INTRODUCTION
Legionellae are Gram-negative bacteria found naturally in freshwater environments such as rivers, lakes, and reservoirs and, as such, may also inhabit the municipal water supply and be found in industrial, commercial, and domestic hot and cold water systems (
1–3). Legionellosis, primarily Legionnaires' disease (a serious form of pneumonia) but also including nonpneumonic legionellosis and Pontiac fever (a self-limiting flu-like illness), can result when droplets or aerosols containing legionella bacteria are inhaled and deposited in the lungs.
Legionnaires' disease is most commonly, but not exclusively, caused by
Legionella pneumophila. It is a rare but serious disease. In 2009, there were 344 confirmed cases in England and Wales (
4), while in the United States, 3,522 cases were reported to the Centers for Disease Control and Prevention (
5). It is likely, however, that many more cases are unreported and/or undiagnosed, and it is estimated that Legionnaires' disease accounts for between 3% and 6% of the community-acquired pneumonias that occur in the United Kingdom and United States each year (
4,
6). Even with treatment, the case fatality rate of community-acquired Legionnaires' disease ranges from 10% to 15% (
4). It is, therefore, important to investigate and assess any potential risk.
A number of legionellosis outbreaks have been associated with the use of a spa pool (
7–17). Transmission is facilitated by a relatively high water temperature (30°C to 40°C), which favors the growth of
Legionella spp. (and other human pathogens) and an air injection system which, when operated, causes increased turbulence and the generation of water droplets and aerosols from the surface. The water droplets generated include those of appropriate size (between 1 μm and 8 μm) to transport legionellae deep into the respiratory tract (
7,
18). Outbreak investigations have found an association between time spent in a spa pool and likelihood of illness, suggesting a dose-response effect (
7,
9). However, aerosols released from the pool can be dispersed by air currents or ventilation systems, meaning that people outside the spa pool may also be at risk of acquiring a
Legionella infection (
9,
11). One of the world's largest outbreaks of Legionnaires' disease was linked to a spa pool being exhibited at a Dutch flower show (
12). Simply pausing at the pool was deemed the most important risk factor for infection, confirming that a contaminated spa pool, even if not used for bathing, can cause illness in susceptible people.
Spa pool ownership is increasing (
19), and domestic style spa pools, while on display in retail outlets, have been implicated in a number of
Legionella outbreaks (
10,
14,
17). In these cases, as in the majority of spa-associated outbreaks, the pools were considered the most likely source of infection in view of poor maintenance and/or inadequate disinfection. Domestic users may be more vigilant than retail staff in changing and treating spa water (
14). However, aeration, the high water temperature, and the large and variable organic load mean that disinfectant levels are difficult to maintain (
7).
The aim of this study was to investigate the aerosols released from a domestic spa pool and, using two surrogate organisms (MS-2 coliphage and Pseudomonas aeruginosa), to assess the potential risk to those in and around the pool should there be failures in disinfection practice.
RESULTS
The size distributions of the water droplets sampled using the APS at increasing heights and distances from the pool are shown in
Fig. 1 and
2, respectively. The results are expressed as mean mass concentrations (i.e., the mean mass of the particles per unit volume of air sampled). When the air injection system was inactivated, the mean mass of droplets between 2 and 10 μm that were detected 10 cm above the water line was 0.46 μg cm
−3 (
n = 9). When the air injection system was switched on, the mean mass concentration at this height increased almost 100-fold (
Fig. 1). However, as sample height increased, mean mass concentration decreased. The mean mass of droplets between 2 and 10 μm that were detected 1 m above the water line was 0.04 μg cm
−3 (
n = 9). At this height, mean mass concentration did not significantly change with increasing distance from the pool (
Fig. 2). Closer to the water surface, the mean mass of droplets between 2 and 10 μm that were detected 50 cm from the pool was 1,000-fold lower than that detected at the pool edge (38.6 μg cm
−3). Increasing the sample distance to 1 m increased the mean mass concentration from 0.01 μg cm
−3 to 0.06 μg cm
−3.
The mean numbers of MS-2 coliphage and
Pseudomonas aeruginosa organisms recovered from the pool water were 7.1 × 10
7 PFU liter
−1 and 3.2 × 10
5 CFU liter
−1, respectively (
Table 1). Neither organism was recovered from air samples taken when the air injection system was inactivated (i.e., the mean air concentrations of MS-2 and
P. aeruginosa were <2 PFU m
−3 and <3 CFU m
−3, respectively). When the air injection system was activated, the highest numbers of the contaminating organism (regardless of surrogate) were recovered close to the water surface (i.e., from air samples taken 10 cm above the water line at the edge of the pool). The mean air concentration of MS-2 coliphage was significantly greater than that of
P. aeruginosa at 20 cm and 1 m above the water line (
P < 0.05). However, the comparatively high concentration of MS-2 in the pool meant that the proportion of the suspension aerosolized (i.e., the spray factor) was higher for
P. aeruginosa (
Fig. 3). For both surrogate organisms, as sample height increased, air concentration and, as a consequence, spray factor decreased (
Table 1;
Fig. 3).
