Cryptosporidium and Giardia in recreational water in Belgium

Cryptosporidium and Giardia are important parasites in the aetiology of diarrhoea worldwide. In industrialised countries Cryptosporidium spp. are detected in up to 54.2% and Giardia spp. in up to 74.4% (Homan & Mank 2001; Geurden et al. 2009) of immunocompetent patients with diarrhoea. In non-outbreak-related cases of diarrhoea in Belgian patients, Giardia was identified as the second most prevalent pathogen. In the same study Cryptosporidium was also within the five most commonly detected pathogens, indicating that both parasites occur frequently in patients with gastro-intestinal symptoms (Geurden et al. 2009). In Belgium, the national incidence of cryptosporidiosis and giardiosis is 2.5 and 10.8 per 100,000 inhabitants, respectively, with the highest incidence in children (<5 years of age) and young adults (25–44 years old) (Wetenschappelijk Instituut Volksgezondheid [WIV] 2010).

Transmission of Cryptosporidium spp. and Giardia spp. occurs from either humans or animals by the faecal-oral route. In humans worldwide C. hominis and C. parvum are the most commonly detected Cryptosporidium species. C. hominis is human-specific, while C. parvum infects ruminants and humans. Giardia duodenalis assemblage A and B are responsible for giardiosis in humans, and are also found in a wide range of mammals. Within assemblage A, sub-assemblage AI is mostly found in animals, whereas sub-assemblage AII is predominantly found in humans (Sprong et al. 2009). Sub-assemblage AIII is almost exclusively found in wild hoofed animals. In contrast, there is no clear subgrouping within assemblage B (Sprong et al. 2009; Feng & Xiao 2011). Next to direct contact with an infected host, Cryptosporidium and Giardia can be transmitted through faecal contamination of food or water. Because of the parasites' ability to survive in the environment, their relative resistance to disinfection and the low number of (oo)cysts needed for infection, Cryptosporidium and Giardia are considered as important waterborne infections (Smith et al. 2006). From 2004 to 2010, 199 outbreaks of human diseases due to the waterborne transmission of parasitic protozoa were reported worldwide. Cryptosporidium spp. were the aetiological agent in 60.3% of the outbreaks and Giardia spp. in 35.2% (Baldursson & Karanis 2011). In a third of the outbreaks, recreational water was detected as the source of infection, (Baldursson & Karanis 2011). Swimming in contaminated waters and swimming pools is now recognised as an important transmission route for Cryptosporidium (Karanis et al. 2007). Outbreaks of cryptosporidiosis have been associated with recreational water in the USA (e.g. Craun et al. 2005; Causer et al. 2006; Wheeler et al. 2007; Boehmer et al. 2009; Cantey et al. 2012), Canada (Hopkins et al. 2013), Australia (e.g. Dale et al. 2010; Waldron et al. 2011), Japan (Takagi et al. 2008), Sweden (Insulander et al. 2005; Mattsson et al. 2008) and the UK (e.g. Smith et al. 2006; Coetzee et al. 2008). Although few reports can be found on Giardia infections related to recreational water (e.g. Porter et al. 1988), a marked seasonality in the onset of giardiosis occurs in summer to early autumn in many countries, including Belgium (WIV 2010). This increase coincides with increased outdoor activities (e.g. swimming) (Hlavsa et al. 2005) and with increased travelling during summer holidays.

In Belgium, Cryptosporidium and Giardia were detected in surface water that is used for drinking water production (Ehsan et al. 2014), but no data are available for recreational water, and the importance of water recreation in the transmission of Cryptosporidium and Giardia in Belgium is unknown. The objective of this study was to investigate the presence of Cryptosporidium and Giardia in public swimming pools, recreational water bodies and splash parks in Flanders (Northern Belgium) and to estimate the infection risk associated with swimming and other recreational activities. Positive samples were genotyped in an attempt to identify the source of infection (human vs. animal).

In total, 99 samples from recreational water bodies in Flanders, Belgium were analysed. From March to October 2010, 36 public swimming pools were sampled. The swimming pools were selected based on an increased risk for faecal accidents or external contamination (i.e. paddling pools, therapy pools frequently visited by mentally disabled persons and outdoor swimming pools). Convenience samples were collected from three types of water samples, including pool water (60 L) in 20 swimming pools, filter backwash water (2–60 L) in 16 pools and water from continuous flow centrifugation (2 L) in one swimming pool.

In August 2011, 40 convenience samples (30 L) were collected from 10 recreational lakes. Each lake was sampled four times with weekly intervals.

In July and August 2012, 23 convenience samples (30 L) were taken from neighbourhood water fountains (n = 16), splash parks (n = 7) and a water fountain in a boating lake (n = 1).

