Assessment of Methodological Bias in Studies Reporting Racial Differences in Retinopathy of Prematurity in the United States.

Abstract

Purpose: To assess methodological biases in studies reporting racial and ethnic differences in retinopathy of prematurity (ROP).

Methods: Systematic review of peer-reviewed studies published between 2014 and 2024 that reported on ROP outcome measures by race, ethnicity, or social determinants of health (SDOH). Three reviewers independently assessed each observation for selection and collider bias using definitions derived from perinatal epidemiology literature. Findings were also compared using a structured comparative synthesis between studies with and without identified methodological bias.

Results: A structured PubMed search identified 78 articles; 13 met inclusion criteria, with one study contributing two distinct analytical approaches, yielding 14 total observations. Survivorship bias was identified in 6 of 14 observations (42.9%), primarily due to the exclusion of infants who died prior to ROP screening. Potential collider bias was most common, found in 9 of 14 observations (64.3%), and was introduced through adjustment or stratification by gestational age and/or birthweight. Three studies did not exhibit either assessed biases. Among studies with identified bias, 8 of 10 observations reported lower ROP risk among Black versus White infants, whereas 3 of 4 observations without identified bias reported higher ROP risk or incidence among Black infants.

Conclusion: Methodological biases in ROP studies investigating race or ethnicity are prevalent. Adjustment for gestational age or birthweight may introduce spurious race-ROP associations and contribute to paradoxical findings. Further exploration of the impact of SDOH on disease outcomes may reduce the misattribution of race as a biological risk factor and improve the interpretation of ROP disparities.