Does the Reliability and Accuracy of the Judet and Letournel Classification System for Acetabular Fractures Increase using a Novel Algorithm?
Abstract
Introduction: We have devised an algorithm to assist classifying acetabulum fractures using plain radiographs. This study aimed to test if the accuracy and reliability of fracture classification increases using our algorithm in resident doctors.
Materials and methods: Seventy-two residents of eight tertiary care institutes took part in our survey. These residents were divided into three groups, Groups A, B, and C, with 31, 20, and 21 residents, respectively. Two different Collections (1 and 2) containing radiographs of twenty patients each, with known classification from CT and intra-operative findings, were prepared. Collection 1 radiographs were given to Group A and B, and Collection 2 radiographs were given to Group C. Group A residents were asked to classify the fractures using our algorithm, and Group B and C residents were asked to classify the fractures according to their understanding. Intra-observer and interobserver reliability were estimated.
Results: A total of 1411 unique responses were made. The accuracy of group A, B, and C residents was 53.8%, 34.9% and 28.3%, respectively (p-value 0.001). The interobserver reliability for fracture classification was fair with an algorithm (κ = 0.32) and slight without an algorithm. The intra-observer reliability among five observers was moderate (κ = 0.43).
Conclusion: Our algorithm improves the accuracy and reliability for classifying acetabular fractures according to the Judet-Letournel classification for resident doctors with two to four years of experience.
Abstract | Reference