All samples were transported to the laboratory, stored at 4 °C and analysed within 72 hours.

A protocol was optimised to detect Cryptosporidium and G. duodenalis in water samples, based on the United States Environmental Protection Agency (USEPA) method 1623 (USEPA 2005). Water samples were filtered through Filta-Max Xpress filters (IDEXX Laboratories, Inc., Westbrook, ME, USA) with the aid of a peristaltic pump with recommended flow rates of 2 L/min. The Filta-Max Xpress filters were washed with the Filta-Max Xpress automated washing station for elution of the filters following the manufacturer's instructions. The eluate was centrifuged and the volume of sediment was measured. Between 0.5 and 2 mL of sediment was used for immunomagnetic separation (IMS) of the (oo)cysts. Oocysts and cysts in the sediment were purified by IMS using Cryptosporidium- and Giardia-specific antibody-coated magnetic beads according to the manufacturer's protocol (Dynabeads^® GC-Combo, Invitrogen Dynal, A.S., Oslo, Norway). IMS-purified cysts and oocysts were stained on well slides by fluorescein isothiocyanate (FITC)-conjugated anti-Cryptosporidium and anti-Giardia MAbs FITC-conjugated monoclonal antibodies (EasyStain™) (BTF Pty Ltd, Macquarie Park, Australia). Slides were examined using a Leica Leitz DMRB fluorescence microscope. The well surface was scanned at 200 or 400 times magnification using a FITC fluorescence filter (450–590 nm Chroma technology corp.). Cryptosporidium oocysts and Giardia cysts were identified and counted based on their size, morphology and fluorescence. Results were expressed as count per litre. Slides containing (oo)cysts were kept at 4 °C for DNA extraction.

DNA was extracted from water samples from recreational lakes that were positive by microscopy for Cryptosporidium and/or Giardia. Positive samples from indoor swimming pools, neighbourhood fountains and splash parks were not genotyped, as numbers of (oo)cysts were too low.

Genomic DNA was extracted from (oo)cysts that were scraped from the microscope slides using the QIAamp DNA Mini Kit (Qiagen GmbH, Hilden, Germany) or from sediment using the QIAamp Stool Mini Kit according to the manufacturer's instructions, incorporating an initial step of three freeze-thaw cycles (freezing in liquid nitrogen for 5 minutes and heating at 95 °C for 5 minutes) in the protocol to maximise disruption of (oo)cysts. Previously described polymerase chain reaction (PCR) protocols were used to amplify the 18S rDNA gene (Ryan et al. 2003) and the heat shock protein (hsp)-70 gene (Morgan et al. 2001) of Cryptosporidium. For the identification of Giardia the β-giardin gene (Lalle et al. 2005) was used in a nested PCR. For assemblage-specific amplification of Giardia, the triose phosphate isomerase (tpi) gene was used (Sulaiman et al. 2003; Geurden et al. 2008; Levecke et al. 2009). For all PCR reactions, negative (PCR water) and positive controls (genomic DNA) were included. The PCR products were visualised in agarose gel (1.5%) stained with ethidium bromide under ultraviolet (UV) light. PCR products were fully sequenced by the BIG Dye Terminator V3.1 Cycle sequencing Kit (Applied Biosystems). Sequencing reactions were analysed on a 3100 genetic Analyzer (Applied Biosystems) and assembled with the program SeqMan II (DNASTAR, Madison, WI, USA). To determine the genotypes/assemblages the fragments were aligned with homologous sequences available in the GenBank database, using MegAlign (DNASTAR, Madison, WI, USA).

Backwash water samples from two swimming pools were found positive for Cryptosporidium, with 0.23 oocysts/L and 0.03 oocysts/L, respectively. In one of these samples, as well as in one pool water sample from another swimming pool, Giardia was detected (0.23 cysts/L and 0.07 cysts/L, respectively). Only the results from the pool water were used for calculating the infection risk. The risk of Giardia infection in the investigated swimming pools varied from 1.13 × 10⁻⁶ to 2.49 × 10⁻⁶ per swim per person (Table 1).

