Application of Hydrogen Peroxide as an Innovative Method of Treatment for Legionella Control in a Hospital Water Network

Objectives: To evaluate the effectiveness of hydrogen peroxide (HP) use as a disinfectant in the hospital water network for the control of Legionella spp. colonization. Methods: Following the detection of high levels of Legionella contamination in a 136-bed general hospital water network, an HP treatment of the hot water supply (25 mg/L) was adopted. During a period of 34 months, the effectiveness of HP on Legionella colonization was assessed. Legionella was isolated in accordance with ISO-11731 and identification was carried out by sequencing of the mip gene. Results: Before HP treatment, L. pneumophila sg 2–15 was isolated in all sites with a mean count of 9950 ± 8279 cfu/L. After one-month of HP treatment, we observed the disappearance of L. pneumophila 2–15, however other Legionella species previously not seen were found; Legionella pneumophila 1 was isolated in one out of four sampling sites (2000 cfu/L) and other non-pneumophila species were present in all sites (mean load 3000 ± 2887 cfu/L). Starting from September 2013, HP treatment was modified by adding food-grade polyphosphates, and in the following months, we observed a progressive reduction of the mean load of all species (p < 0.05), resulting in substantial disappearance of Legionella colonization. Conclusion: Hydrogen peroxide demonstrated good efficacy in controlling Legionella. Although in the initial phases of treatment it appeared unable to eliminate all Legionella species, by maintaining HP levels at 25 mg/L and adding food-grade polyphosphates, a progressive and complete control of colonization was obtained.


Introduction
Legionella spp. is a Gram-negative aerobic opportunistic bacterium, responsible for a severe form of pneumonia called Legionnaires' disease (LD) [1]. It is a waterborne human pathogen frequently associated with nosocomial infections, particularly among immunosuppressed patients such as those with acquired immune-deficiency syndrome (AIDS), transplant patients, and those undergoing aggressive chemotherapy [2].
Literature data report several references about the properties, the germicidal effectiveness, and the potential uses for HP in healthcare water networks. Published reports ascribe HP as a good reagent for bactericidal, virucidal, and fungicidal activity.
Continuous treatment of water distribution systems using HP, generally in combination with silver salts, is now allowed in many countries. In France, the Ministère de la Santé Publique considers this formulation as part of the guidelines for the treatment of Legionella pneumophila [27].
HP is a strong oxidizing agent that oxidizes microorganisms' enzymatic systems, releasing free oxygen atoms (nascent oxygen). HP production occurs naturally in the human body as a defense against antigens.
HP as a water disinfectant offers various advantages. It is a strong oxidizer, bactericidal at 3% solution (D value E. coli: 0.57 min), a sterilant at 6% with six hours of exposure, more powerful than chlorine dioxide, and more stable at high temperatures and pH compared to chlorine-based disinfectants.
Furthermore, HP is non-toxic to humans and the environment, tasteless, and is not mutagenic or carcinogenic.
HP has the ability to form powerful oxidants such as hydroxyl radical and singlet oxygen. They appear to present no danger by toxicity or other hazards, when diluted in water to their effective concentration as a disinfectant. HP decomposes rapidly in different environmental compartments, due to biotic degradation (microbial catalase and peroxidase enzymes), and abiotic degradation (transition metal, heavy metal, oxidation or reduction reactions, with organic compounds or inorganic substances). The lack of toxicity of HP to people and animals and its lack of environmental impact has been confirmed by the U.S. Food and Drug Administration (FDA) and the U.S. Environmental Protection Agency [28]. European directives do not establish a concentration limit for HP in drinking water, although the German and British version of the EN 902: 2016: "Chemicals used for treatment of water intended for human consumption. Hydrogen peroxide" provides a dosage up to 17 mg/L. HP still presents numerous advantages from both an economic and operative point of view; operating costs are much lower than traditional water disinfection systems, as are investment costs and expenditure for equipment. HP solution is readily available and can be stored long-term (maximum loss of concentration 3% per year). Moreover, it has a low corrosive effect, while pipeline corrosion is a frequent and no-negligible problem with chlorine-based disinfection systems. To prevent corrosion polyphosphate compounds may be added during disinfection, although they can remove metals. Removal of metals would lead to Legionella deficient in superoxide dismutase activity and thereby increased susceptibility to toxic oxygen metabolites, including hydrogen peroxide. Superoxide dismutases (SODs) are metalloenzymes that catalyze the decomposition of superoxide radical. Although SODs of the manganese (MnSOD) or iron (FeSOD) class are widespread among bacteria, SOD containing both copper and zinc (CuZnSOD) has been demonstrated in a small number of gram-negative bacterial species, such as Legionella pneumophila. CuZnSOD plays a role in the survival of L. pneumophila during the stationary phase of growth [29].
HP-based disinfection has recently been applied, with satisfactory results in the reduction of microbial contamination in dental unit water; recent studies have demonstrated the efficacy of HP treatment versus several microorganisms, including Legionella [30][31][32].

