Controlling Legionella pneumophila growth in hot water systems by reducing dissolved oxygen levels

Abstract Legionella pneumophila, the leading cause of Legionnaires’ disease in the United States, is found in lakes, ponds, and streams but poses a health risk when it grows in building water systems. The growth of L. pneumophila in hot water systems of healthcare facilities poses a significant risk to patients, staff, and visitors. Hospitals and long-term care facilities account for 76% of reported Legionnaires’ disease cases with mortality rates of 25%. Controlling L. pneumophila growth in hot water systems serving healthcare and hospitality buildings is currently achieved primarily by adding oxidizing chemical disinfectants. Chemical oxidants generate disinfection byproducts and can accelerate corrosion of premise plumbing materials and equipment. Alternative control methods that do not generate hazardous disinfection byproducts or accelerate corrosion are needed. L. pneumophila is an obligate aerobe that cannot sustain cellular respiration, amplify, or remain culturable when dissolved oxygen (DO) concentrations are too low (< 0.3 mg/L). An alternative method of controlling L. pneumophila growth by reducing DO levels in a hot water model system using a gas transfer membrane contactor was evaluated. A hot water model system was constructed and inoculated with L. pneumophila at DO concentrations above 0.5 mg/L. Once the model system was colonized, DO levels were incrementally reduced. Water samples were collected each week to evaluate the effect of reducing dissolved oxygen levels when all other conditions favored Legionella amplification. At DO concentrations below 0.3 mg/L, L. pneumophila concentrations were reduced by 1-log over 7 days. Under conditions in the hot water model system, at favorable temperatures and with no residual chlorine disinfectant, L. pneumophila concentrations were reduced by 1-log, indicating growth inhibition by reducing DO levels as the sole control measure. In sections of the model system where DO levels were not lowered L. pneumophila continued to grow. Reducing dissolved oxygen levels in hot water systems of healthcare and other large buildings to control L. pneumophila could also lower the risk of supplemental chemical treatment methods currently in use.


