M Beerens

86 CHAPTER 5 Sample size To assess the influence of casein phosphopeptide-amorphous calcium fluoride phosphate (CPP-ACPF) on the reduction of WSL, a power analysis was conducted as described by Beerens. (Beerens et al. , 2010). Based on a previous observational study at the Orthodontic Department at ACTA (Boersma et al. , 2005). A statistically significant, but clinically irrelevant, natural reduction in fluorescence loss, of 0.9 % (SD = 0.9 %) was found, during a 24 week time period. A clinically relevant change in fluorescence loss was considered to be an average reduction of 2 %, implying an effect size of 0.55. The sample size was calculated for a more conservative effect size of 0.35. For an effect size of 0.35 with a power of 0.9 to be measured between the two groups, a group size of 27 was needed (G*-power 3.1.0, ANOVA for repeated measures, between factors). Although orthodontic patients, in general, are seen at 4- to 6-week intervals during the active phase of treatment, during the retention phase, subjects often do not show up for their scheduled appointments. At the department of orthodontics at ACTA, this level of no shows is relatively high. To compensate for subject withdrawal, it was aimed to include 30 subjects in each group. Subjects who dropped out before T2 were replaced to meet the required minimum group size of 27. Interim analysis The 3-month data from this study were reported in December 2010 (Beerens et al. , 2010). The trial was not stopped earlier than planned. Data analysis Statistical analysis was performed with SPSS (PASW statistics 21.0; SPSS Inc., Chicago, IL, USA). Change of enamel lesions, assessed by QLF, was the primary outcome measure. The average fluorescence loss for all WSL, total lesion area, and IFL were calculated for each subject and then normalized to 20 surfaces corrected for the number of missing and filled surfaces during the trial. Student’s (two-tailed) t -test was used to determine differences between both groups at baseline and follow-up time points. Lesion and microbial changes in time per subject were determined by repeated-measures ANOVA, followed by pairwise comparisons with Bonferroni correction. Visual lesion changes were assessed with clinical photographs as secondary outcome. The Mann–Whitney