The prevalence of Legionella pneumophila in different water systems: A global systematic review and meta-analysis

The presence of Legionella pneumophila (LP) in aquatic habitat is a global concern. The current study was undertaken to estimate the prevalence of LP in water systems with the aid of a systematic review and meta-analysis. The searching was performed among some international databases, including Scopus, PubMed, and Embase to retrieve the related articles between 1/January/1983 and 25/July/ 2017. Therefore, thirty-six articles (with 43 studies) out of 1,541 articles collected, were included in the meta-analysis. The overall prevalence of LP in water systems was determined as 20% (95%CI: 15-25). Also, the lowest and highest pooled prevalence of LP was observed in Poland (4% (95%CI: 0-13%)) and Kuwait (98% (95%CI: 90-100%)), respectively. The lowest and highest prevalence of LP-based on water resources subgroups was a water reservoir (15% (95%CI: 2-37%) and well (40% (95%CI: 26-50%), respectively. The number of studies that used polymerase chain reaction (PCR) for detection of LP was 16/43 (37.3%) while the culture method was 27/43 (62.7%). Generally speaking, the relatively high prevalence of LP among the investigated water systems was demonstrated, which should be reduced by performing appropriate control actions.


Introduction
Legionella pneumophila (LP) as a heterotrophic bacterium is a gram-negative, aerobic, non-sporogenous, and a mobile [1,2] which can be isolated from various environments such as soil and water systems, including water used in cooling towers and ventilation, surface water, tap water, and spring water [3-6].
The tolerance to high temperature is among the features, which could stimulate the presence, and proliferation of LP in water systems [7][8][9]. Although the range of optimal temperature for proliferation was indicated as 29-40 °C, rarely, it can survive in water at temperatures ranging from 0 to 63 °C [10].
Also, the e ciency of chemical or thermal disinfection can be reduced in the presence of bio lm or amoebae; which can be resulted in further contamination of water systems by LP11].
The droplets of contaminated water withLP could convey the bacterium into the lungs and as a result of swallowing by macrophages [12,13]. Therefore, the approaching of e cient and appropriate methods is crucial to decreasing as well as controlling the prevalence of LP water systems. In this regard, according to the guideline of World Health Organization (WHO), the acceptable limit of LP in the drinking water and cooling tower systems were de ned as 1 CFU/L (WHO 2008) and 1000 CFU/L [14], respectively.
Legionnaires' disease are among issues caused by these water-borne pathogens, commonly known as Legionella. Twenty-ve out of 59 species in the Legionella genus, have been correlated with human disease [15]. According to Cazalet et al. (2008), almost 20 kb of 33-kb locus carrying the genes for the proteins in lipopolysaccharide biosynthesis in LP is highly speci c for L. Pneumophila serogroup 1 (LP-Sg1) [16]. Also, multigenome via comparative hybridization detected 3 genes, including lpp0837/wzm, lpp0831, and lpp0838/wzt strains only LP-Sg1 that used for real-time PCR method for identifying LP-Sg1 [16].
Moreover, Legionella pneumophila serogroup 1 (LP) is responsible for the majority of reported cases of Legionnaires' disease (about 90%) [17]. Legionnaires' disease, for the rst time, was diagnosed as pneumonic form [ [14]. In the PCR method, the limit of detection (LOD) is 2 × 10 2 GU/100 mL for LP (mip gene), while theLOD for culture methods is 1 CFU/100 mL [25,26]. Although culture techniques are used widely, these techniques have several limitations including a long time needed to obtain results, weak recovery, weak sensitivity, inability to detect viable but non-culturable cells (VBNC) [27]. In another hand, the polymerase chain reaction (PCR) techniques due to reproducibility, sensitivity, speci city, highthroughput and reducing in the required time to less than 24 hours are widely approached to detect LP in water systems [28][29][30] However, no systematic review and meta-analysis was performed regarding the prevalence of Legionella pneumophila in different water systems. Therefore, for the rst time, the current study was undertaken to perform a meta-analysis to investigate the prevalence LP in the water systems based on geographical, quality of research as well as the type of water.

