Team members: Pitol, A., Kotlarz, N., Byrne, D., Swanberg, B., Uprety, S.
The Commonwealth Games are an international sporting competition that has been held nearly every four years since 1930 by countries in the Commonwealth of England. The upcoming games will be held in 2018 in Gold Coast City, Australia, and will include men’s, women’s, and mixed men’s and women’s relay triathlon events. A triathlon typically consists of a 1,500 meter swim, 40 kilometer cycle, and a 10 kilometer run (Commonwealth Games Federation).
The proposed location for the swim portion of the event is an estuary at the mouth of the Nerang River known as the Gold Coast Broadwater. The Nerang River catchment is approximately 500 km2, two thirds of which is made up of parks or rural land, and one third of which is urban with industrial, commercial, and medium to high density residential land use areas (Hinze Dam Alliance, 2007). A municipal wastewater treatment plant for the city of Gold Coast discharges treated effluent into the river upstream of the Broadwater. Additionally, rural and urban stormwater runoff discharge into the catchment during storm events.
In order to evaluate water quality and protect athlete health, the International Triathlon Union requires testing for Enterococci and Escherichia Coli in natural waters prior to swimming events (ITU rulebook). These bacteria are traditionally used as indicators of fecal contamination in waterways. However, the use of fecal indicator bacteria alone as an indication of microbial water quality can often result in a mischaracterization of risk to human health. Recent literature suggests that there is not always a clear correlation between fecal indicator bacteria concentrations and human health outcomes (Fewtrell and Kay, 2015). The Australian Government National Health and Medical Research Council, in their 2008 Guidelines for Managing Risks in Recreational Water, suggest that risk managers take a more comprehensive approach to characterization of microbial water quality to avoid over-characterizing risk based on fecal indicator bacteria concentrations alone.
The goal of this risk assessment is to determine which factors drive the majority of health risk associated with unintentional ingestion of water during the triathlon swim events, and to provide any necessary recommendations to mitigate those health risks in order to protect athletes and ensure that the Commonwealth Games are a success for the Gold Coast and Australia.
Several bacteria, viruses, with available completed or partial dose response model, were considered for risk assessment. For bacteria, pathogenic Escheria coli was considered but it was eliminated because of the lack of pathogenic E. Coli data in the area. Enterococci was also considered but was eliminated as it was being monitored as a part of International Triathlon Union (ITU) water standards for the event (ITU, 2012). Finally, Campylobacter jejuni was selected for risk assessment due to the presence of large number of seagulls in the coast which has been reported to carry Campylobacter jejuni and can pose threat to the athletes during the commonwealth event (Moore et al., 2002; Savil et al., 2001; Alonso et al., 1993). For viruses, Adenovirus was considered and chosen for risk assessment for its heavy presence in recreational water and marine water (Maurer et al., 2015; Wyn-Jones et al., 2011).
Campylobacter jejuni is a bacteria found in the intestines of birds and other warm-blooded mammals. Runoff contaminated by droppings from sea birds and mammals such as dogs in the surrounding stormwater catchment are likely sources of Campylobacter in the area of the triathlon. Illness from Campylobacter can result in gastroenteritis, and has been associated with both Guillian-Barre syndrome and reactive arthritis (Bhavsar, 2006). Adenovirus serotype 40 is a common virus that can also cause gastroenteritis (Wilhelmi et al. 2003). The wastewater treatment plant discharges about 3.6 miles upstream of the Broadwater and is a potential source of adenovirus, as are any potential sewage overflows during storm events. These potential sources of microbial contamination are represented on the map below.
Figure 1. Environmental conditions for exposure
Adenovirus are mostly transmitted through the respiratory, fecal-oral route and waterborne transmission and Campylobacter in humans are mainly transmitted through consumption of contaminated food or water. Although not common, person-to-person transmission can also occur for both Adenovirus and Campylobacter. The infectious dose for adenovirus is extremely low, possibly as low as a single infectious particle (Zhi-Yi et al, 1992) and the infectious dose for Campylobacter is thought be small as well; less than 500 organisms can cause a disease (Moore et al., 2006). The infections are typically followed by incubation period of less than 10 days for Adenovirus (Foy 1997; Gaydos 1999) and 1-21 days for Campylobacter (Friedman, 2004). Most gastrointestinal infected person shed viruses in stool prior to other symptoms like cramps, fever and vomiting appear.
