**3. Discussion**

The prevention of HAIs is an important problem, particularly in high-risk patient care. The risk of infections has been linked to interactions between pathogens and hosts which involves the number of microorganisms, their virulence factors, and the host's immune defenses [21]. To reduce the impact on human health as well as to avoid economic, legal, and political issues, particular attention must be directed to a hospital's hygiene and environment. This aim can arise only through the development of a risk assessment plan which is linked to knowledge of the hospital and patient characteristics, the health-care procedures already in place and to be improved, and the hospital environment where

patients and HCS may be in contact with microorganisms through the air, water, and contaminated surfaces [6,22,23].

The new revision of the European Drinking Water Directive, such as the WHO Guidelines for Drinking Water Quality, suggests the approach of the Water Safety Plan to identify the main pathogens involved in waterborne diseases, to understand their pathogenic pathways, and to contain their impact on public health [24–28]. *Legionella* and *P. aeruginosa* are two of the main waterborne pathogens involved in hospital environments associated with nosocomial infections [6,29,30].

This study reports knowledge acquired during a *Legionella* environmental surveillance program performed in hospitals, where high *Legionella* levels were detected in SHWOs with TMVs, some of them with concentrations over the risk level (>2 Log cfu/L), suggesting their critical role in bacterial growth and HAI risk. It has been well documented that temperature is a key factor in microbial growth and that, in particular, the mixing of hot and cold water creates an optimal temperature for bacterial environment, which can occur in SHWOs [8,23,31,32].

To analyze the contamination found in SHWOs, hot- and cold-water data sets were separately studied in terms of percentage of positive samples, level of contamination, and *Legionella* isolates distribution, including temperature as a possible determining factor for data fluctuations in the microbial parameters analyzed.

The results showed a similar percentage of hot- and cold-water samples (44.5% and 42.6%, respectively) contaminated by *Legionella*, with the same trend regarding samples over the *Legionella* risk level (73.7% hot vs. 68.0% cold). A nonsignificant difference in terms of *Legionella* contamination between hot and cold samples (*p* = 0.34) demonstrates how hot- and cold-water circuits are not separate with continuous mixing between two pipelines, creating an environment capable of supporting *Legionella* growth.

These results are supported by the residues of H2O2 disinfectant found also in cold-water samples. This disinfectant introduced in hospitals is injected only into the return line of the hot-water distribution system, and generally, when the two main distribution systems (e.g., hot and cold) are well separated, the cold water is expected to be free of disinfectant residues. This observation can be attributed to damage on the TMV cartridge because, during cold-water sampling, although the TMVs were deactivated, we found disinfectant residues in all samples. Moreover, damage in the TMV device was supported by the temperatures measured, which revealed a decrease in hot-water values and an increase in cold-water values, as demonstrated by the large ranges of temperature: 21.9–60.1 ◦C and 9.2–44.7 ◦C for hot and cold, respectively.

Considering the distribution of *Legionella* isolates, a significant difference was found between hot- and cold-water-positive samples (*p* = 0.001), showing that the characteristics of the mixed water produced are able to influence the distribution of isolates. According to knowledge about *Legionella* ecology and epidemiological data, the main positive samples found in hot and cold water (64.7% vs. 42.7%) belonged to *L. pneumophila* (serogroups 1, 3, 4, 6, and 8). In a low percentage of hotand cold-water samples, we found isolates belonging to *Legionella* non-*pneumophila* species (*L. anisa*, *L. rubrilucens*, *L. tauriniensis*, *L. nautarum*, and *L. steelei*), with high values in cold water compared to hot water (28.1% vs. 13.7%). The same differences were found between cold and hot samples regarding the percentage of positive samples contaminated by both species (*L. pneumophila* and other *Legionella* spp.).

These data required supplementary analysis regarding the level of contamination found inside each distribution system and between them. In hot-water samples, we found a higher *Legionella* contamination in samples contaminated by both isolates, with a significant difference with respect to the level of contamination found in samples with only other *Legionella* spp. (*p* = 0.00012). A significant difference was, therefore, found in terms of the level of contamination between *L. pneumophila* and other *Legionella* spp. (*p* = 0.03).

