Retinopathy of Prematurity: Advances in the Screening and Treatment of Retinopathy of Prematurity Using a Single Center Approach

Abstract

Purpose

To focus on the longitudinal evaluation of high-risk infants for the development of retinopathy of prematurity (ROP) at a single tertiary neonatal intensive care unit (NICU) and to evaluate evolving demographics of ROP and the transition of treatment-warranted disease.

Methods

Consecutive retrospective review was performed of all infants screened for ROP between 1990 and 2019 at the Jackson Memorial Hospital neonatal intensive care unit. All inborn infants meeting a birth criteria of <32 weeks gestational age (GA) or a birthweight (BW) of 1500 grams were included. Longitudinal demographic, diagnostic, and treatment data were reported.

Results

Between January 1, 1990 and June 20, 2019, a total of 25,567 examinations were performed and 7,436 patients were included. Longitudinal trends over three decades demonstrated decreasing incidence of ROP (p<0.05). While the mean BW and GA increased over three decades, patients with ROP demonstrated lower BW and GA over time (p<0.05). The prevalence of micro-premature infants (as defined by BW <750 grams) continues to rise over time. Micro-preemies demonstrated increasing severity of zone and stage grading, plus disease, and propensity to require treatment (p<0.05). The rate of progression of ROP to stage 4 and 5 disease has decreased over time, and there has been an associated increased adoption of intravitreal bevacizumab as primary and salvage therapy.

Conclusion

Understanding the evolution of ROP infants and treatment over time is critical in identifying high-risk infants and reducing the incidence of severe stage ROP. Micro-prematurity is one of the significant risk factors for treatment-warranted ROP that continues to increase as neonatal care improves.

INTRODUCTION

Retinopathy of prematurity (ROP) is a vascular disease primarily affecting the premature retina that can lead to severe visual impairment and blindness. Terry initially described the disease entity in 1942 as “retrolental fibroplasia” due to the appearance of eyes in end-stage ROP presenting with complex retinal detachment.^1–3 Since then, our understanding of ROP has transformed dramatically through landmark trials such as Cryotherapy for ROP (CRYO-ROP), the Early Treatment for ROP (ETROP), and Bevacizumab Eliminates the Angiogenic Threat of Retinopathy of Prematurity (BEAT-ROP).^4–6

ROP remains one of the leading causes of pediatric vision loss in the United States and worldwide, resulting in lifelong legal blindness. This disease blinds approximately 550 infants in the US and 50,000 infants worldwide per year.^7,8 The global prevalence of this devastating disease has spurred significant research in the area of prevention, including understanding the epidemiological landscape of the disease, optimizing neonatal intensive care unit (NICU) management, discovering newer technologies for screening and advances in treatment capabilities.^9–12 Despite these advances however, ROP continues to be a challenge in both developed and undeveloped nations. Increasing rates of infant prematurity have been complicated by a focus on micro-premature infants in the developed world and atypical presentations in less developed countries. Previously, these births had been deemed non-viable but now are routinely cared for within the advanced NICU system. The survivability of these infants has been a major driver for the increasing proportional incidence of both any stage ROP and treatment warranted threshold ROP.^13,14 For many recent studies this has been a major confounding variable. In this study, we were able to stratify for micro-prematurity and its associated risks and outcomes.

EPIDEMIOLOGY

ROP is a disease that primarily affects preterm infants born before 31 weeks gestational age (GA) and exacerbated by low birth weight (BW ≤1250g).^4,8 From 1981 to 2006, preterm births had been steadily increasing to a rate of 12–13% of all births in the United States.¹⁵ From 2007–2014, preterm birth rates saw a gradual decline driven by a reduction in births by teens and younger women.¹⁶ But recently, from 2014–2018, rates have again risen to 10% of all births in the United States.¹⁶

Multiple studies have examined the trends of ROP incidence across time in the U.S. typically focusing on public health hospitals and academic centers.^13,17–19 Most studies have reported ROP incidence as remaining steady or gradually increasing. Conversely, mean BW and GA in ROP infants have decreased over time.^13,17 Medical and technological advances in neonatal care are partly responsible for the increased survival of younger and smaller infants as well as the rising popularity of artificial reproductive technology (ART) leading to premature birth and multiple gestation leading to lower birth weight infants.^4,20–23 As a result, these factors have led to an increased prevalence of at-risk infants. Quinn and colleagues established that in the 1980s, 15.8% of preterm births weighed <750g and 43.8% were born at ≤ 27 weeks GA. These rates have risen dramatically in the past decade (2010–2019) to 33.4% and 68.1% respectively. These studies have determined that up to 68% of these low BW infants (<1250g) consistently develop some stage of acute ROP.^13,24

ROP CLASSIFICATION

The International Classification of ROP (ICROP) was developed in 1984 and aimed to classify ROP by stage, grade, and zone.²⁵ This classification established a uniform language for ophthalmologists. Moving forward, advances in classification have been led by consensus agreement among the ROP community. Using the ICROP criteria enabled a standardization of ROP diagnoses and ROP severity grading. Demonstration and illustration of ROP zone and stage severity are outlined in Figures 1 and 2, respectively.

Figure 1.

Retinopathy of prematurity zones. Fundus photograph demonstrating zone I as defined by a circle centered upon the optic nerve with a radius of twice the distance from the nerve to the macula, zone II as defined by the area from the edge of zone I to a circle with a radius extending from the optic nerve to the nasal ora serrata, and zone III as defined by the residual temporal crescent anterior beyond zone II.

Figure 2.

Retinopathy of prematurity staging criteria. (A) Stage 1 defined as a demarcation line. (B) Stage 2 defined as an elevated ridge of retinal tissue. (C) Stage 3 defined as a ridge of retinal tissue with extraretinal neovascularization or fibrovascular proliferation. (D) Stage 4 defined as extrafoveal retinal detachment or subtotal retinal detachment. (E) Stage 5 defined as total retinal detachment.

The ETROP defined treatment guidelines for a newly recognized threshold.⁶ The ETROP used a complex algorithm to stage ROP coupled with a treatment guideline. Prethreshold ROP was defined as zone I, any stage; zone II, stage 2 with plus; zone II, stage 3 without plus; zone II, stage 3 with plus but fewer than 5 contiguous or 8 composite clock hours of disease and threshold ROP was defined by zone I or II, stage 3 with 5 contiguous or 8 composite clock hours of disease with plus; zone I, any stage with plus. ETROP recommended treatment for type 1 ROP which was defined as zone I, any stage with plus disease, or zone I, stage 3 +/− plus disease, or zone II, stage 2 or 3 with plus disease were treated.⁶ Follow-up screening intervals were conducted according to the recommendations by the AAO/AAP as outlined by Table 1.

Table 1.

Follow-up Screening Guidelines.

1 Week or Less	1 to 2 Week	2 Week	2 to 3 Weeks
• Immature retina to zone I	• Immature vascularization to posterior zone II	• Stage 1 ROP in zone II	• Stage 1 or 2 ROP in zone III
• Immature retina extends into posterior zone II, near the boundary of zone I	• Stage 2 ROP in zone II	• Immature vascularization to zone II	• Regressing ROP in zone III
• Stage 1 or 2 ROP in zone I	• Unequivocally regressing ROP in zone I	• Unequivocally regressing ROP in zone II
• Stage 3 ROP in zone II
• The presence or suspected presence of aggressive posterior ROP

ROP - Retinopathy of prematurity

Adapted from the recommendations from the American Academy of Pediatrics and American Academy of Ophthalmology

Historically, the presence of vessel tortuosity and engorgement was noted to be associated with ROP as early as 1949.^26,27 As more authors began including the designation of ROP “plus” or “+” disease in their classification schemes, the popularity of the term increased. Finally, in 1984, the ICROP defined plus disease as “the vascular changes are so marked that the posterior veins are enlarged and the arterioles tortuous,”²⁵ and incorporated it into their classifications for ROP. Thereafter, the definition of plus disease underwent several modifications and quickly became as important as the zone and stage of ROP.²⁷ In the ETROP trial, plus disease was recognized for the first time as being critical in the evaluation of treatment warranted threshold ROP.^6,27 It was defined by comparison of the perineural vasculature to “gold” standard images. The presence of plus disease became a driving indication for treatment of ROP beginning in 2003.^6,27

Other examination risk factors for the development of treatment-warranted ROP have been identified by our group and others. The presence of a persistent tunica vasculosa lentis (TVL) has been noted to have higher rates of zone 1 disease and treatment-warranted disease in a case-matched study.²⁸ Light (versus medium or dark) fundus pigmentation has also been found to be associated with higher risk ROP.²⁹ Specifically, a lighter fundus pigmentation was associated with a diagnosis of plus disease, more posterior zone, higher stage disease, and treatment for ROP.²⁹ These additional risk factors help to identify at-risk infants and make NICU screening more timely and efficient.

CHANGES IN NEONATAL CARE

Advances in neonatal care have significantly impacted infant survival. Low birthweight and very low birthweight infants are now commonly surviving within the NICU resulting in a clear shift towards a unique at-risk subpopulation.^13,30,31 Furthermore, this population demonstrates significant comorbidities including multiparous gestation, bronchopulmonary dysplasia, intraventricular hemorrhages and necrotizing enterocolitis and coexisting infections that were much less commonly associated with prematurity in the prior two decades.^32–35

In 1956, Kinsey first discovered the link between prolonged exposure to oxygen rich environments and the proliferation of retrolental fibroplasia.³⁶ Multiple large-scale studies have since corroborated his findings. The Surfactant, Positive Pressure, and Pulse Oximetry Randomized Trial (SUPPORT)³⁷ found that lower oxygen target saturations of 85%−89% were not associated with a significant increase in mortality and were in fact associated with a lower rate of ROP as compared to the higher oxygen target saturation group of 91%−95%. Similarly, the Canadian Oxygen Trial (COT)³⁸ in 2013 confirmed that rates of death were similar among lower target and higher target oxygen saturation groups. In contrast, two major studies, the Benefits of Oxygen Saturation Targeting (BOOST II) study and the Neonatal Oxygenation Prospective Meta-analysis (NeOProM) collaboration have shown that mortality rates were higher in the lower target oxygen saturation group versus the higher saturation target group, although rates of ROP were improved in the lower target group.^39,40 These studies have not defined a consensus to oxygen target saturations, though current recommendations suggest targeting 91% to 95% saturation range as a way to both minimize mortality and maintain acceptable rates of ROP.^3,41

More widespread use of surfactant and maternal antenatal steroids has also been attributed to increased infant survival, and potentially a decreased incidence and severity of ROP over time.^42,43 Surfactant has been found to decrease postnatal time on mechanical ventilation, increase pulmonary stability, and decrease ROP incidence.^43,44 Antenatal steroids have also been proven to decrease rates of pulmonary complications as well as intraventricular hemorrhage (IVH), which in turn improves infant survival.^32,33 While these risk factors may increase the number of surviving pre-term infants, some studies suggest that reducing pulmonary complications and IVH may also reduce the incidence of ROP.^32,33