Adenovirus Genotype and Campylobacter Species and Illness Characterization
Adenovirus serotype 40 and 41 have been reported to cause gastrointestinal illness in both children and adults (Wilhelmi et al. 2003). Adenovirus is considered to be second only to rotavirus in terms of its significance as a cause of childhood gastroenteritis (Crabtree et al. 1997). Studies conducted in Australia from 1981–1996 indicate that adenoviruses contributed 3% to 9% of the gastroenteritis cases. The majority of the cases were associated with young children and involved serotype 41 (40% to 80%) and to a lesser extent, serotype 40 (less than 20%). Seasonal patterns of the virus genotypes were evident, with type 41 being prevalent in late autumn and type 40 remaining prevalent year-round (Grimwood et al. 1995; Palambo and Bishop 1996). Camplobacteriosis, caused by a bacterial infection of Campylobacter jejuni in the human body, is the most common cause of diarrhea in the United States. Approximately 14 cases for each 100,000 persons in the population every year are infected. However, there are over 1.3 million persons per year with unreported cases. About 76 infected persons die each year (Bhavsar, 2006).
Exposure Scenario Athletes will be exposed to Adenovirus and Campylobacter via direct accidental ingestion while swimming in the river during the triathlon event. Exposure is estimated using data for the consumption of water during swimming and time spent in the water. The dose of pathogens ingested takes into consideration this exposure rate and concentration of pathogens in the river where the athletes are swimming. Figure 2 describe the sources (orange), microbial agents (blue), and barriers (green) that contribute to the final exposure route of direct ingestion (yellow) for each of the scenarios evaluated.
Figure 2. Exposure scenarios evaluated. a) Baseline scenario, b) Campylobacter in the wastewater, and c) Rainfall causing urban runoff
Wastewater Treatment Process:
The sewage treatment process in the Gold Coast region starts with pre-treatment of sewage by rotating drums to remove organic waste followed by grit removal to eliminate sands and stones. The third steps include bioreactors which allows bacteria to grow and remove contaminants from the water and bioreactor is flowed by clarifier. The effluent then is put in contact with chlorine for 30 minutes for disinfection and then is used to water the golf course and other sporting field. Surplus water is released to the ocean through the channel in the city. Adenovirus data for the effluent from wastewater treatment was collected from the treatment plant in the US with similar treatment process including chlorine contact time.
Ultraviolet Disinfection (UV Solar Disinfection):
Three mechanisms of solar disinfection include optical inactivation, thermal inactivation and interaction between optical and thermal effects. Laboratory studies have looked at different physical aspects such as light quality and intensity, temperature, type of container and turbidity on different microbes, mostly bacteria, to study the inactivation, loss of viability and degradation (Reed, 2004). UVA and UVB which compromises 99% of total UV from sunlight reaching the sea level is proven to have higher destruction of coliform bacteria at 290-235nm (Acra, 1984). Oates et al (1984) reported that effective solar disinfection requires around 3–5 hours of strong sunlight at an intensity above 500 W m−2 which are mostly achieved by areas lying between 15 and 35 degrees from the equator, where there is over 3000 h sunshine per year, with those areas lying between the equator and 15 degrees, having around 2500 h sunshine per year being the next most favorable locations. However, Goldcoast in Queensland, where Commonwealth games are being held in 2018 lie 350 from the equator and only have an average of 1200h 13 of sunshine per year with average yearly temperate with 260. Considering these facts, we have decided that UV disinfection is a negligible barrier for inactivation of bacteria (campylobacter). For viruses, most of the research have been concentrated on lab scale and with different intensities and wavelength than the natural UV wavelength and intensity. No assumption for virus activation could be made with confidence given the available data.
The microbial agents considered in this assessment are Adenovirus 40/41 and Campylobacter jejuni. Because there was not an existing dose response model for Adenovirus 40/41, a dose response model for a surrogate pathogen will be used. Three surrogate pathogens were compared based on their contribution to the uncertainty in risk of infection: Adenovirus 4 (inhalation pathway), Rotovirus, and Norovirus. The following table describes equations, parameters, and uncertainty characteristics describe the dose response models for the three surrogate pathogens as well as for Campylobacter jejuni.
