# TROCHANTERIC FRACTURES: ARE THE CLASSIFICATIONS RELIABLE?

## TL;DR

This study finds that current classifications for trochanteric fractures have low agreement among observers and suggests improvements or alternatives.

## Contribution

The study introduces a bimodal classification proposal and evaluates its reproducibility using interobserver agreement.

## Key findings

- The interobserver agreement for classifying trochanteric fractures was weak (kappa = 0.272).
- Classifications with low reproducibility should be improved or abandoned.
- A minimum kappa value of 0.8 is recommended for reliable classifications.

## Abstract

To evaluate classification use based on the assessment of agreement in trochanteric fractures between different observers through a simple proposal of bimodal classification.

A total of 50 radiographic images of femur trochanteric fractures were selected and classified by 22 evaluators, 10 traumatologists and 12 residents, as STABLE or UNSTABLE. The assessment of reproducibility was done using the kappa statistical index. After the evaluation, the groups were isolated and submitted to the ANOVA test with Bonferroni correction to evaluate the statistical differences between them.

When evaluated by the kappa index, the results of the 50 fractures assessed by the 22 evaluators were k = 0.272. The reproducibility of the classification proposal was considered statistically weak.

Based on this study, it is recommended that the classifications be extensively tested and reach a minimum level of interobserver reproducibility (kappa > 0.8). It is suggested that classifications that do not achieve this result be improved or abandoned.
Level of evidence II, Prospective study.