Internationally the World Health Organization (WHO) has targeted infant mortality broadly within multiple economically challenged regions. The focus of many of these interventions targets educational strategies that typically require minimal resources. In 1983, a focus on enhanced skin-to-skin contact, exclusive breastfeeding, and early recognition of perinatal complications was proposed to decrease infant mortality. Beginning in the early 2000s, the WHO published guidelines recommending this practice.^45,46 Utilizing these simple measures, multiple studies have demonstrated up to a 51% reduction of preterm neonatal deaths. These studies did not specifically investigate the impact of these strategies on the development of ROP.^45,46

Artificial reproductive technology (ART) has also been a significant factor in the increase in preterm births, multiple gestational pregnancies, and infants at-risk for ROP.^22,47 Each year, there is an annual increase in the rate of ART by at least 10% globally.⁴⁷ A number of investigations have confirmed the association between ART and the development of ROP. In one such study, 23% of all ART infants developed some stage of ROP.⁴⁸ Others have found that ART is associated with nearly a 5-fold increase in the risk of developing treatment-warranted ROP.²²

As these trends in neonatal medicine continue, we can expect rising number of pre-term infants at risk for ROP. Further, these advances are associated with earlier gestational age and lower birth weight premature infants.^13,14,49 This data supports the increasing need for targeted ROP screening. Like many areas within ophthalmology, the increasing burden will require unique solutions. One such approach is the incorporation of remote screening, both with and without the integration of artificial intelligence to evaluate both the quality of the obtained images and to classify these images based on risk.¹²

SCREENING GUIDELINES

Screening guidelines for ROP have been a persistent focus of discussion. In 1997, the American Academy of Pediatrics (AAP) and the American Academy of Ophthalmology (AAO) recommended routine screening for ROP in all infants younger than 28 weeks GA or less than 1500g BW.⁵⁰ This recommendation changed in 2006 when the criteria was set for all infants of GA 32 weeks or younger, or less than 1500g BW.⁵¹ In 2013, the AAP and AAO recommendations were updated again to include screening for infants 30 weeks or younger or less than 1500g BW.⁵² Presently, these guidelines remain essentially unchanged, with the caveat that high risk infants outside of the traditional screening parameters may benefit from screening at the discretion of the neonatologist.⁵³

Critically, the first ophthalmic screening is recommended at either six weeks post-conceptual age (even if earlier than 31 weeks) or at 31 weeks gestational age. Follow-up screening intervals are also pivotal in the determination of progression and/or risk. The AAP/AAO recommend 1 week or less follow-up for those infants with highest risk attributes, with extended screening intervals for moderate and lower risk examination findings (see Table 1).

CHALLENGES IN SCREENING

ROP is more than a complex medical disease, it is also a significant public health challenge. As our knowledge of the pathophysiology of ROP and its at risk patient population evolves, there continues to be an ongoing discussion on optimal screening and treatment guidelines.

Multiple different issues contribute to the challenges in ROP screening. The difficulty in determining the optimal guidelines is reflected by the range of different screening protocols across countries. For example, the Royal College of Paediatrics and Child Health (RCPCH) recommends screening all infants less than 32 weeks GA or less than 1501g BW.⁵⁴ In Canada, the recommendation is to screen any infant ≤ 30 and 6/7 weeks GA regardless of birthweight and any infant with a BW ≤1250g.⁵⁵ This screening is recommended to begin based on the post-menstrual age (GA plus chronological age).⁵⁵ Both of these recommendations, from independent economically developed nations, are different than those of the AAP/AAO.⁵³

An additional factor that has complicated ROP screening is the rapid advancement of neonatal care and evolving demographics of preterm infants in NICUs across the U.S. and the world. Longitudinal tracking of these demographics is difficult. Adding to the complexity is that many studies evaluating ROP risk factors are often regionally focused and have differed in outcomes suggesting that even regional factors may influence ROP risks. For example, the incidence of ROP has been reported to vary from as low as 15–20% to as high as 68%.^{13,17,24,56,57} Generalizability, therefore, becomes an issue.⁵⁸

Further complicating the challenge of ROP is the existence of both known and unknown variables as predictors of risk. This has led to the lack of universal algorithms or protocols that fully capture and predict with high sensitivity and specificity which infants will eventually develop ROP.^59,60 Factors beyond GA and BW have been proposed as helpful predictors for developing ROP, such as insulin-like growth factor and weight gain, but still need further validation.⁶¹

The complexity of screening has led multiple investigators to propose algorithms capable of predicting high risk ROP utilizing both natal and postnatal factors. Well known algorithms include CHOP-ROP (Children’s Hospital of Philadelphia-ROP), OMA-ROP (Omaha-ROP), WINROP (weight, insulin-like growth factor, neonatal ROP), DIGIROP (Digital ROP), and CO-ROP (Colorado-ROP).^62–66 In 2009, Lofqvist and colleagues validated a screening algorithm called WINROP using weekly measures of body weight, serum insulin-like growth factor (IGF-1) levels, GA, BW, and IGF-binding protein.⁶¹ Although utility of the complex WINROP algorithm was validated, recent external revalidation studies have demonstrated sensitivities as low as 55%.^61,67 The CHOP-ROP algorithm was described in 2011 by Binenbaum and colleagues. It utilizes a combination of BW, GA, and daily weight gain in order to stratify risk.^58,68 The sensitivity of predicting type 1 or 2 ROP was reported to be 98%, but the authors recommended further validation before widespread adoption.⁶⁸ More recently in 2019, the DIGIROP algorithm was proposed as a simplified method using BW and GA to predict ROP risk and was both internally and externally validated to be comparable in its accuracy to CHOP-ROP, OMA-ROP, and WINROP (utilizing area under the curve comparative analysis).⁶⁵ The abundance of ROP screening algorithms is helpful but clearly suggests that no single algorithm has yet been identified to meet both the clinical need and to maintain excellent sensitivity and specificity.⁶⁵

Lastly, generalized guidelines that aim to maximize the inclusion of at-risk infants can expose infants to a stressful exam^59,60 while more targeted guidelines may allow certain infants at risk for ROP to be delayed in their screening.⁵⁸ This same logic applies to screening intervals. Clearly, screening guidelines are critical to identify those at-risk infants who will require initial and ongoing screening to eliminate the late consequences of advanced ROP.

THE EVOLUTION OF ROP TREATMENT

In the 1970s, Japanese investigators first published several trials documenting the positive treatment impact of cryotherapy in acute stage ROP.^5,69,70 Prior to this, attempts at treating and preventing progression of ROP with other approaches had not succeeded⁷¹ with surgical intervention being the only treatment in very severe cases of what was then considered retrolental fibroplasia.⁷²

The CRYO-ROP trial in 1988 demonstrated a 50% reduction of unfavorable outcomes in ROP eyes treated by cryotherapy versus untreated eyes.⁵ Soon thereafter, several factors led to the adoption of laser photocoagulation over cryotherapy for treatment of ROP including both the complications of cryotherapy and the ease of treatment by way of indirect laser ophthalmoscope to maintain favorable anatomic and visual outcomes.^5,72–76 In our institution the transition from cryotherapy to targeted laser therapy occurred in the late 1990s. In the early 2000s, the ETROP study supported the importance of early treatment, utilizing laser photocoagulation to reduce the rates of unfavorable outcomes in high-risk ROP.⁶ Recognizing this outcome, the AAO shifted its treatment recommendations to laser photocoagulation as an initial therapy whenever possible.⁷⁶ As with any invasive treatment, long term complications may occur. For laser photocoagulation, visually unfavorable outcomes included permanent restriction of visual field, increased incidence of high myopia, amblyopia, cataracts, nyctalopia, pachyphakia, microcornea, and angle closure (PMAC).^77–80 These complications highlighted the need for alternative therapies and prompted much of the advanced investigations leading to the adoption of anti-VEGF therapy initially as a salvage therapy then ultimately as primary therapy for treatment-warranted ROP.⁴

Current treatment paradigms often include both anti-VEGF and/or laser photocoagulation. Although safety and efficacy trials are ongoing, several recently published trials, including BEAT-ROP and Ranibizumab versus laser therapy for the treatment of very low birthweight infants with retinopathy of prematurity (RAINBOW), have demonstrated positive structural outcomes from treatment with anti-VEGF.^4,81,82 Initial results looking at adverse effects have reported lower rates of post-treatment refractive error, anterior segment complications, and visual field constriction.⁸³ The major controversy in the direct use of intravitreal anti-VEGF has focused on the theoretical risk on developing vascular systems within the premature infant.⁴ Internationally, the standard of care of advanced treatment warranting ROP is intravitreal anti-VEGF while in the United States several academic centers continue to focus on laser ablative approaches to primary therapy.^83,84

PURPOSE OF THE STUDY

To focus on the longitudinal evaluation of high-risk infants for the development of ROP at a single tertiary NICU and to evaluate evolving demographics of ROP and the transition of treatment-warranted disease utilizing an integrated screening/treating surgeon approach enabling reduction in bias often associated with extended study timelines.

MATERIALS AND METHODS

This study was compliant with the Health Insurance Portability and Accountability Act of 1996, adhered to the tenets of the Declaration of Helsinki, and was approved by the Institutional Review Board of the University of Miami Miller School of Medicine.

SUBJECTS

A consecutive retrospective review was performed including all infants who received any examination(s) for ROP at the Jackson Memorial Hospital (JMH) Neonatal Intensive Care Unit (NICU). Eligible infants included all those screened between January 1, 1990 and June 20, 2019. All included infants were screened and evaluated for ROP by a trained retinal physician from the Bascom Palmer Eye Institute (BPEI)/University of Miami with the assistance of a trained nursing staff. A neonatologist and a dedicated ROP nurse identified all infants requiring screening. This dedicated ROP nurse was present at all times during the screening process alongside the physician. Infants deemed to be critically unstable and at high risk for mortality underwent deferred screening until cleared by the neonatology team. If an infant born from a multiple gestation birth required ROP screening, all NICU supported siblings were screened.

For the analysis of ROP characteristics, all inborn infants meeting a birth criteria of less than 32 weeks gestational age and/or a birth weight of 1500 grams (g) were included. All outborn infants, those born at outside institutions and subsequently transferred to our institution, were excluded in primary analysis and were reported separately. For the purpose of this analysis, micro-premature infants were defined as infants weighing <750g BW.⁸⁵

DATA COLLECTION

The BPEI/JMH data set demographic data was abstracted to include BW, GA, GA at time of examination, dates of initial examination and follow-up examinations, multiple gestation birth status, survival status, and birth location at primary institution or transfer from another institution. Screening data collected included laterality of eye, presence of ROP, zone, stage, number of clock hours of involvement, presence of plus disease, presence of tunica vasculosa lentis, degree of fundus pigmentation, concurrent ocular comorbidities, and clarity of view during examination. Each child was identified with the most advanced stage and zone reached at any time during the study window. Fundus pigmentation was graded on a scale of 1 to 3, with 1 indicating light fundus pigmentation, 2 indicating medium fundus pigmentation, and 3 indicating dark fundus pigmentation (see Figure 1).

