A Prediction Model for Types of Treatment Indicated for Patients with Temporomandibular Disorders

Abstract

© 2019, Quintessenz. Aims: To identify potential predictors of types of treatment indicated for patients with temporomandibular disorders (TMD) and to develop, validate, and calibrate a prediction model for type of treatment. Methods: The derivation cohort at baseline was comprised of 356 adult patients with TMD. Patient and disease characteristics were recorded at baseline as potential predictors. Types of treatment indicated for TMD patients were the end points of the model, classified into no treatment, physical treatment only (including splint and/or physiotherapy), and combined physical and psychological treatment. Multinomial logistic regression analysis was used to develop the prediction model. The internal validation, calibration, discrimination, and external validation of the model were determined. For practical use, the prediction model was converted into score charts and line charts. The score of each included predictor was produced based on the shrunken regression coefficients. Results: Patient age, gender, anxiety, sleep bruxism, pain-related TMD, function-related TMD, stress, passive stretch of maximum mouth opening, and depression were significantly associated with the type of treatment indicated for TMD patients. The multinomial model showed reasonable calibration and good discrimination, with area under the curve values of 0.76 to 0.86. The internal validity of the model was good, with a shrinkage factor of 0.89. The external validity of the model was acceptable. Conclusion: Potential predictors in patient profiles for prediction of type of treatment indicated for TMD patients were identified. The internal validity, calibration, discrimination, and external validity of the model were acceptable.