Legionella is a genus of Gram-Negative bacteria, which contains over 50 different species and is a part of the gamma proteobacteria class (Schwake et al., 2015). Legionella species are the causative agents of either Pontiac Fever, a flu like non-fatal disease, or Legionnaires’ Disease (LD), an atypical form of pneumonia with much more severe symptoms which can be fatal. Legionella pneumophila is the commonest cause of LD, accounting for almost 85% of cases (Papagianeli et al., 2021; Kermani et al., 2022). The most common transmission route of LD is through inhaling of Legionella contaminated aerosols, resulting in subsequent pulmonary infection (Kermani et al., 2022) (Volker et al., 2016). As the notification rate for Legionellosis has risen to 1.8/100.000 which is the highest ever recorded in the EU, it is critical to prevent the Legionella contamination of artificial water systems, which pose the biggest threat for human infection (Papagianeli et al., 2021; Kermani et al., 2022).

Legionella species’ natural reservoirs are freshwater and wet soil, where the bacteria survive mostly as intracellular parasites of free-living protozoans, especially amoeba species such as Acanthamoeba (Rogers et al., 1994; Borella et al., 2005). However, Legionella’s most important reservoirs are man-made water environments including piped drinking water, cooling towers, fountains and humidifiers, with hot water systems being colonized more often than cold water and hotel hot water systems being related to Travel Associated LD (Borella et al., 2005; Volker et al., 2016; Rasheduzzaman et al., 2020). In artificial aquatic environments, Legionella colonization has been associated with the presence of biofilms (Borella et al., 2005; Bargellini et al., 2011), which protect Legionella from disinfectants and contain the protozoa in which Legionella replicates. In the biofilm environment, the abundance of nutrients allows for Legionella to grow extracellularly, also (Schwake et al., 2015).

Legionella in hotel water systems may pose a great risk especially to immunocompromised and debilitated individuals. Furthermore, hotels often have a high occupancy rate and a constant turnover of guests. This means that if Legionella is present in the water system, many people may be exposed to it over time, increasing the risk of an outbreak. Of crucial importance is the complexity of plumbing systems in the hotels, including numerous water fixtures, pipes, and tanks. Legionella bacteria thrive in warm water, and these systems can provide ideal conditions for their growth and spread of the pathogen. Moreover, water in some parts of the system may stagnate, creating pockets of stagnant water where Legionella can multiply. Since Legionella bacteria can become aerosolized in showers, hot tubs, water spectacles, etc., guests may get infected by inhalation of contaminated water droplets. Finally, there are always legal and reputational concerns since if a guest contracts Legionnaires’ disease while staying at a hotel, it can lead to legal liability and significant damage to the hotel’s reputation. Lawsuits, investigations, and negative media coverage can result from outbreaks associated with hotels.

To mitigate the risk of Legionella in hotel water systems, many jurisdictions have implemented regulations and guidelines for hotel operators to follow, including regular monitoring, water treatment, and maintenance practices to ensure the safety of their guests. Proper water management and Legionella prevention measures are essential to protect both guest health and the hotel’s reputation. To ensure that the above-mentioned criteria are met, several factors need to be considered when trying to minimize the risk of water systems from Legionella.

Various studies have associated several physio-chemical factors with the survival of Legionella and the formation of biofilms in pipeline surfaces with the colonization of Legionella in an attempt to create mathematical models that predict the risk of the bacterium’ s growth and human exposure, models which would help develop control strategies to prevent legionellosis (Yunana et al., 2021). Despite the fact that most “traditional” entero-pathogens, enter artificial water systems in the rare case of sewage system malfunction or greater water age (Falkinham et al., 2015; De Giglio et al., 2021), Legionella are considered opportunistic environmental pathogens, a distinct category of pathogens, similarly to Mycobacterium spp., and Pseudomonas aeruginosa. If certain conditions are met, these pathogens are able to survive disinfection procedures and grow in the oligotrophic environment of engineered water, even in low concentrations of oxygen, posing a health threat especially for the immunocompromised people (Lu et al., 2017; Zhang et al., 2021; Zhang et al., 2021a). The most important risk factors identified so far include high water temperature (Rasheduzzaman et al., 2020), water stagnation, pipeline material and roughness, structure age, nutrient concentration, such as Zn and Mn, disinfectant concentration, pipeline length and corrosion and Heterotrophic Platelet Count (HPC) (Bargellini et al., 2011; Volker et al., 2016).

