Abstract
Summary
Biomechanical studies comparing fixation constructs are predictable and do not relate to the significant clinical problems. We believe there is a need for more careful use of resources in the lab and better collaboration with surgeons to enhance clinical relevance.
Introduction
It is our impression that many biomechanical studies invest substantial resources studying the obvious: that open reduction and internal fixation with more and larger metal is stronger. Studies that investigate “which construct is the strongest?” are distracted from the more clinically important question of “how strong is strong enough?”. The aim of this study is to show that specific biomechanical questions do not require formal testing. This study tested our hypothesis that the outcome of a subset of peer reviewed biomechanical studies comparing fracture fixation constructs can be predicted based on common sense with great accuracy and good interobserver reliability.
Patients & Methods
Between 2000 and 2012, we found 254 peer reviewed biomechanical studies in prestigious orthopaedic journals comparing construct ‘A’ versus construct ‘B’ to evaluate load to failure in order to determine ‘which construct is the strongest?’. Eleven studies comparing fracture fixation constructs were randomly selected from different journals based on our sense that the answer was obvious prior to performing the study. Three-hundred independent observers; including orthopaedic- and general- surgeons affiliated with the Science of Variation Group (www.scienceofvariation.org), predicted the outcome of these biomechanical studies. Observers were presented the original published illustrations of different treatment modalities and were asked to answer one question: “which construct is the strongest?” Sensitivity, specificity and accuracy were calculated according to standardised formulas. The agreement among the observers was calculated by using a multirater kappa, described by Siegel and Castellan. The kappa values were interpreted as proposed by Landis and Koch.
Results
Accuracy was the same or greater than 80% for all studies except for study 1. The level of experience had no influence on the accuracy of predicting outcomes. Sensitivity averaged 84%, ranging from 60% (for study 1) to 99% (for study 7), specificity 86%, ranging from 60% (for study 1) to 99% (for study 7), and accuracy averaged 86% from 60% (for study 1) to 99% (for study 7). The overall categorical rating of inter-observer reliability according to Landis and Koch was moderate (κ = 0,53; SE = 0.01), ranging from κ = 0,03 (SE = 0.01) to κ = 0,95 (SE = 0.01). Analyses of SOVG subgroups identified excellent agreement among Canadian surgeons. Moderate and substantial agreement were found in most of other subgroups: ranging from first year medical students to specialists 20 years or more in practice; and specialists who practice in Australia, Europe and United States. Study 5 was easiest to predict based on common sense (Accuracy 97%, inter-observer reliability 0,88). Study 1 was predicted with least accuracy 61% and the lowest kappa value 0,04.
Conclusions
The outcomes of biomechanical studies comparing fracture reduction and fixation constructs are highly predictable with good inter-observer reliability.