Results
Before the start of the HP treatment, high Legionella concentrations were detected in all the examined water points. Concentrations ranged from 3000 to 20,800 cfu/L, with a mean value of 9950 ± 8279 cfu/L in flushed water samples, indicating significant colonization of the entire hospital building water network (sampled 3 July 2013). The bacterial species was identified as Legionella pneumophila 2-15.
This potentially critical situation led to the application of a HP-based disinfection strategy on 19 July 2013, commencing with a "shock treatment" of the water network using a HP solution of 100 mg/L and silver ions at 100 µg/L for 12 h, followed by a continuous treatment with 10 mg/L of HP and 10 µg/L of silver ions. On 29 July 2013, an initial sampling was performed after the beginning of the treatment. Culture analysis demonstrated the absence of Legionella pneumophila 2-15 in all four sampling sites. However, Legionella species not previously cultured before was present: Legionella pneumophila 1 was isolated in one out of four sampling sites (2000 cfu/L) and other non-pneumophila species were present in all the sites (mean load 3000 ± 2887 cfu/L).
Samples tested by PCR showed results positive for the presence of Legionella, and where non-pneumophila Legionella species were identified, the sequence of the mip gene showed the presence of Legionella longbeachae serogroup 1 (strain NSW150).
These results were considered unsatisfactory, and led to a change in disinfectant product; from September 2013, HP and silver ions were replaced by a formulation of HP only, at higher concentration (25 mg/L), and with the addition of polyphosphates as a film-forming product.
This disinfection strategy showed tangible results after six months of treatment.
The efficacy of the new disinfection was observed from January 2014; a progressive reduction of non-pneumophila Legionella species loads to less than 500 cfu/L occurred, and the complete disappearance of L. pneumophila 1 was observed.
In July 2014, a reduction in HP concentration from 25 to 10 mg/L, probably due to a failure in the HP dispensing device, caused an increase in Legionella concentrations and the reappearing of over 3000 cfu/L L. pneumophila 2-15. However, the mean bacteria load remained lower compared to the initial situation (800 ± 1524 cfu/L for L. pneumophila 2-15 and 220 ± 178 cfu/L for non-pneumophila Legionella).
With the HP concentration maintained at 25 mg/L, at the subsequent sampling test  (Figures 1 and 2). The higher levels of contamination in this point may be due to its location, as the most distal area in the water network from the HP dispensing device, and due to its infrequent use and consequent water stagnation in the terminal portion.
Statistical analysis demonstrated that the abatement in Legionella loads after HP treatment was significant. Application of the Nemenyi test gave significant p-values from a comparison of the latest test to sampling on November 2014 (p-value < 0.05) ( Table 1).
The application of the new disinfectant formulation with 25 mg/L HP and the addition of filming product demonstrated a positive effect on corrosion reduction, suggesting that decreased intervention for water system maintenance is possible with correct treatments. The detection of dissolved iron concentrations was used as a proxy for pipelines corrosion; this was influenced by seasonal variability, due to different sources of water supplied from the municipal aqueduct (an additional water supply is required in summer). In autumn, sampling displayed the highest iron ion values compared to other periods of the year, despite a trend in reduction of iron ion values during the monitoring period, suggesting a possible decrease of corrosion levels. The same trend was observed with turbidity ( Table 2). Further physico-chemical parameters (temperature, pH, conductivity) and detected HP concentration as assessed at point of use, are shown in Table 3.