Introduction
Legionella bacteria can be found in lakes, ponds, and streams but have been shown to pose a health risk when it is allowed to grow or amplify in building water systems (US EPA 2001;NASEM 2019;CDC 2021).Legionella pneumophila is the species of Legionella believed to cause >90% of cases of Legionnaires' disease in the United States (Fields et al. 2002;Yu et al. 2002).Two distinct forms of legionellosis have been described, an infectious pneumonia called Legionnaires' disease, and a milder flu-like illness without pneumonia called Pontiac Fever (NASEM 2019).The primary route of exposure leading to legionellosis is the inhalation of aerosolized water containing the bacteria which is not transmitted from person to person.Of greatest risk is exposure to L. pneumophila from contaminated hot water systems, particularly in hospitals and long-term care facilities (US EPA 2001;Fields et al. 2002;Percival and Williams 2014;NASEM 2019;CDC 2021).
A 650% rise in Legionnaires' disease cases in the United States has been reported over the last two decades, peaking at 9,933 confirmed cases in 2018 (CDC 2018).Healthcare facilities (hospitals and nursing homes) account for 76% of reported Legionnaires' disease cases (CDC 2017; Clopper et al. 2021).Mortality rates for healthcare-acquired Legionnaires' disease are estimated to be 25% while community-acquired cases are approximately 10% (CDC 2017).The growth of L. pneumophila within hot water systems in healthcare facilities has consistently been identified as a risk to patients, staff, and visitors.Contaminated hot water systems have also been found to cause disease outbreaks in hotels, resorts, apartment buildings, and condominium buildings (Clopper et al. 2021).
The risk of healthcare-acquired disease prompted the Centers for Medicare and Medicaid Services (CMS) to issue a June 2017 directive that mandated all hospitals and skilled nursing facilities immediately implement water management programs to control and prevent Legionnaires' disease and other waterborne pathogens (CMS 2017).The voluntary standard, ASHRAE 188-2015 that has been mandated for healthcare facilities remains optional for most other buildings and facilities.
Controlling Legionella growth within water heaters, the hot water distribution system and fixtures have historically relied upon increasing hot water temperatures and supplemental treatment with disinfectants such as chlorine, chlorine dioxide, or monochloramine.Increasing hot water temperatures pose a scalding risk if delivery temperatures exceed 54.5 � C (130 � F) (US EPA 2016a; ASHRAE 2020).Injecting the premise plumbing of building water systems with oxidizing disinfectants accelerates corrosion of plumbing materials and equipment in addition to generating disinfection byproducts (DBPs) that can pose chronic health risks (US EPA 2016a, 2016b).Adding supplemental chemical disinfectants to potable drinking water, including hot water systems, often triggers regulatory oversight under the Safe Drinking Water Act (40 CFR Part 141 Subpart L 141.130 of the US EPA SDWA).Even though supplemental disinfection of building water systems currently offers the most effective and reliable control method for Legionella, it creates hazardous disinfection byproducts and offers diminished efficacy when water usage rates are low.
Other treatment methods sometimes used to control bacteria, including L. pneumophila, in drinking water and hot water systems include ultraviolet (UV) light, ozone, and copper-silver ionization (CSI).Each of these treatment methods can inactivate or inhibit bacterial growth but have significant limitations.Ozone injection rapidly dissipates, leaving no residual disinfectant past the point of injection, and can promote corrosion of plumbing materials.UV treatments offer no residual disinfectant beyond the point of treatment, can become ineffective as they age and produce less energy, or when particles foul the lamps and block UV energy from reaching bacteria in the water.
Routine maintenance and cleaning of the UV reactors is critical to their effectiveness (US EPA 2016a).
CSI involves injecting copper and silver ions into a water system and maintaining them within a narrow band of concentrations.The US EPA review concluded that the majority of published studies on its effectiveness were case reports.While laboratory studies have indicated Legionella can be inactivated by copper and silver ions, the presence of biofilm and amebae can protect the bacteria.Several studies indicate that Legionella may develop resistance over time (US EPA 2016a).Use of CSI may result in corrosion especially when water quality is poor and plumbing materials are incompatible.Additionally, high levels of copper and silver can result from treatment if not effectively controlled.Ongoing sample collection and laboratory analysis of treated water using inductively coupled plasma mass spectrometry (ICP-MS) is necessary to monitor system efficacy and safety (US EPA 2016a).
An alternative to injecting chemical disinfectants, treating with UV light, or increasing water temperatures has been proposed.Reducing DO in building hot water systems could be used to inhibit L. pneumophila growth (Krause 2022) without generating hazardous byproducts.Dissolved oxygen is essential to the survival and growth of aerobic bacteria, including L. pneumophila, in natural and human-made aquatic environments (ISO 11731 2017;Feeley et al. 1978;Wadowsky et al. 1985;Fields et al. 2007;ASTM D5952-08 2008;Percival and Williams 2014).Legionella growth is inhibited when DO is too low to support metabolic respiration, making oxygen availability a critical growth factor for all Legionella species (Feeley et al. 1978;Fliermans et al. 1979Fliermans et al. , 1981;;Wadowsky et al. 1985;US EPA 1999).Experimental data on the feasibility and efficacy of controlling Legionella growth in a building water system have not been previously reported in the scientific literature and are needed to develop a practical system for treating building water.Viable Legionella has been found in natural reservoirs at DO concentrations ranging from 0.3 to 9.6 mg/L, suggesting that below 0.3 mg/L, Legionella growth is not supported (Fliermans et al. 1981).Experimental data to determine if this threshold is consistent within a pressurized premise plumbing or hot water system is needed to evaluate DO reduction as a feasible and reliable method to control L. pneumophila (Krause 2022).
The principle purposes of this study were: first, to determine if dissolved oxygen could be efficiently reduced in a circulating hot water system using a model hot water system; second, to determine if L. pneumophila concentrations were reduced when dissolved oxygen levels were lowered; and third, to identify a possible threshold level of dissolved oxygen that inhibits L. pneumophila growth in a model hot water system that is constructed with materials and equipment used in contemporary premise plumbing systems.A secondary objective of the study was to examine the potential for amplification of Nontuberculous Mycobacteria (NTM) when dissolved oxygen levels are reduced.