In cold-water samples, we observed a different trend, with high samples contaminated by both species showing significant differences with respect to samples having only *L. pneumophila* (*p* = 0.0012) and samples contaminated by only other *Legionella* spp. (*p* = 0.0046).

Considering the comparison of mean concentration found for each isolate between hot versus cold samples, a significant di fference was found only for *L. pneumophila* (*p* = 0.008).

Relevant information comes from these results regarding the ecology of isolates in water distribution systems.

*Legionella* lives in a water environment, with optimal growth in warm environments. Therefore, the abundance of *L. pneumophila* in hot-water samples found was in line with data about the high incidence of this species in human disease. In hot-water environments, there is likely a selective pressure of *L. pneumophila* on *Legionella* non-*pneumophila* species, which is suppressed in cold-water distribution systems, as demonstrated by the high number of samples with both species when the water temperature was mixed. Our hypothesis is also based on observations done during *Legionella* culture, where we generally find a lower *Legionella* non-*pneumophila* species isolation rate, due to their slow growth and late detection after 10–15 days of incubation when *L. pneumophila* is more abundant. When the culture technique was conducted up to 10 days, some of these species were missing and, consequently, underestimated; by contrast, an extension of culture timing permits their detection.

The poor awareness of these species and their underestimation is also associated to the low rate of clinical isolation, to their low correlation with human disease, and to the non-detection by diagnostic techniques (e.g., antigenic urinary tests) [33,34].

Another important point that can explain the high presence of *Legionella* non-*pneumophila* species in cold water is related to the disinfection treatment that often, as seen in this study, is performed on the hot-water circuit, leaving the cold-water distribution system without any type of control (monitoring by culture, temperature measures, flushing, and disinfectant residues measures). This represents a reservoir for other *Legionella* species. The absence of disinfectant or low levels of disinfectant residues measured usually require high temperatures for their activation and are unable to control their growth.

These results were also confirmed by our previous data [35] regarding the ability of *Legionella* to colonize and increase its concentration in cold-water distribution systems, inducing a change of cold water microflora; during renovation works, pipeline, TMV, and faucet damage; or when rapid breakdown of hot temperatures occurs. The presence of a high percentage of positive samples with high *Legionella* concentration contaminated by both species in both distribution systems confirms that SHWOs with mixed water develop an environment favorable to *Legionella* growth.

The high contamination of SHWOs are therefore supported by a wide fluctuation of temperatures found in samples: both low and high temperatures are able to favor bacteria growth. The analysis of contamination levels with respect to temperatures was analyzed by dividing the temperature values measured between four ranges, each of them associated to the ecology of *Legionella*.

The possibility to maintain separation between cold- and hot-water pipelines is one of the strategies suggested by National and European directives in order to contain the proliferation of bacteria. Our data demonstrated an inverse correlation between the temperature and bacteria load: at higher temperatures (>49.6 ◦C), a lower *Legionella* mean concentration (1.64 Log cfu/L) was observed, according with the directive's suggestions about the value of >50 ◦C being able to perform complete control of the level of *Legionella* [12].

The results obtained inside the II and III ranges of temperature (21–45 ◦C and 45.1–49.6 ◦C, respectively) showed approximately the same *Legionella* concentrations with a nonsignificant di fference inside these ranges (*p* = 0.474). These data confirm that samples with temperature close to the optimum *Legionella* growth range (25–42 ◦C) are more contaminated and that an increase of temperature (>49.6 ◦C) leads to control of the *Legionella* proliferation (II vs. IV, *p* = 0.012; III vs. IV, *p* = 0.001) [36].

