Construct Validity and Measurement Invariance of OIDP

As a consequence of nation-wide application, the issue of cross-cultural validation of the oral impact of daily performance questionnaire, OIDP becomes important. Objectives: Using confirmatory factor analysis, CFA, this study: 1) evaluated a proposed 3-factor structure of the OIDP in Tanzanian adolescents 2) cross-validated the 3-factor model in Tanzanian adults and 3) assessed whether a 3-factor model would be replicated among Ugandan adolescents. Methods: Between 2004 and 2007 complete data for the Kiswahili translated-and original English versions of the OIDP were collected in 1601 Tanzanian adolescents (mean age 13 yr), 1031 Tanzanian adults (mean age 63 yr) and 1146 Ugandan adolescents (mean age 16 yr). Model generation analysis was restricted to Tanzanian adolescents and the model achieved was tested without modifications in Tanzanian adults and Ugandan adolescents. Results: Compared to a 1-factor model: Χ2=1048.5, df=20, CFI=0.78, RMSEA=0.18, an initially proposed 3-factor model revealed an improved fit to the Tanzanian adolescent data: Χ2=291.5, df=17, CFI=0.94, RMSEA=0.10. Adjustments of the original 3-factor model gave a 3-factor solution with cross-loading that had even better fit: Χ2=30.11, df=12, p=0.003, CFI=0.99, RMSEA=0.03. Cross-validation of the final model in Tanzanian adults provided a reasonable fit: Χ2= 88.211, df=12, CFI=0.98, RMSEA =0.07. Multi-group CFA with the combined Tanzanian and Ugandan adolescent data demonstrated acceptable fit Χ2= 140.829, df=24, CFI=0.98, RMSEA=0.04. Differences in factor loadings occurred (Χ2 change 32.95, df= 7, p<0.001) when comparing unconstrained and constrained nested models. Conclusion: Construct validity of a modified 3-factor model was confirmed in three independent samples suggesting that OIDP conforms to the theoretical considerations that physical, psychological, and social aspects are measured in oral health related quality of life. Measurement invariance across Tanzania and Uganda was not confirmed at the level of factor loadings making cross-national comparison of scores questionable.