.24; P = 9.7 × 10−5; language, OR, 1.85; 95% CI, 1.21-2.83; P = .004; motor, OR, 3.11; 95% CI, 1.85-5.24; P = 1.9 × 10−5), and female sex was associated with a lower risk of moderate to severe NDI in all domains (cognitive, OR, 0.36; 95% CI, 0.22-0.57; P = 1.7 × 10−5; language, OR, 0.52; 95% CI, 0.36-0.76; P = 6.2 × 10−4; motor, OR, 0.40; 95% CI, 0.26-0.64; P = 1.2× 10−4). Increased age at time of Bayley assessment was also associated with an increased risk of cognitive and language impairment (cognitive, OR, 1.34; 95% CI, 1.05-1.71; P = .02; language, OR, 1.73; 95% CI, 1.41-2.11; P = 1.1 × 10−7; motor, OR, 1.18; 95% CI, 0.94-1.47; P = .15) (Table 5). Anti-VEGF treatment was not associated with increased risk of moderate to severe NDI in any of the Bayley domains when compared with no treatment (Table 5). Laser treatment, when compared with no treatment, was not associated with an increased risk of adverse NDI in the cognitive and language domains; however, laser treatment was associated with increased risk of moderate to severe motor impairment (OR, 3.28; 95% CI, 1.17-9.19; P = .02) (Table 5). BPD, NEC, PDA, and sepsis were not independently associated with lower Bayley scores and were thus not included in the final adjusted model.
Discussion
A prior study from our group highlighted the association of socioeconomic factors, such as insurance type and household income, with ROP diagnosis and severity.13 We found that although Hispanic neonates had more severe ROP, lower GA rather than race itself was the primary driver of this health care disparity. The aim of the current study was to evaluate the association of socioeconomic factors with neurodevelopmental outcomes in this same cohort that included preterm infants from a large, safety net county hospital in Los Angeles (Harbor-UCLA Medical Center) as well as a community NICU with an ethnically diverse population (Cedars Sinai Medical Center) as a better reflection of Los Angeles County’s racial and ethnic community.
Results suggest that infants with public health insurance were approximately 4 times more likely to have worse neurodevelopmental outcomes in the cognitive and language domains as compared with their counterparts with private health insurance, and approximately 4 times more likely to have worse neurodevelopmental outcomes in the motor domain. In our population of vulnerable infants screened for ROP, insurance type was more associated with neurodevelopmental outcomes than proxy median household income. Although certainly public health insurance may come with various barriers to timely access to health care, it is important to acknowledge that public health insurance status may reflect social and historical disadvantages other than simply access to care, including financial instability, parental educational opportunities, race and ethnicity, and low familial support.14,15,16,17 A study by Juul et al18 examining 21 risk factors found that higher maternal education was the most significant predictor of higher Bayley scores. Furthermore, lower household income has been shown to be associated with poorer neurocognitive performance in infants and adolescents due to lack of educational stimulation, highlighting the association between socioeconomic factors and neurodevelopmental outcomes.19,20 In summary, a neonate’s public health insurance status is reflective of a myriad of socioeconomic factors that likely in turn influence neonates’ access to not only health care but also to resources that may in turn affect their neurodevelopment, including access to early childhood education, long-term rehabilitation and therapy services, food security, and familial support.
Obtaining a better understanding of factors correlating neurodevelopmental outcomes in infants screened for ROP, particularly infants found to have severe ROP requiring treatment, can help better inform both clinical decision-making as well as caregiver counseling. Our study results suggest that lower BW, diagnosis of IVH, increased GA at time of Bayley assessment, and male sex were independently associated with increased risk of adverse neurodevelopmental outcomes. There are conflicting studies regarding the association of the diagnosis of ROP itself with neurodevelopmental outcomes, with a study by Ahn et al6 demonstrating the association of ROP with poor neurodevelopmental outcomes independent of other factors associated with prematurity (such as GA, sex, BPD, and IVH). Alternate studies7,9 have disputed the idea that ROP is a risk factor for NDI; rather, these studies argue that factors including GA, BW, and comorbidities of prematurity such as IVH are the determining factors of NDI in infants with ROP. Our study supports these findings, as we found that, after adjustment for potential confounders, infants who were treated for type 1 ROP did not demonstrate increased risk for neurodevelopmental impairment in cognitive or language domains when compared with untreated infants (ie, infants without type 1 ROP), underscoring the association of other clinical factors with NDI in premature infants, including BW and diagnosis of IVH.
Furthermore, our study examined if ROP treatment (ie, primary laser or primary anti-VEGF treatment) affected neurodevelopmental outcomes. Previous studies21,22 have shown that infants with ROP treated with anti-VEGF injections are at increased risk of severe neurodevelopmental disabilities but these studies did not adjust for socioeconomic factors, including insurance type, and comorbidities of prematurity, including IVH, and did not monitor infants over a period of time after treatment. Multiple recent studies23,24,25,26,27 have shown that infants treated with anti-VEGF monotherapy were not at increased risk of adverse neurodevelopmental outcomes and furthermore showed that infants who received anti-VEGF treatment had similar neurocognitive outcomes as treatment-naive infants 2, 4, and 6 years into childhood. Using a substantially larger, ethnically diverse cohort with over 700 infants in a multicenter approach, our study results further suggest the absence of an association between anti-VEGF treatment and adverse neurodevelopmental outcomes in infants with ROP. These findings are integral in further supporting the safety of anti-VEGF injections for ROP and in providing assurance to the relatively low risk of treatment.