Treatment related data included fluorescein angiography photography, fundus RetCam photography (Natus Medical Incorporated, Pleasanton, CA), type of treatment (cryotherapy, laser photocoagulation, intravitreal anti-vascular endothelial growth factor, surgery), whether a combination of treatments was required, need for retreatment, date of additional follow-up treatment, and time to retreatment if deemed necessary by the treating physician. All data for infants were collected at each examination continuing until time of discharge from the NICU, until ROP had regressed, or until the infant expired.

CLASSIFICATION AND TREATMENT

Disease classification and diagnosis were determined according to ICROP standards.²⁵ Plus disease was classified based on standards utilized by the ETROP and CRYO-ROP, which defined plus as a degree of vascular dilation and tortuosity of the posterior retinal blood vessels in more than two quadrants.^5,6 During this three decade study interval two primary therapies were utilized. Laser photocoagulation was the primary therapeutic approach during the first two study windows.⁶ Intravitreal bevacizumab injections became the primary therapeutic treatment during the third decade. Laser photocoagulation procedures were performed under intravenous sedation with ongoing NICU monitoring at the bedside. Intravitreal injections were performed at the bedside using topical anesthesia only following strict standardized protocols.^4,86,87 IVB treatment was delivered using a protocol focused on minimizing infection risk, while minimizing treatment related discomfort. The treatment protocol for bevacizumab utilized a staged approach. Infants were treated with anti-VEGF upon diagnosis with treatment-warranted ROP. Infants were then screened serially post-injection to assess either a positive response to treatment and/or a potential for advancement of high-risk characteristics (re-development of plus disease, secession of vascularization, alterations to higher stages, or inability to follow care due to social or geographic limitations). The treatment protocol utilized a sterile eyelid speculum that was placed under topical tetracaine hydrochloride 0.5% or proparacaine hydrochloride 0.5% for anesthesia. The eye(s) to be treated was then prepped with povidone iodine 5% and the injection of 0.625mg in 0.05ml of bevacizumab was given 1.5 mm posterior to the corneoscleral limbus utilizing direct control of the globe. To accomplish this the infant is held by the NICU nurse and given a pacifier for comfort. Beginning in 2014, a custom short 32-gauge 3/16-inch needle was used to inject (TSK Steriject, TSK Laboratory, Japan). Analysis of our injection data documented the safety of the use of these needles to eliminate risk to damage either the retina or the lens. For ROP with progression to subtotal or total retinal detachment, pars plana vitrectomy with or without lensectomy was performed.⁸⁶

STATISTICAL ANALYSIS

Comparisons of categorical variables was performed using the chi-squared test or Fisher’s exact test. Continuous variables were compared using Student’s t-test or analysis of variance (ANOVA) test. Correlations were analyzed using logistic regression with determination of odds ratio for categorical dependent variable or multivariable linear regression for continuous dependent variable. A p-value <0.05 was considered statistically significant. Statistical analysis was performed using StataIC 15.1 (College Station, TX).

RESULTS

Between January 1, 1990 and June 20, 2019, a total of 25,567 examinations were performed. For this analysis, 7,436 infants were included that met our study criteria requiring birth demographics and inborn status. For these 7,436 infants, the mean BW and mean GA were 1,101g and 28.4 weeks, respectively. This final cohort included 1,322 multiple gestational births (17.8%). ROP of any severity was diagnosed in 2,198 (29.6%) infants (see Table 3). For infants weighing ≤ 1250g, the proportion of infants diagnosed with ROP was 43.1%. The overall mean zone and stage severity was zone 2.0 and stage 1.8, respectively. The mean BW and mean GA for infants with ROP were significantly lower at 822.9 and 26.3 weeks, respectively (p < 0.001, Table 3), compared to the non-ROP population.

Table 3.

Characteristics of All Patients Diagnosed with ROP

Number of patients with ROP	2,198.0
ROP Incidence	29.6%
Gender (Number of male/total, %)	183/353	51.8%
Multiple Gestation (n, %)	354	16.1%	p = 0.015#
ROP Severity
Zone (mean, std dev)	2.0	0.37
Stage (mean, std dev)	1.8	0.75
Birth Weight
Mean, std dev (g)	822.9	407.3	p < 0.0001*
Proportion (n, %)
< 750g	935	44.6%	p < 0.0001#
751–1000g	725	34.6%
1001–1250g	314	15.0%
1251–1500g	101	4.8%
>1500g	23	1.1%
Gestational Age
Mean, std dev (weeks)	26.3	2.3	p < 0.0001*
Proportion (n, %)
<27w	1,275	60.8%	p < 0.0001#
27–32w	763	36.4%
>32w	60	2.9%

^*t-test

^#chi-square, all p-values comparing patients with ROP versus those without ROP

Abbreviations: n - number, g - grams, w - weeks, std dev - standard deviation

STRATIFICATION OF INFANTS BY AGE AND WEIGHT

Screened infants and ROP infants were stratified by BW (<750g, 750–999g, 1000–1249g, 1250–1500g, and >1500g) and GA (<27 weeks, 27–31 weeks, and >32 weeks, Tables 1 and 2). Comparing ROP infants to non-ROP infants, a significantly higher rate of births at lower BW and earlier GA were noted. Micro-premature infants (<750g BW)⁸⁵ comprised 44.6% of all infants with ROP. Conversely, a significantly lower proportion of ROP infants were born at larger BW (1.1% for >1500g BW). Evaluation of the stratification of GA in ROP are also presented in Tables 3. ROP infants demonstrated a higher proportion born at younger GA (60.8% for GA < 27 weeks versus 2.9% for GA >32 weeks) compared to the non-ROP cohort (18.8% for GA < 27 weeks versus 24.1% for GA > 32 weeks). In multivariable logistic regression including independent variables of BW, GA, TVL, and fundus pigmentation, both BW and GA were associated with increased association with ROP incidence (OR .998, p < 0.001 for BW and OR 0.54, p < 0.001 for GA).

Table 2.

Characteristics of All Study Patients

Total number of examinations	25,567
Total number of patients	9,124
Number of included patients	7,436
Gender (Number of male/total, %)	714/1348	53.0%
Outborn (n, %)	312	4.0%
Multiple Gestation (n, %)	1,322	17.8%
Birth Weight
Mean, std dev (g)	1,101	407.5
Proportion (n, %)
< 750g	1,692	23.5%
751–1000g	1,439	20.0%
1001–1250g	1,449	20.1%
1251–1500g	1,411	19.6%
>1500g	1,206	16.8%
Gestational Age
Mean, std dev (weeks)	28.4	3.1
Proportion (n, %)
<27w	2,231	31.0%
27–32w	3,676	51.1%
>32w	1,291	17.9%

Abbreviations: n - number, g - grams, w - weeks, std dev - standard deviation

EVOLUTION OF BIRTH WEIGHT AND GESTATIONAL AGE

For all infants enrolled and screened over a 30-year period, mean BW increased each decade cohort (1990–1999, 2000–2009, and 2010–2019) from 1076.3g to 1091.0g to 1150.4g, respectively. Mean GA for all infants also increased across decade cohorts from 28.3 weeks in the first two decades to 28.6 weeks in the last decade (Table 4). For infants diagnosed with ROP, the mean BW decreased with each decade (858.0g to 815.1g, to 780.7g, respectively). Similarly, the mean GA also decreased (26.7 weeks to 26.3 weeks to 25.8 weeks, respectively, Table 4).

Table 4.

Change in Patient Characteristics Over Time

		1990–1999		2000–2009		2010–2019
All Patients
Number of Total Patients	2626		2929		1881
Multiple Gestation (n, %)	401	15%	470	16%	451	24%	p < 0.001ǂ
Birthweight (grams, mean)	1076.3		1091.0		1150.4		p < 0.001*
Proportion (n, %)
< 750g	527	21%	730	26%	435	23%
750–999g	555	22%	551	19%	333	18%
1000–1249g	619	25%	514	18%	316	17%
1250–1500g	469	19%	567	20%	375	20%
>1500g	317	13%	485	17%	404	22%
GA (weeks, mean)	28.3		28.3		28.6		p < 0.001*
Proportion (n, %)
<27w	756	30%	912	32%	563	30%
27–32w	1317	53%	1401	49%	958	51%
>32w	415	17%	522	18%	354	19%
ROP Patients
Number of Patients (% total)	831	32%	894	31%	473	25%	p < 0.001ǂ
Multiple Gestation (n, %)	123	15%	139	16%	92	19%	p = 0.075ǂ
Birthweight (grams, mean)	858.0		815.1		780.7		p < 0.001*
Proportion (n, %)
< 750g	267	35%	411	47%	257	55%
750–999g	304	40%	283	33%	138	29%
1000–1249g	143	19%	123	14%	48	10%
1250–1500g	38	5.0%	40	4.6%	23	4.9%
>1500g	6	0.8%	12	1.4%	5	1.1%
GA (weeks, mean)	26.7		26.3		25.8		p < 0.001*
Proportion (n, %)
<27w	408	54%	529	61%	338	72%
27–32w	310	41%	325	37%	128	27%
>32w	39	5.2%	16	1.8%	5	1.1%
Non-ROP Patients
Number of Patients (% total)	1795	68%	2035	69%	1408	75%
Multiple Gestation (n, %)	278	15%	331	16%	359	25%	p < 0.001ǂ
Birthweight (grams, mean)	1275.5		1212.2		1172.0
Proportion (n, %)
< 750g	260	15%	319	16%	178	13%
750–999g	251	15%	268	14%	195	14%
1000–1249g	476	28%	391	20%	268	19%
1250–1500g	431	25%	527	27%	352	25%
>1500g	311	18%	473	24%	399	29%
GA (weeks, mean)	29.6		29.2		29.0
Proportion (n, %)
<27w	348	20%	383	19%	225	16%
27–32w	1007	58%	1076	55%	830	59%
>32w	376	22%	506	26%	349	25%

^ǂChi-square analysis comparing change across decades

^*ANOVA comparing change across time

Abbreviations: BW - birthweight, g - grams, GA - gestational age, n - number, std dev - standard deviation, w - weeks

In the non-ROP infant cohort, the proportion born at lower BW remained stable over time while the proportion of infants born at heavier BW increased across the 30-year period (Table 4). However, over the same 30-year period, infants with ROP born at lower BW increased over time. Within each decade, rates of multiparous births showed a continuous increase.

PLUS DISEASE

Plus disease was identified in 416 (18.9%) ROP infants (Table 5). The mean BW and GA of this group were 670.2g and 24.9 weeks, respectively. Both BW and GA were significantly reduced among infants with plus disease (p < 0.001) compared to all ROP infants. Additionally, a higher proportion of infants in the plus disease cohort were born at a BW < 750g (75.8%) and GA <27 weeks (87.4%) compared to the overall ROP cohort (44.6% with BW <750g and 60.8% GA <27 weeks). 935 infants weighing <750g BW were diagnosed with ROP and 298 of those infants were found to have plus disease (31.8%).

