Sample size and statistical analysis
According to the study by Lacroix et al.11, at least 63 elderly people in each group will be needed to meet a statistical level of 80% in detecting a difference of 2 points adopting an alpha error of 5%, considering a standard deviation of 4 points. The principle of intention-to-treat analysis was considered, and the sensitivity analysis was performed through simple imputation using the mean of the variables. To test the data normality, the histogram, mean and median, standard deviation, skewness and kurtosis were analyzed, and the Shapiro-Wilk normality test was used for confirmation. Due to the normal distribution of the variables, parametric statistics were used, and the intragroup comparisons were made through the Student test for paired samples. The analysis of subgroups of age groups, which was pre-specified in the study protocol, as it was done with more than two groups that presented parametric distribution, one-way ANOVA was used, followed by the one-way ANOVA Tukey’s multiple comparison test. The significance level established for all analyzes was p <0.05 and statistical procedures were analyzed and processed in the Statistical Package of the Social Sciences - SPSS version 21.0 (IBM, Armonk, NY, USA).