Table 1. Dose-Response Models and Associated Parameters
|Adenovirus 4||Exponential P (infection) = 1-exp(-k*dose)||K = 0.0151||Couch et al., 1969|
|Rovorirus||Beta-Poisson P (infection)=1-[1+dose*((2^(1⁄α)-1))/N_50 ]^(-α)||α = 0.253 N50 = 6.17||Ward et al., 1986|
|Norovirus||Fractional Poisson P (infection)= P*(1-e^((-dose)/μ) )||P = 0.72 μ = 1,106||Messner et al., 2014|
|Campylobacter-jejuni||Beta-Poisson P (infection)=1-[1+dose*((2^(1⁄α)-1))/N_50 ]^(-α)||α = 0.144 N50 = 890||Black et al., 1988|
The three surrogate dose response models that were evaluated had different units for dose; therefore, the following conversion assumptions were made:
Most Probable Number per Liter (MPN/L) to Colony Forming Units per Liter (CFU/L): Campylobacter dose response model used CFU/L as dose and the data collected was in MPN/L. Several model represented the similarity between MPN and CFU (Cho et al., 2010; Noble et al., 2004) and the model showed strong positive correlation between the method of enumeration of bacteria and thus the MPN data from literature was directly used as CFU data. Viruses per liter (Viruses/L) to Focus Forming Unit (FFU) and genome per liter (Genome/L): Rotavirus dose response model used as a surrogate model for Adenovirus serotype 40/41 used FFU as a dose whereas the wastewater effluent of Adenovirus was found in viruses/L. All the Viruses/L was taken as FFU assuming all viruses are infective. Similarly, genome copies/L were assumed to be in 1:1 ratio with Viruses/L.
Scenario 1: Baseline
Under our baseline risk scenario, the mean risk of Campylobacter illness to athletes by swimming was estimated to be 0.13%. It is unclear what level of risk of Campylobacter illness is acceptable for recreational waters because there are currently no guidelines for Campylobacter levels in recreational waters. However, the risk of Campylobacteriosis from a single swimming event is very similar to the incidence of Campylobacteriosis in Queensland broadly. The incidence was 0.13% based on notified cases of Campylobacteriosis (6,220 cases) in 2014 given the population of Queensland (4.7 million). It is important to note that athletes may be coming to the Gold Coast from areas with lower levels of Campylobacter illness and they could be more susceptible to Campylobacter infection although there is limited data on this.
The mean risk of adenovirus illness estimated by our model was 26.6% using a Rotavirus dose- response relationship as a surrogate for Adenovirus 40/41 (Table 2). The mean risk of illness decreased to 15% when an Adenovirus 4 dose-response model was used. The lowest risk was estimated to be 2.25% using a Norovirus dose response model. This risk estimate would satisfy the 3.6% risk considered tolerable by the current recreational criteria for gastrointestinal illnesses (US Environmental Protection Agency).
It has been suggested that using of Adenovirus 4 dose-response model for predicting gastrointestinal illness through ingestion is likely to overestimate the risk of illness, which leads to conservative estimations of risk for Adenovirus 40/41. Adenovirus 4 dose-response model has been used previously in other QMRA studies (Soller et al., 2010, McBride et al., 2013).
Risk of illness caused by Campylobacter (0.12%) was significantly lower than risk of illness by Adenovirus, even for the lowest mean risk estimated for Adenovirus (2.19%). This difference in risks is because, in the baseline scenario, the dose of Adenovirus was two orders of magnitude higher than the dose of Campylobacter. Given the same doses of Campylobacter and Adenovirus, the mean risks of illness would be within the same order of magnitude (5.6 and 2.19% for Campylobacter and Adenovirus, respectively).
The model is not very sensitive to the time spent swimming (duration of swim) because we used times of swimming from triathletes in past Commonwealth Games. These values are very similar because athletes finish events with milliseconds of each other. Duration of swim may become more important in a model where the participants have a wider distribution of time spent swimming. In addition, considering how sensitive the risk is to the concentration, we expect that the risk would be lowered considerably when assuming that only a small proportion of Adenovirus 40/41 (3%) was as infectious as Adenovirus type 4, compared to the assumption that all genomes were Adenovirus 4.
There is high uncertainty in the risk of illness caused by Adenovirus due to the variation on Adenovirus concentration in the water and the mismatch between the prevalence data of the pathogen of interest and the dose-response used to model the risk.
Scenario 2: Campylobacter in Wastewater
In the second scenario, Campylobacter was considered to be in the background river concentration (baseline scenario) as well as in the wastewater effluent (along with the Adenovirus in the wastewater effluent). Table 2 describes the changes in risk resulting from this inclusion. The mean risk increase significantly from 0.13% to 9.3% and the 95th percentile risk increased from 0.36% to 18.9%.