Table 5.

Characteristics of Patients Diagnosed with Plus Disease

Number of patients (% of ROP patients)	416	18.9%
Multiple Gestation (n, %)	75	18.0%	p = 0.236
Birth Weight
Mean, std dev (g)	670.2	160.4	p < 0.001*
Proportion (n, %)
< 750g	298	75.8%	p < 0.001#
751–1000g	76	19.3%
1001–1250g	18	4.6%
1251–1500g	-	0.0%
>1500g	1	0.3%
Gestational Age
Mean, std dev (weeks)	24.9	1.6	p < 0.001*
Proportion (n, %)
<27w	346	87.4%	p < 0.001ǂ
27–32w	47	11.9%
>32w	3	0.8%

^*t-test, birthweight and gestational age copmarison between patients with versus without diagnosis of plus disease

^#chi-square, incidence of plus disease in ROP infants with birthweight <750g versus >750g

^ǂchi-square, incidence of plus disease in ROP infants with gestational age <27w versus >27w

Abbreviations: BW - birthweight, g - grams, GA - gestational age, n - number, std dev - standard deviation, w - weeks

Plus disease evaluated with trend analysis over time is presented in Table 6. The incidence of plus disease is seen to rise from the 1990s to the 2000s but has stabilized in the last decade. Similar to all infants with ROP, those with plus disease have shown a gradual trend towards lower BW (p = 0.005 for BW over time) and earlier GA (p < 0.001 for GA over time) with an increasing percentage of <750g BW and <27 weeks GA infants. Significantly, a higher percentage of infants with plus disease were diagnosed with zone I ROP (48%) showing an increasing trend into the most recent decade (p < 0.001).

Table 6.

Characteristics Over Time for ROP Patients with Plus Disease

	1990–1999		2000–2009		2010–2019
Plus Disease
N (% of ROP patients)	98	12%	241	27%	77	16%	p < 0.001ǂ
BW (grams, mean, SD)	721.9	193.6	674.0	152.9	604.5	118.0
Proportion (n, %)
< 750g	53	65%	175	74%	70	91%	p = 0.005ǂ
750–999g	21	26%	48	20%	7	9.1%
1000–1249g	6	7.4%	12	5%	0	0%
1250–1500g	1	1.2%	0	0%	0	0%
GA (weeks, mean, SD)	25.4	2.2	24.9	1.5	24.4	1.1
Proportion (n, %)
<27w	65	78%	205	87%	76	99%	p < 0.001*
27–32w	15	18%	31	13%	1	1.3%
>32w	3	3.6%	0	0%	0	0%
Multiple Gestation (n, %)	17	17%	39	16%	19	25%	p = 0.236ǂ
Severity
% Zone 1 disease	4	4%	41	17%	37	48%	p < 0.001ǂ
% Stage 3+ disease	71	72%	181	75%	27	35%	p < 0.001ǂ

^ǂChi-square analysis comparing change across decades

^*Fisher’s exact comparing change across decades per group

Abbreviations: BW - birthweight, g - grams, GA - gestational age, n - number, SD - standard deviation, w - weeks

ROP SEVERITY

Infants with ROP were classified according to zone and stage. In this series, 145 infants (6.6%) developed ROP in zone I, 1883 infants (85.7%) in zone II, and 162 infants (7.4%) in zone III with a median grade of zone II ROP (Table 7). Zone 1 infants were on average 619.3g BW and 24.4 weeks GA, while zone III infants were, on average, 1023.9g and 28.2 weeks. The mean BW and GA of infants with zone I disease were significantly lower than zone II and zone III (p < 0.001). For all zones, the proportion of zone I infants displayed an increasing trend with each passing decade (p < 0.001), documenting the increasing severity of ROP over time.

Table 7.

ROP Zone and Stage Severity Over Time

Year	1990–1999				2000–2009				2010–2019				TOTAL
	N	%	BW	GA	N	%	BW	GA	N	%	BW	GA	N	%	BW	GA
Zone 1	11	1%	711.7	25.1	64	7%	616.4	24.4	70	15%	611.2	24.3	145	6.6%	619.3	24.4
Zone 2	713	86%	834.6	26.5	774	87%	815.6	26.3	396	84%	808.1	26.1	1883	86%	820.6	26.3
Zone 3	101	12%	1008.9	28.0	54	6%	1065.8	28.9	7	1%	932.1	27.1	162	7.4%	1023.9	28.2
Total	825		858.5	26.7	892		815.5	26.3	473		780.7	25.8	2190
									p < 0.001ǂ						p < 0.001#	p < 0.001#
Stage 1	275	33%	982.1	27.8	310	35%	927.2	27.4	217	46%	877.1	26.6	802	37%	931.6	27.3
Stage 2	423	51%	815.2	26.3	329	37%	800.4	26.2	220	47%	711.2	25.4	972	44%	784.9	26.0
Stage 3	122	15%	708.1	25.2	247	28%	700.3	25.1	36	8%	624.7	24.5	405	18%	694.7	25.1
Stage 4	6	1%	843.3	26.7	4	0%	570.0	24.3	0	0%			10	0.5%	734.0	25.7
Stage 5	1	0%	610.0	24.0	3	0%	595.0	24.3	0	0%			4	0.2%	600.0	24.3
Total	827		858.5	26.7	893		815.5	26.3	473		780.7	25.8	2193		823.2	26.3
									p < 0.001ǂ						p < 0.001#	p < 0.001#

^#ANOVA analysis comparing changes in birthweight and gestational age across zone and stage severity

^ǂChi-square analysis comparing change in incidence of severity groups across all 3 decades

Abbreviations: BW - birthweight, g - grams, GA - gestational age, n - number

Stage 1 disease was noted in 802 infants (36.5%), stage 2 in 972 infants (44.2%), stage 3 in 405 infants (18.4%), stage 4 in 10 infants (0.5%) and only 4 infants (0.2%) progressed to stage 5. Staging severity for ROP in our cohort demonstrated an inverse correlation with respect to mean BW and GA, with stage 3 infants being the smallest and youngest. Stages 4 and 5 demonstrated inconsistent trends likely due to low sample size. Stage 1 mean BW and GA were 931.6g and 27.3 weeks respectively, while stage 3 infants were 694.7g and 25.1 weeks (p < 0.001). Over time, mean BW and GA in infants continuously decreased, with the 2010–2019 decade demonstrating both the youngest and smallest infants diagnosed with stage 1, 2, and 3 ROP. Infants that progressed to total stage 5 retinal detachment (4 infants, 0.2%) and subtotal (stage 4 a/b) retinal detachment (10 infants, 0.5%) occurred only during the first two decades and were never associated to anti-VEGF treatment. No infants progressed to stage 4 or 5 ROP in the most recent decade.

Stratification of ROP severity as a function of time over the three-decade study window is presented in Table 8. Overall, ROP zone shifted more posteriorly along with an increase in initial staging. Within each decade, micro-preemies demonstrated the most posterior zone and the most severe stage of ROP and were noted to be increasing in prevalence particularly over the third decade of our study (Table 7).

Table 8.

Mean Zone and Stage Severity Stratified by BW and GA

	1990–1999		2000–2009		2010–2019
Birthweight (g)	Zone	Stage	Zone	Stage	Zone	Stage
<750	2.01	2.18	1.89	2.24	1.77	1.83
750–999	2.10	1.74	1.99	1.81	1.97	1.43
1000–1249	2.33	1.44	2.19	1.54	2.04	1.25
1250–1500	2.24	1.35	2.22	1.35	2.00	1.17
>1500	2.40	1.50	2.08	1.17	2.00	1.20
					p < 0.001#	p < 0.001#
Gestational Age (w)
<27w	2.06	2.04	1.89	2.19	1.81	1.74
27–32w	2.18	1.57	2.12	1.59	2.02	1.31
>32w	2.22	1.61	2.25	1.25	2.00	1.00
					p < 0.001#	p < 0.001#

^#ANOVA analysis comparing changes in mean zone and stage severity across birthweight and gestational age groups over all 3 decades

Abbreviations: BW - birthweight, g - grams, GA - gestational age, w - weeks

ROP TREATMENT

Treatment warranted ROP occurred in 256 infants. The post-conceptual age of initial treatment was 37.3 weeks for all treated ROP infants. Over time, the post-conceptual age of treatment-warranted ROP decreased from 38.4 weeks in the 1990s to 37.6 weeks in the 2000s to 36.1 weeks in the last decade (p < 0.001). Plus disease was present in 95% of treated infants (p < 0.001). Multiple gestational birth infants accounted for 19.9% of all treated infants. For treated ROP infants, the average BW and GA were significantly lower than averages of infants with ROP who did not require treatment (p < 0.001). The BW stratification of infants demonstrated that 77.4% of infants who were treated for ROP weighed <750g at birth (p < 0.001 compared to non-treated ROP infants). Similarly, 92.0% of infants treated for ROP were born at <27 weeks GA whereas only 1.2% were born at >32 weeks GA (p < 0.001 compared to non-treated ROP infants). Average zone and stage severity were higher for treated infants than non-treated infants with ROP (mean zone 1.75, p < 0.001 and mean stage 2.68, p < 0.001, see Table 9).

Table 9.

Characteristics of Treated ROP Patients

Treated Patients (n)	256
Adjusted Age (w, mean, std dev)	37.3	3.1
Re-Treated Patients (n)	41	16.0%
Adjusted Age (w, mean, std dev)	43.0	6.4
Treated Patients with Plus Disease (n, %)	238	93%	p < 0.001ǂ
Multiple Birth (n, %)	51	19.9%	p = 0.101#
Birth Weight (g, mean, std dev)	659.9	160.5	p < 0.001*
Proportion (n, %)
< 750g	192	77.4%	p < 0.001ǂ
751–1000g	50	20.2%
1001–1250g	4	1.6%
1251–1500g	1	0.4%
>1500g	1	0.4%
Gestational Age (weeks, mean, std dev)	24.8	1.6	p < 0.001*
Proportion (n, %)
<27w	230	92.0%	p < 0.001ǂ
27–32w	17	6.8%
>32w	3	1.2%
Patients by Primary Treatment Type
Laser Photocoagulation	189	73.8%
Anti-VEGF (n, %)	60	23.4%
Cryotherapy	6	2.3%
Combination Cryotherapy/Laser	1	0.4%
ROP Severity
Zone (mean, std dev)	1.75	0.43	p < 0.001*
Stage (mean, std dev)	2.68	0.69	p < 0.001*

^*T -test comparison of mean average of ROP infants who received treatment versus non-treated

^#Chi-square comparison of ROP infants who received treatment versus non-treated

^ǂFisher’s exact comparison of ROP infants who received treatment versus non-treated

Abbreviations: BW - birthweight, g - grams, GA - gestational age, n - number, std dev - standard deviation, VEGF - vascular endothelial growth factor, w - weeks

Primary therapy included 189 infants (73.8% of all ROP treated infants or 8.6% of all ROP infants) treated with laser photocoagulation. Intravitreal bevacizumab (IVB), beginning in 2011, was used as primary treatment in 60 infants (23.4% of all ROP treated infants or 2.7% of all ROP infants). Treatment over the three decades demonstrated an early transition from cryotherapy to laser photocoagulation. A rapid transition of care occurred with the introduction of IVB as primary treatment. This is seen graphically to have occurred within the last decade of our study (Figure 3). The last patient treated with laser photocoagulation as primary therapy for treatment warranted ROP was treated in 2011.

