ROP is a potentially blinding disorder that affects the developing blood vessels of the retina in infants born prematurely. The retina, a light-sensitive layer at the back of the eye, normally receives its blood supply during the final weeks of gestation. When birth occurs extremely early, retinal vessels are immature and exposed to an unstable environment involving fluctuations in oxygen, inflammation, and growth signals. This disruption can cause abnormal blood vessel growth, which may damage the retina and, in advanced cases, lead to retinal detachment and permanent vision loss.

The trial included 100 preterm infants with severe ROP and randomly assigned them to receive either dexamethasone eye drops or placebo. Randomization is designed to distribute known and unknown risk factors evenly between treatment groups, allowing researchers to assess whether differences in outcomes are likely to be associated with the intervention. In this study, 10 infants in the dexamethasone group developed type 1 ROP, compared with 19 infants in the placebo group. The resulting absolute difference was 18 percentage points.

Type 1 ROP is defined by retinal findings that indicate a high risk of progression and generally require invasive treatment. Current interventions may include laser photocoagulation, which destroys abnormal peripheral retinal tissue, or injections of anti-vascular endothelial growth factor drugs, which suppress signals driving pathological vessel growth. These procedures can preserve vision but may require specialized equipment, anesthesia or sedation, and repeated monitoring. A topical therapy that could slow disease progression would therefore represent an attractive, noninvasive addition to neonatal eye care.

Dexamethasone is a synthetic corticosteroid that suppresses inflammation and alters the activity of multiple genes involved in immune responses, vascular signaling, and tissue growth. In premature infants, inflammation is believed to contribute to abnormal retinal vascular development. Corticosteroids may also influence pathways involving vascular endothelial growth factor, or VEGF, a protein that promotes blood vessel formation. Excessive or poorly regulated VEGF activity is central to the abnormal vessel growth seen in severe ROP, although the biological effects of applying corticosteroids directly to the eye in this vulnerable population remain complex.

The study’s numerical results suggest a possible benefit: the proportion of infants progressing to treatment-requiring disease was nearly twice as high in the placebo group as in the dexamethasone group. Expressed relatively, the dexamethasone group experienced approximately a 47% lower rate of type 1 ROP. However, statistical significance depends not only on the size of an observed difference but also on the number of participants and the variability in the data. With only 100 infants, the trial may not have had enough statistical power to distinguish a true treatment effect from a chance finding.

That distinction is crucial for families and clinicians. A result that does not reach statistical significance is not proof that the treatment has no effect, but it also cannot be treated as reliable evidence that the treatment works. Larger trials would be needed to determine whether the apparent reduction persists across different neonatal intensive care units, gestational ages, oxygen-management strategies, and stages of ROP. Researchers would also need to identify the optimal dose, timing, and duration of treatment and determine whether the drops are effective when administered before the disease reaches a severe stage.

No differences in adverse events were observed between the dexamethasone and placebo groups in the trial. This finding provides early reassurance about the safety of the approach, but safety monitoring remains particularly important in extremely premature infants. Even medications delivered as eye drops can potentially be absorbed into the bloodstream, and corticosteroids may influence blood pressure, glucose regulation, infection risk, and growth. The immature eye may also respond differently from the eyes of older children or adults, making long-term follow-up essential.

The investigators, led by Ann Hellström, MD, PhD, of the Institute of Neuroscience and Physiology at the University of Gothenburg in Sweden, describe topical dexamethasone as a potentially safe strategy for reducing progression to treatment-requiring ROP. The findings offer a promising signal in a disease where prevention and early intervention are urgently needed, but they stop short of changing clinical practice. Until larger, adequately powered studies confirm the benefit, dexamethasone eye drops should be regarded as an investigational approach rather than an established replacement for the screening, monitoring, and invasive treatments currently used to protect premature infants’ vision.

Subject of Research: The potential of topical dexamethasone eye drops to reduce progression of retinopathy of prematurity to treatment-requiring disease in preterm infants.

Web References: https://doi.org/10.1001/jamapediatrics.2026.3247

References: JAMA Pediatrics study, DOI: 10.1001/jamapediatrics.2026.3247

Keywords: Retinopathy of prematurity; dexamethasone; preterm infants; premature birth; eye drops; type 1 ROP; clinical trial; randomized trial; placebo; neonatal medicine; ophthalmology; retinal disease; corticosteroids; infant eye health

Retinopathy of prematurity, a potentially blinding vasoproliferative disorder primarily affecting premature infants, remains one of neonatal medicine’s most vexing enigmas. The condition arises during a critical window of vascular development in the retina, where aberrant angiogenesis leads to scarring and detachment of the retina. Traditional epidemiological studies have heavily leaned on p-values to designate variables as risk factors, often prioritizing statistically significant associations without deeper interrogation of biological plausibility or mechanistic insight. This approach, while providing a simplistic threshold for significance, may obscure the complex pathobiology underlying ROP.

