RESUMO
OBJECTIVES: To examine the McGill Denture Satisfaction Questionnaire (MDSQ) in terms of dimensionality, item reduction and construct validity in a binational sample of complete denture wearers. MATERIALS AND METHODS: We conducted secondary analyses of baseline data from two studies on implant-retained overdentures: a quasi-experimental study in the United States (n = 145) and a randomised trial in Brazil (n = 120). All participants wore upper/lower dentures and responded at baseline to the MDSQ items concerning their original mandibular dentures. A putative model of the MDSQ items resulted in two question subsets: (a) overall satisfaction, retention/stability, aesthetics, cleaning, speech and comfort, plus general chewing ability; (b) mastication of specific foods. Analyses focused on the internal consistency of each subset and possible item reduction, using Cronbach's alpha (Cα), inter-item correlation and exploratory factor analysis (EFA). RESULTS: The 1st subset showed high inter-item correlation for most question combinations and no redundancy (r ≤ .8). An item on cleaning had low correlation, but its removal does not increase internal consistency (Cα ≥ .83). Results were similar for both studies, with EFA showing a single significant factor (namely "overall satisfaction, lower denture") able to explain nearly 54% of the variance. The 2nd subset also shows strong internal consistency (Cα ≥ .95) and inter-item correlation, with a single factor representing 65% of the variation. CONCLUSIONS: This study discloses the reliability and construct validity of the MDSQ for patient-centred evaluation of complete dental prostheses in the edentulous mandible. Findings also support the use of both "overall satisfaction" and "masticatory ability" as summary scores, for improved outcome assessment.