The contamination in SHWOs and the wide range of temperatures found can be explained, moreover, by taking into account the SHWO technology provided by hospitals. All of them are characterized by the presence of magnetic valves, which are the principal part of electronic/non-touch/sensor tap systems. Cold and hot water from the junctions of the central water pipeline system are mixed to provide an acceptable and comfortable setting temperature, generally, around 36 ◦C. The magnetic valves in the cartridge are made of material membranes—for example, made of rubber, plastic, or Polyvinyl Chloride (PVC)—which are very hard to disinfect and easily enhance bacterial growth and biofilm development, which can become a protective envelope against biocides and disinfectants. Furthermore, in these tap systems, flushing procedures are forbidden by the presence of a photocell system, leading to low water pressure and flow [18]. These considerations were supported by data about positive samples located in the range of the TMVs' working temperature (21–45 ◦C), which is very close to the temperature associated with optimal *Legionella* growth, where we did not find a di fference between hot and cold samples.

Our hypothesis is strengthened furthermore by the observation that, in three hospitals which implemented a substitution program from sensor-activated faucets with TMVs to manual clinical valves without TMVs, an increase of mean temperature was measured, corresponding to a significant reduction trend in *Legionella* concentration levels.

As concerning *P. aeruginosa* SHWO contamination, the lower presence of positive samples coming from the eleven hospitals suggests the general good performance of disinfection procedures applied by hospital sta ff on faucets. The choice of tapware provided by faucet aerators guarantees low pressure without an internal thread, and descaling and disinfection procedures are applied weekly, permitting to avoid bacterial growth on outlets and preventing biofilm development [32].

The data regarding the higher *P. aeruginosa* contamination in cold-water samples can be explained by the same consideration as for sensor-activated faucets in *Legionella* contamination due to the sharing of these bacteria in the same habitat.

These findings led to the following considerations:


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

The eleven hospitals of this study, numbered 1 to 11, were involved in a *Legionella* environmental surveillance program from 2013 to 2019. After the introduction of last version of the Italian Guidelines in 2015, the 11 hospitals developed a risk assessment plan for *Legionella* control, considering the locations of buildings, their types of patients, and the water distribution system characteristics. All hospitals were supplied by municipal water that, after softener treatment, was heated by a heat-exchanger along with a hot-water return line.

All hospitals performed a six-month plan of *Legionella* environmental monitoring and active surveillance to control nosocomial *Legionella* infection by urinary-antigen test. Therefore, a complete program of maintenance procedures by measuring and recording temperatures, flushing outlet points, continuous disinfection of the system by H2O2/Ag<sup>+</sup>, and a fortnightly plan regarding aerator disinfection and/or replacement was undertaken.

During environmental monitoring, we found a higher *Legionella* concentration in SHWOs with respect to other hospital outlets involved in monitoring, indicating the necessity of a supplementary investigation.

In Table 5, the number of SHWOs (*n* = 52) in each hospital is reported, all of them equipped with sensor-activated faucets with TMVs (Figure 3). The main distribution system supplied hot-water outlets in a temperature range between 40–50 ◦C, while the cold-water outlets showed a temperature range of 15–20 ◦C. In SHWOs, the presence of TMVs produced a continuous mixed water at a set temperature around 36 ◦C.


**Table 5.** Number of SHWOs/hospitals.

**Figure 3.** SHWOs with sensor-activated faucets and TMVs (**a**) and a sensor-activated faucet (**b**).

During the study, eight hospitals had not implemented any replacement in SHWOs; however, three hospitals (1, 8, and 11) implemented a substitution program for their surgical hand preparation points regarding the faucet apparatuses: sensor-activated faucets with TMVs were removed and substituted with elbow-operated manual faucets without TMVs (Figure 4).

**Figure 4.** SHWOs with elbow-operated manual faucets without TMVs.

The environmental surveillance program consisted of *Legionella* and *P. aeruginosa* monitoring, according to the risk assessment plans provided by hospital healthcare directives.

The hot-water circuit in all hospitals was treated by H2O2/Ag+ disinfectant, which was added by a pump proportionally to the volume of cold water supply at a concentration around 50 mg/<sup>L</sup> in order to allow a residue at outlets between 10–20 mg/L, following manufacturer's instructions.