Experimental design and construction
Experimental results were obtained using a hot water model system.The model system was constructed using an in-line gas transfer membrane contactor to test the feasibility of reducing DO levels (Figure 1).The primary purpose of the hot water model system was to gather data on L. pneumophila growth when dissolved oxygen concentrations are reduced by measuring culturable concentrations under otherwise favorable conditions.Once it was determined that the model system was colonized and L. pneumophila was growing, indicated when L. pneumophila in samples exceeded make-up water levels by at least 1-log, concentrations were measured over 42 days (6 wk) to see how they changed when dissolved oxygen levels were incrementally lowered.The hot water model system was also used to gather data on a possible threshold level of DO that reliably inhibited L. pneumophila growth under otherwise favorable conditions.The hot water model system received makeup (feed) water that contained L. pneumophila concentrations at increasing levels throughout the experiment.This condition, while unusual for many building water systems, ensured that any diminution of L. pneumophila concentrations in the model system was most likely a result of changes in DO levels.Colonization of the makeup water lines that supplied water to the hot water model system was indicated by sample results that are presented later.

Hot water model construction
The hot water model system was constructed using an 189 liter (50 gallon) 240 volt, 4,500 Watt electric water heater (Rheem Model # XE50M06ST45U1) to heat the water.The water heater temperature was maintained between 41 � C and 44 � C (106 � F to 111 � F), a range that is ideal for biofilm development and L. pneumophila growth (ASHRAE 2020; AIHA 2022).Plumbing lines used to circulate water through the test system components consisted of copper, stainless steel, and chlorinated polyvinyl chloride (CPVC).Cold makeup water was supplied from the Pittsburgh Water and Sewer Authority via the building's potable water system at a pressure of 345 to 414 kPa (50 to 60 psi).A check valve was installed to prevent backflow of water into the building's water system.The municipal water supply was chlorinated and free chlorine measurements confirmed levels were low, ranging from 0.04 to 0.21 mg/L.A gas transfer membrane contactor was installed in line with a bypass line (3MTM Liqui-CelTM EXF-4 � 28 Series) to reduce dissolved oxygen levels in the circulating water after the system was determined to be colonized.The membrane contactor was operated under vacuum mode and with high purity nitrogen gas sweep (99.99% purity N2 at 0.24 L/sec (0.5 cubic feet per minute)) to efficiently remove DO.A high-flow 5 mm absolute polypropylene particulate filter, housed within a stainless-steel filter housing was used to protect the membrane contactor from debris and fouling (American Melt Blown Filters, 625 HF Series Pleated Cartridge).A variable speed pump (dp Pumps, 1 hp �60 Hz, part #290022362060GND) was used to circulate water throughout the test system at a rate of 15 L/min (4 gallons/min) and 303 kPa (44 psi).A 15.24 m (50 feet) long recirculating loop and a 3 m (10 feet) branch line feeding the distal outlet were constructed using CPVC pipe and flushed daily using an electronically operated solenoid valve and timer.The model system was operated under low flow conditions with long interim periods of stagnation during the colonization phase (Days 0-42).Once per day, five days each week, approximately 3.8 liters (1 gallon) of water was discharged from the model test system.On weekends, no water was discharged or flushed and the system was placed into circulation-only mode.A suspension of L. pneumophila was added to the system to facilitate colonization on Day 0. Water samples for Legionella analysis were collected on Days 23,29,36,43,50,57,64,71,and 78.Water samples for Nontuberculous Mycobacteria (NTM) analysis were collected on Days 23 and 78. Figure 1 depicts the hot water model system components and their relative order.

Inoculation
Inoculation of the hot water model system was achieved via two mechanisms.An in-line port was constructed to allow 500 mL of makeup supply water to be valved off and drained.Five hundred mL of water containing L. pneumophila serogroup 1 (ATCC 33152), at approximately 1,000 CFU/mL, was then added to the makeup water port.Once re-sealed, the valves were opened, and the water was flushed into the hot water tank.Based on the total system volume, the estimated beginning concentration of L. pneumophila was 2.5 CFU/mL.Water sample analysis also revealed the presence of L. pneumophila serogroup 5, believed to be present in the municipal water supply received from the Pittsburgh, PA water system.The hot water model system was operated for 43 days (6 wk) under conditions favorable for Legionella before sample results confirmed that colonization and amplification of L. pneumophila had occurred.

