626 - Quality of Reporting in Neonatal Randomized Controlled Trials: A 10-Year Analysis

Publication Number: 2611.626

Kim Ruiz-Arellanos, Beth Israel Deaconess Medical Center, Boston, MA, United States; John Zupancic, Harvard Medical School, Boston, MA, United States; Susanne Hay, Harvard Medical School, Boston, MA, United States; Roger F. Soll, Robert Larner, M.D., College of Medicine at the University of Vermont, Hinesburg, VT, United States; Haresh Kirpalani, McMaster University Michael G. DeGroote School of Medicine, Burlington, ON, Canada; Maria Laura Mourão, Beth Israel Deaconess Medical Center, Boston, MA, United States; Sofia Gnecco, Beth Israel Deaconess Medical Center, Brookline, MA, United States; Luna Hurtado-A, Beth Israel Deaconess Medical Center, Brighton, MA, United States; Santiago Reyes, Beth Israel Deaconess Medical Center, Boston, MA, United States; Rhys Johnstone, Beth Israel Deaconess Medical Center, Westerly, RI, United States; Adrian Baca-Arzaga, University of Miami, Miami, FL, United States; Marianna B. Castellanos, The Children's Hospital at Montefiore, Bronx, NY, United States; Grace E. Fuller, University of North Carolina at Chapel Hill School of Medicine, Chapel Hill, NC, United States; Kristyn S. Beam, Beth Israel Deaconess Medical Center, Needham, MA, United States; Brian King, Harvard Medical School, Boston, MA, United States

Background: Randomized controlled trials (RCTs) are essential for evidence-based neonatology, but judgement of validity depends on transparent and comprehensive reporting. Despite reporting guidelines, the overall quality of neonatal RCT reporting has not been systematically evaluated in an unselected group of publications.

Objective: To assess the quality of reporting in neonatal RCTs according to CONSORT criteria and examine its association with study characteristics, including year, region, funding source, sample size, and intervention type.

Design/Methods: We analyzed data from NeoCanon, a validated database of over 5,000 neonatal RCTs; we included infants ≤31 days old or prior to first discharge home. A random 20% annual sample (n=581) was evaluated across 32 predefined criteria (see table) in three domains: Study Design (25%), Trial Conduct (62.5%), and Transparency (12.5%). The 32 criteria assessed compliance with CONSORT criteria, with one point assigned to each criteria for an overall score across all domains. Scores were summarized as medians (IQR) and compared using nonparametric statistics.

Results: The median overall reporting quality was 81% (IQR 68-90). Domain medians were 86% for Study Design, 84% for Trial Conduct, and 50% for Transparency. Only 4.5% of studies scored below 50%. The lowest reported items per domain were inclusion of "randomized" in the title; specification of blinding; and protocol citation/availability as supplementary material (Fig 1). Reporting quality differed significantly by region, funding, and sample size. Multinational and inter-regional studies had the highest scores, followed by North America, while Middle East/North Africa and Sub-Saharan Africa scored lowest (p < 0.01). Foundation-funded trials outperformed all other sources, whereas unfunded and industry-sponsored studies scored lower (p < 0.01). Studies with sample sizes >5000 or < 100 participants demonstrated poorer reporting (p < 0.01). Year of publication, intervention type, gestational age, and presence of significant result in abstract were not significantly associated with reporting quality.

Conclusion(s): Over the past decade, neonatal RCT reporting quality has been moderately complete but uneven across domains and study characteristics. While methodological reporting has improved, transparency remains critically low, particularly regarding protocol accessibility. Strengthening adherence to reporting standards, especially for smaller and regionally constrained studies, is essential to enhance reproducibility in neonatal research.

Table 1. Reporting quality of 32 methodological items across 3 domains in 581 neonatal randomized controlled trials (2013-2022).
Color gradients reflect relative performance within each domain, from red (lowest) to green (highest) based on the range of "Yes" responses.