At the heart of Dammann et al.’s argument is the assertion that p-values, as sole arbiters of risk, inadequately capture the dynamic interplay of genetic, environmental, and clinical factors influencing ROP. Indeed, the dichotomization of risk factors based solely on arbitrary significance levels tends to reduce rich, continuous data into binary judgments, potentially overlooking important subthreshold contributors. Moreover, the probabilistic nature of p-values is frequently misunderstood, leading to misconceptions that statistically non-significant results imply absence of effect or clinical irrelevance.

This paradigm shift is particularly salient in neonatal research, where multifactorial etiologies dominate clinical presentations. Prematurity itself introduces a constellation of vulnerabilities—immature organ systems, oxygen therapy protocols, inflammatory insults, and fluctuating metabolic states—all of which can modulate susceptibility to ROP. Dammann and colleagues argue for a comprehensive analytic framework that integrates advanced statistical modeling, biomarker discovery, and longitudinal clinical data to unravel these complex interactions.

One technical advancement advocated in the study is the use of Bayesian approaches and machine learning algorithms to analyze neonatal datasets. These methods allow for probabilistic assessments that incorporate prior knowledge and manage uncertainty more effectively than traditional null hypothesis significance testing. By shifting toward models that produce risk estimates rather than binary cutoffs, clinicians and researchers can better stratify at-risk infants and tailor interventions dynamically.

The implications of moving beyond p-values extend to clinical trial design as well. The field has historically been mired in replication crises and controversies over sample size and power calculations. Dammann et al. suggest that future studies consider effect sizes, confidence intervals, and model-based inference alongside or in place of p-value thresholds. Such an approach fosters a richer interpretation of data and encourages transparency about the strength and limitations of evidence.

Biologically, the study underscores the importance of understanding the molecular pathways mediating retinal vascular development and injury. New insights into the roles of vascular endothelial growth factor (VEGF), oxidative stress markers, and inflammatory cytokines illustrate the multilevel regulation of retinal angiogenesis. Critically, these pathways do not operate in isolation; they are influenced by systemic conditions such as bronchopulmonary dysplasia and sepsis, which complicate the attribution of causality to any single factor based on p-value significance alone.

The authors also highlight recent advances in imaging technologies and biomarker profiling that enable more granular phenotyping of ROP progression. Optical coherence tomography (OCT), for instance, reveals subtle microstructural changes preceding overt retinal detachment. Coupled with proteomic and genomic analyses, these methods promise to identify early signatures of disease risk that transcend simplistic statistical categorizations.

In examining clinical practice, Dammann et al. draw attention to the heterogeneity in oxygen supplementation strategies and their complex influence on neonatal outcomes. Conventional protocols aiming to maintain specific oxygen saturation levels have been scrutinized for their unintended contribution to fluctuating retinal oxygenation, a key driver of pathological angiogenesis. The study advocates for individualized approaches informed by real-time monitoring and computational models that integrate patient-specific data to optimize therapeutic windows.

The paper’s critique extends to epidemiological studies that have identified risk factors like low birth weight or mechanical ventilation through statistically significant associations but without adequate adjustment for confounding variables or assessment of interaction effects. The authors call for methodological rigor and the adoption of causal inference tools to better delineate true risk factors from correlates or bystanders.

Furthermore, the discussion delves into the broader implications for neonatal care beyond ROP. The researchers propose that the reliance on p-values permeates many realms of pediatric research, often hindering progress by fostering rigid conclusions and impeding the exploration of complex disease mechanisms. Embracing a paradigm that values comprehensive risk modeling, biological plausibility, and reproducibility may catalyze advances across multiple domains including neurodevelopmental outcomes and chronic lung disease.

This transformative perspective compels funding agencies, journal editors, and clinicians to reevaluate criteria for research quality and translational potential. The study argues that rewarding mechanistically informed research and innovative analytic methodologies will accelerate discovery and improve patient care.

In conclusion, Dammann, Chui, and Stansfield’s work represents a clarion call for a paradigm shift in neonatal research, urging the community to transcend the limitations of p-value based significance testing. Their insights into retinopathy of prematurity underline the necessity of integrating statistical innovation with biological understanding to unravel the multifactorial nature of neonatal diseases. As neonatal care continues to evolve, embracing this comprehensive framework promises to refine risk stratification, enhance personalized interventions, and ultimately improve outcomes for the most vulnerable patients.

Subject of Research: Retinopathy of prematurity and the limitations of p-value based risk factors in neonatal clinical research.

Article Title: Retinopathy of prematurity and beyond: P values don’t make risk factors.

Article References:

Dammann, O., Chui, K. & Stansfield, B.K. Retinopathy of prematurity and beyond: P values don’t make risk factors.
Pediatr Res (2025). https://doi.org/10.1038/s41390-025-04251-2

Image Credits: AI Generated