Boil Water Advisory

Campylobacter

Campylobacter is considered the most important bacterial agent in waterborne diseases in many European countries (Stenström et al. [4]; Furtado et al. [5]) - a large number of outbreaks of Campylobacter have been reported in Sweden for example, involving over 6000 individuals (Furtado et al. [5]). Issues with Campylobacter in drinking waters tend to be a post-treatment problem, due to situations such as broken sewers. Campylobacter spp. have been isolated from rivers (Arvantidou et al. [6]; Obiri-Danso and Jones [7]), lakes (Arvantidou et al. [6]), groundwater (Savill et al. [8]) as well as drinking water (Alary and Nadeau [9]; Savill et al. [8]; Vogt et al. [10]). The occurrence of the organisms in surface waters has also proved to be strongly dependent on rainfall, water temperature and the presence of waterfowl (WHO [1]).

Cryptosporidium

Cryptosporidium has been well described as one of the most important pathogens in waterborne outbreaks in many different countries (Proctor et al.1998; MacKenzie et al. 1995). The most important one is the well known outbreak in Milwakee (USA) in 1993, where more than 400,000 people were infected. Cryptosporidium is a protozoan parasite which can infect both animals and humans. The parasite infects the epithelial cells of the digestive and respiratory tract causing diarrhoea. The incubation period is around 7 days (range 1-14 days) and the disease is often self-limiting with a duration of 6-9 days. Longer duration is often found in AIDS patient where the effect of the infection can be lethal (WHO [11]). All age group can be infected. Infection can happen through consumption of infected water or food, from animals and humans can give the infection to other humans (WHO [11]; Fayer [12]). After ingestion of an oocyte four sporozoites are released and infect the epithelial cells. Cryptosporidium only survives outside a host in the oocyst state where it can live month in cold water (Fayer [12]).

Exposure Assessment

For the modelling, due to the given time, the next assumptions are made:

• Mixing between treated and untreated water is not considered: the model is for a worst case event
• Boiling for the full 3 minutes is effective in killing C. jejuni and Cryptosporidium (Morbitity and Mortality Weekly Report (MMWR) [13]).
• All isolates have equal chance of causing illness
• All inhabitants are equally exposed
• Concentrations of reference pathogens were sourced from literature
• The only route of exposure is drinking water.

The purpose of the exposure assessment is to determine the amount, or numbers of organisms that correspond to a single exposure (dose) or the total amount or number of organisms that constitute a set of exposures (Haas et al. [14]). Amongst three main routes of exposure (ingestion, inhalation, and dermal absorption), only the ingestion route was taken into consideration in this study as faecal-oral transmission is considered the main route of spreading of enteric diseases. The volume of tap water ingested via drinking is much higher than that for bathing, brushing teeth or preparing food.

Volume ingested via drinking

In this study, we used data published Teunis et al. [15]. Based on the data from Netherlands, the human daily consumption of unboiled drinking water is an average 0.27 liter, and the 95%-interval of the consumption is 0.017 – 1.27 liter.

Fraction of population following the boil water notice

In this study, it was assumed that 100 % of the population will know about the boil water advisory after 24 hours from the beginning of the contamination. Furthermore, it was assumed that 80 % of the population will comply with the advisory and 20 % will not. This assumption was based on the data presented by Karagiannis et al. [16] were 81.8% of population responded to comply with the advisory after a faecal contamination detected by the presence of E. coli in tap water in Netherlands. Most respondents still used unboiled water to brush teeth, wash salads and fruits; a result that is supported by the findings from UK reported by O’Donnell et al., [17] where only 38 % of the population followed the given advices precisely. However, as there was no data available of the tap water volumes ingested through these activities, there were excluded from the assessment.

Dose Response

The risk of infection was calculated using @RISK, applying 10,000 iterations in the Monto Carlo simulations. The modelling is done for C. jejuni and Cryptosporidium. The dose-response models used were β-Poisson model for rotavirus and EHEC infections and the exponential model for helminth ova (Haas et al. [14]). The equations are:

(a) β-Poisson dose-response model

\begin{align*} P_{inf} = 1 - [1+(d/{ID}_{50})(2^{\frac{1}{\alpha}}-1)]^{-a} \end{align*}

where P_inf is the risk of infection of an individual exposed to a single pathogen dose d

ID₅₀ is the median infective dose

α is a pathogen “infectivity constant”

5.1 Campylobacter

In table 1 the model input for campylobacter is given.

Table 1. Model input for campylobacter

5.2 Cryptosporidium

In table 2 the input for cryptosporidium is given.

Table 2. Model input for cryptosporidium

Risk Characterization

The risk of infection was calculated using @RISK and the results are presented as probability of infection (P_inf) per exposure and number of infected people in the studied population.

6.1 Campylobacter

In the graph, figure 1, the results are shown.

Figure 1. Modelling of campylobacteriosis cases

The results are also shown in table 3.

Table 3. Campylobacteriosis case results

From the table and the graph is shows that probability of infection, given the consumption of contaminated water is 0.258. For the reactive scenario it means that 8,527 people, out of 100,000 people, get infected. In the proactive scenario the number of people is lower, 1,705 people will get infected from drinking the contaminated water.

6.2 Cryptosporidium

In the graph, figure 2, the results are shown.

Figure 2. Modelling cryptosporidium cases

The results are also shown in table 4.

Table 4. Cryptosporidium case results