Eight out of 10 sampled recreational lakes were positive for one or both parasites on at least one sampling occasion. In seven out of 10 lakes Cryptosporidium oocysts were detected once (n = 4) or twice (n = 3), with oocyst counts ranging from 0.07 to 0.60 oocysts per L (Table 2). Genotyping was only successful for two out of 10 samples. Both sequences were identified as Cryptosporidium andersoni (National Centre for Biotechnology Information [NCBI] accession numbers KM455082, KM455083). In the other cases either no DNA could be obtained from the slides (n = 1), no PCR product was obtained (n = 6) or no sequence was obtained from the PCR product (n = 1). Giardia cysts were detected once (n = 5) or twice (n = 2) in seven out of 10 lakes. Apart from one lake, these were the same lakes that were also positive for Cryptosporidium. Giardia cyst counts ranged from 0.23 to 0.70 cysts/L (Table 2). In seven out of nine positive samples a PCR product was obtained with the β-giardin gene and/or the tpi gene. Sequencing results showed G. duodenalis assemblage AI in six out of seven cases (NCBI accession numbers KM455069-KM455071, KM455074-KM455078, KM455080, KM455081), either as the only assemblage (n = 4) or in combination with assemblage AII (n = 1, NCBI accession number KM455079) or assemblages BIII (NCBI accession number KM455072) and E (NCBI accession number KM455073) (n = 1). In one positive sample assemblage ‘BIV-like’ was present (NCBI accession number KM455068). The infection risk for men, women and children for swimming in recreational lakes is shown in Table 3. The estimated infection risk for Cryptosporidium varied from 2.79 × 10⁻⁵ to 5.74 × 10⁻⁵ per swim per person, while for Giardia the infection risk was between 7.04 × 10⁻⁵ and 1.46 × 10⁻⁴ per swim per person. For other outdoor water recreation activities the estimated infection risk was 5.71 × 10⁻⁶ for Cryptosporidium, and 1.47 × 10⁻⁵ for Giardia.

No Cryptosporidium oocysts were detected in any of the sampled water fountains or splash parks. However, water samples from three fountains and two splash parks contained Giardia cysts. Cyst counts were 0.03, 0.07 and 0.20 cysts/L in the water fountains and 0.13 cysts/L in both splash parks. The infection risk of Giardia infection during a 3.5-minute visit to a splash park for children equalled 1.68 × 10⁻⁴ (95% confidence interval: 0–1.57 × 10⁻³).

Swimming is one of the most popular recreational activities worldwide with over 350 million person-events estimated to take place annually in the USA alone (Fayer 2004). Health risks for swimmers may arise from exposure to bathing waters of poor quality. They may suffer from various diseases such as gastro-enteritis caused by bacteria, viruses or parasites of faecal origin (World Health Organization (WHO) 2003). Among protozoan parasites, Giardia and Cryptosporidium are associated with waterborne outbreaks worldwide from recreational water, including swimming pools and recreational lakes. In this study, three out of 37 swimming pool samples (8.1%) tested positive for Cryptosporidium, Giardia or both (two filter backwash samples and one swimming pool water sample). Similar contamination rates were reported in non-outbreak-related pools in France, the Netherlands, the USA and Egypt (Fournier et al. 2002; Schets et al. 2004; Shields et al. 2008; Abd El-Salam 2012). In Italy, a higher proportion of pools were positive for Cryptosporidium and/or Giardia (Briancesco & Bonadonna 2005; Oliveri et al. 2006), while in Greece no (oo)cysts were found in five swimming pools (Papadopoulou et al. 2008). Based on the observed (oo)cyst concentrations, an attempt was made to estimate the infection risk associated with swimming in these pools. Since oocysts were only detected in filter backwash water, no infection risk for Cryptosporidium could be calculated for the investigated swimming pools. Using the methodology of Schets et al. (2011b), the infection risk for Giardia was estimated as 1.13 to 2.49 × 10⁻⁶ per swim per person. In comparison, in a swimming pool in the Netherlands, an infection risk of 1.3–2.8 × 10⁻⁵ for Giardia was estimated per swimming event per person (Schets et al. 2004, 2011b). The higher infection risk in the Dutch study was due to a faecal contamination incident and filter malfunctioning during the time of sampling. Although no recent faecal contamination incidents were reported in the swimming pools in the present study, the selection of ‘high risk’ swimming pools may have introduced a bias, and the calculated infection risk may not be representative of all swimming pools in Flanders. An increased infection risk has been associated with swimming pools that are frequently visited by young children (Stafford et al. 2000; Hlavsa et al. 2005). Moreover, as cyst viability was not assessed, total cyst counts were used in our risk assessment, assuming 100% viability. Although Giardia cysts can survive in water for a considerable time (Olson et al. 1999), the infection risk is possibly overestimated because of this assumption. However, it should be noted that these are baseline contamination levels and that the infection risk can increase dramatically in the case of a faecal contamination incident or filter malfunctioning.

