A convenience sample of 50 water and 58 swab samples were obtained from eight indoor swimming pools and three hot tubs from March, 2009 through April, 2010 from throughout the central Ohio area in the United States. All swimming pools were public Class A or B competition or large recreation pools (Class A pools are large public pools intended primarily for competition; Class B pools are public recreation pools). One hot tub was a small residential model and two were public, large capacity systems. All public hot tubs and pools were operated by either an educational institution or municipality. Water samples were collected in sterile, polypropylene tubes, at approximately 0.30–0.50 meters below the water surface to best simulate an exposure zone; this depth is similar to a sampling regime recommended for beach water [19]. Sterile rayon swabs were used to collect biofilm from swimming pool and hot tub jets, strainers, and side drain locations. Swab samples were rotated (spun) both along the swab axis and moved laterally to collect an appropriate biofilm sample. All samples were transported to the laboratory on ice for immediate analysis. Aseptic technique was observed throughout sampling activities. Temperature readings were obtained while onsite using a digital thermometer.

Water samples were analyzed for free chlorine, total chlorine, and turbidity. Free and total chlorine levels were measured immediately according to DPD Method 10069 and 10070 using a Hach 2800 spectrophotometer (Hach, Loveland, CO, USA). The method reporting range is 0.1–10.0 mg/L (although >10.0 mg/L is provided by the Hach). Turbidity readings were measured using a Hach 2100P portable turbidimeter (Hach, Loveland, CO, USA).

Water samples were concentrated by membrane filtration using a sterile 0.45-μm-pore-size, 47-mm-diameter cellulose membrane filter (Millipore, Bedford, MA, USA) [19]. Swab samples were processed similarly, also using membrane filtration. In order to maximize bacterial recovery from a swab, the swab-containing tube was filled with approximately 5mL of buffer, vortexed for 10s, and the buffer passed through a 0.45-μm-pore-size, 47-mm-diameter cellulose membrane filter. All membrane filters were placed directly onto Pseudomonas Isolation Agar (Becton Dickinson, Sparks, MD, USA) and incubated at 37 °C for up to 48 hours. After incubation, up to three randomly selected colonies with appropriate morphological characteristics were sub-cultured for enrichment and purification using R2A agar with 0.5% yeast extract (in order to maximize the growth of the bacteria isolated from disinfectant-treated source). R2A plates were incubated at 37 °C for up to 48 hours. R2A was used because recovered bacteria may have been injured or stressed due to the presence of chlorine, and R2A is an efficient medium for recovering bacteria from disinfectant-containing samples by providing a low concentration of diverse nutrients [20]. In previous work, we modified this method by adding a low concentration of yeast extract, which showed good recovery of bacterial isolates from potable water [21]. Up to three isolates were obtained from purified, suspected P. aeruginosa samples, and confirmed using polymerase chain reaction (end-point and real time qPCR). Negative and positive controls (ATCC 27853) were utilized for all reactions. All end-point PCR was conducted using a MultiGene Thermal Cycler (Labnet International Inc., Edison, NJ, USA), under the following conditions: 95 ºC for 1 min; 40 cycles of denaturation at 95 ºC for 15 s, annealing at 58 ºC for 20 s; final extension at 68 ºC for 40 s; with primer pa722F (5′-GGCGTGGGTGTGGAAGTC-3′) and pa899R (5′-TGGTGGCGATCTTGAAC TTCTT-3′). For the real-time PCR, we used ABI Pseudomonas detection kit (Applied Biosystems, Foster City, CA, USA) with a 48-well StepOne™ Real Time System (Applied Biosystems, Foster City, CA, USA) following the manufacturer’s instructions.

The minimum inhibitory concentration (MIC) to 22 antimicrobial agents was determined using a TREK Sensititre Sensitouch microdilution system (TREK, Cleveland, OH, USA). Twenty-three randomly selected isolates (at least one per P. aeruginosa-positive sample) were chosen for analysis and processed according to the manufacturer’s recommended protocol and the Clinical and Laboratory Standards Institute methods [22]. The 23 screened isolates represent 20 of the 23 total P. aeruginosa-positive samples; isolates from three samples failed to be subcultured sufficiently for microdilution analysis.

The findings of this study indicate that P. aeruginosa contamination is common in swimming pools and hot tubs, even where chlorine concentrations are well above recommended levels. Recommendations for adequate chlorine levels are 2 to 4 mg/L for swimming pools, and 3 to 5 mg/L for hot tubs, measured as free chlorine [23]. According to NSPF recommendations, facilities should consider immediate closure when free chlorine levels fall below 1 mg/L, because contamination can quickly reach unsafe concentrations. With regard to hot tubs, prior research has demonstrated that even when hot tubs are drained, cleaned, and refilled, P. aeruginosa may return with a level of 10⁴–10⁶ cells/mL within 24 to 48 hours when adequate chlorine levels are not maintained [15]. In the current study, chlorine concentrations were higher than in past studies; and while P. aeruginosa was routinely isolated, the regularly high chlorine levels found here likely reduced prevalence. This confirms the importance of not only adequate disinfectant levels, but also constant monitoring and adjustment to ensure chlorine levels never drop below recommended levels. It should be noted that the facilities in this study conducted hourly monitoring of pool and hot tub chlorine concentrations, except the residence, which performed weekly monitoring. The less frequent monitoring of the residential hot tub may have resulted in greater halogen fluctuations, and thus higher prevalence of P. aeruginosa.