The aim of the current study was to implement different mathematical approaches to try and obtain conclusive results on the effect of several physico-chemical and microbiological factors on the presence of Legionella in water distribution systems.

A total of 663 samples were tested over the four-year period of 2018-2022. The samples were collected from 38 different hotels from the four prefectures of the island of Crete distributed as follows: seven (7) from Lassithi (eastern Crete), nine (9) from Heraklion (central Crete), four (4) from Rethymno (western Crete) and 18 from Chania (western Crete).

Of the samples tested, 107 (16%) were positive for Legionella (regardless the serogroup, or species), 12 (1.8%) were positive for E. coli, 102 (15.4%) were positive for total coliforms, 73 (11%) were positive for P. aeruginosa and 107 (16%) for Pseudomonas species.

Apart from L. pneumophila sg 1, five (5) other serogroups were also, detected in the water samples tested; these were serogroups 3, 6, 8, 9, and 10. The dominant of all serogroups including 1, was 3 (it was detected in 39 out of the 663 samples tested; 5.9%). In fact, during the last decade, serogroups 3, 6 and 8 have dominated the isolation rate in water samples tested in Crete. Serogroup 14 has almost disappeared while several species are not frequently isolated any more. A large metagenomic analysis is undertaken to figure out the reasons hiding behind that. As regards Legionella species, L. anisa, L. londiniensis, L. taurinensis and L. erythra were isolated with their rates ranging not more than one percentage of the total samples. These results along with the range of the cfu/L are shown in Table 1.

The distribution of the water samples collected according to their temperature is shown at Figure 1. Some of the samples collected from hot water supplies were far away from being considered as hot ones since their temperatures were very low. On the other hand, very few cold-water samples fulfilled the criterion of less than 20°C.

In this study we evaluated the effect of several factors that may influence the presence of Legionella in water distribution systems. We focused on temperature, since this is considered the most crucial one, chlorine, the presence of other pathogenic (E. coli, coliforms, Pseudomonas, P. aeruginosa) and non-pathogenic microorganisms (HPC), the type of operation of the hotel (annual or seasonal), other epidemiological factors (number of rooms, number of stars), and the type of sample collected (direct or following a two-minute flush). We also applied a holistic approach by implementing different mathematical analyses and models, to evaluate the potential significance and cross talk between the major risk factors.

We also performed individual analyses of the results on the presence of different Legionella spp., into separate for L. pneumophila sg 1, L. pneumophila (including all serogroups) and total Legionella (including any species or serogroup isolated). Moreover, we made our calculations based on both the LOD of our laboratory (50 cfu/L) and on the upper acceptable limit according to our National Legislation for Legionella (1000 cfu/L).

As regards temperature, Legionella species demonstrate the ability to survive in a wide spectrum of temperatures, ranging from lower than 20°C, maintaining their ability to multiply in slow rates in some of these temperatures, to as high as 70°C, although for short periods of time, while displaying differences in their preferred temperature ranges for growth and survival depending on the species and the serogroup. Legionella pneumophila can be isolated from water temperatures ranging from 5.7 to 63°C, while proliferating at temperatures ranging between 25 and 45°C, peaking at 35 ± 2°C (Rasheduzzaman et al., 2020; Papagianeli et al., 2021). Even though the multiplication rate of Legionella increases in temperatures close to the optimal, the maintenance of a stable, albeit lower, population seems to be steadier in lower temperatures such as 25°C. This can be explained by the lower metabolic rates of the bacteria in these temperatures (Schwake et al., 2015). These low metabolic rates also seem to account for the low numbers of Legionella in biofilms forming in temperatures as low as 20°C, where the high number of Acanthamoeba theoretically provides a substrate for Legionella growth (Rogers et al., 1994).

