**3. Discussion**

In this study, the e ffectiveness of WTP 828 was evaluated in a MCH water system because of its unique layout (i.e., constructed as three separate buildings). The water distribution system is characterized by a single tap water output, and each building is equipped with its own hot water return line and water disinfection treatment system. Before the introduction of WTP 828, MCH implemented a disinfection approach of continuous treatment with a ClO2 mixture (dosage of 0.5 mg/L). This type of treatment led to corrosion of some parts of the plant and a visible decrement of the e fficiency of *Legionella* colonization containment, as demonstrated by the high number of *Legionella*-positive samples in the three buildings (114/120, i.e., 95.0%) and the presence of *P. aeruginosa* in some water outlets, found during not routinely control (data not shown). In October 2013, the MCH Health Director decided to introduce WTP 828 into Building 2, as well as at available sampling points in Buildings 1 and 3.

The results obtained were discussed, for each MCH building, in relation to the period of WTP 828 introduction, as follows:


Introduction of WTP 828 during the WTP1 phase (led to an overall reduction of the percentage of *Legionella*-positive samples (from 95% to 60.0%) when compared with the ClO2 mixture. Preliminary results showed that WTP 828 treatment led to a marked reduction of contamination in Building 2 (*p* = 0.046); these results can be explained by the observation that Building 2 was the first building to undergo WTP 828 treatment. Additionally, this building has never been refurbished or otherwise altered since it was built. By contrast, Buildings 1 and 3 underwent an upgrade of the water distribution system and the construction of new accommodation sites, thus accounting for the small number of samples collected due to the absence of outlets.

These results are in line with Italian Guidelines and other authors [7,10], regarding the needs to have a broad knowledge of the buildings characteristics, the water distribution system, the pipelines material, and the disinfectant interaction with them, before to choose the disinfection method to use. As described by other authors, many actions undertaken during renovation works can induce a mobilization of biofilm and alter the flushing of disinfectant at distal outlets due to the lower water consumption and the closing of some outlets. The particulate and the increase of water turbidity, produced by structural works, can induce a decomposition of oxidants such as H2O2 [10,47,48]. Moreover, distal outlets in some parts of the hospital are seldom used: In particular areas of the hospital (surgeries room, intensive care units, etc.) the sterile water is preferred to the tap water, therefore the consumption of tap water is lower too.

The conclusion of accommodation works and the completion of the final structures within Buildings 1 and 3 allowed us to implement the risk assessment plan for *Legionella* control and to increase the number of sampling sites and the frequency of sampling, according to several studies indicating that routine cultures of the hospital water supply for *Legionella* may provide an important strategy for the prevention of legionellosis outbreaks [49].

To assess the e ffectiveness of WTP 828, we compared data obtained during the WTP1 phase with those obtained during the WTP2 phase. We observed a reduction both in the percentage of positive

samples and the mean *Legionella* levels in all buildings during WTP2. In detail, a significant reduction in the amount of *Legionella* contamination was observed in Buildings 2 and 3 (*p* = 0.001 and 0.037, respectively). *Legionella* control was then maintained for the entire duration of the study.

The observed di fferences in *Legionella* colonization between the buildings can be ascribed to the di fferent uses and water consumption in these buildings. Risk factors that should not be overlooked are, in fact, the scale of the extension, connection of existing pipes within the newly constructed branched networks, presence of dead branches, pipe characteristics (e.g., materials, age), treatment of the water system (e.g., water softening and disinfection), intended utility, and maintenance procedures [20]. In light of these considerations, we also investigated our results in relation to data concerning annual water consumption in each building size, number of water outlets, pipe materials, and the timing of renovation works.

Building 1 has six levels and covers an area of 18,539.93 m2. It mainly comprises o ffices, surgery rooms, operating rooms, and diagnostic rooms, some of which only require the use of sterile water; therefore, overall water consumption is limited. In this building, the third floor hosts a technical room for air treatment without water outlets; therefore, some closed pipes are present. The water consumption (1913 m<sup>3</sup>/year) indicated a much lower use than in Building 2 (3017 m<sup>3</sup>/year), suggesting lower water flushing from the outlets. It is evident that low use and stagnation of water may a ffect the activity and delivery of disinfectant, reducing its e ffect on the microorganisms [50,51]. The renovation works were completed in 2015. The pipelines that made up the water network comprised mainly multilayer PVC, which increases biofilm formation [52–57]. Our data revealed that, despite a reduction in the percentage of *Legionella*-positive sites and mean *Legionella* levels, WTP 828 was not completely effective in this building, demonstrating continuous fluctuations in the amount of *Legionella* spp. colonization. Corrective measures have since been implemented; these include two chemical shock treatments as above described and the implementation of maintenance hospital procedures such as increasing the flushing time once a week and during weekends, anti-scale procedures at each distal outlet (every fifteen days), and strictly cold and hot water temperature control weekly [12]. The long-term e ffects of our interventions resulted in the maintenance of *Legionella* contamination levels below the range of alert prescribed by the Italian Guidelines (101–1000 CFU/L); this will limit the risk of exposure and preserve the health of patients and workers.

