Quantification and validation of predictive values of occlusal variables in temporomandibular disorders using a multifactorial analysis.

Statement Of Problem: A consensus is lacking on the association between occlusal variables and temporomandibular disorders (TMDs).

Purpose: This study estimated the maximum potential power of occlusal variables to differentiate patients with TMD from asymptomatic normal adult subjects.

Material And Methods: The occlusal characteristics in 2 sets of female patients with intracapsular TMD (1993, n = 257, and 1998, n = 124) differentiated into disk displacement and osteoarthrosis subdiagnoses were compared with asymptomatic female controls (n = 51 and 47) with multiple logistic regression analysis. Significant variables and total contribution to the log likelihood were compared with the predictive value of univariate analysis, including sensitivity and specificity.

Results: Occlusal factors in the females (1993, 1998) explained no more than 4.8% to 27.1% of the log likelihood. In comparison to the logistic regression analysis, univariate analysis was less predictive of patients with TMD, due to notably lower sensitivity. Patients with disk displacement were mainly characterized by unilateral posterior crossbite and longer RCP-ICP slides. Patients with osteoarthrosis were most consistently characterized by longer RCP-ICP slides and larger overjet, and in part to reduced overbite. Significant relative risk for disease (odds ratio > 2:1) was mainly associated with infrequent, more extreme ranges of occlusion measurements.

Conclusion: Occlusal factors may be cofactors in the identification of patients with TMD, but their role should not be overstated. Some occlusal variation may be a consequence of rather than a cause for TMD. Single variables have more limited value and it takes sets of adverse variables to model TMD. Combinations of variables appear to be disease specific. Some extreme ranges of occlusion were the domain of patients with TMD, but most patients were within the normal ranges.