Figure 3.

Grading of Fundus Pigmentation. Reproduced with permission from Fan et al. Ophthalmology. Examples of fundus pigmentation grading to assess all screened patients with (A) representing examples of light (grade 1) fundus pigmentation, (B) medium (grade 2) fundus pigmentation, and (C) dark (grade 3) fundus pigmentation. Figure and caption reprinted with permission.

Intravitreal bevacizumab was the primary therapy for treatment warranted ROP in 60 infants (see Table 10). In this series, all 60 infants showed an immediate response to IVB when evaluating both plus disease and staging. Vascularization progressed peripherally in all 60 infants. 42 of our 60 infants required no further staged therapy, 18 of our infants met our criteria for supplemental IVB treatment and were treated within our protocol. In this series of 60 infants, no patients progressed to stage 4/5 ROP and no patients required surgical intervention via vitrectomy/scleral buckle. For this entire IVB cohort, the mean BW and GA was 593.4g and 24.4 weeks. This represented a significant decrease in mean BW and GA in comparison with infants receiving laser photocoagulation as primary treatment (see Table 10, p < 0.001 for both). Micro-prematurity (BW <750g) was present in 93.3% of infants treated with IVB which was markedly different from the laser photocoagulation group which included 67.4% of micro-premature infants (p = 0.001).

Table 10.

Characteristics of Eyes Receiving Anti-VEGF Versus Laser

Primary Treatment	Laser (n=190)	IVB (n=60)	p-value
Mean Birthweight (grams, std dev)	679 ± 169.3	593.4 ± 114.7	p = 0.0003*
Mean Gestational Age (weeks, std dev)	24.9 ± 1.6	24.4 ± 1.1	p = 0.031*
Mean Adjusted Age of Treatment (weeks, std dev)	37.7 ± 3.25	35.9 ± 2.28	p < 0.001*
Micropremature Infants (<750g BW)	128 (67.4%)	56 (93.3%)	p = 0.001#
Progression to Stage 4 or 5 Disease	3.4%	0.0%

^*T-test comparison of mean average of ROP infants who received treatment versus non-treated

^#Chi-square comparison of ROP infants who received treatment versus non-treated

Abbreviations: BW - birthweight, g - grams, IVB - intravitreal bevacizumab, n - number, std dev - standard deviation, VEGF - vascular endothelial growth factor

ROP screening intervals can be seen in Table 11. The average screening interval for all patients across the entire study was 14.3 days. Micro-premature infants were screened more frequently than all other age groups (12.8 days vs. 15.1 days, p < 0.001, Table 11).

Table 11.

Screening Interval Timing for Inborn ROP Infants

	Number of Patients	Time Between Screenings (Days)
All Patients	2,053	14.3
Birthweight Groups
< 750g	915	12.8	p < 0.001*
≥ 750g	1,050	15.1
Gestational Age
Groups
< 27w	1,216	13.0	p < 0.001*
≥ 27w	746	15.8

^*T-test analysis between groups

Abbreviations: g - grams, ROP - retinopathy of prematurity, w - weeks

TUNICA VASCULOSA LENTIS

Beginning in 2010, the presence of tunica vasculosa lentis on examination was recorded consistently by a single retina specialist. In the last decade of this study cohort, 102 infants (4.7% of ROP infants) were noted to have examination characteristics consistent with persistent TVL (Table 12). ROP was diagnosed in 74 of these infants (83.8%), with a mean BW and GA of 615.5g and 24.9 weeks, respectively. Plus disease was concomitantly present in 26 (25%) of infants with TVL.

Table 12.

Characteristics of All Patients with TVL on Exam

Total number of patients	102
Number of patients with ROP	74	73%
Number of patients with plus disease	26	25%
Birthweight
Mean, std dev (g)	615.5	138.7	p < 0.001*
Proportion (n, % of plus patients)
< 750g	62	83.8%	p < 0.001#
751–1000g	12	16.2%
1001–1250g	0	0.0%
1251–1500g	0	0.0%
>1500g	0	0.0%
Gestational Age
Mean, std dev (weeks)	24.9	1.5	p < 0.001*
Proportion (n, % of plus patients)
<27w	68	91.9%	p < 0.001#
27–32w	6	8.1%
>32w	0	0.0%
Multiple Gestation (n, % total)	11	10.8%	p = 0.374ǂ
Zone/Stage Severity
Zone (mean, std dev)	1.62	0.5	p < 0.001*
Stage (mean, std dev)	1.75	0.7	p = 0.369*

^*T-test comparison of tunica patients versus non-tunica patients

^#Fisher-exact comparison of tunica patients versus non-tunica patients

^ǂChi-square comparison of tunica patients versus non-tunica patients

Abbreviations: g - grams, n - number, std dev - standard deviation, TVL - tunica vasculosa lentis, w - weeks

In univariate analysis, there was an increased association with ROP when TVL was detected on examination (p < 0.001) that did not remain significant in multivariate analysis (independent variables including BW, GA, and fundus pigmentation). Mean zone of ROP in this patient subpopulation was 1.62 (as compared to mean zone 2.0 for ROP patient cohort). Mean stage of ROP was 2.19 (as compared to mean stage 1.8 for ROP patient cohort). Both zone and stage were more severe in comparison to the overall ROP infant cohort.

OUTBORN INFANTS

Outborn infants, those not born at our tertiary care hospital, were excluded from our primary analysis. The characteristics of transferred infants are shown in Table 13. The indication for transfer in all 312 infants was the need for advanced ophthalmic evaluation and care. The mean BW was 1,048.4g and mean GA was 27.9 weeks, which is lower than the mean averages for inborn screened infants (p = 0.026 and p = 0.009, respectively).

Table 13.

Characteristics of Outborn Patients

Total number of patients	312
Number of patients with ROP	103	33%	p = 0.191
Number of patients with plus disease	48	15%	p < 0.001#
Multiple Gestation (n, % total)	97	31.1%	p < 0.001#
Total number of outborn patients treated	46	44.7%	p < 0.001*
Number of infants requiring anti-VEGF	22	21.4%	p < 0.001*
Number of infants requiring laser	37	35.9%	p < 0.001*
Number of infants requiring surgery	5	4.9%	p < 0.001*
Average Screening Interval (days)	8.2
Birthweight
Mean, std dev (g)	1,048.4	381.3	p = 0.026*
Proportion (n, mean (g))
< 750g	81	26.6%	p = 0.453#
751–1000g	66	21.7%
1001–1250g	61	20.1%
1251–1500g	54	17.8%
>1500g	42	13.8%
Gestational Age
Mean, std dev (weeks)	27.9	3.2	p = 0.009*
Proportion (n, mean (w))
<27w	117	37.9%	p = 0.018#
27–32w	150	48.5%
>32w	42	13.6%
Zone/Stage Severity
Zone (n, % total ROP & outborn)
1	18	17.6%	p < 0.001ǂ
2	82	80.4%
3	2	2.0%
Overall mean, std dev	1.84	0.42	p < 0.001*
Stage (n, % total ROP & outborn)
1	31	30.4%	p < 0.001ǂ
2	22	21.6%
3	32	31.4%
4	13	12.7%
5	4	3.9%
Overall mean, std dev	2.38	1.16	p < 0.001*

^*T-test comparison of mean average of outborn versus inborn infants

^#Chi-square comparison of outborn versus inborn infants

^ǂFisher’s exact comparison of outborn versus inborn infants

Abbreviations: BW - birthweight, g - grams, GA - gestational age, n - number, ROP - retinopathy of prematurity, std dev - standard deviation, VEGF - vascular endothelial growth factor, w - weeks

Of the 312 transfers, 103 (33%) infants were diagnosed with ROP, and of those infants, 46.6% were found to have features consistent with plus disease on examination. ROP severity of outborn infants showed a higher proportion diagnosed with posterior zone (mean 1.84) and stage (mean 2.38) ROP; 17.6% of infants were diagnosed with zone 1 disease (p < 0.001), and 16.6% of infants with stage 4 or 5 disease (p < 0.001, see Table 13). Ultimately, 46 total infants (44.7% of all outborn infants with ROP, OR 59.4, p < 0.001) required treatment. In this cohort, 22 (21.4% of outborn infants with ROP) received treatment with anti-VEGF and 37 (35.9%) received laser photocoagulation, both representing a significantly higher rate of ROP treatment (p < 0.001) in comparison to our inborn infants. In 13 cases, severity of clinical presentation require some combination of IVB, laser, or surgical intervention.

DISCUSSION

The importance of understanding evolving trends in the high-risk ROP population is illustrated in our current longitudinal study of over 9,000 infants spanning 30 years. Previous investigators have studied demographic patterns in ROP infants, but these studies have been limited by either small numbers or shorter follow up leading to significant variations in the reported findings.^{13,17,18,30,49,56,88–90} While prior studies have employed meta-analyses to aggregate and compare data, such studies are limited by variations in study design, patient populations, and examiner characteristics.¹³ The present study is the largest to date analyzing trends in ROP focused on patient screening, diagnosis, and treatment within a single-center population over a 3-decade study window.

OVERVIEW

Study Infants

For all included and screened infants in the study, mean BW and GA was 1,101g and 28.4 weeks respectively. In the first two decades, a reclassification caused an apparent shift in both BW and GA (p < 0.001) that did not mirror our clinical experience with high-risk disease; essentially, amendments in screening protocols were expanded by the AAO/AAP, thereby shifting the screening age from a gestational age of 28 weeks or less to infants up to 32 weeks or less beginning in the early 2000s.^51–53 With this expanded screening criteria, a marked shift in the development of micro-premature infants was lost in the larger data set. Further, this screening change shifting both the mean BW and GA also minimized the recognition of multiple gestational births and the increased survivability of both our youngest and smallest infants.

