Abstract
Background The AO comprehensive pediatric longbone fracture classification system describes the localization and morphology of fractures, and considers severity in 3 categories: (1) simple, (2) wedge, and (3) complex. We evaluated the reliability and accuracy of surgeons in using this rating system.
Material and methods In a first validation phase, 5 experienced pediatric (orthopedic) surgeons reviewed radiographs of 267 prospectively collected pediatric fractures (agreement study A). In a second study (B), 70 surgeons of various levels of experience in 15 clinics classified 275 fractures via internet. Simple fractures comprised about 90%, 99% and 100% of diaphyseal (D), metaphyseal (M), and epiphyseal (E) fractures, respectively.
Results Kappa coefficients for severity coding in D fractures were 0.82 and 0.51 in studies A and B, respectively. The median accuracy of surgeons in classifying simple fractures was above 97% in both studies but was lower, 85% (46–100), for wedge or complex D fractures.
Interpretation While reliability and accuracy estimates were satisfactory as a whole, the ratings of some individual surgeons were inadequate. Our findings suggest that the classification of fracture severity in children should be done in only two categories that distinguish between simple and wedge/complex fractures.
Contributions of authors
TS and LA: contributed equally to this work and wrote the manuscript. TS: initiator of the project, conducted the case collection, developed the study protocol, participated in the classification sessions, and provided the clinical expertise as principal investigator. LA: developed the study protocol, coordinated the project, and provided the methodological expertise, including data analysis. The other 5 authors are members of the AO Pediatric Classification Group and were involved in the development of the classification and the clasification sessions as experienced surgeons. All authors approved the current version of the manuscript.