Unlike in our baseline risk scenario, the probability that infection becomes illness contributed significantly more to the variance in risk than did the concentration of Campylobacter in the river (Table 3). The concentration of Campylobacter in the river under Scenario 2 conditions is low relative to Campylobacter concentrations in the wastewater and this is the reason that river concentrations became less important in describing the variance in the risk estimate. As expected, the concentration of Campylobacter in the wastewater effluent played an important role in the variance of the calculated risk.
Rainfall Causing Urban Runoff
In scenario three, we simulated the change in risk as a result of a rainfall event that increases runoff to the river and therefore increases the concentration of Campylobacter in the river. In this case, the mean risk of Campylobacter illness increased to 14.3% and is within the same order of magnitude as risk of Adenovirus illness using a Rotavirus dose-response model (the most conservative model for Adenovirus risk based on the estimated mean risk). The variance described by the probability that infection becomes illness increases significantly from our Baseline Scenario to Scenario 3 and this is in parallel with an increase in Campylobacter doses. This suggests that a highly infectious strain of Campylobacter will be more important to risk under Scenario 3 than in the baseline scenario.
Comparison of the risks of Campylobacter illness from all the three scenarios is presented in Figure 3.
Table 2. Mean and 95th percentile risks for Scenarios 1, 2 and 3
|Hazard||Dose Response Model||Mean Risk (%)||95th Percentile Risk (%)||Best-Fit Distribution||Meets or Exceeds Tolerable Risk Level|
|C. jejuni for baseline scenario (Scenario 1)||C. jejuni||0.13||0.36||Lognormal||No guidelines|
|C. jejuni when wastewater C. jejuni flow is considered (Scenario 2)||C. jejuni||9.3||18.9||Beta||No guidelines|
|C. jejuni when wastewater C. jejuni flow is considered and rainfall causes urban runoff (Scenario 3)||C. jejuni||14.3||24.6||Beta||No guidelines|
|Adenovirus 40/41||Norovirus DR||2.25||12.9||Lognormal||Meets|
|Adenovirus 40/41||Adenovirus 4 DR||15.3||50.0||Gamma||Exceeds|
|Adenovirus 40/41||Rotavirus DR||26.6||41.4||Min Extreme||Exceeds|
|Hazard||Conc. in River||Conc. in WW||Conc. in Urban Runoff||Water Ingestion Rate Swimming||Probability that Infection Becomes Illness||Duration of Swim|
|C. jejuni for baseline scenario (Scenario 1)||47.8||N/A||N/A||27.5||22.0||2.4|
|C. jejuni when wastewater C. jejuni flow is considered (Scenario 2)||0.0||31.8||N/A||7.0||60.3||0.9|
|C. jejuni when wastewater C. jejuni flow is considered and rainfall causes urban runoff (Scenario 3)||0.0||0.0||2.8||3.7||93.0||0.5|
Table 3. Contribution to variance for parameters with uncertainty
Stakeholders with interest in the games include athletes, spectators, organizing groups including the Commonwealth Games Federation and Australian government, sponsors and local businesses, and citizens of Gold Coast.
Athletes will be most concerned with how water quality will affect their performance at the games. They may be more willing to accept a high risk of contracting an illness during the event itself, however they will be concerned with becoming ill immediately before the event.
Prior to the 2016 Olympic Games in Rio de Janeiro, there has been significant media concern regarding water quality and athlete safety during water events (Fistarol et al, 2015). The high profile of pollution at the Rio Olympics may raise stakeholder concerns about microbial risk at the Commonwealth Games. The Gold Coast is a popular tourist destination which attracts nearly 11 million visitors each year (Gold Coast Tourism Report). If athletes become ill, it may lead to the perception that the water quality at Gold Coast is low and damage the reputation of the Gold Coast and other Australian beaches. If the damage to the Coast’s reputation prevents tourists from visiting the coast, all stakeholders could be affected by a loss of revenue. Athletes, the Commonwealth Games Federation, and spectators may perceive Australia as an unsafe place to hold high profile swimming events.
Risk Communication Strategy
Risk communications strategies should be tailored to each stakeholder group. Athletes will need to be able to train in open water in advance of the games. However, athletes and coaching staff should understand how the total risk of illness increases with repeated exposure so they can manage their training strategies.