In Building 2, the presence of multiple outlets (336) and some facilities with high water consumption (e.g., cafes, restaurants, and markets) suggested that water flushing facilitated the circulation of the disinfectant in the plumbing system, reducing the number of bacterium-positive samples and the *Legionella* concentration, in accordance with a study by Douterelo et al. [50]. The water distribution system consists of mainly galvanized iron, which, as suggested in the literature, on the contrary to plastic material, such as polyethylene (PE) and polyvinylchloride (PVC) [54,57], together with prolonged use of the WTP 828 disinfectant, may help to inhibit *Legionella* colonization and enable the maintenance of this inhibition over long periods.

Building 3 is the smallest structure of MCH, covering an area of 1271.06 m2. The total annual water consumption in this building is 589 m<sup>3</sup> per 129 outlets. The services (two food preparation areas) and in-patient rooms allow the daily circulation of disinfectant within the plumbing system, thereby contributing to the e ffectiveness of WTP 828 in controlling *Legionella* contamination levels in this building.

The impact of the disinfectant used (WTP 828 or ClO2 mixture) and the type of approach applied in MCH to reduce the risk of acquiring *Legionella* disease was therefore studied by calculating OR and RR epidemiological measures. Significant results were obtained comparing WTP 828 versus ClO2 by calculating the OR measure, which showed that the introduction of WTP 828 after replacement of the ClO2 mixture was a good strategy to decrease risk in MCH, increasing control of the *Legionella* contamination level. By comparing the two di fferent phases of our study, WTP1 versus WTP2, in a prospective approach using the RR measure, showed that the approach implemented during the WTP2 phase characterized by a new monitoring plan, the increase in the number of samples and adoption of a new protocol of flushing, cleaning and disinfectant monitoring, can help to decrease the risk of acquiring the disease.

The risk of Legionellosis is linked to different factors such as personal characteristics and immunodeficiency status, but such personal risk factors can also enhance the risk of acquiring the disease when environmental control is not correctly performed or is underestimated [58,59].

The serotyping and genotyping data revealed different colonization patterns in MCH buildings, but we did not find a significant association between the presence of some *Legionella* strains in MCH buildings.

Different authors have suggested that changes in the disinfection treatment regime (e.g., the type of disinfectant) or the dose (e.g., shock treatment) might influence the type of *Legionella* strains that become prevalent [42,47,60–62]. In agreemen<sup>t</sup> with these observations, the increased WTP 828 dosage used during the shock treatment performed two times in Building 1 resulted in a reduction in *L. pneumophila* SG1 and increases in *L. species: L. anisa* and *L. rubrilucens* (data not shown).

The absence of *P. aeruginosa* from water samples during the study period seems to indicate a good effect of WTP 828 on the containment of these bacteria with respect to previous evidence provided by the MCH Health Director. During the previous treatments with ClO2, the release of pipelines materials, the presence of accommodation works, and a lack of cleaning procedures favored the growth of *P. aeruginosa* (data not shown). The introduction of routine cultures of this bacterium in hot and cold-water samples suggested in the MCH study, also helped to control the efficiency of the cleaning procedures, other than evaluate the biofilm presence, where *Legionella* and other microorganisms become more resistant to antibiotics and disinfectants [48].

In detail, the protocol undertaken by the cleaning staff every fifteen days, i.e., cleaning and flushing procedures (e.g., disinfecting the taps and showers and flushing cold and hot water outlets) played an important role in preventing biofilm formation [63,64], which can support *Legionella* growth. Semiannual meetings with the stakeholders and hospital staff to inform them of the bacterial infection risk and the procedures undertaken to reduce such risks were also useful.

HPC is an indirect indicator of water quality and is often used to assess the efficacy of water treatment and to measure the amount of heterotrophic bacteria colonization in distribution systems. Despite some studies showing the absence of a correlation between levels of HPC bacteria and human infection, suggesting that HPC levels are not highly predictive of *Legionella* colonization, the control of this parameter could help to understand whether the water system contains potentially infectious organisms [33,44]. Our results indicated that WTP 828 performed well with respect to HPC containment during the entire study period, maintaining levels below directive limits [46].

A weakness of this study is that we were unable to demonstrate that WTP 828 treatment did not affect the physical and chemical parameters of the water in all study periods. Data regarding previous disinfection treatment with the ClO2 mixture, along with data from the WTP1 phase, were missing; therefore, we could not compare changes in water quality that occurred during all study periods, underscoring the important role played by environmental monitoring of physical and chemical parameters when demonstrating the efficacy of a disinfectant. By contrast, the control of these parameters during the WTP2 phase allowed us to monitor the effect of the disinfectant on water pipes, take measures to prevent damage to the water network, and maintain the quality of "drinking water" to prevent risks to human health.