As Legionella species exhibit limited survival in temperatures higher than 50°C, a constant high temperature can be used as a means of thermal disinfection. However, the current guidance for hot water temperature (50 to 55°C) seems inadequate as it is rejected by a number of studies which favor a range of temperatures from 55°C to 60°C (Rasheduzzaman et al., 2020; Papagianeli et al., 2021). While species such as L. longbeachae, a species found commonly in potting soil as well as in water systems are susceptible enough to a water temperature of 50°C, L. pneumophila, and especially the common serogroup 1, seems to be resistant to this temperature, whereas serogroup 7 seems to be resistant to temperatures as high as 55°C (Borella et al., 2005; Cervero-Arago et al., 2015). When it comes to temperatures higher than 55°C they were found to be significantly associated with decreased Legionella colonization coming in contrast with the positive correlation between water temperature <55°C and Legionella contamination (Rasheduzzaman et al., 2020). Above 60°C, all L. pneumophila serogroups exhibit similar behavior, surviving for several minutes (30 min 70°C). Legionella show higher tolerance for heat when associated with amoeba and biofilms, thus showing the need to examine other prevention factors such as disinfectants and nutrient availability (Cervero-Arago et al., 2015). Knezevic et al. recorded a statistically significant difference in the median temperature among the samples positive for Legionella and those negative for the bacterium, whether the water was warm or cold (Knezevic et al., 2022). The same study also concluded that L. pneumophila serogroup 1 showed a preference for higher temperatures compared to the rest serogroups.

Based on the results of our study, according to the generalized linear model both hot and cold-water supply seemed to affect the presence of Legionella when tested in separate at the cut-off proposed by the national legislation (1000 cfu/L). When applying a univariate approach, the exposure to temperature (under 25°C for cold water and over 47°C for hot water) also showed a high association with Legionella.

Moreover, according to the logistic regression and the binomial model created, exposure to different water temperatures seemed to be the covariate with the greater association with Legionella. In particular, cold water seemed to have positive linear relationship with Legionella while hot water seems to have negative linear relationship with the bacterium. Similar results were obtained when Count and Hurdle models were also applied.

In general, it seems that it is agreed that a hot water temperature of <50°C is a risk factor for Legionella contamination, but there still seems to exist a controversy when it comes to the temperature range between 50°C and 60°C. There are studies that have suggested that contamination at temperatures between 50°C and 60°C or >60°C don’t differ significantly and accept the guideline of >55°C (Volker and Kistemann, 2015), while others have pointed out the survival of a colonization of water distribution systems even at >60°C and hold temperatures as high as 57°C as a risk factor if water chlorination/hygiene is inadequate. Thus, these studies strictly suggest a temperature of over 60°C unless these other parameters can be regularly and closely monitored (Domenech-Sanchez et al., 2022).

Of interest, temperature also affects both the physico-chemical parameters which include chlorine decay and the absorption of chemicals and determine the formation of biofilms as well as the dynamics of formed biofilms allowing Legionella to detach and spread through the water system (Calero Preciado et al., 2021).

As regards disinfection, chlorination of hot water systems is one of the two major preventive measures against Legionella contamination, alongside thermal disinfection, and it can be used either as a secondary method of disinfection or as an installation disinfection method (hyper-chlorination). It is recommended by the WHO that drinking water has a minimum chlorination of 0.2 mg/L or 0.5 mg/L, depending on the risk of said water source for contamination (Cervero-Arago et al., 2015). The chlorine levels of the water seem to correlate negatively with L. pneumophila growth except for serogroup 1 which seems to be more resistant to the disinfectant. Knezevic et al. reported a simultaneous presence of chlorine along with Legionella, however the presence of the bacterium was reported at low to very low levels of the disinfectant (Knezevic et al., 2022).

According to our results and regardless of the models used, only total Legionella, dis-regarding the cut-off value (50 or 1000 cfu/L, revealed a statistically significant value and only when chlorine was treated as a qualitive parameter.