Legionella sampling and analysis
Planktonic (i.e., "free swimming") concentrations of L. pneumophila were collected from the model system to determine if amplification or inhibition was indicated.Water samples were collected into sterile bottles containing 0.5 mL of 0.1 N sodium thiosulfate from valved sampling ports installed during the initial construction of the test system.
Water samples (250 mL) collected from the hot water model system were immediately prepared for analysis at the Special Pathogens Laboratory (SPL) in Pittsburgh, PA.Note: The hot water model experiments were performed at the SPL Product Evaluation facility in Pittsburgh, PA.This facility is in the same building as the SPL microbiology laboratory, precluding the need to ship samples.Each sample was analyzed for Legionella using a modified ISO 11731:2017 method by plating onto buffered charcoal yeast extract (BCYE) agar with and without antibiotics.Culture plates were aerobically incubated at 36 ± 2 � C for seven days.Legionella isolates were serotyped using direct fluorescent antibody (DFA).Results were reported in CFU/mL.

Nontuberculous mycobacteria (NTM) sampling and analysis
The hot water model system was sampled for NTM, but not artificially inoculated as part of the experiment.Naturally occurring NTM in the municipal water supply was relied upon as the source of any NTM seeding the test system.Each sample was analyzed for NTM at the Special Pathogens Laboratory by plating 100 mL of the sample onto rapid growth media (RGM) agar.Another 2 mL of sample was treated with a 1:1 ratio of 0.2 M KCl-HCl for 3 min and 100 mL was plated onto Middlebrook 7H10 and Mitchison 7H11 agar.Culture plates were aerobically incubated at 30 þ/-0.5 � C for six weeks.Samples were speciated and quantified using matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS).Results are reported in CFU/mL.

Measuring water parameters in the hot water model system
Water parameters were measured several days per week and each time that samples were collected, immediately after the water sample for Legionella analysis was collected.Water parameters were also measured at interim periods during each week.Free chlorine was measured using the DPD Method (Hach SL 1000, Loveland, CO, USA).Using Chemkey reagents for free chlorine, the method measures in the range of 0.04 to 4.0 mg/L with a precision of 2.05 þ/-0.03mg/L (95% confidence interval).Dissolved oxygen and temperature were measured using a Rosemount dissolved oxygen probe (Model #499ATrDO) (Emerson/Rosemount Chanhassen, MN, USA).The Rosemount dissolved oxygen probe detects and measures dissolved oxygen using amperometric sensing at concentrations between 0.1 and 20.0 mg/L (ppm) with a reported accuracy of ±5% of reading.A handheld thermometer (Thomas) was used to measure water temperature with a reported range of −30 to 70 � C (−22 to 158 � F) with a reported accuracy of ±0.3 � C (±0.54 � F), and a resolution of 0.01 � C (0.18 � F).Sidestream lines with valves were used to direct water from different sampling locations to the DO probe so all measurements were taken using the same measurement probe, avoiding variability between measuring devices.

Calculating the D-value and interpreting the results
The rate at which bacteria sterilization methods kill or inactivate pathogens is characterized by calculating the decimal reduction time or D-value (Wadowsky 1985;Mazzola et al. 2003).The D-value is the time required to reduce viable counts of a bacterial suspension by 90% (1-log) at a given temperature and is calculated using the equation: where: The temperature is provided as a subscript to the reported value (e.g., D ¼ 1 100C ) (Wadowsky 1985;Korczynski et al. 1981).Because concentrations of L. pneumophila in the makeup water supplied to the model system varied over time, ranging from 40 to 670 CFU/mL, D-values were calculated using the sample result from the circulating hot water samples compared with the L. pneumophila concentration in the makeup cold water measured the prior week.The geometric mean (GM) of L. pneumophila concentration from two hot water sample locations within the test system was used to calculate the D-Value and reported at 42 � C (108 � F) for each sampling period.A decimal reduction in L. pneumophila concentration was deemed to indicate inhibition.Likewise, amplification was indicated when a decimal increase in L. pneumophila concentration was observed.

