Guidance on use of rainwater tanks

Microbial hazards

Page last updated: March 2011

Rainwater collected and stored in domestic tanks will contain a range of microorganisms from one or more sources. While most will be harmless, the safety of rainwater will depend on excluding or minimising the presence of enteric pathogens. Enteric pathogens include types of bacteria, viruses and protozoa. These organisms do not grow or survive indefinitely in water environments and are introduced into drinking water supplies by contamination with faecal material.

Tank rainwater can contain organisms referred to as opportunistic pathogens such as Aeromonas spp and Pseudomonas aeruginosa (Sinclair et al. 2005). Except for the severely immuno-compromised these organisms are not considered to represent a significant risk through normal uses of drinking water supplies (WHO 2008).

Most domestic rainwater tanks are installed above ground and collect run-off from roofs via guttering. Likely sources of enteric pathogens include:

  • faecal material (droppings) deposited by birds, lizards, mice, rats, possums and other animals
  • dead animals and insects, either in gutters or in the tank itself.
Less commonly, rainwater is collected in underground tanks. If these tanks are not fully sealed or protected against ground run-off, then microorganisms associated with human and animal excreta may also contaminate stored rainwater.

More recently new houses being built in Australia have rainwater tanks fed by buried pipe work (so-called ‘wet systems’) in order to maximise yield while minimising the number of separate tanks. This also reduces the need for house walls to have ‘unsightly’ sloping, external downpipes. However, there are risks from this design approach.

Rainwater systems incorporating buried pipe can be susceptible to inadvertent cross-connections or contamination from external sources such as septic systems. In South Australia one outbreak of gastroenteritis was attributed to buried rainwater pipework being installed in the same trench as pipework to a septic system (SA Health, unpublished).

Microbial quality of drinking water is commonly measured by testing for Escherichia coli (E. coli), or alternatively thermotolerant coliforms (sometimes referred to by the less accurate term faecal coliforms), as indicators of faecal contamination and the possible presence of enteric pathogens. In the past, the broader total coliform group has also been used for this purpose. However, this group includes non-pathogenic organisms that can grow in water environments and be present in the absence of faecal contamination. Accordingly, total coliforms are no longer recognised as being a suitable indicator of faecal contamination or having health significance (ADWG).

Thermotolerant coliforms or E. coli have been commonly identified in domestic tanks (Fuller et al. 1981; Dillala & Zolan 1985; Fujioka & Chin 1987; Haeber & Waller 1987; Wirojanagud 1987; Gee 1993; Edwards 1994, Thurman 1995; Victorian Department of Natural Resources and Environment 1997; Simmons et al. 2001; Sinclair et al. 2005; Chapman et al. 2006 and 2008; Evans et al. 2007; Abbott et al. 2007; Rodrigo et al. 2009). This implies that enteric pathogens could often be present in rainwater tanks.

However, when surveys have included testing for specific pathogens, detection has been relatively infrequent (Sinclair et al. 2005; Chapman et al. 2008; Rodrigo et al. 2009). In Australia Salmonella and Campylobacter have been detected in small numbers of samples (Sinclair et al. 2005; Chapman et al. 2006; Rodrigo et al. 2009). Atypical enteropathogenic E. coli were detected in a survey of metropolitan Adelaide tanks (Rodrigo et al. 2009). In contrast, studies using PCR-based analyses detected relatively high frequencies of Salmonella, Campylobacter and Giardia in rainwater tanks in Queensland (Ahmed et al. 2008, 2009). A quantitative microbial risk assessment based on these analyses calculated risks of disease that are higher than the currently reported incidence of disease in Queensland (Ahmed et al. 2009). As indicated by the authors, further investigations into the significance of these results are required.

In New Zealand, Campylobacter was identified in nine of 24 tanks, but the maximum concentrations were less than one organism per 100 mL (based on average concentrations) and it was concluded that the risk of illness from drinking this water was low (Savill et al. 2001). A second New Zealand survey found faecal coliforms in 70 of 125 rainwater tanks but Salmonella was only detected in one tank. Cryptosporidium oocysts of unknown species were detected in two of 50 tanks that contained at least 30 faecal coliforms or 60 enterococci per 100 mL (Simmons et al. 2001). Campylobacter or Giardia was not detected in any tanks. Similarly Wirojanagud (1987) reported the detection of faecal coliforms in 43 of 156 samples from rainwater tanks in Thailand, but Salmonella was only detected in one sample and Shigella in none.

An exception to this trend was the detection of Cryptosporidium and Giardia in 400 L samples, collected from a large number of rainwater cisterns in the Virgin Islands, where installation of rainwater storage is compulsory due to a lack of fresh water resources (Crabtree et al. 1996). The health significance of this finding was not established.

Top of Page

Legionella

There has been increasing interest in potential risks associated with Legionella in rainwater tanks. Legionella causes disease through inhalation and not through drinking contaminated water. Rainwater tanks have been proposed as potential sources of organisms that could be amplified in hot water systems in buildings (Chapman et al. 2008).

