Aim
This study aimed to investigate if in the 49-item Oral Health Impact Profile (OHIP): (i) more missing data occurred when participants answered more questions, (ii) more missing data occurred in a particular item or set of related items, and (iii) item missingness was associated with the demographic characteristics and oral health-related quality of life (OHRQoL) impairment level.
Methods
We used OHIP data from the Dimensions of OHRQoL (DOQ) project, which consolidated data from 35 individual studies. Among these studies, we analyzed OHIP data from 19 studies (4,847 surveyed individuals, of which 3,481 were completed under supervision and 1,366 were completed unsupervised) that contained some missing information. We computed descriptive statistics to investigate the OHIP missingness. We also used logistic regression analyses, with missing information as the dependent variable, and number of questions filled in (OHIP item rank) as the independent variable for samples with and without supervision. To investigate whether missing data occurs more in a particular item or set of related items we fitted regression models with individual OHIP items and the OHRQoL dimensions as indicator variables. We also investigated age, gender, and OHRQoL level as predictor variables for missing OHIP items.
Results
We found very low levels of missingness across individual OHIP items and set of related items, and there was no particular item or set of related items that was associated with more missing data. Also, more missing data did not depend on whether the participants answered more questions. In studies without supervision, older persons and females were 5.47 and 2.66 times more likely to have missing items than younger persons and females. However, in studies with supervision, older persons, and participants with more OHRQoL impairment were 1.70 and 2.65 times more likely to have missing items.
Conclusion
The study participants from general and dental patient populations did not find OHIP-49 burdensome. OHIP item missingness did not depend on a particular OHIP item or set of related items, or if the study participants responded to a greater number of OHIP items. We did not find a consistent pattern of the influence of sociodemographic and OHRQoL magnitude information on OHIP missingness. The amount of missing OHIP information was low making any potential influence likely small in magnitude.