Other studies have also pointed out the negative correlation between chlorine concentration and Legionella growth however, they note that inside Acanthamoeba vesicles, where Legionella grow parasitically especially in biofilms, there is an increase in chlorine resistance. However, increasing chlorination is not a viable option as it would cause odor and taste problems to the water while also posing a health threat itself (Falkinham et al., 2015).

Apart from the physico-chemical factors, we also investigated the potential influence of other pathogenic microorganisms on the presence of Legionella. The pathogens whose presence and dynamics are currently studied to assess the water quality, mainly include Enterococci, E. coli and other coliforms. However, these pathogens’ presence or absence fails to determine the growth of opportunistic pathogens (OPs), which are usually not of Enteric or Fecal origin (Felfoldi et al., 2010; Mapili et al., 2020). It has also been suggested that the presence of certain other opportunistic pathogen species or other Legionella species in biofilms promote Legionella growth through increased resistance to temperature and disinfectants by horizontal gene transfer (Girolamini et al., 2022). This however is still controversial as studies suggest both an association and a lack thereof between the growth of Legionella and other OPs such as P. aeruginosa (Masaka et al., 2021; Atnafu et al., 2022).

According to the generalized linear model that we created, E. coli, coliforms and P. aeruginosa seemed to have a positive correlation with L. pneumophila sg 1, L. pneumophila and total Legionella. When testing our results by univariate analysis, Pseudomonas species and total coliforms showed a statistically significant impact on the presence of Legionella regardless of the species. It seems that increased numbers of E. coli, coliforms and P. aeruginosa may be present with increased numbers of Legionella. It is not clear whether the parallel presence of E. coli and coliforms is due to adequate presence of nutrients or the presence of biofilm that can provide a rescue room for these bacteria. To assess this issue the study of biofilm presence and perhaps of amoebae should have been necessary.

Our multivariate analysis showed that P. aeruginosa seems to prefer lower temperatures than Legionella for its survival and multiplication; a similar observation was recorded following PCA analysis. On the contrary, hot water exerts a negative effect on all pathogens tested when chlorine and temperature were not considered as variables. Free chlorine seems to exert a similar effect on all pathogens.

It has been suggested that Legionella could be used as an alternate indicator of water quality. On the other hand, the need for specific culture media, the different sensitivity to certain factors such as disinfectants between Legionella and other opportunistic pathogens and Fecal bacteria and the existence of viable but not culturable Legionella limit the pathogen’s potential usage as a sole indicator of water quality and it is suggested that it should be used in conjugation with other “typical” pathogens (Zhang et al., 2021b).

As far as HPC is concerned, it is considered an early indicator of microbial growth within a water distribution system since it signalizes the availability of organic nutrients, formation of biofilms and the presence of amoebae (Stojek and Dutkiewicz, 2011; Wingender and Flemming, 2011). HPC is used to monitor and assess the quality of water, as its increase is indicative of a post treatment biological activity in a hot water system, essentially signifying the existence of biofilms or the nutritional conditions needed for their formation (Bargellini et al., 2011; Volker et al., 2016). Formation of biofilms allows for Legionella to grow both intracellularly, as parasites to protozoan hosts, and extracellularly, being supported by various cyanobacterial species (Rogers et al., 1994). Specifically, Legionella grows intracellularly in vesicles contained in various Acanthamoeba spp. and extracellularly by harvesting carbon and energy sources excreted by other bacteria (Declerck et al., 2009; Stojek and Dutkiewicz, 2011). Biofilms also provide protection from disinfectants and temperature while serving as a substrate for exchange of genetic material and play a pivotal role in the maintenance of Legionella through the water system as parts of them get detached as a result of many factors such as temperature (Whiley et al., 2019; Calero Preciado et al., 2021; Girolamini et al., 2022).

