Factor structure and changes in oral health-related QOL among Japanese: comparison of nationwide surveys in 2006 and 2022

Oral health-related quality of life (OHRQoL) reflects the impact of oral conditions on daily functioning and well-being. As oral health behaviors and awareness of oral frailty have changed in Japan, it is important to examine whether the measurement properties of established instruments remain stable over time. This study examined the structural validity and cross-temporal measurement invariance of the Japanese version of the General Oral Health Assessment Index (GOHAI) and compared score distributions using nationwide survey data collected in 2006 and 2022. A repeated cross-sectional analysis was conducted among adults aged 20–79 years (2006: n = 1,120; 2022: n = 1,543). Confirmatory factor analysis (CFA) and multigroup CFA were used to evaluate one-, two- and three-factor models and to assess configural, metric, and scalar invariance across survey years. A decrease in CFI of ≤ 0.01 between successive models was used to support invariance. Latent means were compared under the scalar invariance model. GOHAI scores were compared overall and by sex and age group, with Bonferroni correction applied separately to the sex, age-group, and sex-by-age comparison families. The three-factor model yielded mixed fit indices in both survey years. Although the two-factor model showed slightly better statistical fit, the original three-factor structure was retained for theoretical consistency. Multigroup CFA supported configural invariance (CFI = 0.902, RMSEA = 0.078), as well as metric and scalar invariance according to the ΔCFI criterion, with successive CFI decreases of 0.003 and 0.008, respectively. No statistically significant differences in latent means were observed for any of the three factors. Overall GOHAI scores did not differ significantly between survey years; however, scores were significantly lower in 2022 among women overall and participants in their 60s after correction. No sex-by-age subgroup differences remained significant after correction. This study supported configural and metric invariance of the Japanese GOHAI across the two survey years. Scalar invariance was supported according to the ΔCFI criterion, although the limited absolute model fit warrants cautious interpretation. These findings highlight the importance of continued monitoring of OHRQoL across demographic groups.