Microbial risks to swimmers are typically communicated via E. Coli or Enterococci concentrations reported as indicators of fecal contamination. The International Triathlon Union requires sampling for these parameters prior to holding a triathlon event; however, those data were not considered as part of this risk assessment. If the E. Coli or enterococci data suggest a high level of microbial contamination, stakeholders may perceive the risk to athletes as high, even if this risk assessment suggests that the actual risk is low. Risk communication plans may need to address any discrepancies between the two risk assessment methods. It will be especially important to monitor media and public perception of microbial risks to triathletes following the Rio Olympic Games in order to develop a communication strategy which will appropriately address stakeholder perception of risks and concerns.
A strong relationship with stakeholders is important for risk management and communication. Understanding the concerns of each stakeholder group is key to effective risk communication, and communication strategies should be developed with consideration to areas of concern for each stakeholder group. The goal of each communication strategy should be to ensure that the risk assessors are perceived as a trustworthy source of accurate, timely information.
The media are an important partner in risk communication. Media portrayals can have significant impacts on stakeholder perception of risk. In the days leading up to the Rio Olympics, multiple media articles have highlighted discrepancies between communications from the Brazilian government and the opinions of other “experts” (Block 2015). However, there is a lack of actual water quality data available to the public and stakeholders. In the time before the Commonwealth Games, it will be important to provide water quality data and important qualifying information to stakeholders and the media as a way to maintain transparency and build trust. A well-designed webpage with information for a variety of stakeholders and up-to- date water quality data would be an effective tool for communicating risk to many groups, and help to control the narrative around those risks.
The incubation period for Campylobacter is typically between 2-4 days (CDC). The incubation period for adenovirus serotype 40 is longer, typically from 3 to 10 days, with symptoms lasting for approximately ten days (Adenovirus). Therefore, one strategy for mitigation of risk to athletes is to recommend that athletes do not enter the water at the Gold Coast for at least 20 days prior to the event. If athletes do choose to train at the Gold Coast, it is recommended that they do not enter the water for three days after any heavy rainfall event (Recreational Water Quality). It is important to ensure that coaches and athletes are aware of the risk incurred by swimming in these water systems. Additionally, it can be recommended to athletes that they try to minimize swallowing during swims.
The wastewater treatment plant upstream of the Broadwater is assumed to be the major source of adenovirus in the estuary. The wastewater treatment plant effluent is used for irrigation and dust suppression, or is otherwise stored in lagoons. Excess effluent is discharged to the ocean. It is recommended that organizers work with wastewater treatment plant personnel to minimize discharge from the treatment plant to the estuary in the days leading up to the triathlon event. For instance, a greater percentage of water could be recycled, or the level in the storage lagoons could be temporarily raised.
Additionally, it is recommended that organizers raise public awareness of the importance of keeping beaches clean prior to the events. Organizers should ensure that adequate sanitation facilities are provided for the influx of domestic and international visitors that will be attending the games. Signage should be provided in multiple languages or include illustrations to facilitate the appropriate usage of sanitation facilities in the beach area.
Prior to the games, the City of Gold Coast can take efforts to reduce the quantity or improve the quality of runoff into the Broadwater. Rain gardens, vegetative buffer zones, or treatment wetlands could be employed in areas near the water to intercept and treat runoff before it enters the estuary. Organizers should investigate whether selection of a secondary site would be a viable option in the case of an event which causes significant microbial contamination, such as a major malfunction of the wastewater treatment plant or very significant flooding.
Organizers should also investigate whether delaying the swims after such an event would be a viable option to prevent athlete illness. A conceptual decision tree has been developed as an initial framework for both of these risk mitigation strategies. However, in order to quantitatively evaluate such an option, additional studies will be required.
Figure 4: Conceptual Decision Tree for Risk Mitigation Trade-Offs
Many athletes will arrive at Gold Coast early in order to acclimate to the area and train. This quantitative risk assessment has focused on the probability of an athlete contracting an illness during a single 1.5 kilometer swim. If an athlete chooses to swim multiple times in the days preceding the race, the total risk of illness prior to the games will increase according to the following equation:
Total Risk = 1- prob(1 - SwimRisk)N
Where SwimRisk is the probability of contracting an illness during a single swim as calculated in this risk assessment, and N is the total number of 1.5 kilometer swims taken by the athlete. Future studies should be performed to evaluate the risk to triathletes during training prior to the event and provide additional recommendations to mitigate those risks.
Finally, this risk assessment found that rainfall has a significant effect on the risk of illness to athletes. It is recommended that additional studies to investigate and quantify this risk be performed to better understand the hazard that is presented to athletes by swimming in floodwaters, and provide more specific guidance to mitigate those hazards.