Results
Sample data from the hot water model system demonstrated that L. pneumophila growth was inhibited when dissolved oxygen levels were reduced using a gas transfer membrane contactor and other control measures were absent.Circulating hot water within the model system was maintained at a temperature of 42 � C (108 � F) ± 1 � C and non-detectable free chlorine levels (< 0.01 mg/L) throughout the experiment.With the membrane contactor operating at low efficiency to maintain DO levels above anoxic conditions, the test system was determined to be colonized and amplifying L. pneumophila 43 days after inoculation, as indicated by a geometric mean concentration of 24 CFU/ mL, with a makeup water concentration of 0.5 CFU/ mL.This result demonstrated that bacteria were not being filtered or otherwise physically removed from the model system by the membrane contactor.After inoculation of the test system, during the colonization period (Weeks C1 to C3) and the first scoping period (Week S1) circulating hot water DO concentrations ranged from 4.75 to 0.09 mg/L with a geometric mean of 0.69 mg/L.Operation over the next seven days (Week S2) at DO concentrations with a geometric mean of 0.41 mg/L, L. pneumophila growth continued as indicated by a geometric mean hot water concentration of 134 CFU/mL and 1.0 CFU/mL in the makeup water.Further operation over the next seven days (Week S3) at DO concentrations of 0.82 mg/L L. pneumophila growth continued, as indicated by a geometric mean concentration of 615 CFU/mL in the hot water and 40 CFU/mL in the makeup water.These data demonstrated the system's design and operation were conducive to L. pneumophila colonization, absent any effective control measures, at DO concentrations with a geometric mean of 0.40 mg/L or above.
When the model system was operated at DO concentrations with a geometric mean of less than 0.3 mg/L (0.29, 0.27, and 0.25 mg/L), L. pneumophila inhibition was observed.Inhibition was indicated when the geometric mean of L. pneumophila concentrations was at least 1-log less than the concentration found in the makeup water.During a 7-day period (Week T1) when the system was operated at DO concentrations with a geometric mean of 0.29 mg/L, the L. pneumophila concentration was 14 CFU/mL (GM) with a makeup water concentration of 250 CFU/mL, resulting in a calculated D-Value of 5.6 42C days.Operating the system for another 7-day period (Week T2) at DO concentrations with a geometric mean of 0.25 mg/L, the L. pneumophila concentration was 1 CFU/mL (GM) with a makeup water concentration of 20 CFU/mL resulting in a calculated D-Value of 5.4 42C days.Operating the system for a final sevenday period (Week T3) at DO concentrations with a geometric mean of 0.27 mg/L, the L. pneumophila concentration was 32 CFU/mL (GM) even when the makeup water concentration dramatically rose to 670 CFU/mL resulting in a calculated D-Value of 5.3 42C days.The arithmetic mean of the three calculated D-Values was 5.4 42C days and a standard deviation of 0.15 Days.Data for all test periods are summarized in Table 1 and depicted in Figure 2, with complete data reported in supplemental materials.D-Values were calculated for each period when a reduction in L. pneumophila concentrations greater than or equal to 1-log was observed.
Water samples were analyzed for NTM as part of a limited assessment to see if operating the hot water model system at low dissolved oxygen levels promoted the colonization or growth of these pathogens.Samples were collected at the beginning and end of the study, on Day 23 at the end of Week C1, and 55 days later at the end of Week T3.While the concentration of NTMs in the makeup water supplied to the test system increased by 2-log over the 55 days, from 5 CFU/mL to 324 CFU/mL, the initial concentration of NTMs in the hot water system dropped from a geometric mean of 66 CFU/mL to non-detectable levels (< 1 CFU/mL) where DO levels were maintained below 0.30 mg/L for the final 21 days.At the incoming supply water and distal fixture sampling point, where DO levels were not lowered, NTM concentrations increased.These data indicate that before DO levels were reduced, NTM had colonized the hot water system.Makeup water contained 5 CFU/mL and the hot water samples had a geometric mean concentration of 66 CFU/mL when DO levels were 1.79 mg/L, equating to a 1-log increase.After DO levels were reduced below 0.30 mg/L for the final 21 days the makeup water contained 324 CFU/mL and the hot water samples had no detectable NTM.When the non-detection results are assigned a value of < 1.0 CFU/mL, this equates to at least a 2-log reduction.While the results of this limited-scope study must be interpreted cautiously, they do not suggest that maintaining a DO concentration below 0.3 mg/L promoted NTM amplification.