Discussion
This is one of the few studies of its kind performed in a hospital setting, although previous experiences reported in the literature used HP either as a formulation with silver salts, or with acetic and peracetic acid, in different concentrations.
In an Israeli study, performed on a 50-bed ward of a great hospital, a formulation of HP and silver salts applied for 24 months, demonstrated a significant reduction in Legionella contamination, from mean count values of 200-14,000 cfu/L to total absence [33].
Another two experimentations, performed in Italian long-term care facilities, did not yield a total abatement of the contamination, although an approximate 2-log reduction in Legionella mean count was attained [34,35].

Discussion
This is one of the few studies of its kind performed in a hospital setting, although previous experiences reported in the literature used HP either as a formulation with silver salts, or with acetic and peracetic acid, in different concentrations.
In an Israeli study, performed on a 50-bed ward of a great hospital, a formulation of HP and silver salts applied for 24 months, demonstrated a significant reduction in Legionella contamination, from mean count values of 200-14,000 cfu/L to total absence [33].
Another two experimentations, performed in Italian long-term care facilities, did not yield a total abatement of the contamination, although an approximate 2-log reduction in Legionella mean count was attained [34,35].

Discussion
This is one of the few studies of its kind performed in a hospital setting, although previous experiences reported in the literature used HP either as a formulation with silver salts, or with acetic and peracetic acid, in different concentrations.
In an Israeli study, performed on a 50-bed ward of a great hospital, a formulation of HP and silver salts applied for 24 months, demonstrated a significant reduction in Legionella contamination, from mean count values of 200-14,000 cfu/L to total absence [33].
Another two experimentations, performed in Italian long-term care facilities, did not yield a total abatement of the contamination, although an approximate 2-log reduction in Legionella mean count was attained [34,35].
Remarkable results were obtained in a recent study by Modena-Reggio Emilia University (Italy), from using a treatment based on the association of HP with acetic and peracetic acid, performed on a highly-contaminated building. Initially, Legionella counts were rapidly reduced, remaining stable at low levels in the first seven months, until complete removal of bacteria in more than 90% of samples during the subsequent eighth month. During the treatment, an interesting observation was the progressive substitution of L. pneumophila (serogroups 1, 6,9) with environmental species of Legionella, such as L. jamestowniensis [36].
The results of this 36-month field study performed at the Villafranca General Hospital demonstrated good efficacy of HP treatment in controlling Legionella. In particular, the dosage of HP 25 mg/L with the addition of polyphosphates emerged as a valid alternative to the more frequent use of silver salts.
Overall, our data suggest that the effectiveness of HP is evident in the long-term. In the early stages of treatment, some Legionella species other than L. pneumophila appear to resist treatment by disinfectant and subsequently replace L. pneumophila; these species are less pathogenic and are almost never associated with human disease. Legionella longbeachae was identified as responsible for 43 cases between 2005 and 2012 by the European Surveillance System (Tessy) and for a cluster of six cases from an amateur gardener in 2013 in Scotland [37], although infection by this species is unusual. These findings demonstrate the importance of accurate environmental surveillance in order to monitor the presence of all Legionella species, including non-pneumophila ones.
This characteristic trend, with a first stage of L. pneumophila reduction followed by its substitution by less pathogenic Legionella, and long-term efficacy in the reduction of all Legionella species, was highlighted by Marchesi et al. [36], in the use of HP disinfection alongside peracetic acid.
As with other water disinfection methods, it was demonstrated that maintaining an appropriate and uninterrupted concentration of HP was very important, as new increases of Legionella loads were observed during the study period due to a lower doses of disinfectant. Through our data, we can assert that 25 mg/L HP levels ensure good control of Legionella colonization and its almost complete disappearance in the long-term, qualifying this method as a reliable new option for water treatment.
This study is the first to evaluate the application of HP as the only disinfectant in Legionella colonization control, and demonstrated good performance using HP-based disinfection. This product could represent a valid alternative to chlorine-based disinfectants, with at least comparable results in efficacy, and advantages in cost reduction, with direct costs for installation of the dosing device and disinfectant, and indirect costs associated with the reduction of pipeline corrosion and maintenance fees.

Setting
The Villamarina General Hospital of Piombino (Leghorn, Italy), is a 136-beds hospital (120 ordinary and 16 day hospital) with a catchment area of about 60,000 inhabitants. Built in the early 1990s as an extension of a previous structure, it has been active since 1992.
The architectural structure of the hospital is a monoblock with a central plate with three levels (−2, −1, ground floor) and three vertical towers, one containing five floors and the other two containing four floors.
In July 2013, within the water safety plan (WSP) implementation program, hot water system disinfection, and a systematic monitoring program with sampling of final points of use within the water network was commenced.