Multiparous births increased from a rate of 16% to 24% between 1990 and 2019 (p < 0.001). This increase reflects the enhanced survivability of premature infants by way of advancements in neonatal care including the use of perinatal corticosteroids and antibiotics,⁹¹ enhanced ventilator technology, improved fetal and infant monitoring, and the understanding of the importance of optimization of supplemental oxygenation.⁹² Furthermore, an increased focus on artificial reproductive technology and improved high risk obstetrical support has increased multiple gestational births with the associated increase in preterm delivery.²²

ROP Infants

ROP incidence in our inborn screened cohort was 29.6% (2198/7436). When stratifying by infant BW, using <1250g as the cut-off (as was done in the ETROP trial),^6,72 the overall incidence in our cohort increased to 43.1%. This rate mirrors those found in other U.S. based studies,⁵⁷ yet remains lower than that seen in the CRYO-ROP study.^5,6 Across time, the incidence of ROP infants was essentially stable from the first to the second decade, then decreased in the last decade. This gradual decrease in ROP incidence has been seen in other longitudinal studies both in the U.S. as well as internationally when adjusted for differential classification criteria.^{13,18,30,56,90}

However, despite this overall decreasing incidence, a significant increase in infants with lower BW and younger GA categories were diagnosed with ROP over time (p < 0.001). This increase in lower BW and younger GA infants was, to a degree, masked by the extension of screening criteria to include older infants with associated greater BW. With enhanced infant survival, we have seen a shift toward micro-prematurity (BW <750g, irrespective of GA) within our institutional framework. In particular, 55% of micro-preemies and 72% of infants with GA < 27 weeks were diagnosed with ROP in the last decade (Table 4). This indicates that premature infants at-risk for ROP has shifted dramatically over time to those born smaller and earlier even though the overall incidence of ROP has not changed due to a decrease in ROP among larger and older neonates. Additionally, this data suggests a GA of less than 27 weeks irrespective of BW is a strong indicator of high-risk ROP.

The rate of plus disease (Tables 4 and 5) mirrored the incidence of ROP. Overall, 18.9% of ROP infants were found to have plus disease. Analyzing trends of plus disease, there was a significant increase over time in the incidence of micro-preemies among all plus disease infants (65% in 1990s to 91% in 2010s, p < 0.005). The incidence of infants born with GA <27 weeks and plus disease also increased from 78% to 99% (p < 0.001). This severity of disease in smaller infants was also reflected by a significantly higher rate of zone 1 disease in plus infants seen over time (Table 6).

EMERGENCE OF THE MICRO-PREMATURE INFANT

At first glance, the increasing BW and GA of screened infants over time in the face of largely stable or even lower ROP incidence is falsely reassuring.^12,13 In our opinion, this pattern is driven by the emergence of the micro-premature infant (as defined by BW < 750g). In the 1990s, 2000s, and 2010s, the average GA and BW of infants screened in the NICU increased significantly over time, while the average GA and BW of infants diagnosed with ROP decreased from 26.7 weeks and 858.5g to 26.3 weeks and 815.5g to 25.8 weeks and 780.7g, respectively (p < 0.001). Furthermore, the percentage of infants diagnosed with ROP that weighed <750g at birth significantly increased from 35% to 55%. In other nationwide studies, infants with BW < 750 g have increased in incidence from 15.8% in CRYO-ROP to 24.9% in ETROP, and to 28.0% in G-ROP and 33.4% in e-ROP.^5,6,13,57 A similar evolution is seen in infants born at <27 weeks GA. Correspondingly, in the older, more mature infant population, ROP incidence has remained mostly stable at 6% for all screened infants with a BW > 1250g, while for infants with GA 32 weeks or older there has in fact been a decrease in ROP incidence (5.2% to 1.1%).

In other areas of the world, trends of micro-preemies and extremely low birthweight infants (ELBW) are consistent with our findings. In Taiwan, the overall rates of ROP have remained constant over a ten-year period, but ELBW infants have seen a 2–4-fold increase in the incidence of ROP.⁵⁶ In Australia and Sweden, studies from the 1990–2000s have also demonstrated an unchanged incidence in ROP over time.^18,89

As a result, the micro-preemie has become responsible for more than 50% of ROP diagnoses in the NICU in the last decade (Table 4). The alarming rise of the micro-preemie requires more attention. It is clear that this increasing prevalence of micro-premature infants with ROP has been counterbalancing the increasingly lower rates of ROP in larger, healthier infants. Further, this analysis reveals significant (and cautionary) trends for this specific demographic that highlights the need for a greater focus on these most at-risk infants.

MICRO-PREMATURE INFANTS EXPERIENCE MORE SEVERE ROP AND PLUS DISEASE

The overall increase in micro-preemies diagnosed with ROP has been accompanied by an increase in infants with higher risk characteristics. In our population, zone I disease became more prevalent over time and was increasingly associated with smaller and smaller infants. As can be seen in Table 7, more micro-preemies developed severe ROP as defined by more posterior zone and higher stage ROP as compared to larger infants across all three decades. The average BW and GA of zone I disease infants was 619.3g and 24.4 weeks respectively, significantly lower than that of infants diagnosed with zone II disease (820.6g and 26.3 weeks). In fact, the average infant with zone I ROP can be classified as a micro-preemie. A somewhat similar pattern can be seen in higher stage ROP infants. For those infants with stage 3 ROP, the average BW and GA were 694.7g and 25.1 weeks. In the most recent decade, 36 infants were diagnosed with stage 3 ROP with a mean BW and GA of 624.7g and 24.5 weeks, while 220 infants were diagnosed with stage 2 ROP with a mean BW and GA of 711.2g and 25.4 weeks, respectively (Table 7).

Examining the average zone and stage severity has also revealed concerning trends. The average zone for micro-preemies decreased over time from 2.01 to 1.77. This finding essentially indicates that between 1990–1999, the anatomical location of active ROP in the average patient was located in zone II, while between 2010–2019, the ROP progressed to an anatomical location somewhere between zone I and zone II (Table 8). For ROP staging across 30 years, the mean stage severity for micro-preemies was 2.14, indicating that those ROP infants weighing < 750g at birth progressing to stage 2 to 3 ROP. In summary, micro-preemies demonstrated on average more posterior zone and more severe stage than all other BW groups and ROP zone became significantly more severe over time (p < 0.001).

Severity of ROP also mandates a discussion of plus disease. Based on physician experience, plus disease in micro-premature infants was often noted to be more subtle as compared with more mature infants. The micro-premature infant in the most recent decade comprised 91% of all infants with plus disease between 2010 to 2019. In fact, over 90% of infants diagnosed with plus disease were born weighing less than 1000g across all 30 years of the study, which was significantly different than those infants with ROP without plus disease (p < 0.001). In the most recent decade, the difference is even more apparent; 100% of all infants diagnosed with plus disease weighed < 1000g BW. Over time, the average BW of plus disease infants dropped precipitously from 721.0g to 604.5g, both of which fall within the parameters of micro-prematurity. This analysis documents that infants diagnosed with plus disease are predominantly concentrated in the micro-premature demographic and that these infants are becoming smaller and smaller over time.

As such, not only has the rate of ROP in micro-preemies been increasing, but the severity and risk of disease (as demonstrated by disease zone, stage, and plus disease) has progressed as well. This highlights the importance of identifying micro-preemie infants as an independent high-risk population.

MICRO-PREMATURE INFANTS REQUIRE MORE TREATMENT

Overall, 77.4% of infants requiring treatment were micro-premature at birth (Table 9). In fact, <1% of all infants receiving treatment weighed more than 1250g at birth. For all types of ROP treatment, the average BW and GA of infants were 659.9g and 24.8 weeks, which is significantly lower than the average BW/GA of infants diagnosed with ROP but not requiring treatment (823.2g and 26.3 weeks, respectively, p < 0.001). Furthermore, in our experience, infants who received intravitreal anti-VEGF injections demonstrated even lower BW and GA (593.4g, p < 0.001, and 24.4 weeks, p = 0.031, respectively). For all patients receiving intravitreal anti-VEGF, virtually all were micro-preemies (93.3%, Table 10).

MICRO-PREMATURE INFANTS UNDERGO MORE FREQUENT SCREENING

The clinical relevance of increasing ROP incidence combined with more aggressive disease in micro-preemies is reflected by the changing patterns of diagnosis and treatment in these infants. In evaluating ROP screening, the mean number of days in between examinations for all inborn infants with ROP was 14.3 days (Table 11). This indicates that, on average, infants were screened in the NICU approximately two weeks apart, which is consistent with AAO/AAP ROP screening guidelines of screening infants between 1 to 3 weeks between examinations. Further analysis demonstrates that significant trends were associated with both BW and GA. In the micro-premature population with ROP, the average screening interval was 2.3 days less than that of infants born >750g (p < 0.001). Therefore, when infants are born micro-premature, the screening physician has tended to screen them sooner and more often based on clinical findings and features of the examination. This data suggests that micro-premature infants may benefit from a different screening algorithm than that currently recommended for at-risk NICU infants.

COHORT OUTCOMES

For all included infants within our inborn cohort, only 10 infants developed stage 4 ROP and an additional 4 infants progressed to stage 5 ROP over our three-decade study window. Overall, 3 infants (0.1%) received surgical intervention due to progression of ROP to retinal detachment. In the most recent decade, there were no infants that progressed to stage 4 or 5 ROP and, subsequently, no infants required surgical intervention. In multiple longitudinal demographic studies in the U.S. and Taiwan, approximately 0.1% to 0.2% of infants required surgical intervention.^17,93 In the BEAT-ROP trial, 15 out of 143 infants with stage 3+ ROP (10.4%) required vitrectomy.⁴

In our opinion, in the last decade of the study, the decrease in incidence of stage 4 and 5 ROP has been highly encouraging. The authors hypothesize that the reasons for these improving outcomes over the last decade are multifactorial. As mentioned already, focusing on high-risk infants with known risk factors is crucial in identifying those infants most likely to require treatment. In particular, micro-prematurity may be a defining risk factor for early treatment. Additionally, plus disease, fundus pigmentation, and TVL may independently increase risk. Finally, our ROP team has established a strict and robust protocol for the screening of infants that has been reproducible over many decades. As outlined in the methods, a dedicated ROP nurse and neonatologist identifies all infants requiring screening. The screenings are subsequently performed weekly without fail. The ROP nurse rounds with the ophthalmologist for all infants and finally the neonatologist is informed of all pertinent findings. The team follows strict adherence to screening protocols without extending days between exams. Finally, the early adoption of anti-VEGF has allowed our practice to avoid anatomical progression to stage 4 and 5 disease, eliminating the need for vitrectomy in the last decade. In the BEAT-ROP trial, the rate of recurrence in stage 3+ zone 1 ROP was significantly higher for laser photocoagulation, with 13 of 33 laser photocoagulation infants requiring vitrectomy versus 0 of 31 infants who received IVB.⁴

SCREENING GUIDELINES AND ROP DETECTION ALGORITHMS

Since micro-preemies require more frequent screening intervals, and since they are also at risk of requiring more ROP treatment than the general ROP population, should there be more instructive guidelines for micro-preemies? Presently, BW and GA are critical in determining when screening should be performed, but there is no specific role for BW or GA in the determination of the frequency of screenings. Current guidelines by the AAO/AAP indicate that infants should be seen within 1 week or less if there is any stage ROP in zone 1, stage 3 or worse ROP in zone 2, or the presence of AP-ROP. If the patient has stage 1 or 2 disease in zone 3, however, they are recommended to be screened between 2 and 3 weeks from the time of exam,⁵³ regardless of birthweight or gestational age.

