Validation and improvement of a predictive model for treatment outcome in patients with temporomandibular disorders

OBJECTIVE: To examine whether it was possible to improve individual prediction of treatment outcome in patients with temporomandibular disorders (TMD) through continuous quality improvement registrations and implementation of changes in clinical routines. MATERIAL AND METHODS: All 5777 patients referred to a specialist clinic for stomatognathic physiology; 2187 TMD patients started treatment. The patients were divided into Muscle or Mainly TMJ symptom groups. Actual treatment outcome was predicted as Good or Dubious based on patient history and clinical findings. The degree of improvement of initial symptoms was graded using a numeric rating scale 0-100. Improvement in initial complaints of 50% or more was judged as a clinically important difference (CID); 989 patients (76%) completed treatment 1992-1998 (Sample 1) and 769 (86%) treatment 1999-2004 (Sample 2). RESULTS: For patients with Muscle symptoms in Sample 1, CID was reached by 90% of those predicted Good and by 56% of those predicted Dubious. In Sample 2, the figures were Good 93% and Dubious 57%. In subgroup Mainly TMJ symptoms, CID was reached by 94% of those predicted Good and by 88% if predicted Dubious in Sample 1. In Sample 2, the figures were the same for those predicted Good, i.e. 94%, and Dubious, 73%. CONCLUSION: Continuous quality improvement registrations and implementation of changes in clinical routines improved the ability to predict individually the actual treatment outcome between patients predicted Good or Dubious, respectively. Rheumatic disease turned out not to be a negative predictor for treatment outcome.