In seven out of 10 of the investigated recreational lakes (10 out of 40 samples) Cryptosporidium was detected at least once, with relatively low oocyst concentrations (0.07–0.6 oocysts/L). Similar contamination rates and oocyst counts were obtained in recreational waters in France, the Netherlands, Luxembourg and in central Spain, (Coupe et al. 2006; Schets et al. 2008; Helmi et al. 2011; Galvan et al. 2014), while higher contamination rates and oocyst counts were reported in northern Spain (Castro-Hermida et al. 2010) and Canada (Loganthan et al. 2012). In most of these studies, contamination rates with Giardia were slightly higher, with higher cyst counts, compared to Cryptosporidium. Similarly, in the present study Giardia cyst concentrations were 0.23–0.7 Giardia cysts/L. Based on the observed (oo)cyst counts, the estimated infection risk for Cryptosporidium varied from 2.79 to 5.74 × 10⁻⁵ per swim per person, while for Giardia the infection risk was between 1.46 × 10⁻⁴ and 7.04 × 10⁻⁵ per swim per person. For the limited contact water recreation activities the estimated infection risk was lower than for swimming for Cryptosporidium (5.71 × 10⁻⁶), and comparable for Giardia (1.47 × 10⁻⁵). Assuming only one visit to one of the recreational lakes per year, the infection risk for Giardia associated with swimming in recreational lakes was already above the generally accepted criterion of <1 infection per 10,000 individuals (USEPA 1989) despite the fact that all the investigated lakes had at least good water quality, according to the criteria of Directive 2006/7/EC. In this directive, bathing water quality is defined by threshold values for microbiological parameters, corresponding to four bathing qualities (‘excellent’, ‘good’, ‘average’ and ‘poor’). As parasites are not covered by this directive, they are not routinely monitored. The results of this study and other studies (Schets et al. 2008, 2011a, 2011b) suggest that infection risk for Cryptosporidium and Giardia cannot be extrapolated from the commonly used parameters for bathing water quality. However, it should be noted that the infection risk for Cryptosporidium and Giardia might be prone to bias. First, we used the total (oo)cyst concentrations to estimate this risk, assuming that all (oo)cysts were assumed to be viable in the risk assessment. Second, not all Cryptosporidium and Giardia species that were detected are infectious to humans. Regarding Cryptosporidium, only C. andersoni was identified. Although C. andersoni has been reported in human patients (e.g. Jiang et al. 2014), it is usually associated with cattle. Similarly, the frequently identified zoonotic Giardia duodenalis assemblage AI (6/7 samples) is mostly found in animals (Sprong et al. 2009). Other assemblages that were identified are either livestock-specific (E) or found predominantly in animals (BIV-like), and only in two samples were human-specific assemblages (AII, BIII) identified. Although it cannot be excluded that other species or genotypes were overlooked, e.g. due to poor recovery of DNA, these data suggest that animals, possibly livestock, were the predominant source of contamination for the investigated recreational lakes, indicating that the risk of these protozoa might be overestimated. Finally, we would like to underscore that the choice of the dose-response parameter p_m has an important impact on the final risk assessment. This is particularly the case when the dose-response varies considerably between different Cryptosporidium species and G. duodenalis (sub)assemblages. For example, within C. parvum a large variation in dose-response estimates was observed, ranging from 5.3 × 10⁻³ (Iowa-isolate) to 59.0 × 10⁻³ (‘Ungar C. parvum’ isolate) (Messner et al. 2001). In the present study, we allowed for this variation in the risk assessment of Cryptosporidium, but not for Giardia as currently little is known about the variation in dose-response between different isolates and (sub)assemblages.

Giardia cysts were detected in water samples from three out of 17 fountains and two out of seven splash parks. Although no Cryptosporidium oocysts were found, it cannot be excluded that low concentrations of oocysts were missed, given the limited sensitivity of the USEPA 1623 method. The infection risk during a splash park visit equalled 1.68 × 10⁻⁴, and hence, of all water-related activities, poses the highest risk for a Giardia infection. Outbreaks of cryptosporidiosis and giardiasis related to visits to water fountains have been documented in the USA (e.g. Centers for Disease Control and Prevention (CDC) 1998, 2000; Eisenstein et al. 2008; Kirian et al. 2008), suggesting that the presence of Giardia in fountains may constitute a real risk for waterborne infection.

Cryptosporidium oocysts and/or Giardia cysts were detected in swimming pools, recreational lakes, splash parks and water fountains in Belgium. Although in recreational lakes (oo)cysts were frequently present, most positive samples belonged to species/genotypes that are either animal-specific or predominantly found in animals, suggesting that the risk of infection during recreation is relatively low. Lower contamination rates were found in swimming pools, splash parks and water fountains, but assuming that humans are the most probable source of contamination for these water bodies, these findings suggest a risk for human infection.