Prospective assessment of inter-rater reliability of a neonatal adverse event severity scale.

Introduction: To ensure the quality of clinical trial safety data, universal data standards are required. In 2019 the International Neonatal Consortium (INC) published a neonatal adverse event severity scale (NAESS) to standardize the reporting of adverse event (AE) severity. In this study the reliability of AE severity grading with INC NAESS was prospectively assessed in a real-world setting. Methods: Severity of AEs was assessed by two independent observers at each of four centers across the world. In each center two series of 30 neonatal adverse events were assessed by both observers: in a first phase with a generic (Common Terminology Criteria for Adverse Events, CTCAE) severity scale not specific to neonates, and in a second phase with INC NAESS (after a structured training). Intraclass correlation coefficients (ICC) were calculated to express inter-rater agreement in both phases, and bootstrap sampling was used to compare them. Results: 120 AEs were included in each of both phases. The ICC with the use of INC NAESS in phase 2 was 0.69. This represents a significant but modest improvement in comparison to the initial ICC of 0.66 in phase 1 (confidence interval of ratio of ICC in phase 2 to phase 1 = 1.005-1.146; excludes 1). The ICC was higher for those AEs for which a diagnosis specific AE severity table was available in INC NAESS (ICC 0.80). Discussion: Good inter-rater reliability of the INC NAESS was demonstrated in four neonatal intensive care units (NICUs) across the globe. The ICC is comparable to what is reported for scales with similar purposes in different populations. There is a modest, but significant, improvement in inter-rater agreement in comparison to the naïve phase without INC NAESS. The better performance when reviewers use AE-specific NAESS tables highlights the need to expand the number of AEs that are covered by specific criteria in the current version of INC NAESS.

Inter-rater reliability of the neonatal adverse event severity scale using real-world Neonatal clinical trial data.

OBJECTIVE: The Neonatal Adverse Event Severity Scale (NAESS) was developed to improve scoring of neonatal adverse events (AEs) and accelerate neonatal drug development. This is the first validation study of the novel tool. STUDY DESIGN: Retrospective validation study assessing the inter-rater reliability (IRR) of the NAESS. Reviewers used real-world AE data from a neonatal trial. Intra-class correlation (ICC) statistical analysis was performed. RESULT: Sixty AEs were randomly assigned to twelve reviewers for a total of 240 severity scores. Generic and AE-specific NAESS tables were assessed. The ICC was 0.63 (95% confidence interval 0.51 to 0.73). Percent variation due to reviewer and residual error was 0.03 and 0.34, respectively. CONCLUSION: In this first study of the NAESS tool, an ICC of 0.63 indicates moderate reliability. Results highlight the need for improved data collection on neonatal AE forms, augmented training on the NAESS tool, and will inform the prospective validation studies.