Our findings demonstrate that micro-preemies required more frequent screening and treatment (Table 11). Since current guidelines depend on examination features (such as zone, stage, and extent of vascularization) to determine screening intervals, there may be utility of incorporating BW and/or GA as an indicator for more severe disease in instances where the suspicion for ROP progression is high, the examination is difficult due to corneal haze or poor pupillary dilation, or the patient has significant systemic comorbidities precluding an adequate examination. In these cases, ophthalmologists should consider micro-premature status as a risk factor.

While performing external validation of certain prediction algorithms like the WINROP algorithm, some authors have already identified the areas where prediction tools have failed.^59,62 For example, Wirth and colleagues⁵⁹ investigated the utility of WINROP in very premature infants and found that out of 570 infants, the sensitivity and specificity was only 57.1% and 46.0%. However, when looking at infants born >31 weeks GA or those weighing over 1250g BW, the sensitivity and specificity rose to 100% and 95.7%. They concluded that infants born earlier than 31 weeks GA or those weighing less than 1250g BW do not perform well with the algorithm and fundus examinations are necessary.⁵⁹ Therefore, special attention still must be paid to extremely premature infants despite newer, more advanced prediction algorithms. Although many of these algorithms have demonstrated abilities to reduce the burden of frequent screening exams on the healthcare provider and patient while detecting treatment-warranted ROP at a high rate,^59,60,62,65 ophthalmologists should use caution when relying on these algorithms for the most vulnerable group of ROP infants.

TREATMENT OUTCOMES

256 total infants received treatment, defining a treatment incidence of 11.6% over 30 years (Table 9). Other studies have reported incidence of treatment from 5% to as high as 25.4% globally.^17,18,90 189 total infants received laser photocoagulation as primary therapy, while 60 received IVB as primary therapy; all IVB were performed in 2011 and later (Figure 3). Only 6 infants received cryotherapy and 1 infant received a combination of cryotherapy and laser as primary therapy during the first two decades of the study. The mean BW and GA were higher for those receiving photocoagulation than those receiving IVB (Table 10, p < 0.001).

In our current practice, there has been a shift from laser photocoagulation monotherapy towards early adoption of IVB (Figure 3). With the shifting increase to treatment-warranted ROP, the incidence of laser treatment in our study for ROP infants increased from 6.8% to 16.7% between the first and second decade. In the third decade, with the transition to intravitreal anti-VEGF, laser showed a significant decline (5.8% incidence) and in most cases was used as a consolidating therapy only. IVB was employed as primary treatment starting in 2010, and rapidly transitioned as a therapy for all treatment-warranted ROP. Within this decade, for the inborn cohort, no infant progressed to stage 4 or 5 ROP after treatment with IVB, in contrast to 3.4% who progressed after primary treatment with laser photocoagulation over the entire study window.

There are several reasons for the increased adoption of IVB over laser photocoagulation. Most importantly, the positive outcomes of avoiding stage 4 and 5 disease when anti-VEGF has been utilized as a primary therapy for treatment-warranted ROP. One key factor in the use of anti-VEGF as a primary therapy for ROP is the overall experience of the treating surgeon with pediatric anti-VEGF use outside of ROP. Advances in the delivery of anti-VEGF have facilitated both the ease and safety of the injection protocol. Additionally, the increased adoption of IVB is driven by the recognized concerns associated with laser including but not limited to severe visual field constriction,⁹⁴ high myopic refractive error,^95,96 macular dragging,⁹⁷ vitreous hemorrhage,⁹⁸ and development of cataract.^96,99 Furthermore, multiple studies exploring the outcomes of IVB versus laser have confirmed that IVB has a similar efficacy while limiting permanent avascularity of the peripheral retina.^4,95,99 In fact, supporting our analysis, a recent meta-analysis suggested that eyes undergoing intravitreal anti-VEGF therapy experienced longer time to retreatment/recurrence and a lower likelihood of undergoing surgical intervention.¹⁰⁰ Some authors have suggested that laser causes blood-retinal barrier breakdown which in turn induces increased VEGF leakage from the immature retina, thereby worsening ROP.⁹⁹ Our group has previously described the phenomenon of pachyphakia, microcornea, and angle closure (PMAC), in which infants undergoing laser photocoagulation have developed increased lens thickness, smaller corneal diameter, and increased axial myopia.⁸⁰ As compared to IVB treatment for ROP, retinal ablation has been noted to induce anterior segment ischemia resulting in zonular laxity, corneal maldevelopment, and pachyphakia. This can result in iridocorneal adhesions, angle-closure, and poorer visual outcomes.⁸⁰ Of course, the authors do acknowledge the need for long-term studies regarding systemic safety of IVB in the use of ROP although current studies have not yet demonstrated any significant evidence of increased mortality or adverse neurological events.^4,101,102

Our IVB ROP treatment protocol employed a staged approach to treatment and a defined follow-up with inclusion criteria for supplemental IVB treatment. This defined protocol approach to the use of IVB enabled early adoption of IVB therapy with what we believe was a significant minimization of risk for failure. To this end, 60 infants meeting treatment-warranted ROP criteria received primary IVB therapy and targeted follow-up. 40 IVB primary treated infants required no staged therapy throughout the entirety of their ROP course. 20 infants received a supplemental IVB injection and, for this subset, all infants achieved a favorable outcome. No patients progressed to stage 4 or 5 ROP and no patients required surgical intervention. Other investigators have published recurrent ROP incidence after initial IVB monotherapy ranging from as low as 6% in BEAT-ROP to as high as 41% in more recent studies.^4,103 In the last decade, as IVB has gained popularity and increased use in ROP, preemies have also become progressively smaller and higher risk. This may be influencing the wide range of recurrent ROP in infants receiving IVB.

An ongoing controversy associated with the use of intravitreal anti-VEGF for primary treatment is the recognition of long-term persistent avascular retina. In our experience many infants do not fully vascularize the far temporal periphery and this has led to a concern for potential risk for late complications. In our institution, we consider consolidation laser ablation to this small residual avascular zone based on the individual characteristics of the infants or the concerns of ongoing follow up post-NICU discharge.

Mean PCA of treatment has decreased over time. Over the last 30 years, the mean PCA has decreased from 38.4w to 36.1w (p < 0.001) indicating a shift towards smaller and higher risk infants requiring earlier intervention. The mean amount of time between treatments was 5.7 weeks. This contrasts with BEAT-ROP where the average time between initial IVB and recurrence requiring retreatment was between 14.4 to 19.2 weeks.⁴ However, recent studies have demonstrated more frequent retreatment that is more consistent with our results. One such prospective study in 2018 experienced 5% of 61 infants requiring retreatment after IVB within 4 weeks, with 22 infants (36%) requiring retreatment for persistent disease or recurrence as indicated by disease present beyond 4 weeks.¹⁰³ Two other studies published in 2016 and 2017 demonstrated a range of 5.1 to 12.7 weeks for retreatment of recurrent ROP after initial IVB.^101,104 The shortening interval between treatments with IVB is likely representative of multiple aspects of the evolution of ROP and ROP treatment. The most apparent reason noted in our study is the increased severity of ROP amongst younger, smaller micro-preemies requiring more aggressive and earlier treatment. Another consideration is the increased adoption of IVB amongst providers as standard of care for treatment of zone I and posterior zone II ROP. The comfort of utilizing IVB has led to more frequent use of IVB for mono and dual therapy.

In summary, over time a clear trend towards primary usage of IVB has been adopted at our institution. Over 90% of infants requiring treatment were micro-premature at birth. Though there was a higher rate of requiring second treatment in infants receiving IVB as initial treatment, there was no progression to stage 4 and 5 since the start of using IVB at our institution in 2011. The authors recognize the need for future studies evaluating the efficacy and safety of IVB over time.

OTHER RISK FACTORS

Tunica Vasculosa Lentis

Tunica vasculosa lentis (TVL) was recognized as a potential risk factor for ROP in the early 2000s.¹⁰⁵ The persistence of tunica vasculosa lentis (TVL) on screening examination has been utilized by the present authors as an informal bedside indicator for a higher likelihood of ROP as well as more severe ROP features.^28,106 In a recent case-matched study by our group, we established that TVL is a risk factor for an increased risk of zone I disease and for treatment-requiring ROP.²⁸ Other studies have found a relationship between presence of TVL and lower BW, lower GA, and higher rates of unfavorable outcomes.^2,107

The current study collected data on presence of TVL in the most recent decade and found 102 total infants with persistent TVL at the time of screening (Table 12). Our investigations showed that infants with TVL are primarily micro-premature at birth and carry a higher risk of ROP diagnosis and plus disease. Specifically, 83.8% of TVL infants weighed <750g at birth and 91.9% were born at <27 weeks GA. The average BW and GA of TVL infants were 615.5g and 24.9 weeks, respectively. Both are significantly lower than BW and GA of non-TVL ROP infants (p < 0.001). 74 of 102 infants (73%) with TVL were diagnosed with ROP and 26 (25%) were diagnosed with plus disease (as compared to an incidence of plus disease of 18.9% for all ROP infants). Overall, average zone and stage of the 74 infants diagnosed with ROP were 1.62 (p < 0.001 as compared to 2.0 for all ROP infants) and 1.75 (p = 0.37, as compared to 1.8 for all ROP infants), respectively.

From our analysis, TVL has been shown to be associated with both ROP incidence as well as ROP treatment (p < 0.001 for both, Table 12). Several investigators have suggested an intimate link between ROP and profuse TVL on examination in ROP infants as driven by vasoactive factors.^28,106,108 In the treatment of ROP, IVB has demonstrated its utility in inducing regression of TVL, thus corroborating the relationship between VEGF, ROP, and TVL.¹⁰⁶ In summary, providers should be highly suspicious of ROP and treatment-warranted ROP in those infants noted to have TVL on the initial screening exam.

Outborn Versus Inborn Infants

Outborn infants were analyzed separately to exclude potential biases in the primary cohort analysis. The authors’ experience with outborn infants (those transferred from external institutions) for ROP management is that these infants require higher level care, more frequent screening, and/or treatment with laser photocoagulation, IVB, or surgery. Despite this common experience, rarely has this finding been explicitly confirmed. In one study by Wade and colleagues, inborn infants were more likely to be ROP-free at discharge in a subset of mature infants.³¹ The present study has corroborated this observation of a predisposition for more acute and aggressive ROP within outborn transferred infants. Furthermore, there was a nearly 3-fold increase in the proportion of outborns over time. Quinn and colleagues found a similar two-fold increase over a 27-year period, which was attributed to a higher rate of more critically ill infants.¹³ In our experience, an additional potential factor includes the increased centralization of pediatric care to larger, better equipped hospitals in our geographic region.

