External Validation of a Clinical Prediction Tool for the Use of Manual Therapy in Patients With Temporomandibular Disorders

TL;DR

This study validates a model to predict which patients with jaw disorders will benefit from manual therapy, showing that pain patterns and expectations are key factors.

Contribution

The study externally validates a clinical prediction model for manual therapy outcomes in TMDs and identifies a novel interaction between treatment expectations and central sensitization.

Findings

The original model showed strong discrimination (AUC = 0.95) and fit (R² = 0.75) in predicting pain reduction after manual therapy.

A significant interaction was found between treatment expectations and central sensitization inventory (CSI), with high CSI negatively affecting outcomes in those with positive expectations.

New models predicting one-month outcomes showed excellent discrimination (AUC = 0.96) using similar baseline predictors.

Abstract

Manual therapy (MT) is frequently used to manage temporomandibular disorders (TMDs), yet patient responses vary significantly. Predictive models may help clinicians tailor treatments to individual patients.

The primary aim of this study was to externally validate a previously developed prediction model for identifying patients with TMDs who are more likely to benefit from MT. Additionally, new prognostic models to predict outcomes at a one‐month follow‐up were developed.

A cohort of 124 adults with a diagnosis of a TMD received a four‐week MT program (one session per week) applied to craniomandibular structures. Predictors collected at baseline included clinical and psychosocial variables: pain during mouth opening, pain localisation, treatment expectations, and the central sensitisation inventory (CSI). The primary outcome was a ≥ 30% pain reduction post‐treatment. Model performance was assessed using discrimination, calibration, and decision curve analysis. New models predicting one‐month outcomes were developed and internally validated via bootstrapping.

The original model showed strong discrimination and overall fit in the validation cohort (AUC = 0.95, R
2 = 0.75). Except for pain location, the predictors in the original model also showed excellent discrimination in the developed model based on the outcome at the one‐month follow‐up (AUC = 0.96). A significant interaction between treatment expectations and CSI was found, with high CSI negatively affecting the outcome in people with positive treatment expectations.

This study externally validated a clinical prediction model for pain reduction in people with TMDs following MT, confirming previously identified predictors of good outcome such as pain during mouth opening, positive treatment expectations, localisation of pain in the craniocervical region, and lower CSI scores. A link to a web‐based calculator of the prediction model is provided. The interaction between treatment expectations and CSI suggests that CSI plays a key role in shaping treatment response, modulating the influence of treatment expectation. Future studies with a control group are needed to confirm these results and distinguish true treatment effect modifiers from general prognostic factors.

ClinicalTrials.gov ID: NCT03990662

Predicting the response to manual therapy for people with temporomandibular disorders.

Figures