58 Chapter 3 ally” on the OHIP-14 items. The OHIP-14 has been demonstrated to be reliable in the Netherlands [24]. Internal reliability in our sample was moderate (Cronbach’s alpha 0.71|0.77). Analysis Descriptive statistics were used to report the demographic information and risk factors of oral health-related quality of life. Differences in demographics and risk factors between both study groups were analysed using Chi-square tests (χ2) and independent t-tests. Significant group differences were analysed post hoc with Bonferroni correction. Subscale scores of the dimensions of the OHIP-49 were calculated by summing the responses to subsets of items. The assumption of normality was tested, leading to the conclusion that data were non-normally distributed. Mann-Whitney U-tests were conducted to compare dimensions of the OHIP-49 and the OHIP-49 total score between the study groups. To build a model with risk factors as predictors for OHRQoL, a multiple linear regression was conducted. The predictors that were added in the model had never been studied in other studies. Therefore, forced entry was used as a method (Field, 2014). The sum score of the OHRQoL-49 was used as the dependent variable. Case and control group were entered in the first stage of the regression. Risk factors were entered at the second stage, to assess the degree to which the model could explain the variance in total OHRQoL. Preliminary analyses were performed to ensure there was no violation of the assumption of normality, linearity, multicollinearity, and homoscedasticity (Field, 2014). Chi square test of independence (Phi) were performed to examine the strength of the association between binary and dichotomized risk factors (Appendix 1) When associations between variables were < 0.60 and the variance inflation factor (VIF) <2, variables were included in the final two models (Field, 2014). There were no associations between risk factors >.60. The assumption of normal distribution was violated; therefore bootstrap was used, and the 95% CI Bias was corrected and accelerated. To calculate the estimation of prevalence of impact on OHRQoL in case and control group, the outcomes of the OHIP-14 items scale were dichotomized, 0= no impact on OHRQoL (score OHIP-14= 0), 1= negative impact on OHRQoL (score OHIP-14 ≥1). Next, cross-tabulation was used on the outcomes of impact on OHRQoL, measured with the Fisher’s exact test. Odds ratios and confidence intervals (CIs) were calcu-
RkJQdWJsaXNoZXIy MTk4NDMw