Legionella is a common environmental organism which survives and grows in sludges and slimes. Risks of waterborne legionellosis (Legionnaires’ disease and Pontiac Fever) are typically associated with amplification in water between 25íC and 50íC (WHO 2007). Above ground rainwater tanks have been identified as a potential source of Legionella because they tend to accumulate sludges and in Australian summers are likely to contain water between 25íC and 50íC. Legionella has been detected by culture or PCR in rainwater tanks (Sinclair et al. 2005; Chapman et al. 2008; Ahmed et al. 2009).

In New Zealand there was a reported outbreak of Legionnaires’ disease that may have been associated with rainwater fed hot water tanks (Simmons et al. 2008). Using sequence-based typing, it was concluded that aerosols from a nearby marina water blaster was a likely source of contamination of nearby rooftops and rainwater tanks. Nearly two-thirds of the rainwater tanks investigated had never been cleaned, providing environments to support survival and growth.

While rainwater tanks can provide environments for Legionella, they are common environmental organisms. Infection normally follows amplification in warm water and dissemination in aerosols. There is little evidence that rainwater tank supplies are associated with increased public health risk. Tanks should be kept clean and the rainwater, when used in hot water systems, stored and delivered to reduce the likelihood of the growth of Legionella bacteria, and also reduce the likelihood of burns and scalds.

Illness and rainwater tanks

The relatively frequent detection of faecal indicator bacteria is not surprising, given that roof catchments and guttering are subject to contamination by bird and small animal droppings. However, despite the prevalence of indicator organisms, reports of illness associated with rainwater tanks are relatively infrequent. Although traditional under-reporting of gastrointestinal illness will contribute to a lack of evidence, epidemiological investigations undertaken in South Australia have failed to identify links between rainwater tanks and illness (Heyworth et al. 1999; Heyworth 2001; Rodrigo et al. 2010).

Investigations in the late 1990s compared rates of gastrointestinal illness in South Australian children who drank rainwater collected in domestic tanks, compared to those who drank filtered and disinfected mains water that was fully compliant with the ADWG. Overall, the investigations found no measurable increase in illness associated with drinking rainwater. In the first part of the investigation, a questionnaire was supplied to the parents of about 9500 children undertaking a general health check before enrolling at school for the first time. There was a slightly higher, but non-significant, incidence of gastrointestinal illness reported for rural children who drank rainwater rather than mains water (Heyworth et al. 1999).

There was no difference in rates of illness between children drinking rainwater or mains water in urban areas.

The second part of the investigation expanded on the results through use of a diary with parents of about 1000 rural children. The results were reversed and the study found a small but significant decrease in illness associated with consumption of rainwater compared to mains water (Heyworth 2001). The investigations included questions about rainwater tank maintenance. As found in most surveys, maintenance was generally poor.

A double-blind controlled study was conducted in 2007-2008 in South Australia to compare rates of gastrointestinal illness between filtered and unfiltered rainwater (Rodrigo et al. 2010). Three hundred households that drank untreated rainwater were selected for the study from metropolitan Adelaide and adjacent hills suburbs with a total of 1352 participants. All households included at least two children. Half of the households had an active water filter fitted and the balance had an inactive filter. Householders recorded incidence of illness weekly with the exception of 5 weeks over the Christmas period. Two hundred and seven households completed the study with similar dropout rates for both groups.

The overall trend was that the inactive filter group had a slightly lower rate of illness. The study found that untreated rainwater does not contribute significantly to community gastrointestinal illness for either adults or children (Rodrigo et al. 2010).

Water quality testing was also performed and 30% of 974 samples were positive for E. coli with levels ranging from 1 to 2400 cfu/100 mL. There was no relationship between water quality and illness.

Top of Page

There have been a few reports of illness associated with Campylobacter and Salmonella in rainwater tanks. In four of these reports the contaminating organisms were detected in both those infected and their rainwater sources (Koplan et al. 1978; Brodribb et al. 1995; Simmons & Smith 1997; Taylor et al. 2000). Brodribb et al. (1995) reported on an investigation into recurrent infections of an elderly immuno-compromised woman by Campylobacter fetus, where the organism was also isolated from the patient’s rainwater tank, which served as her sole source of drinking water. No further infections occurred after the patient started boiling the tank water before consumption. It was postulated that an outbreak of 23 cases of campylobacteriosis at a Queensland island resort was probably associated with contamination of rainwater tanks, even though Campylobacter was not isolated from rainwater samples (Merritt et al. 1999). A study of risk factors for campylobacteriosis in New Zealand associated consumption of rainwater with increased risk in a small number of cases (23 cases, 11 controls; odds ratio 2.2) (Eberhart-Phillips et al. 1997).

