Intra-observer Reliability And Inter-observer Reproducibility Of GATA Classification For Spine Tuberculosis

Objective: To determine the reliability and reproducibility of GATA classification for spine tb using clinical/radiographs data that had not been premeasured. Methods: A total of 100 case of spine tb were distributed to 4 attending spine surgeons. including clinical/radiographic data, details of the GATA classification, and a evaluating sheet in which cases would be evaluated using the system. The results were determined by calculating the average percentage of intra-observer and inter-observer agreement. Reliability and reproducibility was quantified using kappa coefficients and spearman rank-order correlation. Results: Inter-rater reliability assessed by mean percentage and Kappa coefficients was 61.92%, 0.43 for vertebral collapse,91.83%,0.519 for abscess formation,82%,0.513 for disc degeneration, 60.08%,0.381 for kyphosis, 78.25%,0.536 for sagittal index, 63.3%, 0.317 for instability,52.33%, 0.343 for classification. Respective results using Spearman correlation were 0.392,0.526,0.528, 0.503,0.569,0.392, 0.576 .In- trarater mean percentage and kappa coefficients were 72.75%, 0.574for vertebral collapse,91%,0.453 for abscess formation,81.75%,0.613 for disc degeneration, 70.75%, 0.523 for kyphosis, 88.75%,0.724 for sagittal index, 85.25%, 0.627 for instability,73%, 0.591 for classification .Respective results using Spearman correlation were 0.748, 0.473, 0.613, 0.73, 0.73, 0.632, 0.762. Conclusion: The GATA classification for spine tb has good Inter-observer Reproducibility and low intra-observer Reliability.