*4.8. Data Analyses*

Bacteriological data were converted into log10 x values to normalize the distribution for the correlation analysis. The normality of continuous variables was assessed using the Shapiro-Wilk test, and data are presented as the mean ± SD. Continuous variables were evaluated using one-way ANOVA and a post-hoc test (Bonferroni), and categorical variables were compared using the χ2 and Mann Whitney test. One-way ANOVA and the post-hoc test (Bonferroni) were conducted to assess differences between disinfectant treatments and between buildings.

To test the changes in *Legionella* risk between treatments, we used odds ratios (ORs) in a retrospective analysis and relative risk (RR) in a prospective analysis. In detail, OR was calculated for WTP1 versus the ClO2 mixture and WTP2 versus the ClO2 mixture, and RR was calculated for the prospective treatments (WTP2 versus WTP1). Statistical analyses were performed using STATA version 10.0 (Stata Corp., College Station, TX). A *p*-value < 0.05 was accepted as significant.

#### *4.9. Hospital LD Surveillance*

MCH performed active legionellosis surveillance beginning in 2013. Data were collected throughout the entire study period (WTP1 and WTP2 phases). The symptoms of legionellosis are consistent with an acute infection of the lower airways, with clinical and/or radiological signs of focal pneumonia. A preliminary diagnosis was routinely confirmed by a urine antigen test (*Legionella* Urine Antigen EIA, Biotest, Milan, Italy) and a serological immunofluorescence test (*L. pneumophila* IFA, Meridian Diagnostic Europe, London, UK).