To assess the complete monitoring of water microbiological quality supplied by SHWOs and to evaluate di fferences in terms of contamination between hot- and cold-water distribution systems, both circuits were sampled.

For the three hospitals that implemented a substitution program with manual clinical valves without TMVs, the data of cold-water samples were not available, as the risk assessment plan after replacement involved only hot SHWO samples; therefore, comparison in terms of *Legionella* contamination before and after the substitution program was considered only in hot-water circuits.

## *4.1. SHWO Sampling*

According to the Italian Guidelines for the Prevention and Control of Legionellosis [12], analysis of *Legionella* contamination was performed by collecting two liters of cold and hot SHWO samples. In particular, in order to determine the quality in the main distribution system, post-flushing sampling was applied, which consisted of removing the filter or faucets, disinfection of taps with ethanol (70%), open taps, flushing for 2 min, and collection of cold before hot samples [38]. For cold samples, the TMVs were deactivated; by contrast, TMVs were reactivated to collect hot samples at the setting temperature for SHWOs (36 ◦C).

From the 52 SHWOs, 669 samples were collected (427 hot and 242 cold), where two liters of water were sampled using 1-liter sterile polytetrafluoroethylene (PTFE) bottles containing sodium thiosulphate (20 mg/L) [38,39].

The samples were processed by a membrane-filtration technique using polyethersulfone membrane filters with a porosity of 0.22 μm (Sartorius, Bedford, MA, USA), according to the International Standard Organization (ISO) 11731:2017 procedure [40].

#### *4.2. Legionella and P. aeruginosa Culture and Typing*

The *Legionella* culture was performed on Glycine-Polymyxin B-Vancomycin-Cycloheximide (GVPC) plates (Thermo Fisher Diagnostic, Basingstoke, UK) and subsequently incubated at 36 ± 1 ◦C with 2.5% CO2. *Legionella* growth was evaluated every 2 days for a total of 15 days of culture.

After the incubation period, the colonies with morphologies associated to the *Legionella* genus were enumerated and five suspected colonies for each morphology, as indicated by ISO 11731:2017, were subcultured on Bu ffered Charcoal Yeast Extract (BCYE) agar with l-cysteine (cys+) and without l-cysteine (cys−) as supplement, which is a selective media used for *Legionella* isolation. The positive *Legionella* colonies were those that grew on *Legionella* BCYE cys+ agar but failed to grow on *Legionella* BCYE cys− agar.

The isolates grown on BCYE cys+ were serologically typed by an agglutination test (*Legionella* latex test kit, Thermo Fisher Diagnostic, Basingstoke, UK). The isolates identified as *L. pneumophila* were then processed for serogroup identification by polyclonal latex reagents (Biolife, Milan, Italy).

Colonies identified by the agglutination test as belonging to *Legionella* non-*pneumophila* species were subsequently analyzed by *mip* gene sequencing and by Polymerase Chain Reaction (PCR) using degenerate primers (as described by Ratcli ff et al. [41]) and modified by M13 tailing to avoid noise in the DNA sequence [42]. Gene amplification was carried out in a 50-μL reaction containing DreamTaq Green PCR Master Mix 2× (Thermo Fisher Diagnostic) and 40 picomoles of each primer; 100 nanograms of DNA extracted from the presumptive colonies of *Legionella* was added as a template. The same amounts of DNA from *L. pneumophila* type strain EUL00137, provided by the European Working Group for *Legionella* Infections (EWGLI) [43], and fetal bovine serum were used as positive and negative controls, respectively.

Following purification, DNA was sequenced using BigDye Chemistry and analyzed on an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Specifically, *mip* amplicons (661–715 base pairs) were sequenced using M13 forward and reverse primers (mip-595R-M13R caggaaacagctatgaccCATATGCAAGACCTGAGGGAAC; mip-74F-M13F tgtaaaacgacggccagtGCTGCAACCGATGCCAC) to obtain complete coverage of the sequenced region of interest. Raw sequencing data were assembled using the CLC Main Workbench 7.6.4 software (QIAGEN, Redwood City, CA, USA). The sequences were compared with sequences deposited in the *Legionella mip* gene sequence database using a similarity analysis tool. Identification on the species level was done based on ≥98% similarity to a sequence in the database [44].

