Objectives: CUB is a new mixture model for analyzing ordinal data. This model contains two components called “feeling” and “uncertainty”, and the effect of covariates on these components can be evaluated. The response of WOMAC questionnaire is ordinal data. We compared PRP and HA treatments for OA by modeling on WOMAC items.

Patient and methods: The effect of treatment group, age, sex, BMI and baseline WOMAC score of patient with osteoarthritis was evaluated by modeling on final response of WOMAC questionnaire.

Result: All covariates except BMI had significant effect on feeling but no effect on uncertainty, Therefore, the value of uncertainty was estimated for each item separately.

Conclusion: Positive coefficient of treatment group on feeling component approves that patient in PRP group had less pain, stiffness, limitation than HA group. The results of this model are accordant to the result of popular test. But we evaluated the treatment by control of the other covariates and split the randomness of the answers from the individual professional judgment.

CUB model, knee osteoarthritis, WOMAC questionnaire.