While it stands to reason that Legionella occurrence could correlate with HPC, studies have shown both the existence and the lack between HPC and Legionella occurrence and it should be noted that the studies which do suggest a correlation point out that their results are limited by other factors such as temperature, with >50°C not showing any correlation between HPC and Legionella occurrence (Serrano-Suarez et al., 2013; Whiley et al., 2019). Some other studies have shown a significant correlation between Legionella contamination and HPC at specific temperatures, for example at 30°C, while others have only associated the two parameters in certain temperatures and only for very high HPC values (>100 CFU/ml) revealing the need for HPC and biofilm to be examined with more parameters such as water stagnation (Bargellini et al., 2011; Volker et al., 2016). A recent study (Knezevic et al., 2022) reported that HPC was positively associated with Legionella only when HPC was recorded following incubation on blood agar plates at 36°C for four days. At less days of incubation or when yeast extract was used as a medium, no significant correlation was reported. These studies perhaps indicate that HPC may not be a rigid parameter that can be used to monitor the presence of Legionella.

Stagnation of water occurs when there is an inadequate water flow rate, often due to a lack of a circulation pump or the existence of a physical dead leg in the hot water system. The stagnation zones allow bacteria to create biofilms providing both the time necessary for their formation and by making it difficult to maintain high temperatures and disinfectant concentrations (Volker et al., 2016). Although we did not study the role of water stagnation in our study, there have been studies that agree upon the correlation of water stagnation and Legionella contamination. Some authors attribute it to the dissipation of chlorine, going as far as to point out the correlation of winter and spring, the seasons with the lowest water usage and thus the highest water stagnation, with Legionella growth (Al-Bahry et al., 2011; Zhang et al., 2021b).

In our study we chose to test for the effect of number of stars and rooms upon the presence of Legionella in the water samples collected. The choice of these two factors was made as an indirect way to test for the effect of the complexity of a water distribution system assuming that the higher the number of stars and rooms the bigger the building and the more complex the water distribution system. However, we failed to show any correlation between the presence of Legionella compared to the number of stars and rooms of the hotel tested irrespective of the model, or the analysis chosen.

Other studies have revealed that the age and height of a building are positively correlated with Legionella contamination of the water distribution system, possibly because they make it difficult for proper water temperatures to be maintained throughout the entirety of its extent, especially at higher floors. An age of 20 or more years and a number of floors over 10 are also considered risk factors (Masaka et al., 2021; Atnafu et al., 2022; Girolamini et al., 2022). On the contrary, a newer study failed to exhibit a correlation with Legionella growth with the age of a water distribution system pipes (Wingender and Flemming, 2004). In general, however, it seems that lower floors of taller buildings are not protected, as Legionella spreads to them from the higher floors (Borella et al., 2005; Gamage et al., 2022). Furthermore, the outlet distance has also been associated with Legionella contamination, as the outlets further away from temperature decrease, biofilm formation and stagnation of water (Volker et al., 2016).

According to the analysis that we performed, the proximity of the sample point to the boiler seemed to exert an effect (positive or negative based on the distance from the boiler) to L. pneumophila sg 1 but not total Legionella regardless of the cut-off used.

Although we did not test for the effect of nutrients, Legionella species have very low requirements to survive and grow. The presence of copper seems to limit the bacterium’s growth. The hardness and the pH of the water seem to influence Legionella’s growth, both directly and indirectly, as a low pH leads to corrosion of pipe walls which allows for biofilms to develop (Borella et al., 2005; Volker et al., 2016).

Other studies agree that the release of nutrients in the water distribution system, possibly by like rubber coated valves favor the formation of biofilms and some of these nutrients such as ammonia have been correlated with Legionella contamination (Wingender and Flemming, 2004; Zhang et al., 2021b).

We also tried to test for the effect of direct or indirect sampling. The results that we revealed were ambiguous since no relationship was calculated at a cut-off of 1000 cfu/L for Legionella and of 50 cfu/L for L. pneumophila sg 1 while a negative effect was calculated at a cut-off of 50 cfu/L for Legionella and a positive effect at a cut-off of 1000 cfu/L. Certainly, more studies need to be carried out to study the potential effect of flushing on the detection of Legionella.

Lastly, we did not reveal any correlation between the presence of Legionella compared to the number of hotels situated at the region of sampling when applying complicated analysis, while when we focused strictly on the number of hotels tested in the area from which the samples were collected, a statistically significant p-value was calculated regardless the type of analysis followed. Again, more studies need to be carried out to reveal any potential effects lying behind.

