Related Experiment Video
Updated: Dec 22, 2025
07:58
An Automated Differential Nuclear Staining Assay for Accurate Determination of Mitocan Cytotoxicity
Published on: May 12, 2020
7.3K
Interrater Reliability in Toxicity Identification: Limitations of Current Standards
Andrew T Fairchild1, Jarred P Tanksley1, Jessica D Tenenbaum2
1Department of Radiation Oncology, Duke University, Durham, North Carolina.
Summary
Interrater reliability for toxicity identification using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) showed moderate agreement. Strategies to improve accuracy in oncology clinical trials are recommended.
Area of Science:
- Oncology
- Clinical Trials
- Health Informatics
Background:
- The National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) v5.0 is widely used for oncology toxicity assessment.
- Limited validation exists for CTCAE's interrater reliability (IRR).
Purpose of the Study:
- To assess the interrater reliability (IRR) of toxicity identification using CTCAE v5.0.
- Evaluate the consistency of toxicity grading in electronic health records.
Main Methods:
- Two independent reviewers assessed 100 oncology patient notes from radiation therapy visits.
- Discrepancies were resolved by a third reviewer; term harmonization was performed.
- Interrater reliability was calculated using Cohen's kappa coefficients.
Main Results:
- Unweighted kappa was 0.68, and weighted kappa was 0.59, indicating moderate agreement.
- Reliability was consistent for symptoms identified as present or absent.
Conclusions:
- Significant discordance highlights the complexity and potential errors in retrospective toxicity identification.
- Training and CTCAE criteria simplification are suggested to minimize IRR in clinical trials.
