Reliability Assessment Of The Four-column Classification Of Pilon Fractures

Objective: The purpose of this study is to evalue the inter-observer and intra-observer reliability of Four-Column classification of Pilon fractures,and compare with the traditional classifications of Ruedi-Allgower and AO to demonstrate the reliability of Four-Column classification and provide the statistical basis for the further promotion of Four-Column classification.Materials and Methods: From January 2008 to December 2015,162 patients of Pilon fractures were treated in the first Affiliated Hospital of Dalian Medical University(including 5 cases of bilateral Pilon fractures,a total of 167 Pilon fractures),96 male patients,66 female patients,mean age 45.9 years(18-87 years),53 patients of falls injuries,25 patients of car accident injuries,and other 84 patients,51 patients associated with other fractures,83 left Pilon fractures,84 right Pilon fractures,9 open Pilon fractures.According to the Ruedi-Allgower classification,type?of 25 fractures(14.97%),type ? of 116 fractures(69.46%),type ? of 26 fractures(15.56%)were involved.According to AO classification : type B1 of 20 fractures(11.97%),type B2 of 103 fractures(61.68%),type B3 of 9 fractures(5.39%),type C1 of 6 fractures(3.59%),type C2 of 12 fractures(7.19%),type C3 of 17 fractures(10.18%)were involved.According to the Four-Column classification,single column of 70 fracture(41.92%),two columns of 56 fractures(33.53%),three columns of 24 fractures(14.37%),four columns of 17 fractures(10.18%)were involved.Among all the 167 Pilon fractures,anterior column of 59 fractures(35.33%),posterior column of 103 fractures(61.68%),medial column of 97 fractures(58.08%),lateral column of 63 fractures(37.73%)were involved.167 Pilon fractures were randomly ordered with a random number table,and then randomly selected 60 Pilon fractures.Every patient were carried out X-ray(including anteroposterior,lateral),CT scan and 3D reconstruction before surgery.Six orthopedists were selected,and they had from 5 to 10 years of experience.In the week before typing,all three classifications of the graphic and text description were distributed to six orthopedists,and these six orthopedists were trained about the Ruedi-Allgower classification,AO classification and Four-Column classification.A week later,the 60 cases of the imaging data were distributed to six orthopedists to read.The imaging data included anteroposterior and lateral DR,CT scan and three dimensional reconstruction.Each case of radiological data included : AO classification(only X-ray),Ruedi-Allgower classification(only X-ray),Four-Column classification(including X-ray,CT scan,3D reconstruction).All cases were the same,and 60 cases of Pilon fractures were anonymous,only a digital ID,covered all identifiable data(in addition to mark the left and right outside the label).The order of three folders of all cases were random number table random order,and each was different.Every observer had sufficient time to think about the case within the specified time,according to their own rhythm in electronic form of the study.Six orthopedists participating in the study were familiar with the three classifications,but during the typing process,they still referred to the three type of illustrations and text to increase the accuracy of the results.Six doctors writed down their opinions on the predesigned form.All image data was arranged randomly labeled 1-60,and included all types Pilon fracture.After the first round type,the six orthopedists would not get these imaging data,and would not get any feedback.Reproducibility of eight weeks later,the same cases were used,without replacing staff participating in typing and no re-training.The kappa coefficients(?)was interpreted to assess the inter-observer reliability and intra-observer reliability.SPSS22.0 statistical software was Applied for statistical data analysis,? <0.05 considered statistically significant.Results: The average ? value for inter-observer reliability using AO classification was 0.370(range 0.254-0.485),representing “fair agreement”,the reliability of the Ruedi-Allgower classification of inter-observer average ? value was 0.344(range 0.161-0.501),representing “fair agreement”,whereas the average ? value was 0.739(range 0.615-0.832),representing “substantial agreement” by using the four-column classification.The average ? value for intra-observer reliability using AO classification was 0.383(range 0.283-0.472),representing “fair agreement”,the reliability of the Ruedi-Allgower classification of inter-observer average ? value was 0.385(range 0.229-0.582),representing “fair agreement”,whereas the average ? value was 0.757(range 0.715-0.812),representing “substantial agreement” by using the four-column classification.Conclusion: On the basis of X-ray and CT scan,the Four-Column classification which is mainly based on the three-dimensional reconstruction,is a simple and comprehensive classification theory.The Inter-observer reliability and intra-observer reliability of this classification has reached substantial agreement,and it can make up for the lack of the traditional type.Therefore,We suggest that give priority to the use of four column classification in clinical work.