2.6.1. Sample Size Calculation

Primary outcomes were counts and frequency of detection of each target bacterial species. A sample size calculation could not be made in light of the heterogeneity of the published microbiological results. Hence, a convenience sample was selected, being larger than those reported in previous studies: 30 periodontitis patients with DS [20], 40 [19], 67 [37] or 70 [23] subjects with DS.

#### 2.6.2. Data Analysis

Quantitative results were expressed as CFU per mL, in median values and interquartile range (IQR) and mean values and standard deviation (SD). Secondary outcomes were other microbiological variables (proportions of target pathogens) and clinical variables of the sampled sites and Ramfjord teeth, expressed as means and SD. The unit of analysis was the patient.

To assess the normality of the distribution, Shapiro-Wilk (in group with sample sizes below 30) or Kolmogorov-Smirnov test (in group with sample sizes above 30) were performed. Differences between three groups (periodontal health, gingivitis and periodontitis) were compared by one-way ANOVA test or Kruskal-Wallis test for quantitative variables and chi-square or Fisher tests for categorical variables.

The level of statistical significance was set at *p* < 0.05. A statistical software package IBMR SPSS Statistics 25.0 (IBM Corporation, Armonk, NY, USA) was used for data analysis.