Material And Methods
In this study, the systematic review was conducted in accordance with the Cochrane protocol and Prisma protocol [31].

Search strategy
The search strategy was performed on the Legionella pneumophila studies in water systems. Searching was performed in the main international databases, including Scopus, PubMed, and Embase. title-abs-key ( microbe) or title-abs-key ( bacteria))) and ( ( title-abs-key ( water) or title-abs-key ( water and system) or title-abs-key ( drinking and water) or title-abs-key ( tap and water) or title-abs-key ( groundwater))) and (c) Embase: 'legionella':ab,ti OR 'legionella pneumophila':ab,ti OR 'microbe':ab,ti. Also, the references articles were reviewed to obtain more articles. Seventeen years (1/January/1983 and 25/July/ 2017) was chosen as the searching period.

Inclusion and exclusion criteria
Full-text articles were downloaded and then reviewed carefully to check if they meet the proposed criteria [32-34] such as (1) original study; (2) cross-sectional data; (3) published in English; (4) published online between 1/January/1983 and 25/July/ 2017; (5) full-text available articles; (6) existence of exact total sample size and positive samples; (7) the de ned type of water was examined; and (8) accurate methods including culture or PCR techniques were mentioned. The articles were excluded when not meet our criteria.

Data extraction
The obtained information from each study can be summarized as study characteristics; (the rst author, year of study); total sample size; the number of positive samples; the geographical study (countries); study methodology (Culture or PCR); type of water systems (tap drinking water, hot water, water reservoir, cold water, well and spring).

Meta-analysis of data
The ratio of the positive samples (p i ) to the total sample (n i ) de ned as prevalence (P = p i /n i ), which is between 0 to 1 value [32, 35-37]. Estimation of the prevalence of LP in water systems was performed using the binomial distribution model [38]. The heterogeneity (I 2) statistics was used to determine the variation between the prevalence of Legionella penompophia among the included studies [39]. In the current study, when the heterogeneity was higher than 50%, the random effect model (REM) was used to estimate prevalence based on the de ned subgroup. The Begg's and Egger's test was used to estimate publication bias [21,40]. A meta-analysis of data was conducted using Stata 12.2 intercooled version (Stata Corp, College Station, TX). All statistical analysis was signi cant at P-values< 0.05.
According to Begg's (p-value = 0.029) and Egger's (p-value = 0.024), signi cant publication bias among studies was noted ( Figure 4A-B); Hence, to remove the effect of publication bias, the metatrim test was performed. The pooled prevalence of LP in a water system based on metatrim in the random effect model (REM) was 21% (95%CI: 15-26%) ( Figure 5).

