Development of the Orthodontic Treatment Impact Questionnaire: Cross-sectional validation

Abstract

Introduction: The need to involve patients in developing and evaluating health care interventions is now well-recognized. This study assesses and refines the Orthodontic Treatment Impact Questionnaire for use as a patient-reported outcome in an interventional clinical trial to evaluate and compare any orthodontic interventions.

Methods: The face and content validity of a previously developed questionnaire were tested in 2 focus groups involving adolescents aged 11-17 years. They were wearing a range of orthodontic appliances and at different treatment stages. A similar cross-sectional convenience sample completed the questionnaire during routine appliance adjustment appointments. A Rasch model, using item response theory, was used for item reduction, assessment of the response format, and differential item functioning. Spearman's rank correlation was used to assess construct validity, Cronbach α for internal consistency and reliability, and intraclass correlation coefficient for test-retest reliability.

Results: Seven adolescents (4 females, 3 males) were involved in the initial testing; 181 (117 females, 64 males; mean age, 14.7 ± 1.5 years) completed the questionnaire once and 41 twice. The initial measure demonstrated a misfit to the Rasch model. Ten of the original 31 items had disordered thresholds and were removed. The 5-point scale was changed to a 3-point scale. None of the participants demonstrated a misfit to the model. Construct validity (P = 0.480), internal consistency (Cronbach α = 0.827) and test-retest reliability (intraclass correlation coefficient = 0.85; 95% confidence interval, 0.73-0.92) were good.

Conclusions: The initial Orthodontic Treatment Impact Questionnaire was tested and modified using item response theory. The modified questionnaire demonstrated good construct validity, reliability, and internal consistency. Further testing to assess generalizability and longitudinal responsiveness is required.