Is there an alternative to the Delbet-Colonna classification? Introduction and reliability assessment of a new classification system for paediatric femoral neck fractures: preliminary results.

Abstract

The purpose of this study was to introduce a new classification system for paediatric femoral neck fractures (PFNFs) and to evaluate its reliability. Two hundred and eight unilateral PFNFs (mean patient age: 9.0 ± 4.8years) were included. Based on preoperative radiographs, the new classification system distinguished PFNFs without anterior or posterior translation (Type I), PFNFs with anterior (Type II) or posterior (Type III) translation, PFNFs with a comminuted medial or posterior column (Type IV), and subtrochanteric femoral fractures (SFFs; Type V). Radiographs were evaluated twice with an interval of two weeks by 19 raters with different specialties, experiences and geographical origins. The results were compared with a selection of 50 patient age-matched unilateral PFNFs and SFFs (mean patient age: 9.1 ± 4.9years). These were graded twice by the same graders according to the Delbet-Colonna (D-C) classification. Four radiologists and 15 paediatric orthopaedic surgeons from Europe and Asia graded the radiographs. Fair agreement was found between radiologists (κ = 0.296 ± 0.01) and surgeons (κ = 0.3 ± 0.005) (P = 0.17), although more experienced surgeons performed better than less experienced ones; a similar fair assessment was found for raters from Europe (κ = 0.309 ± 0.021) and Asia (κ = 0.3 ± 0.006) and for type II, III and IV fractures; the κ value in the first evaluation (0.309) was similar to that in the second evaluation (0.298). The overall κ value of the D-C classification subtypes was significantly higher (0.599 ± 0.217) than that of the new classification, 0.326 ± 0.162 (t = 3.190 P = 0.005). The new classification system showed fair reliability relative to the D-C classification. The reliability of the new classification system was not affected by the specialty or geographic origin of the rater or the evaluation round, only by rater experience level. The concordance was worse for PFNFs with anterior or posterior translation or with a comminuted medial or posterior columns.