The mean air concentrations of MS-2 coliphage and
P. aeruginosa in samples taken 10 cm above the water line but 50 cm away from the pool were 4.6 PFU m
−3 and 3.1 CFU m
−3, respectively, 100-fold lower than that recovered from samples taken at the same height but at the pool edge (
Table 2). Mean air concentration increased with increasing distance from the pool. However, this increase was not significant (
P > 0.05), and at 1 m from the pool, the spray factors were just 3.98 × 10
−4 ml m
−3 and 3.8 × 10
−2 ml m
−3 for MS-2 and
P. aeruginosa, respectively. At this distance, neither organism was recovered from air samples taken 1 m above the water line (
Table 2).
DISCUSSION
During 2007 and 2008, 18 of the 116 outbreaks associated with treated recreational water that were reported to the CDC were caused by bacteria. Ten (55.5%) of these outbreaks were caused by
Legionella spp. All were linked epidemiologically to spa pools and resulted in 122 persons developing legionellosis (
25). Key to reducing the microbiological contamination of a spa pool is an appropriate and effective management system. Control measures should include a sufficient concentration of disinfectant (≥2 ppm of free chlorine) and frequent emptying and cleaning of the pool (
26). However,
Legionella spp. have been recovered from public spa pools that have been treated and/or emptied appropriately (
11,
26,
27), implying that the concentration of disinfectant necessary to achieve microbial control may be higher than that normally recommended. Thus, domestic spa pool users may be at risk from
Legionella despite complying with manufacturers' maintenance instructions.
During this investigation, the manufacturer's instructions for spa pool maintenance were not followed. At no point during the 2-month study was the spa pool water changed or disinfected, nor were the surfaces of the pool cleaned. This lack of a control regimen was primarily to ensure the viability of the MS-2 coliphage but, in terms of the pool hygiene compliance of the domestic user, represented the worst-case scenario. Bacterial colonization of the spa pool was rapid.
P. aeruginosa has very simple nutritional requirements, and trace elements naturally present in water are sufficient to enable the organism to survive and to multiply (
28). Although not detected in the main water supply, high numbers of
P. aeruginosa (10
5 CFU liter
−1) were recovered from the pool water within 3 days of filling.
P. aeruginosa may have been present in the source water at levels of <1 CFU 100 ml
−1 (i.e., lower than the detection limit of the membrane filtration assay). Alternatively, bacteria may have entered the pool in rainwater. During the current study,
P. aeruginosa together with other pseudomonads (e.g.,
P. fluorescens,
P. antarctica, and
Brevundimonas vesicularis [identified using MALDI-TOF]) were recovered from rainwater that had collected on the pool cover.
L. pneumophila, although previously recovered from rainwater (
29), was not detected. Regardless of the source of contamination, once the organism was transferred to the spa pool, the elevated temperatures facilitated the survival and growth of
P. aeruginosa.
Outbreaks of
P. aeruginosa folliculitis and otitis externa are frequently associated with spa pools (
30). Unlike legionellosis, these types of
Pseudomonas infection result from contact with contaminated water rather than the inhalation of aerosols. However, the virulence of
L. pneumophila depends upon the size of the particles in which the cells are contained and their site of deposition in the respiratory tract (
31).
P. aeruginosa is similar in size to
L. pneumophila (approximately 2 μm in length and 0.3 to 0.9 μm in width) (
18); therefore, when investigating the aerosolization of water droplets, it may be a more appropriate surrogate for
L. pneumophila than MS-2 coliphage (ca. 25 nm in diameter) (
32).
L. pneumophila induces a lethal infection in guinea pigs when administered as a small-particle aerosol (<5 μm in diameter) (
31). Droplets of this size are large enough to contain the bacteria (i.e., >2 μm) but small enough to deposit in the respiratory tract. The mass of the droplets relates to the number of bacteria contained within (
18). Aerosolization of droplets from the spa pool when the air injection system was inactivated was minimal. However, when the air injection system was turned on, the mean mass of droplets within the respirable range generated close to the water surface increased significantly (
Fig. 1), as did the mean air concentration of
P. aeruginosa (
Table 1).
Published quantitative microbial risk assessment models (
23,
24) have used aerosol mass measurements to estimate the microbiological concentration in the air above a whirlpool spa. The results of the current study, which reports corresponding aerosol mass and microbial concentration data, suggest these models may overestimate bacterial aerosol emission.
Armstrong and Haas (
23) used particle size data generated by Baron and Willeke (
18) to estimate the aqueous aerosol load 35 cm above a turbulent 40°C spa pool (total mass of aerosolized particles [5 mg m
−3] × water density = 5 × 10
−3 ml m
−3). By multiplying the aqueous aerosol load by the
Legionella concentration in the bulk water (assumed in this model to be 3.6 × 10
3 CFU ml
−1), they estimated the concentration of
Legionella organisms in the aerosol released by a whirlpool spa to be 18 CFU m
−3. Subjecting the particle size data generated during the current study to this model results in an estimated aqueous aerosol load 20 cm above the turbulent spa pool of 350 ml m
−3 (350 μg cm
−3 [mass of 2 to 10 μm aerosol] [
Fig. 1] × water density). The concentration of
P. aeruginosa in the water was 3.2 × 10
2 CFU ml
−1, leading to an estimated concentration of
P. aeruginosa in the air of 1.12 × 10
5 CFU m
−3, 10,000-fold higher than the actual mean air concentration (47 CFU m
−3) (
Table 1).