Discussion
L. pneumophila colonization of hot water systems in hospitals, long-term care facilities, hotels, office buildings, and other multi-story buildings has been identified as the cause of many Legionnaires' disease outbreaks over the past 40 years.These systems tend to favor the colonization and growth of L. pneumophila, due to the temperatures present, the accumulation of sediment, nutrients, and biofilm, and the accelerated degradation of chlorine-based disinfectants.Adding oxidizing supplemental disinfectants to premise plumbing systems in buildings poses many challenges, unavoidably generates DBPs, and often promotes the corrosion rate of plumbing materials and equipment.Identifying an alternative control method for L. pneumophila and other Legionella species that avoids the corrosive side effects of chlorinebased oxidants, and does not produce potentially carcinogenic disinfection byproducts, could give building water system operators another tool in their efforts to maintain hot water systems in a hygienic condition and control waterborne pathogens.
Early published studies reported that low DO levels in natural water reservoirs and laboratory media prevented L. pneumophila from growing (Fliermans et al. 1979(Fliermans et al. , 1981;;Wadowsky et al. 1985).Because Legionella are obligate aerobes, they require a sufficient amount of dissolved oxygen for survival and growth (Wadowsky et al. 1985;US EPA 1999, 2001;Fields et al. 2002;Percival and Williams 2014), however, a threshold of DO at which anaerobiosis inhibits metabolism or growth in a hot water system has not been previously reported in the literature.L. pneumophila has been described as a micro-aerophile, indicating that it may thrive in habitats with lower DO concentrations (Mauchline et al. 1992;Brazeau and Edwards 2013).As with temperature and pH, L. pneumophila may favor a range of DO concentrations for its growth (Wadowsky et al. 1985).Identifying the lower threshold of DO concentration, below which growth is inhibited or cellular respiration ceases, is needed to design a system that uses anaerobiosis as a control measure.
The experiments performed as part of this research attempted to answer three questions.(1) Could DO levels be effectively reduced in a hot water system, using an in-line gas transfer membrane contactor?(2) In a water heating system colonized with L. pneumophila, would concentrations be reduced when DO levels were lowered?(3) Is there a threshold level of DO below which L. pneumophila growth is inhibited as indicated by a 1-log reduction in culturable concentrations?
Data produced from the hot water model system demonstrated that L. pneumophila growth was established 43 days after a one-time inoculation (see Table 1).However, the incoming makeup water was found to be a persistent source of L. pneumophila throughout the experiment, indicating that areas of the makeup water supply lines had likely become colonized.L. pneumophila growth within the model systems persisted for another 14 days until dissolved oxygen levels were reduced to less than 0.30 mg/L (geometric mean).Maintaining the circulating hot water at DO concentrations ranging from 0.25 to 0.29 mg/L continued to inhibit L. pneumophila growth for another twenty-one days.These data suggest that a threshold level of DO in hot water exists below which L. pneumophila growth is inhibited, most likely due to suppressed cellular respiration.Areas within the model system that had become colonized with L. pneumophila but were not maintained at DO concentrations below 0.30 mg/L, continued to support its growth as indicated by higher culturable concentrations measured in samples from the fixture discharge.
The threshold level of DO that inhibited L. pneumophila growth in this study was 0.30 mg/L, which coincides with the lower range of DO reported by Fliermans et al. in a 1981 published study. Fliermans et al. reported that samples from natural water sources that were positive for L. pneumophila (47 of 793) came from sources with DO levels ranging from 0.3 to 9.6 mg/L.Ninety percent of positive samples had DO levels above 1 mg/L, with only 10% of L. pneumophila-positive samples having DO levels between 0.3 and 1 mg/L (Fliermans et al. 1981).This concurrence of findings supports the likelihood of a biological threshold of DO below which L. pneumophila cannot sustain growth or remain culturable in natural or man-made water sources.
Based upon calculated D-values, the efficacy and rate of reducing culturable L. pneumophila concentrations was much less than what is reported for oxidant-based disinfectant methods such as free chlorine or ozone (Muraca et al. 1987;US EPA 2016a).Traditional disinfectant and control methods applied at concentrations above those allowed in drinking water (e.g., hyperchlorination) would be expected to act much more rapidly but would interrupt drinking water use because these concentrations are hazardous to consume.After substantial and effective flushing of treated premise plumbing systems, water can be consumed and used with little risk, however, corrosive processes can sometimes be initiated and impact system materials following prolonged or repeated treatments.Based upon the mechanisms of action and timeframe of efficacy, traditional chlorine and oxidant methods remain appropriate for acute treatment of building water systems suspected of being a source of Legionnaires' disease.
However, regrowth of Legionella and reestablishment of protective biofilm is an ongoing concern after any short-term treatment is performed.Long-term control of conditions and the subsequent growth of Legionella often requires ongoing injection of supplemental disinfectants and other modifications to building water systems.Supplemental treatment of hot water systems with free chlorine or other oxidants tends to have limited efficacy and carries with it the same drawbacks, namely production of disinfection byproducts, corrosion of plumbing and equipment, and the need for regular flushing of treated water from all fixtures.As an alternative control measure, reducing DO within a circulating hot water system would avoid the drawbacks of traditional methods, with the benefit of not having to regularly flush water from distal branch lines, as there are no sources of oxygen production within the drinking water premise plumbing lines that are constructed of copper or other gas impermeable material.
D-values calculated from the hot water test model experimental data indicate a decimal (1-log) reduction of L. pneumophila concentrations in 5.3 to 5.6 days at DO levels below 0.3 mg/L, ranging from 0.25 to 0.29 mg/L, at 42 � C.These data reflect conditions defined as "anoxic," with DO concentrations maintained below 0.3 mg/L.Under the operating conditions described for the hot water model system, a control threshold of 0.3 mg/L DO is supported with an average D-value of 5.4 days at 42 � C. To determine if the strength of this inhibitory effect is different at lower DO levels additional experimental data would be necessary.
Makeup water supplied to the hot water model system contained elevated levels of L. pneumophila (ranging from 0.5 to 670 CFU/mL), yet concentrations of the bacteria were consistently more than 1-log lower in system areas where DO levels were maintained below 0.3 mg/L, indicating a robust inhibitory effect.Levels of L. pneumophila in the makeup water supplied to the hot water model system were much higher than those normally seen in municipal water distribution systems.It has been well documented that Legionella species, including L. pneumophila, can be found at very low levels in all parts of municipal water distribution systems.Municipal water distribution systems inoculate building water systems and cooling towers at low levels where conditions are favorable to its amplification (Voss et al. 1986;Campo and Apraiz 1988;Colbourne et al. 1988;Colbourne and Dennis 1989;US EPA 1999).This experimental design of the model system attempted to replicate the persistent low-level introduction of L. pneumophila to a hot water system, however, the test organism managed to colonize the makeup water supply lines serving the model system and began introducing the bacteria at high concentrations.Because efforts to disinfect the makeup water supply lines would likely have impacted the model system, it was decided to leave the contaminant in place and observe the impacts it had on the downstream concentrations where DO was being reduced.Ultimately the results demonstrated that even with unusually high levels of L. pneumophila in the makeup water supply, inhibition of growth within the hot water distribution system could be achieved when DO levels were reduced below 0.3 mg/L.
Data from the model system demonstrated that over one week, L. pneumophila concentrations were reduced by 1-log or more, and inhibition was observed even when makeup water was a significant and persistent source.As a long-term control measure to inhibit L. pneumophila growth in hot water systems, reducing dissolved oxygen holds much promise.Reducing DO levels using a gas transfer membrane contactor does not add potentially hazardous chemicals, does not require raising temperatures to levels that pose a scalding risk, inhibits rather than promotes corrosion, and does not trigger regulatory compliance under the US EPA Safe Drinking Water Act.Reducing DO levels also does not conflict with water conservation efforts and mandates.Current control technologies that rely upon adding supplemental chemicals inherently require that low-use fixtures and branch lines be periodically flushed to move freshly treated water into the lines.This typically involves flushing water down the drain.On the contrary, once dissolved oxygen has been removed from water it does not become re-oxygenated while in the plumbing lines.The brief period after it has been discharged from the fixture does not allow sufficient time for the re-growth of L. pneumophila.Careful placement of fixture valves and installation of fixtures that prevent water from being held past the valve, in contact with the atmosphere, is necessary to avoid re-oxygenation of the water and favorable growth conditions in the fixture discharge body.Further examination of equipment and modifications to fixtures that prevent reoxygenation of the water held within the fixture merits further research.