Water Disinfection
Considering the water network characteristics, with old and galvanized steel-made pipelines, the need to minimize the risk of corrosion and related maintenance costs led to the discarding of chlorine dioxide disinfection methods and the uptake of HP as a disinfection product.
The first performed action was a "shock" disinfection with a solution of HP and silver ions for 12 h. Subsequently, a continuous disinfection system, based on a formulation of 10 mg/L HP and 10 µg/L silver ions, at points of use, was applied.
In September 2013, the HP-silver ions disinfection was replaced by the use of HP alone with 25 mg/L at points of use. Furthermore, polyphosphates were added to disinfectant as a film-forming product to reduce pipeline corrosion. This formulation was employed from September 2013 to June 2016, with the only exception of July 2014, when HP concentration went down to 10 mg/L, probably due to a dosing device malfunction.

Sample Collection and Detection of Legionella spp.
Between July 2013 and June 2016, 59 hot water samples were collected. Four different target points, one for each floor, were selected, considering their degree of representation of the hospital water network: an emergency room bathroom (basement), an intensive care room (ground floor), an Oncology Day Hospital room (first floor) and the hand basin of the Obstetrics and Gynecology delivery room (fourth floor). Since July 2014, a fifth sampling point was added at the circulation circuit of the hot water network.
Water samples were assayed for Legionella and tested for the following physico-chemical parameters: temperature ( • C), pH (pH units), conductivity (µS/cm), turbidity (NTU), iron ions (µg/L Fe). Turbidity and iron ions were assayed only for two sampling points, at the intensive care room (ground floor), and at the Oncology Day Hospital room (first floor).
At the same time, the disinfectant concentration (hydrogen peroxide, or hydrogen peroxide and silver ions during the period of application) in water was regularly measured at the point of use to determine an accurate indication of HP levels being introduced into the water supply. The HP concentration was determined by colorimetric test Merckoquant ® test strips.
During the first sampling test, to detect basal levels of Legionella colonization, two water samples were collected from each sampling point, one immediately and one after 5 min of flushing. In the subsequent sampling tests, only flushed samples, that were more representative of overall water network contamination levels, were collected.
The isolation of Legionella spp. in hot water samples was performed in accordance with standard procedures [38,39]. One liter of water was filtered through a membrane with a porosity of 0.2 µm diameter (Millipore, Billerica, MA, USA). After filtration, the membrane was immersed in 10 mL of the same water and subjected to sonication for 5 min, allowing the detachment of cells from the membrane and their suspension in water. The suspension was subjected to a thermal inactivation treatment at 50 • C for 30 min with the aim to select for Legionella spp., while inactivating all microbial species not resistant to high temperature. Afterwards, 0.1 mL of the suspension was seeded on Legionella BMPA selective medium (Oxoid Ltd., Basingstoke, Hampshire, UK), and the plates were incubated at 37 • C for 7-10 days within jars under a modified atmosphere (2.5% CO 2 ) Finally, Legionella colonies grown on the medium were subjected to species and serogroup identification analysis using a multi-purpose latex agglutination test (Legionella Latex Test, Oxoid Ltd., Basingstoke, Hampshire, UK). Identification of Legionella species was carried out by sequencing of the mip gene (558 bp) [40].
After the electrophoretic run, (110 V for 30 min), mip gene amplification results were visualized in a UV transilluminator.
Amplified mip gene was sequenced by outsourcing (GATC, Biotech, Jakob-stadler-plaz 778467 Constance, Germany), and sequence alignment was performed by BioEdit Version 7.0.0. Sequences identification was obtained using a Basic Local Alignment Search Tool (BLAST) Database.

Statistical Analysis
Data are shown as means, standard deviations and medians. Normality of distribution was assessed using the Kolmogorov-Smirnov test, the variable was not Gaussian. The Nemenyi test for paired data was performed to compare the values of different times. Post-hoc power tests were conducted to estimate the sample size, 1-beta values of significant data were >0.8, assuring an appropriate sample size. The statistical analysis was carried out using the IBM SPSS software package, version 17.0.1 (IBM, Armonk, NY, USA).

Conflicts of Interest:
The authors declare no conflict of interest.