A model for the prediction of time to union in fractures of the tibia

Background and Purpose This report covers the second arm of a trial, the main focus of which was to assess the effect of interferential currents (IFCs) on time to union in tibial fractures. No significant improvement was found with the use of IFCs for the parameters employed (Fourie & Bowerbank, 1997). Methods The focus of this study is the development of logistic regression models which used subject characteristics to predict non‐union of fracture within 24, 32 and 40 weeks. Such models could be used to identify clients for consideration of alternative interventions, for example, electric current stimulation, bone grafting or the injection of bone morphogenic proteins (BMPs). The usefulness of such models would depend on their ability to correctly identify subjects whose fractures do or do not unite. These models were validated with respect to their sensitivity and specificity to predict non‐union for a separate data set. Results The results indicated, for instance, that use of a model to predict non‐union of fracture within 24 weeks would lead to 27% of subjects being correctly classified (as union or non‐union), 51% of subjects whose fractures did not unite within 24 weeks were identified and 65% of subjects (diagnosed as non‐union) whose fractures did not unite within 24 weeks (the closer these values are to 100% the more accurate the model). Conclusion The conclusion reached was that the models may have value for low‐cost, non‐invasive interventions, but that they could not be used to predict those cases where surgical involvement would be necessary, due to the high false‐positive rate. Copyright © 1998 Whurr Publishers Ltd.