Limitations
This study and its findings had several limitations and did not examine all relevant factors and potential impacts of reducing DO levels in hot water systems.This was an experimental test using a model in a laboratory setting and there may be unforeseen differences in results when using DO to control Legionella in operational building water systems.Variables that may influence how L. pneumophila and water chemistry respond to reduced DO in hot water systems may include the type of pipe material, the type of water heater, the presence of established biofilm, the presence of air hammer arrestors, or the levels of DO and other common constituents of municipal water.The need to modify how building water systems are constructed, operated, maintained, or periodically disinfected is not fully known and was not part of this study.Identifying how plumbing fixtures can be modified to prevent reoxygenation of water is needed since data from this study revealed that L. pneumophila growth was not inhibited at water discharge fixtures.This was assumed to be caused by the reoxygenation of water held in the fixture past the valve while it was in contact with the atmosphere.The model hot water system was constructed with modern plumbing materials that are routinely used in residential, commercial, and healthcare facilities and full-scale equipment (i.e., water heater, filter casing, contactor, circulating pump, etc.).However, scaling up the equipment and system operations to meet the needs of a large building's hot water system is needed to fully assess its efficacy and the potential for adverse impacts.Further study of DO reduction treatment in building hot water systems should also consider possible shifts in microbial communities (i.e., microbiomes), impacts on chemical disinfectants introduced from municipal water supplies, and the potential creation of an environment favorable for the growth of anaerobic pathogens (Krause 2022).