Limitations
A limitation of our study was that only a subset of infants screened for ROP had Bayley data available, which may introduce selection bias for more compliant families. In addition, we found that older age at the time of neurodevelopmental assessment was associated with an increased risk of NDI in the cognitive and language domains, which likely reflects the fact that infants considered higher risk for NDI are the infants who are followed up to older ages. Furthermore, when examining income data, we used median household income via zip code as a proxy for the infants’ household income, which may overestimate or underestimate an infant’s true household income; as such, it is difficult to discern the exact influence of income on neurodevelopmental outcomes. Lastly, there is inherent bias that is present in BSID scores, as evaluators are not masked to patients’ medical histories, and infants may be evaluated by different practitioners at each BSID evaluation, although generally interrater reliability for BSID scoring is high.28
Conclusions
In summary, results of this cohort study suggest that both clinical and socioeconomic factors were associated with neurodevelopmental outcomes in infants screened for ROP. Clinical factors include those associated with prematurity, such as lower BW and diagnosis of intraventricular hemorrhage. Insurance type was the most influential risk factor on neurodevelopmental outcomes in our cohort of infants screened for ROP. Furthermore, our study aligns with previous studies that suggest that infants with severe ROP requiring anti-VEGF treatment were not at increased risk of early neurodevelopmental impairment, and our results suggest the safety of this treatment from a neurodevelopmental aspect.
eFigure 1. Flow Chart Representing Study Cohort Inclusion and Cohort Breakdown by ROP Classification, Birth Hospital, and Insurance Type
eFigure 2. Box Plots of Bayley Composite Scores by ROP Outcome
eFigure 3. Box Plots of Bayley Composite Scores by Insurance Type
Data Sharing Statement
References
- 1.Colvin M, McGuire W, Fowlie PW. Neurodevelopmental outcomes after preterm birth. BMJ. 2004;329(7479):1390-1393. doi: 10.1136/bmj.329.7479.1390 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Bolisetty S, Dhawan A, Abdel-Latif M, Bajuk B, Stack J, Lui K; New South Wales and Australian Capital Territory Neonatal Intensive Care Units’ Data Collection . Intraventricular hemorrhage and neurodevelopmental outcomes in extreme preterm infants. Pediatrics. 2014;133(1):55-62. doi: 10.1542/peds.2013-0372 [DOI] [PubMed] [Google Scholar]
- 3.Hadders-Algra M. Early diagnostics and early intervention in neurodevelopmental disorders-age-dependent challenges and opportunities. J Clin Med. 2021;10(4):861. doi: 10.3390/jcm10040861 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Linsell L, Malouf R, Morris J, Kurinczuk JJ, Marlow N. Prognostic factors for poor cognitive development in children born very preterm or with very low birth weight: a systematic review. JAMA Pediatr. 2015;169(12):1162-1172. doi: 10.1001/jamapediatrics.2015.2175 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Choi YJ, Hong EH, Shin YU, Bae GH, Kim I, Cho H. Severe retinopathy of prematurity associated with neurodevelopmental disorder in children. Front Pediatr. 2022;10:816409. doi: 10.3389/fped.2022.816409 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Ahn JH, Lee KM, Kim MJ, et al. Neurodevelopmental outcomes in very low birthweight infants with retinopathy of prematurity in a nationwide cohort study. Sci Rep. 2022;12(1):5053-5053. doi: 10.1038/s41598-022-09053-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Beligere N, Perumalswamy V, Tandon M, et al. Retinopathy of prematurity and neurodevelopmental disabilities in premature infants. Semin Fetal Neonatal Med. 2015;20(5):346-353. doi: 10.1016/j.siny.2015.06.004 [DOI] [PubMed] [Google Scholar]
- 8.Sveinsdóttir K, Ley D, Hövel H, et al. Relation of retinopathy of prematurity to brain volumes at term equivalent age and developmental outcome at 2 years of corrected age in very preterm infants. Neonatology. 2018;114(1):46-52. doi: 10.1159/000487847 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Altendahl M, Sim MS, Kokhanov A, Gundlach B, Tsui I, Chu A. Severe retinopathy of prematurity is not independently associated with worse neurodevelopmental outcomes in preterm neonates. Front Pediatr. 2021;9:679546. doi: 10.3389/fped.2021.679546 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Fierson WM; American Academy of Pediatrics Section on Ophthalmology; American Academy of Ophthalmology; American Association for Pediatric Ophthalmology and Strabismus; American Association of Certified Orthoptists . Screening examination of premature infants for retinopathy of prematurity. Pediatrics. 2018;142(6):e20183061. doi: 10.1542/peds.2018-3061 [DOI] [PubMed] [Google Scholar]