Consistent with previous research, a relationship between high chlorine levels, biofilm growth (as indicated by swab samples, which contain higher levels of biofilm-embedded bacteria, versus water samples, which are more likely to represent planktonic bacteria), and the presence of P. aeruginosa, was noted. LeChevallier et al. (1988) have described several mechanisms leading to disinfectant resistance, of which attachment to surfaces was most important [24]. They note that low-nutrient environments enhanced free chlorine resistance by two to three times. Thus, it is not surprising that this study indicated higher prevalence of P. aeruginosa in pools/hot tubs with higher chlorine levels. There was also a relationship between higher water temperature and the presence of P. aeruginosa. This finding is consistent with the thermo-tolerant nature of P. aeruginosa.

There are several possible reasons why P. aeruginosa was more likely to be isolated from hot tubs over swimming pools. Hot tub piping systems are complex and inaccessible for cleaning; this complexity may enhance biofilm formation [9]. In addition, hot tub capacity is lower (per bather) than swimming pools (0.757 m³/person for a hot tub versus 6.81 m³/person for swimming pools); this may act to concentrate microbial load present in the water [23]. Finally, high pressure jets are an integral component of hot tubs, with bathers often using jets to enhance relaxation. Contact with jets in this manner may enhance sloughing of skin squames, further fouling hot tub piping systems.

The results of this study also indicated that environmental P. aeruginosa has considerable levels of antibiotic resistance. Isolates demonstrated resistance to a wide range of clinically relevant antimicrobial agents, including antipseudomonal agents aztreonam (22%), gentamicin (9%), and imipenem (26%); and intermediate resistance to amikacin (9%), meropenem (4%), and tobramycin (9%). Resistance was also noted to alternate agents ceftriaxone (4% resistant, 30% intermediate), ticarcillin/clavulanic acid (4%), and trimethoprim/sulfamethoxazole (13% resistant, 17% intermediate). Although P. aeruginosa is a model of antimicrobial resistance (due to a number of intrinsic factors), past research has found that the degree of resistance to antipseudomonal agents varies considerably. The current results are similar to past studies, which have found analogous prevalence levels of resistant P. aeruginosa in both the hospital and community [4,10].

The origins of multidrug-resistant strains in the swimming pool and hot tub environment are unknown, but may have developed through several mechanisms. Since P. aeruginosa is intrinsically resistant, it is possible that the frequency with which we isolated resistant P. aeruginosa is representative of natural background levels. Alternately, in some cases resistant P. aeruginosa strains may ultimately be traced to a human source, either direct shedding from colonized humans or indirect transmission through the contamination in upstream aquatic reservoir. P. aeruginosa is not a predominant component of normal human flora (colonization rates are low: up to 2% on the skin and to 24% in fecal matter) [4,25]. However, during hospitalization, colonization rate may reach 50% [4,25]. Also, immune suppressed individuals or those whom have recently undergone antimicrobial treatment may have higher colonization rates [4,25,26]. In addition, in clinical settings, there is a strong relationship between the P. aeruginosa strains which infect patients and those in the immediate patient environment. Rogues et al. [27] pointed-out that carriage of P. aeruginosa by patients was “both the source and the consequence of (tap) water colonization”. That is, a colonized or infected patient may lead to in-room and neighboring area contamination; and contaminated water, fomites, and aerosols may infect susceptible individuals [27]. A similar phenomenon may occur in recreational water, particularly if frequented by colonized or infected individuals.

In some cases, there may be selective pressure for resistance by the presence of low-level antibiotic concentrations in either sweat secretions, which may enter recreational water through bather use, or from supply water. Prior research has demonstrated the presence of a range of antimicrobial agents in arm and axilla sweat secretions after oral or intravenous dosing [28]. While isolated concentrations were found to be quite low, this is a possible mechanism for the development of resistant strains. Antibiotics have also been routinely isolated in water at various stages of processing, including in post-chlorination drinking water [29]. Importantly, in some cases chlorinated water was found to select for more multidrug-resistant strains over non-chlorinated water [30]. Given that the swimming pool and hot tub supply water in our study begins as municipal drinking water but then becomes highly chlorinated to reach disinfection levels, there may be enhanced selection for resistant strains.

While the levels of resistance among study isolates are similar to those previously reported, it is important to draw the distinction between a nosocomial setting and the non-clinical, non-outbreak, recreational setting of the current study. In clinical scenarios, there is constant selective pressure to enhance the proliferation of multidrug-resistant strains. Given that P. aeruginosa has both intrinsic resistance and a dynamic ability to develop resistance during the course of infection, a high frequency of resistance is now expected in hospitals. However, in the non-clinical environment, the absence of selective pressure may reduce antimicrobial resistance levels. The presence of resistance to front-line antipseudomonal drugs may have important clinical and prevention implications. Given how common swimming pool and hot tub use is, immune-suppressed or other high-risk individuals may wish to exercise caution before use, especially hot tub use. Exposure to resistant P. aeruginosa results in even greater risk for individuals with cystic fibrosis.

Study limitations include a relatively small sample size and the convenience design. Although only one residential hot tub was accessible for sampling in this study, there was a desire to broadly evaluate whether differences existed in the prevalence of P. aeruginosa and the degree of antimicrobial resistance. The residential hot tub owner did not experience previous skin diseases related to the hot tub and the hot tub did not undergo thorough maintenance and flush-fill cycles as did the public hot tubs, which were cleaned daily, drained completely once per week, and have fresh water introduced continually. Further study is required with expanded sample size to determine if the frequency of multidrug-resistant strains remains steady over time. Additional studies should focus primarily on hot tubs, where 53% of samples were positive. Also, isolates from three P. aeruginosa-positive samples were unable to be revived for antimicrobial resistance analysis.