Conclusions
There are few published studies examining the role of DO in the growth of L. pneumophila in natural and man-made aquatic environments, however, its fundamental need for oxygen is well-described.Before these experiments reported here, none of the published studies examined the association of DO in water and L. pneumophila growth in a model hot water system.However, it has been long recognized that because Legionella is an obligate aerobe it requires dissolved oxygen for survival and growth (Feeley et al. 1978;Wadowsky et al. 1985;US EPA 1999;Percival and Williams 2014).
The findings of this study, using a hot water model system demonstrated that dissolved oxygen could be reliably controlled using an in-line gas transfer membrane contactor to levels that inhibited L. pneumophila growth.Using a calculated D-value to quantify the rate of culturable bacteria decline at varying levels of DO in the model system, a threshold concentration of DO that inhibited growth was identified.The threshold DO concentration of 0.3 mg/L coincides with the lower limit of DO in natural habitats of lakes, ponds, and streams where L. pneumophila was present (Fliermans et al. 1981).Agreement with data from a broad examination of L. pneumophila's natural environment supports the conclusion of a similar threshold for L. pneumophila in premise plumbing and hot water systems.Application of dissolved oxygen reduction in hot water systems of healthcare and other large buildings as a prevention control method for L. pneumophila may reduce both the risk of this waterborne pathogen and supplemental chemical treatments currently in use.

Figure 2 .
Figure 2. Hot water model system L. pneumophila inhibition below 0.3 mg/L DO. (þ) indicates amplification of L. pneumophila in the hot water system; (−) indicates inhibition of L. pneumophila in the hot water system; ( � ) indicates no data available from the prior week's sample as this was the first round of sampling; error bars reflect the highest and lowest values used to calculate the geometric mean of L. pneumophila concentrations; data without error bars represents a single sample value.

Table 1 .
Measurement data from hot water model system.Samples and measurements taken from the Incoming/Supply Water represent the water entering the model test system (Figure1Site 4) where DO levels were not measured but not controlled.Samples and measurements labeled as Hot Water represent a combination of water leaving the hot water heater tank (Figure1Site 12) combined with the Hot Water Return (Figure1Site 5) of the model test system; where DO levels were measured and controlled.Samples and measurements taken from the Distal Fixture represent the water discharged from the model test system (Figure1Site 17) through an automated kitchen faucet where DO levels were measured but not controlled.� D-Value is reported as the number of days to reduce L. pneumophila concentrations in the circulating hot water compared to the incoming water by 1-log at 42 � C. þNon-Tuberculosis Mycobacteria (NTM) values reflect total concentrations of M. llatzerense, M. chimaera/intracellulare, and M. gordonae.þþ values reflect total concentrations of M. chimaera/intracellulare and M. gordonae.