Bo Yin1, Yuanming He1, Dong Wang1, Junlin Zhou2. 1. Department of Orthopedic Surgery, Beijing Chaoyang Hospital, Capital Medical University, Beijing 100020, PR China. 2. Department of Orthopedic Surgery, Beijing Chaoyang Hospital, Capital Medical University, Beijing 100020, PR China. Electronic address: junlinzhou_article@outlook.com.
Abstract
INTRODUCTION: Given the drawbacks of a femoral intertrochanteric fracture classification based on 2-dimensional radiographic imaging, an artificial intelligence-based classification system- the Tang classification system-which uses 3-dimensional image analysis, has previously been developed. This study explored the reliability of the Tang classification by comparing the consistency of this classification with the conventional 2-dimensional femoral intertrochanteric fracture classification systems. METHODS: X-ray and computed tomography (CT) data of 258 patients with femoral intertrochanteric fractures were classified by 6 orthopedic surgeons using the Evans, Jensen, AO/OTA, and Tang classification systems on 2 separate occasions, 1 month apart. Kappa statistics were used to evaluate the inter- and intraobserver differences in classifications. RESULTS: When the interobserver reliability was based on X-ray image analysis, the Kappa values for the Evans, Jensen, AO/OTA, and Tang classifications were 0.54 ± 0.03 (moderate agreement), 0.53 ± 0.02 (moderate agreement), 0.46 ± 0.02 (moderate agreement), and 0.63 ± 0.02 (substantial agreement), respectively. When the interobserver reliability was based on CT images, the Kappa values of the Evans, Jensen, AO/OTA, and Tang classifications were 0.49 ± 0.03 (moderate agreement), 0.49 ± 0.03 (moderate agreement), 0.44 ± 0.03 (moderate agreement), 0.64 ± 0.02 (substantial agreement), respectively. For X-ray images, the intraobserver Kappa values for the Evans, Jensen, AO/OTA, and Tang classification were 0.53 ± 0.02 (moderate agreement), 0.54 ± 0.03 (moderate agreement), 0.45 ± 0.03 (moderate agreement), and 0.65 ± 0.03 (substantial agreement), respectively. When intraobserver reliability was based on CT images, the Kappa values for the Evans, Jensen, AO/OTA, and Tang classification were 0.52 ± 0.03 (moderate agreement), 0.52 ± 0.02 (moderate agreement), 0.41 ± 0.02 (moderate agreement), and 0.63 ± 0.03(substantial agreement), respectively. CONCLUSIONS: The current study suggests that the Tang classification system is more reliable than the Evans, Jensen, and AO/OTA classification systems for measuring intertrochanteric fractures of the proximal femur.
INTRODUCTION: Given the drawbacks of a femoral intertrochanteric fracture classification based on 2-dimensional radiographic imaging, an artificial intelligence-based classification system- the Tang classification system-which uses 3-dimensional image analysis, has previously been developed. This study explored the reliability of the Tang classification by comparing the consistency of this classification with the conventional 2-dimensional femoral intertrochanteric fracture classification systems. METHODS: X-ray and computed tomography (CT) data of 258 patients with femoral intertrochanteric fractures were classified by 6 orthopedic surgeons using the Evans, Jensen, AO/OTA, and Tang classification systems on 2 separate occasions, 1 month apart. Kappa statistics were used to evaluate the inter- and intraobserver differences in classifications. RESULTS: When the interobserver reliability was based on X-ray image analysis, the Kappa values for the Evans, Jensen, AO/OTA, and Tang classifications were 0.54 ± 0.03 (moderate agreement), 0.53 ± 0.02 (moderate agreement), 0.46 ± 0.02 (moderate agreement), and 0.63 ± 0.02 (substantial agreement), respectively. When the interobserver reliability was based on CT images, the Kappa values of the Evans, Jensen, AO/OTA, and Tang classifications were 0.49 ± 0.03 (moderate agreement), 0.49 ± 0.03 (moderate agreement), 0.44 ± 0.03 (moderate agreement), 0.64 ± 0.02 (substantial agreement), respectively. For X-ray images, the intraobserver Kappa values for the Evans, Jensen, AO/OTA, and Tang classification were 0.53 ± 0.02 (moderate agreement), 0.54 ± 0.03 (moderate agreement), 0.45 ± 0.03 (moderate agreement), and 0.65 ± 0.03 (substantial agreement), respectively. When intraobserver reliability was based on CT images, the Kappa values for the Evans, Jensen, AO/OTA, and Tang classification were 0.52 ± 0.03 (moderate agreement), 0.52 ± 0.02 (moderate agreement), 0.41 ± 0.02 (moderate agreement), and 0.63 ± 0.03(substantial agreement), respectively. CONCLUSIONS: The current study suggests that the Tang classification system is more reliable than the Evans, Jensen, and AO/OTA classification systems for measuring intertrochanteric fractures of the proximal femur.