The results regarding *Legionella* contamination in the samples were expressed as colony formant unit (cfu) per liter (cfu/L). According to ISO 11731:2017, a negative result (absence of bacteria growth) was expressed as the lower limit of detection, that is, <50 cfu/L [40].

The same samples (*n* = 669) were analyzed to quantify the presence of *P. aeruginosa* due to its role in biofilm formation and to its capacity to inhibit *Legionella* growth during isolation culture, producing inaccurate results [45]. The analysis was performed on a volume of 100 mL of hot and cold samples, filtered using a cellulose nitrate membrane filter with a 0.45-μm pore size (Sartorius, Bedford, MA, USA), according to UNI EN ISO 16266:2008 [46,47].

The membrane was seeded on *Pseudomonas*-selective agar plate (PSA, Biolife, Milan, Italy) and incubated for 48 h in 36 ◦C incubators. Colonies that showed green-blue fluorescence when placed under a Wood's lamp (ultraviolet light at 365 nm) were subcultured on Nutrient agar (NA, Biolife, Milan, Italy) for 18–24 h. Subsequently, the colonies were identified biochemically as *P. aeruginosa* by indole, oxidase reaction tests, and BBL Crystal Enteric/Non Fermenter ID Kit (Becton Dickinson Systems, Cockeysville, MD, USA), according to the manufacturer's instructions.

The results are expressed in terms of cfu/100 mL.

#### *4.3. Physical and Chemical Analyses*

The physical and chemical parameters—the temperature of the water samples as well as the disinfectant residues at SHWOs—were measured during the collection of samples.

The temperature (◦C) (T) was measured by a conductivity meter coupled with a thermistor probe (Temp 6 basic for probe Pt100 RTD from −50 to +199 ◦C; Thermo Fisher Scientific Inc., Eutech Instruments Pte Ltd., Singapore). An on-site commercial kit for the residual hydrogen peroxide component of H2O2/Ag+ (mg/L) was used. The kit uses a colorimetric test based on peroxidase activity to transfer peroxide oxygen to an organic redox indicator, which produces a blue oxidation product.

The hydrogen peroxide concentration was measured semiquantitatively by visual comparison of the result seen on the reaction zone of the test strip with the fields on a color scale in a range of 0.5–25 mg/<sup>L</sup> H2O2.

## *4.4. Statistical Analysis*

The *Legionella* concentration data were converted into Log10 cfu/L (Log cfu/L) to normalize the non normal distributions. According to the Italian Guidelines for *Legionella,* the detection limit corresponding to 50 cfu/L (1.7 Log cfu/L) was used; by contrast, the risk value, >100 cfu/L, was expressed as >2 Log cfu/L.

To compare data of hot- and cold-water *Legionella* concentrations, the Mann–Whitney test was used (Table 1).

The distribution of different *Legionella* isolates between hot and cold positive samples was studied by chi-squared test (χ2). Therefore, the differences in *Legionella* isolate concentrations in hot- or cold-water samples were studied by the Kruskal–Wallis test: the significant data found were then also analyzed by the Mann–Whitney test (Table 2).

Multiple comparisons between *Legionella* concentrations and the four ranges of temperature measured were performed by using the ANOVA test (Table 3).

Regarding the three hospitals that implemented the replacement program (e.g., with elbow-operated clinical valves without TMVs), the data analysis to compare *Legionella* levels before and after replacement was performed by parametric *t*-Student test when considering a number of values *n* > 30 and by nonparametric Wilcoxon test for *n* < 30 (Table 4).

The *P. aeruginosa* results were converted into Log cfu/100mL. The contamination found was studied by Mann–Whitney test to compare hot- and cold-water samples.

Statistical analyses were performed using the SPSS software for Windows version 23 (IBM SPSS, Inc., Chicago, IL, USA).

The data were considered significant for *p* values (*p*) ≤ 0.05.