The findings of the current study further support the belief that simple recording of temperature, chlorine, Legionella concentrations, etc. is no longer enough to minimize the risk of colonization of water distribution systems. A more holistic approach is need to properly survey Legionella presence since the latter seems to be affected by far more factors than just temperature and chlorine (or any other disinfectant). There is a need to take under consideration and routinely test for the presence of opportunistic and other bacteria since their presence may either act as indicator of Legionella presence or may enhance the presence of the bacterium. Other epidemiological factors should also be considered when trying to set up a guidance for the risk factors that affect Legionella. Water stagnation and age/type of pipelines should be included into the risk factors considered. There have been several cases of hotel units that try to get rid or even minimize the presence of the bacterium in their water distribution systems without any success. Simple hyper-chlorination and/or heat shock do not always prove enough to minimize the risk. To keep the system under control and to maintain an equilibrium of the bacterium concentration in the water distribution system, there is a need to remove all these factors that enhance the survival and proliferation of Legionella.

The results of this study add up to the existing knowledge recorded from past studies on the area and could be used as a guide to proper hotel management in implementing effective water treatment strategies to reduce Legionella risk in water distribution systems.

Certainly, the results of the current study may be limited by the specific geographic focus on Crete; further studies in diverse locations may provide more generalized insights on the effect of the above-mentioned factors on Legionella. Moreover, although we tried to implement as many statistical approaches as possible and study as many parameters as possible, it would be inevitable to include all of them. Under this term, we need to point out that part of the study was performed during the SARS-CoV-2 pandemic where several hotels either remained closed or started their operation with great delay during the summers. These prolonged closures may have affected the microbiological quality of the water distribution systems and may have made it more difficult to clean the pipelines from any microorganisms, biofilms, etc. Therefore, if the same study was carried out before the pandemic different results could have been recorded. Moreover, we did not test for the potential effect of stagnation of water or the protective effect that amoebae may confer to Legionella. We did not also study the presence of viable but not culturable forms of Legionella or the potential effect that these forms of the bacteria may have had upon our results. Although we tried to investigate the effect of several epidemiological factors associated with the hotel (for example the number of rooms or stars), we did not study the effect of seasonality upon our results. We did not also test for the presence of chemical agents, such as Mg, Mn, Fe, etc., that may affect the presence of Legionella in the water distribution systems. Of course, in case we needed to do so, we would have to repeat all our analyses for each season separately. Application of different models did not always give us back similar results. Under this concept, processing of our data with neural networking or artificial intelligence may had provided us with different outcomes. In general, more studies could or should focus on these aspects as well.

Plumbing drinking water systems show great short- and long-term variability in terms of the presence and concentration of Legionella. Regular testing by classical culture methods cannot always reliably reveal contamination risks. There is a need for a longitudinal sampling approach, better risk assessment and better specification of the number of samples that need to be collected to be representative of the water distribution system. We need to consider the entire drinking water plumbing system as an ecological system and each outlet as an ecological niche. The analysis of single parameters specific to individual sampling points cannot reliably predict the possibility of Legionella contamination. The application of empirical modeling using logistic regression can provide a valuable contribution to a better risk assessment for hydraulic drinking water systems than conventional culture-based detection methods. Moreover, the use of different models may provide different results. It is crucial to assess all these parameters prior to sampling since this can help: in risk assessment, in the prioritization of risk areas, in identifying suitable sampling points, in determining control measures within a building and in the development of a water safety plan for a specific building.

The results of the current study provide further knowledge on the study of Legionella presence in water distribution systems. Several factors may affect its presence and multiplication and certainly more complex approaches are needed to control the bacterium.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

DC designed the experiment, did the laboratory testing and wrote the manuscript. VS wrote the manuscript and did the laboratory testing. AN wrote the manuscript. M-OD did the statistical analysis. TL did the statistical analysis. NT did the laboratory testing. RM did the laboratory testing. CP collected samples. AP did the supervision. All authors contributed to the article and approved the submitted version.