Main content area

Inter‐ and intra‐rater reliability and agreement in determining subcutaneous tumour margins in dogs

Ranganathan, B., Milovancev, M., Leeper, H., Townsend, K. L., Bracha, S., Curran, K.
Veterinary and comparative oncology 2018 v.16 no.3 pp. 392-398
body condition, dogs, mast cells, patients, photographs, planning, prospective studies, sarcoma, statistical analysis, therapeutics, tissues, veterinarians
The objective of this prospective study was to evaluate agreement and reliability of calliper‐based measurements of locally invasive subcutaneous malignant tumours in dogs. Four raters measured the longest diameter of 12 subcutaneous tumours (7 soft tissue sarcomas and 5 mast cell tumours) from 11 client‐owned dogs during 3 randomized, blinded measurement trials, both pre‐ and post‐sedation. Inter‐ and intra‐rater reliability was evaluated using intra‐class correlation coefficient (ICC) and agreement was evaluated using Bland‐Altman plots. Inter‐ and intra‐rater reliability was good (ICC range of 0.8694‐0.89520) and excellent (ICC range of 0.9720‐0.9966), respectively. For agreement calculations, an a priori clinically relevant limit of agreement of 10 mm was set. Inter‐ and intra‐rater agreement was unacceptable with inter‐rater limits of agreement ranging from 15.9 to 55.6 mm and intra‐rater limit of agreement ranging from 11.9 to 28.1 mm. Review of the measurement trial photographs revealed that calliper orientation changes were frequent, occurring in 9/12 (75%) and 8/12 (67%) pre‐ and post‐sedation cases. No significant correlation was found between inter‐rater measurement standard deviations and calliper orientation changes or dog body condition score. These findings suggest veterinarians may have poor agreement in determining the gross edge of tumours, which is expected to introduce bias and inconsistency in tumour staging, assessing response to therapy, and surgical margin planning. Due to the potential consequences for veterinary cancer patients, future studies are needed to validate the present findings.