An alternative model uses a bacterial water to air partitioning coefficient to predict microbial air concentration. The value suggested by Armstrong and Haas (2.3 × 10
−5 CFU of
Legionella m
−3 of air per CFU of
Legionella liter
−1 of water) was derived from endotoxin data and was considered a reasonable surrogate for Gram-negative bacteria in a hot spring environment (
23). However, supporting data are limited, and the authors acknowledge that uncertainty regarding this input parameter contributes to variability in estimated dose. The results from the current study provide additional data regarding water-to-air partitioning coefficients (i.e., the proportion of a suspension aerosolized, i.e., the spray factor).
The use of spray factor in microbiological risk assessment has been demonstrated (
22). The lower spray factor associated with MS-2 coliphage compared to
P. aeruginosa (
Fig. 3) may be due to their presence in particles with an aerodynamic diameter of <0.5 μm. These particles may not have been captured efficiently by the cyclone samplers used in these experiments (
33), and for this reason, the values obtained for
P. aeruginosa should be the most relevant for
L. pneumophila risk assessment.
When the spa pool air injection system was deactivated (simulating a natural hot spring), the air concentration of
P. aeruginosa 10 cm above the spa pool was <1 CFU m
−3, a spray factor of < 3 × 10
−3 ml m
−3 (i.e., <3 × 10
−6 liters m
−3), 10-fold lower than the water-to-air partitioning coefficient used previously (
23). Activating the air injection system significantly increased the spray factor (1.1 ml m
−3 [i.e., 1.1 × 10
−3 liters m
−3]) and, thus, the air concentration of
P. aeruginosa (3.5 × 10
2 CFU m
−3) (
Table 1) and the risk of aerosol transmission.
When bubbles rise through water, they can collect bacteria on their surface, resulting in bacterial enrichment. When the bubbles burst, they produce film droplets which can range from 1 μm to 10 μm in diameter (
34). The number of bacteria in film droplets can be 20-fold higher than that in the bulk water (
34). The presence of high numbers of
L. pneumophila organisms in spa pool water does not always result in infection, and in previous investigations, the absence or low level of illness has been attributed to the absence of air compressors or failure to use the air injection system (
7,
27). The results of the current investigation support these conclusions.
Dose-response models used to estimate exposure dose (
23) and probability of infection (
24) are based on the exposure of guinea pigs to
L. pneumophila. Subjecting our microbiological data to the same inhalation rates (9 liters min
−1 [male] and 6.7 liters min
−1 [female]) (
24), exposure time (15 min), retention rates (50% deposition of a 5-μm-range aerosol) (
23), and infectivity per
L. pneumophila CFU (0.06) (
24) as assumed in these models results in estimated exposure doses 10 cm above a turbulent spa pool of 23.8 CFU and 17.6 CFU for males and females, respectively. At this height, the estimated probabilities of infection are 0.76 for males and 0.65 for females. When the air injection system is inactivated, the estimated risk of infection is negligible.
The recovery of
L. pneumophila aerosolized from bath taps has been shown to decrease with increasing distance from source (
35). Similarly, microbial air concentration and, thus, exposure dose decreased with increasing height and distance from the spa pool (
Fig. 3;
Table 2), implying that while activating the air injection system may increase the risk of infection for those either in or at the edge of the pool, there is no increased risk to those at any distance from the pool (>1 m). However, it is acknowledged that the spa pool used in the current investigation was in a very sheltered location; the cyclone sampler was generally positioned downwind from the pool, and the effect of wind speed and/or direction on the spread of the aerosolized particles was not investigated. When the air injection system was operated, air was forced through a series of air holes that extended around the inner circumference of the pool. How direction and/or force of the air injection can affect the aerosolization of particles is also not known.
Nonetheless, the results of this study confirm that if not properly maintained, a domestic spa pool rapidly becomes contaminated. The elevated temperature of the water facilitates the growth of human pathogens which become aerosolized during activation of the air injection system. This study provides direct data on the aerosolization of bacteria from a spa pool, including water-to-air partitioning coefficients and corresponding bacterial air concentrations and aerosol mass measurements. These data will help inform quantitative risk assessments, increase the robustness of the model(s) used (by improving estimates and assumptions), and, in doing so, increase confidence in the results. After subjecting our microbiological data to published dose-response models, it can be concluded (assuming that the level of aerosolization of P. aeruginosa is similar to that of L. pneumophila) that susceptible individuals either in or at the edge of the pool may be at risk of acquiring legionellosis. However, under the experimental conditions described here, the risk of infection is reduced with increasing height and distance from the pool.