Related Experiment Videos
Interrater reliability in grading abstracts for the orthopaedic trauma association
Mohit Bhandari1, David Templeman, Paul Tornetta
1Department of Clinical Epidemiology, McMaster University, Hamilton, Ontario, Canada. bhandari@sympatico.ca
Clinical Orthopaedics and Related Research
|July 3, 2004
Summary
Reviewer consistency in grading orthopaedic trauma abstracts was low (0.23-0.27 reliability). Discussions helped reach consensus, but blinded review of full presentations did not improve agreement, impacting final rankings.
Area of Science:
- Orthopaedic Surgery
- Medical Education
- Scientific Publishing
Background:
- Selecting abstracts for podium presentations at the Orthopaedic Trauma Association (OTA) annual meeting is competitive.
- Ensuring unbiased and transparent abstract selection is crucial for investigators.
Purpose of the Study:
- To assess reviewer consistency in grading OTA abstracts for podium presentations in 2001 and 2002.
- To evaluate if podium presentation grades align with initial abstract-only grades.
Main Methods:
- Reviewers independently and blindly graded submitted abstracts.
- Abstracts from panel members were adjudicated by non-panel reviewers.
- Reviewers convened to discuss abstracts before final program decisions.
- Agreement was assessed for both abstract grading and full podium presentations.
Main Results:
- Inter-reviewer reliability for abstract grading was low (0.23 in 2001, 0.27 in 2002).
- Consensus was reached through discussion despite initial disagreements.
- Blinded review of podium presentations did not enhance inter-reviewer agreement.
- A significant number of top-ranked presentations were initially graded lower based on abstracts alone.
Conclusions:
- Initial abstract grading by reviewers shows limited consistency.
- Consensus discussions are vital for program selection.
- Abstract grading may not accurately predict the quality of a full podium presentation.
- The selection process may benefit from further refinement to improve accuracy and transparency.