- 11.United States Census Bureau . Explore Census data. Accessed December 18, 2022. https://data.census.gov/cedsci/
- 12.Albers CA, Grieve AJ. Test review: Bayley, N. (2006). Bayley Scales of Infant and Toddler Development, 3rd ed. San Antonio, TX: Harcourt Assessment. J Psychoeduc Assess. 2007;25(2):180-190. doi: 10.1177/0734282906297199 [DOI] [Google Scholar]
- 13.Karmouta R, Altendahl M, Romero T, et al. Association between social determinants of health and retinopathy of prematurity outcomes. JAMA Ophthalmol. 2022;140(5):496-502. doi: 10.1001/jamaophthalmol.2022.0667 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lee DC, Liang H, Shi L. The convergence of racial and income disparities in health insurance coverage in the US. Int J Equity Health. 2021;20(1):96. doi: 10.1186/s12939-021-01436-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Bettenhausen JL, Hall M, Colvin JD, Puls HT, Chung PJ. The effect of lowering public insurance income limits on hospitalizations for low-income children. Pediatrics. 2018;142(2):e20173486. doi: 10.1542/peds.2017-3486 [DOI] [PubMed] [Google Scholar]
- 16.Hair NL, Hanson JL, Wolfe BL, Pollak SD. Low household income and neurodevelopment from infancy through adolescence. PLoS One. 2022;17(1):e0262607. doi: 10.1371/journal.pone.0262607 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Donohue JM, Cole ES, James CV, Jarlenski M, Michener JD, Roberts ET. The US Medicaid program: coverage, financing, reforms, and implications for health equity. JAMA. 2022;328(11):1085-1099. doi: 10.1001/jama.2022.14791 [DOI] [PubMed] [Google Scholar]
- 18.Juul SE, Wood TR, German K, et al. ; PENUT Consortium . Predicting 2-year neurodevelopmental outcomes in extremely preterm infants using graphical network and machine learning approaches. EClinicalMedicine. 2022;56:101782. doi: 10.1016/j.eclinm.2022.101782 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Dennis E, Manza P, Volkow ND. Socioeconomic status, BMI, and brain development in children. Transl Psychiatry. 2022;12(1):33. doi: 10.1038/s41398-022-01779-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Tomasi D, Volkow ND. Associations of family income with cognition and brain structure in USA children: prevention implications. Mol Psychiatry. 2021;26(11):6619-6629. doi: 10.1038/s41380-021-01130-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.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]
- 22.Arima M, Akiyama M, Fujiwara K, et al. Neurodevelopmental outcomes following intravitreal bevacizumab injection in Japanese preterm infants with type 1 retinopathy of prematurity. PLoS One. 2020;15(3):e0230678. doi: 10.1371/journal.pone.0230678 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Chou HD, Shih CP, Huang YS, et al. Cognitive outcomes following intravitreal bevacizumab for retinopathy of prematurity: 4- to 6-year outcomes in a prospective cohort. Am J of Ophthalmol. 2022;234:59-70. doi: 10.1016/j.ajo.2021.06.034 [DOI] [PubMed] [Google Scholar]
- 24.Zayek M, Parker K, Rydzewska M, Rifai A, Bhat R, Eyal F. Bevacizumab for retinopathy of prematurity: 2-year neurodevelopmental follow-up. Am J Perinatol. 2021;38(11):1158-1166. doi: 10.1055/s-0040-1710556 [DOI] [PubMed] [Google Scholar]
- 25.Wallace DK, Hercinovic A, Freedman SF, et al. ; Pediatric Eye Disease Investigator Group . Ocular and developmental outcomes of a dosing study of bevacizumab for retinopathy of prematurity. J AAPOS. 2023;27(1):10.e1-10.e8. doi: 10.1016/j.jaapos.2022.11.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Zhang DL, Yoon HH, deRegnier RO, Arzu J, Rahmani S. Postnatal growth trajectories and neurodevelopmental outcomes following bevacizumab treatment for retinopathy of prematurity. Clin Ophthalmol. 2022;16:2713-2722. doi: 10.2147/OPTH.S378520 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Rodriguez SH, Peyton C, Lewis K, et al. Neurodevelopmental outcomes comparing bevacizumab to laser for type 1 ROP. Ophthalmic Surg Lasers Imaging Retina. 2019;50(6):337-343. doi: 10.3928/23258160-20190605-01 [DOI] [PubMed] [Google Scholar]
- 28.Ranjitkar S, Kvestad I, Strand TA, et al. Acceptability and reliability of the Bayley Scales of Infant and Toddler Development III among children in Bhaktapur, Nepal. Front Psychol. 2018;9:1265. doi: 10.3389/fpsyg.2018.01265 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure 1. Flow Chart Representing Study Cohort Inclusion and Cohort Breakdown by ROP Classification, Birth Hospital, and Insurance Type
eFigure 2. Box Plots of Bayley Composite Scores by ROP Outcome
eFigure 3. Box Plots of Bayley Composite Scores by Insurance Type
Data Sharing Statement