In the present study, outborn preemies had a slightly higher incidence of ROP compared to the inborn infants (33% vs 29.6%, p = 0.19), and were found to be significantly younger at birth (see Table 13, p = 0.018), but not significantly smaller. Outborn infants diagnosed with ROP were prone to more posterior zone and higher stage ROP as compared to inborn ROP infants (p < 0.001). Of those infants diagnosed with ROP, stage 3+ disease comprised 48.0% of outborn ROP infants as compared to 18.7% of the inborn ROP population. Similarly, 44.7% of all transfer infants received some form of treatment (IVB, laser photocoagulation, and/or surgery), as compared to a treatment incidence of 11.4% for inborn ROP infants (OR 59.4, p < 0.001 in logistic regressions). Stage 4 ROP was found in 13 infants, 4 (30.8%) of which required surgical intervention. One of four (25%) stage 5 ROP infants eventually required surgery. Additionally, the mean number of days between screenings for outborn infants was 8.2 days as compared to nearly twice that interval for the average ROP patient (14.3 days, Tables 11 and 13).

Infants can be transferred to tertiary care hospitals for ROP management for various reasons. Commonly, smaller or lower resourced institutions may experience a shortage of trained ophthalmologists or healthcare providers for screening.¹² Consistent and accurate screening protocols for the detection of ROP may also be insufficient.¹² However, not all transferred infants automatically have high-risk ROP. In our institution specifically, like many other tertiary care facilities, outborn infants can also come from low-resourced countries outside of the United States. These facilities may lack even the basic resources for continuous oxygen monitoring or NICU care.¹² Regardless of the reason for transfer, ophthalmologists must maintain a high index of suspicion for severe ROP in outborn infants. These infants typically have more severe disease, an increased likelihood of requiring treatment, and often require more frequent screening. Outborn infants clearly require specific and focused attention early within their referral to decrease the likelihood of progression to advanced stages of ROP typically requiring surgical intervention.

Multiple Gestations

There is no consensus in the literature on the risk of multiple gestations on the presence of ROP or the severity of the disease. While some authors claim that there is an association between multiple birth status and the development of ROP,^109,110 others claim no such association exists.^14,111–113 In the present study, a total of 1,322 (17.8%) infants were part of a multiple birth gestation (Table 2). Among those infants with ROP and those infants with plus disease, multiple birth infants comprised 16.1% (p = 0.015) and 18.0% (p < 0.24) of their respective cohorts. Of those requiring treatment for ROP of any kind, multiple birth infants comprised 19.9% of the cohort (p = 0.10). Multivariable logistic regression including independent risk factors of BW, GA, and multiple gestations did not find that multiple birth status was a risk factor for ROP diagnosis (p = 0.57). Multiparity was not found to be associated with treatment-warranted ROP in univariate analysis (p = 0.10).

Over 3 decades, incidence of multiple gestations was noted to have increased in both the overall population as well as the ROP population. As previously mentioned, this has been a well-known phenomenon due to the advancement of neonatal care resulting in improved survivability of smaller premature infants combined with the increase in infertility treatment and artificial reproductive technology (ART).^{22,23,114,115} ART, in fact, has been implicated as a risk factor for ROP with one study finding that 26.7% of ART infants were born extremely preterm, 4 (5.3%) of which required retinal surgery due to ROP.²³ Other authors found that ART children are at a five-fold increased risk for developing ROP.²² Similar trends within our cohort were present but statistical significance may have been masked by hospital policy that mandated screening of all multiple gestational infants.

Fundus Pigmentation

By utilizing data from the same cohort, our group has previously established that lighter fundus pigmentation (FP) is independently associated with ROP, more aggressive features of ROP, and treatment of ROP.²⁹ In particular, grade 1 or light fundus pigmentation (see Figure 1), as defined as a blond fundus where choroidal vessels can be visualized in the macula, confers a higher risk of ROP (OR 1.38, p = 0.01), plus disease (OR 4.01, p < 0.01), more posterior zone (OR 2.50, p < 0.01), and higher stage (OR 1.92, p < 0.01, see Table 14, reproduced with permission from Fan and colleagues). Additionally, light FP is independently associated with a higher chance of treatment with IVB (OR 2.67, p < 0.01).

Table 14.

Association of ROP Severity & Treatment with Light Fundus Pigmentation

Fundus pigmentation	ROP Status N (%)#		Plus Disease N (%)		Zone Severity N (%)		Stage Severity N (%)			Treatment Type N (%)
	Yes	No	Yes	No	Zone 1	Zone 2&3	Stage 1	Stage 2	Stage 3	IVB	Laser
Light	162	334	38	458	37	125	70	87	5	38	7
496	(33%)	(67%)	(7.7%)	(92%)	(7.5%)	(25%)	(14%)	(18%)	(1.0%)	(7.7%)	(1.4%)
Medium/Dark	198	565	23	740	25	169	99	86	13	23	8
763	(26%)	(74%)	(3.0%)	(97%)	(3.3%)	(22%)	(13%)	(11%)	(1.7%)	(3.0%)	(1.0%)
Total	360	899	61	1198	62	294	169	173	18	68	15
1259	(29%)	(71%)	(4.8%)	(95%)	(4.9%)	(23%)	(13%)	(14%)	(1.4%)	(5.4%)	(1.2%)
Odds of ROP (OR [95% CI] p-value)	1.38 [1.08–1.77] p = 0.01b		4.01 [2.00–8.04] p < 0.01a		2.50 [1.48–4.23] p < 0.01b	1.22 [0.94–1.60] p = 0.14ǂ,b	1.57 [1.08–2.3] p = 0.02a	1.92 [1.27–2.9] p < 0.01a	*	2.67 [1.57–4.54] p < 0.01b	*

^ǂZone 2 & 3 had reduced odds of light fundus pigmentation as compared to Zone 1 (OR 0.49, p = 0.01)

^*Associations for stage 3 severity and treatment with laser were not statistically significant; all analyses performed with logistic regression

^aforward stepwise multivariable logistic regression

^bunivariable logistic regression

^#percentages expressed respective to fundus pigmentation groups

Abbreviations: ROP = retinopathy of prematurity, OR = odds ratio, CI = confidence interval, IVB = intravitreal bevacizumab

Melanin pigments are known to remove free radicals and ROS during retinal oxidative stress and phototoxicity, and therefore serve a protective role.^116,117 Darkly pigmented fundi have an increased presence of melanin in the retina, RPE, and choroid, and can theoretically reduce the progression of ROP in the premature, avascular retina.¹¹⁶ Similar evidence exists in other diseases, such as AMD and Stargardt disease, whereby melanin serves a role in reducing oxidative stress and decreases mitochondrial output of ROS.^118,119 Consequently, fundus pigmentation is yet another examination feature that providers may use to risk stratify premature infants at risk of developing ROP.

Study Limitations

Our study utilizes a database collected and recorded by a single ROP service over three decades, thereby allowing for potential advantages in relationship to other longitudinal, large-scale studies. A number of recent studies looking at larger populations of ROP infants have utilized either automatically generated inpatient databases or surveys, or utilized an aggregation of multi-institutional trials/studies from different time periods.^13,17,30,111 Some authors have suggested that this approach may be better than single-center studies as they are more representative of practice across the United States as a whole.¹³ Although the comments made above do have merit, we believe that data collected from a single team offers unique advantages, including, but not limited to, a strict uniformity of data collection, higher consistency in clinical ROP grading, and trends that are detectable over time without the confounding of variables geographic location, multiple screening interpretations, and variations in NICU procedures.

Our study was performed in a retrospective review of a consecutive neonatal based infant screening program. Of note, ethnicity was not an identified variable throughout our study though our population is predominantly Hispanic, Caribbean and African American. In our NICU population classification of ethnicity remains complex and is typically determined by self-identification. As with all long-term longitudinal studies some variables were recognized to potentially be important later in the course of our study such as fundus pigmentation and the presence of tunica vasculosa lentis. Ultimately long-term analysis must maintain a core data set over time but maintain the flexibility to include newly identified variables.

CONCLUSION

In this retrospective, consecutive case series over three decades, we have described the evolution of NICU infant survival, the emergence of the micro-premature infant, and the importance of adapting treatment practice patterns to achieve the best possible outcomes for our infants. One of the critical factors has been an integrated ROP team led by a screening/treating ROP ophthalmic specialist. While the team requires the support of the NICU director, inclusion of a dedicated ROP nurse and coordinator has been essential.

Large scale longitudinal analyses, though difficult to perform, enable a comprehensive overview of the ROP landscape, and aid in capturing evolving diagnostic criteria, infant risk factors, and treatments. While all studies strive to minimize or eliminate bias, this single-center approach allowed for control of multiple variables to reduce subjectivity in diagnostic and treatment patterns. Moving forward, even for non-academic institutions, advances in the electronic health record may permit similar frameworks for analysis of their own infant populations for improved neonatal ophthalmic care.

This study has helped us recognize the significant impact of micro-prematurity on virtually every aspect of ROP diagnosis and treatment. We have demonstrated an early recognition of the rising prevalence of our micro-premature infants due to advances in neonatal care. By characterizing this at-risk population of micro-preemies, we have defined an important subgroup for ROP risk. Without a specific focus on the micro-premature infant, ophthalmologists may overlook these high-risk ROP infants who often will progress to treatment-warranted ROP. In our study population, micro-premature infants, and especially those with gestational age <27 weeks, almost exclusively comprised the groups of infants with plus disease and treatment-warranted ROP.

Ongoing advances in NICU infant survival will inevitably continue impacting the patient, the ROP team, and drive new advancements for neonatal care. Adapting to the evolution and increasing incidence of the NICU population mandates better resource allocation and more targeted screening protocols. Ultimately, the resources employed in ophthalmic NICU screening can be made more efficient by refocusing screening criteria, utilization of telemedicine-related remote screening, and incorporating artificial intelligence screening algorithms.

This study also documents the occurrence of a major treatment transition within a tertiary care, academically-focused NICU. To achieve the transition to intravitreal anti-VEGF, a staged treatment protocol was employed for screening, treatment, and post-treatment evaluation, which included supplemental anti-VEGF treatment when necessary. This strategy enabled our team to implement a novel treatment while focusing on infant safety through frequent reevaluation for response both from an ophthalmic perspective as well as a systemic standpoint. We suggest that for institutions not currently employing IVB treatment as a primary therapy for treatment-warranted ROP, the transition to IVB therapy may be accomplished utilizing this protocol in such a way that both the screening and treating ophthalmologist, along with entire NICU team, may feel comfortable.

We passionately believe that ROP care requires a resource-driven approach that focuses on the ROP team and the adoption of evolving technology, screening, and treatment protocols. Ultimately, optimal outcomes for such high-risk infants requires the ability to implement novel treatment approaches rapidly and safely. We hope that this study can assist institutions not currently employing advanced screening and treatment practices to adopt a framework that has virtually eliminated blinding eye disease from retinopathy of prematurity.

Figure 4.

Change over time of BW and GA from 1990 to 2019 demonstrating a gradual increase in both BW and GA over 30 years. Abbreviations: BW – birthweight, GA – gestational age

Figure 5.

ROP Treatment Incidence by Year. Incidence of treatment with anti-VEGF (green) and laser photocoagulation (blue) as defined by number of patients treated as a proportion of all infants diagnosed with ROP. Also displayed is the incidence of stage 3+ disease over time (yellow). Abbreviations: ROP – retinopathy of prematurity, IVB – intravitreal bevacizumab, LPC – laser photocoagulation.