According to Borella et al. [10], the choice of WTP 828 has been carried out also on a careful evaluation of a cost-effective analysis, considering the system of disinfectant production (pump or others), the maintenance costs (disinfectant provision, service, etc.) and the potential dangerous effect on water pipelines, other than the possibility to safety measure disinfectant residues to outlets by colorimetric strips by staff.

#### **4. Materials and Methods**

#### *4.1. MCH Structure and Water Outlet Characteristics*

This study was conducted at Maria Cecilia Hospital (MCH), an Italian hospital founded in 1973 located in Cotignola (RA, Emilia Romagna).

The structure of the hospital is complex as it comprises three separate buildings (Buildings 1–3) built during di fferent years and covering a total area of 27,989.64 m<sup>2</sup> (Supplementary Figure S1). Buildings 1 and 3 were constructed in 2001 and subjected to renovation or enlargement works until 2015. Building 2 is the main MCH building and did not undergo any changes in its structure during the whole study period.

These characteristics permit the study of WTP 828 activity as three separate HWNs and allow modulation of the dose with respect to the level of *Legionella* contamination, water demand, intended use, and renovation works (Supplementary Table S1).

At the end of the renovation works, the final structure of MCH had 122 in-patient rooms, each with one or two beds and an en-suite bathroom. There were 212 beds in total, mainly located in Building 2, and 769 water outlets (e.g., taps and showers) located in in-patient rooms, communal areas, diagnostic and operating rooms, o ffices and services, as follows:


#### *4.2. Hospital Water Network (HWN)*

The hospital plumbing system is very complex, partially antiquated, and (depending on age built) predominantly made up of galvanized iron and polyvinylchloride (PVC) multi-layers. The HWN was coated with an anti-scale treatment to create a protective film on the galvanized iron and PVC surface, as suggested by WHO guidelines in 2011 [32]. It consists of a product based on natural mineral salts such as orthophosphates, polyphosphates, and alkaline silicates dosed at 0.1 mg/L. The MCH structural characteristics, material pipelines, and water consumption for each building were kindly provided by Health Direction and described in Supplementary Table S1.

All buildings are supplied with the same municipal water aqueduct, which brings water from the Ridracoli dam located 53 km from Cotignola. The water is first collected in two 30 m<sup>3</sup> water reservoirs outside the buildings. After filtration through a 150 μm pore size filter, water is fed into two pipelines: one to the cooling towers and refrigerant circuit (a closed loop hydraulic system) and the other to the water treatment station (an open loop hydraulic system). A plan of the water distribution network is shown in Figure 1.

**Figure 1.** The scheme of the MCH water network with main sampling points in technical rooms (\*).

A heat exchanger maintains the temperature of the cold water in the treatment station at <18 ◦C; the hardness of cold water is treated with a general softener to reduce its value between 12–15 ◦f (water moderately hard), which is in line with Italian and European Council directives [45,46]. Some of this water supplies the sterilizers after reverse osmosis treatment, and another portion is used as cold water by the hospital. The cold water is distributed to the substations within each building through a single tap water output. Three different heat exchangers (one at each substation) produce hot water. The cold and hot water circuits are independent of one another, and each building has its own hot water return line.

## *4.3. WTP 828*

Water Team Process 828 (WTP 828) developed by an Italian Company involved in disinfectant production (Water Team S.r.l., Forlì (FC), Italy) is a multi-component oxidizing biocide formulated using a stabilized combination of H2O2 (34%, wt/wt) and Ag+ salts (0.003%, wt/wt) in demineralized water, resulting in a highly effective disinfection solution. The formulation is covered by Italian regulation on intellectual property rights and actually is under investigation to acquire a patent. It is licensed by European and Italian legislation [27,65] for its application in drinking water. The synergistic action of H2O2 and Ag+ salts renders the biocide more powerful than H2O2 alone [66,67]. Ag+ was used to increase the activity of the peroxide, and Ag+ forms an insoluble salt at distal points and is able to attach to pipes and exert bacteriostatic effects on biofilms [41,68].

The WTP 828 is injected into mixed water (hot/cold) after hot water output downstream from the heat exchangers and dosed proportionally to the volume of water supply.

WTP 828 was introduced into MCH for the first time in October 2013 after replacement of a previous disinfection system based on a continuous treatment performed with chlorine dioxide (ClO2 mixture) at a dosage of 0.5 mg/L. This treatment, which was used from September 2009 to September 2013, had compromised the water pipelines and corroded some parts of the plant, thereby reducing efficacy with respect to *Legionella* colonization and supporting the presence of *P. aeruginosa* in some water outlets.

The WTP 828 concentrations during the study were modulated according to the microbiological results for each building. In particular, the initial dose of 30 mg/<sup>L</sup> resulting in a final concentration of 5–10 mg/<sup>L</sup> at distal outlets remained the same in Buildings 2 and 3 throughout the whole study period. By contrast, two shock treatments were required in Building 1 (from February to March and July to August 2015); at these times, the injected dose of WTP 828 increased up to 50–60 mg/L, which resulted in 25–30 mg/<sup>L</sup> of H2O2 at the distal outlets.