An investigation of an outbreak of Salmonella infections in a church group in Trinidad, Jamaica led to detection of the organism in rainwater samples and in food prepared using the rainwater (Koplan et al. 1978). It was reported that the roof catchment was covered with dried and fresh bird droppings. Similarly, Salmonella was isolated from a rainwater tank used by a family of four in New Zealand that had suffered from recurrent infections by the same organism (Simmons & Smith 1997). In an investigation of 28 cases of gastroenteritis among 200 workers at a construction site in Queensland, Salmonella saintpaul was isolated from both the cases and rainwater samples (Taylor et al. 2000). Animal access was suggested as being the source of contamination with several live frogs being found in one of the suspect tanks.

An underground rainwater tank was associated with the only drinking water borne outbreak of cryptosporidiosis/giardiasis recorded in Australia to date (Lester 1992). Eighty-nine people supplied with drinking water from the tank became ill. Investigations revealed the tank had been contaminated by an overflow from a septic tank.

Two explanations have been suggested for the apparent disparity in frequency of faecal contamination and the prevalence of illness. The first is the likely source of contamination. For most rainwater tanks, particularly those installed above ground, faecal contamination is limited to small animals and birds. While faecal contamination from these sources can include enteric pathogens, there is a degree of host group pathogen specificity. Enteric viruses are the most specific; in general, human infectious species only infect humans and animal (non-human) infectious species only infect animals.

The host range for protozoa is a little broader, but except for the severely immuno-compromised, human infections with Cryptosporidium are predominantly associated with genotypes of C. hominis carried by humans and C. parvum carried by livestock (McLauchlin et al. 2000; Chappell & Okhuysen 2002, Hunter & Thompson 2005, Xiao & Fayer 2008). The livestock genotype can be transmitted to some other animals, but the human genotype is very specific for humans. C. meleagridis carried by birds has been associated with low numbers of cases (McLauchlin et al. 2000; Pedraza-Diaz et al. 2001). The evidence for the zoonotic transmission of Giardia is limited (Hunter & Thompson 2005, Xiao & Fayer 2008).

Bacterial pathogens are the least specific and birds, for example, are known to carry and excrete potentially human infectious Campylobacter (Koenrad et al. 1997; Whelan et al. 1983). Birds that live in close proximity to human populations can also transport Salmonella (Hernandez et al. 2003; Refsum et al. 2002).

These limitations are important, as enteric illness mediated by waterborne bacteria requires ingestion of much higher numbers of organisms than enteric illness mediated by protozoa or viruses. Dosing studies have found that while ingestion of between one and 10 virus particles or protozoan cysts can lead to infection, at least 1000 and often more than 100,000 bacteria are required (Haas 1983; Regli et al. 1991; Gerba et al. 1996; Okhuysen et al. 1999).

The second explanation is that ongoing exposure to organisms in rainwater could result in increased immunity (Heyworth 2001, Rodrigo et al. 2010). The protective effect of acquired immunity can last from months to years and for such partially immune people it is possible that larger doses of contaminating organisms may be required to trigger illness (Eisenberg et al. 2001).

In summary, the study conducted in South Australia found no measurable difference in rates of gastrointestinal illness in children who drank rainwater compared to those who drank mains water. However, there are examples of Campylobacter and Salmonella enteritis associated with rainwater tanks and one example of cryptosporidiosis/giardiasis associated with an underground tank. Faults in tank design or poor maintenance were identified as contributory factors in some of the investigations of illness.

Dead animals

Entry by small animals and birds to rainwater tanks can lead to direct faecal contamination, even if the animals escape from the tank. In some cases, animals become trapped in tanks and drown, leading to very high levels of contamination. In the case of larger animals, such as possums and cats, this will almost certainly have a distinctive impact on the taste and odour of the water.

Top of Page

Mosquitoes

Rainwater tanks can provide excellent habitats for mosquito breeding. In addition to causing nuisance, certain types of mosquito can be vectors of arboviruses.

Of particular concern are species of mosquito that can be vectors for dengue virus, which occurs in tropical and subtropical areas of the world. Rainwater tanks have been identified as potential breeding sites for vectors of dengue virus and the World Health Organization (WHO) recommends all tanks have screens or other devices to prevent adult mosquitoes from emerging (WHO 1997).

In Queensland it has long been suggested that rainwater tanks are associated with breeding of the mosquito Aedes aegypti, the primary vector of dengue virus (Kay et al. 1984). This was confirmed in an outbreak of dengue in the Torres Strait Islands in 1996-1997 (Hanna et al. 1998). In addition, a survey conducted in the Torres Strait Islands in 2002 detected adult mosquitoes, including Aedes aegypti, in rainwater tanks with missing or faulty insect screens (Ritchie et al. 2002).

Both the Northern Territory and Queensland have regulations relating to prevention of mosquito breeding in rainwater tanks (Northern Territory 1998; Queensland 2005).

Other mosquitoes that breed in rainwater tanks may also be vectors of arbovirus infections. For example, Aedes notoscriptus could be a vector of Ross River and Barmah Forest viruses (Doggett & Russell 1997). This species is widespread in Australia.