Discussion
The prevalence of LP in the water systems was obtained as an average of 20% (95%CI: 15-25%). The prevalence of LP in the different countries considerably was difference (Figure 2), which can be associated with variation in the quality of disinfection of water, the average life of water facilities, the chemical quality of water and the type of exploitation in each country.
Some factors such as water temperature, stagnation of water in pipes, the formation of bio lms on the interior walls of pipes, the presence of protozoa, the chemical quality of water and pH could affect Legionella growth in aquatic systems [71,72]. However, the proposed temperature for growth of LP is 29-40 °C with an optimum of 35 °C; it could survive in aqueous systems at temperatures of 0-63 °C [10]. The increase in temperature above 55 °C can be considered as an e cient technique for the prevention of bacterial growth and consequently, lower incidence of legionellosis [73]. According to Leoni et al. (2001), no LP growth was observed at temperatures above 41 °C [74]. However, the growth of LP decreased with increase in temperature [75].
Moreover, LP quickly reproduces in hot water systems such as showerheads (45-50 °C) [61]. Based on one recent study, the highest prevalence of LP in hot spring was 17.6% in the 40-48.6 °C [7]. Findings showed that the prevalence of LP in cold water was higher than hot water (Figure 4), which can be correlated with the inhibition effect of increase in temperature on LP growth. Additionally, based on ecological data, protozoans and amoeba can protect this bacteria from disinfectant, osmolality, and pH variations [61,76].
The in uence of metallic elements such as iron and manganese on the growth of Legionella previously also has been demonstrated [77,78]. Due to organophilic properties of LP, its growth is limited in the water with low iron concentrations. Considering to ndings of Portier et al. (2016), the iron pyrophosphate and ferric iron chelator did not affect the persistence of LPin the bio lms, but ferrous iron chelator showed a positive effect because of higher bioavailable ferrous ion. Hence, the growth of LP in the pipes that are made of iron could be stimulated. Also, manganese has an indirect effect on Legionella growth by enhancing the growth of bio lms and plantations [79].
Moreover, the protective role of manganese in enzymatic activity and increasing in resistance to oxidative stress has been con rmed [80]. In this regard, the positive correlation between Mn concentration (> 3µg/L) with a prevalence of LP in water was demonstrated [81]. The presence of a copper ion in water due to its antibacterial nature can resulted in further decline in the prevalence of Legionella [82]. However, bio lms in water systems protect Legionella against disinfection [83]. Therefore, in order to reduce the prevalence of Legionella, washing of pipes and tanks in addition to exhausting of water facilities with chlorinated hot water can be recommended. According to Oberdorfer et al. (2008), the prevalence of Legionella in the old water system of the hospital was higher than new ones [84]. Nowadays, the use of polyethylene (PE) and Polyvinyl chloride (PVC) materials in water facilities is continuously expanding. The growth of bio lm was increased with the releasing of volatile organic compounds in PVC and PE pipes [85]. Therefore, using PE and PVC in water facilities may increase the prevalence of LP in water systems.
pH is another chemical parameter that affects the growth and survival of LP in water systems. The highest prevalence of LP was reported in weakly acid (pH 5-7, 37.5%) [7]. In a report by Ohno et al. (2003), the optimal range of pH for LP in water systems was de ned between 6.0 and 8.0 [8]. However, other data showed that LP had not been explored in extremely acidic water [60,86].
Considering the frequency of detection techniques, PCR (37.3%) approximately was similar to the culture method (62.7%). While in culture techniques, only living bacteria can be detected [87], in the PCR in addition to living bacteria, the bacteria that damaged by the disinfection which their DNA remains also can be detected [88]. Moreover, in the culture technique, the incubation period of 3-14 days is recommended to facilitate the growth of bacterial colonies, while the bacteria can be damaged by used acid treatment in culture technique. In another hand, the observed limitation regarding detection sensitivity (50-60%) is another disadvantage of culture techniques [89]. Due to Culturable None but Viable (CNBV) properties, LP can survey in water for a long time without being identi ed by culture technique. Likewise, data have shown that Legionella prevalence in amoeba-water samples was 25-50% higher than other samples because of symbiosis survival with the amoeba [90,91]. Moreover, one of the most signi cant limitations of PCR techniques is bias due to the presence of inhibitors such as polysaccharide and chlorophyll in the samples [92].
Since the culture and PCR techniques have different ranges of detection limits, however, by using PCR technique, higher level of contamination in pathological samples can be detected quickly and accurately, approaching of both methods in the case of environmental issues such as LP might be recommended.

Conclusion
In the current study, the prevalence of LP in water systems in the de ned subgroups were meta analyzed. A high prevalence of LP in water systems worldwide was demonstrated by the current study as a rst systematic review in this eld. Likewise, the higher prevalence of Legionella pneumophila was considerablein the water systems, particularly in cold water. In this context, approaching of control actions such as avoiding stagnation of water in water systems, use of high-quality water, and continuous puri cation of water using disinfectants factors can be recommended. Continuous monitoring of water  [1] LOD methods for PCR :2 × 10 2 GU/100 mL for LP (mip gene) and culture methods is 1 CFU/100 ml.
[2] Buffered charcoal yeast extract: culture techniques   Forest plot of the prevalence of LP in the water resources based on the type of water resources.

Figure 4
Evaluate publication bias based on Beggs (A) and Eggers (B) tests Figure 5