High interobserver reliability of neutrophil phenotyping based on visual determination of CD16/CD62L expression patterns.

INTRODUCTION

Analysis of neutrophil surface expression markers offers relevant insights into the inflammatory response following trauma. Recently a neutrophil phenotype classification based on CD16 and CD62L expression has been established. A seven-category visual classification system based on pattern recognition has been introduced for bedside use, but interobserver reliability has not been assessed to date.

METHODS

In this study, 40 participants with varying medical backgrounds were asked to classify 50 CD16/CD62L dot plots using the seven-category scoring system. Participants received a standardized instruction manual for pattern recognition and category classification. Classification accuracy was compared to a predefined gold standard, and interobserver agreement was assessed using Fleiss' Kappa. Additionally, categories were grouped into "low" (0-3) and "high" (4-6) inflammatory states to evaluate clinical applicability.

RESULTS

Agreement across the seven-category scale was moderate overall (κ = 0.473), with higher consistency observed in more extreme phenotypes. Categories 0, 1, and 3 were most frequently confused, whereas plots obtained during severe inflammation (4, 5, 6) were rarely misclassified as non-inflammatory. When grouped into binary low versus high inflammatory states, agreement improved to almost perfect (κ = 0.836).

DISCUSSION: Binary classification of neutrophil phenotypes, and thus identification of severe inflammation after trauma, is reliable across different observer levels and appears well suited for clinical application. The full seven-category scoring system offers greater resolution but shows higher interobserver variability. These findings support the use of simplified immune phenotyping at the bedside, while preserving the full classification system for research settings.