Objective: To determine the level of agreement and disagreement in the interpretation of CT scan of pancreas and to identify the factors affecting these interpretations.
Methodology: The study included 200 CT Pancreas scans which were randomly selected from the PACS (Fuji synapse 5), performed at the radiology department of a tertiary care hospital in Peshawar from 1st July 2022 to 30th June 2023. Residents, blinded from final reports of the consultants retrospectively reported the scans. Comparison was made between reports of the residents and experienced consultant radiologists. Statistical analysis performed on SPSS version 26 using kappa coefficient to determine the level of agreement between residents and consultants. A high kappa coefficient indicates strong agreement, while a low coefficient suggests poor agreement.
Results: Substantial agreement was found between junior resident (R3/JR) and consultant radiologists, with a percent agreement of 91.9% (Cohen’s kappa 0.74). An almost perfect agreement was observed between senior resident (R4/SR) and consultant radiologists (CR), with a percent agreement of 98.8% (Cohen’s kappa 0.85).
Conclusion: The concordance rate between residents and consultant radiologists was good, with Cohen’s kappa coefficients of 0.74 and 0.85 for junior and senior resident, respectively.
Key words: Concordance, discordance, interobserver variability, CT scan, pancreas.
|