New Research Project

This project is being undertaken by CRCWQT partners Monash University, the Dept. of Human Services VIC, NCEPH and Melbourne Water Corporation.

In recent years the protozoal parasite Cryptosporidium parvum has emerged as an important cause of waterborne disease outbreaks, however little is known about the causes of sporadic disease due to this organism. Low numbers of Cryptosporidium oocysts are commonly detected in many surface water supplies, but it is not known if this low concentration of oocysts can cause disease. Part of the uncertainty relates to current detection methods being insensitive and not distinguishing between viable and non-viable oocysts, or different species of Cryptosporidium (with only C. parvum being known to cause human disease).

This preliminary study in Melbourne will establish the necessary groundwork and methodology for a full scale study to be undertaken in Adelaide and Melbourne. The first component of the study will involve validation of a questionnaire to assess water consumption. Following the validation, a Pilot study will be conducted to test and refine the case and control finding exercise, the risk factor questionnaire and other procedures for the main study.

Eligible cases will be people with cryptosporidiosis identified from pathology laboratory reports to the Dept. of Human Services VIC. These people will be contacted by research nurses and asked to complete a telephone interview covering exposure to suspected risk factors in the two weeks before their illness developed. Controls will be people without diarrhoeal illness selected by random digit dialling and matched by age and sex to the cases. The same questionnaire will be adminstered to controls covering an equivalent time period as the respective matched case.

The risk factor exposures assessed will include the following:

• Type and quantity of water consumed
• Food and beverage exposure
• Travel
• Recreational water activities
• Gardening activities
• Contact with other people with gastroenteritis
• Domestic and farm animal contact

The Pilot study will run until 20 cases and 80 controls have been recruited, which is expected to take approximately three months.

The Pilot phase of the study will be completed in June 1998, and the full scale study is planned to commence shortly thereafter in Melbourne and Adelaide. These two cities have been chosen to represent the opposite ends of the water quality and treatment spectrum of Australian metropolitan water supplies; Melbourne - high quality source water with minimal treatment (chlorination only), Adelaide - lower quality source water with full conventional water treatment (coagulation, sedimentation, filtration and chlorination).

The Melbourne arm of the study will be funded by the CRCWQT, WSAA, Melbourne Water, City West Water, South East Water and Yarra Valley Water. Funding discussions for the Adelaide arm are expected to be finalised soon.

DBPs and Pregnancy Outcomes

Two recent publications in the journal Epidemiology (1,2) have raised public concerns in the USA, about the possible effect of disinfection byproducts (DBPs) on pregnancy outcomes. Both papers, by the same group of researchers in the California Department of Health Services, described a study of the rates of miscarriage among 5,144 pregnant women living in 3 areas of California. The story was featured in several leading US newspapers and also in the Sydney Morning Herald on 12th February.

One paper examined miscarriage (defined as pregnancy loss at 20 weeks of gestation or less) in relation to the type and amount of drinking water consumed, while the second investigated the association between miscarriage rates and trihalomethane exposure. The three geographic areas under study were not identified but were chosen to represent three types of water supply - mixed surface and ground waters (Region I), surface water (Region II) and ground water (Region III).

In the first paper, the authors reported that in one of the three areas (Region I), women who drank 6 or more glasses of cold tapwater had almost double the risk of miscarriage compared to women who drank no tapwater. In the other two regions (Regions II and III) no association was observed between miscarriage risk and tap water consumption.

The second paper decribes a different set of analyses carried out on the same data. This time the personal exposure of each woman to total trihalomethanes (TTHM) was estimated, and groups of women with different exposure levels were examined in terms of miscarriage risks. For some women it was also possible to estimate exposure to individual trihalomethanes from water utility records. The authors reported a “modest” association between consumption of tapwater containing THMs, with the increase in risk seeming to start at TTHM levels around 75 micrograms /L.

High personal exposure to TTHMs was defined as a high intake of cold tapwater (5 or more glasses per day) with high TTHM level (more than 75 micrograms /L). For women in this category, the risk of miscarriage was about 1.8 times that for women drinking less water with high TTHM levels, or water with lower TTHM levels. Only high individual exposure to bromodichloromethane showed an association with miscarriage risk, when individual THMs were analysed separately.

The authors have pointed out that the results of the studies must be regarded as preliminary, and have advised women not to restrict their fluid intake or switch drinking water sources on the basis of these reports. They have also noted that only 2% of the women in the study fell into the “high personal exposure” category.

What are the limitations of these studies?

The methodology used for these studies represents a significant advance on that used in previous publications which have examined DBPs and pregnancy outcomes. Past studies have used retrospective methods where water “exposure” was generally estimated from maternal address at time of birth, with no assessment of actual water consumption, changes in residence during pregnancy or important factors such as smoking, alcohol consumption or socioeconomic status. The outcomes studied have seldom included miscarriage due to the difficulties in retrospective confirmation of cases.

The new studies have used a prospective design where women were recruited early in pregnancy through a large managed health care organisation, with the eventual pregnancy outcome verified from medical records. Consumption of different types of water was estimated from individual telephone interviews shortly after enrolment in the study, as were a number of other factors which may influence risks of miscarriage.

However the assessment of exposure to DBPs in drinking water was not based on household tap measurements, but on routine distribution samples taken by water utilities. For most women (77%), this was done by averaging all TTHM readings for the relevant water utility that fell within the first trimester of pregnancy. For the remaining 23% of women no measurements within the first trimester were available, so readings within 30 days of the first trimester or annual average TTHM reading from the annual reports of water utilities were used. Thus the estimates of TTHM exposure would not have taken into account any localised or short term variability in DBP levels.

The authors state that the interview covered relevant factors including demographics, previous pregnancy history, employment status, alcohol, tobacco, caffeine and “other factors”. In the data analysis several variables were identified as independent risk factors - these were gestational age at interview, maternal age, smoking, history of miscarriage, maternal race and employment during pregnancy. It is not clear whether maternal diet was assessed, or why two different cut offs for water intake were used in the two papers (6 glasses /day in one paper, 5 glasses /day in the other).

There are a number of observations within the two studies that argue against the idea that DBPs are responsible for the observed differences in miscarriage risks. In the first paper, when women in Region I who drank tapwater were divided into those who drank straight from the tap and those who let water stand before drinking it, the risk appeared higher in the latter group (although the number of women was small). This does not support the idea that risk of miscarriage is related to volatile compounds such as trihalomethanes (THMs) in drinking water.

In the second paper, it is stated that swimming and prolonged showering did not appear to influence miscarriage risks. Previously published studies have shown that dermal and inhalation absorption of trihalomethanes (and perhaps other volatile DBPs) during showering may be as great as drinking water intake. This observation is also at odds with the interpretation that THMs are associated with miscarriage risks.

Indeed, the authors note that the effect on miscarriage is seen in women in Region I who have low levels of TTHMs in their drinking water. In a previous retrospective study in this region the strongest effect was seen in areas served by unchlorinated groundwater. A Reuter’s news agency report quoted the authors as saying they believe a causal agent is present in Region I tapwater, but the effect on miscarriage risks “cannot be explained by exposure to chlorination byproducts, because the association is seen in the absence of high levels of these chemicals”.

Thus it appears that there are a number of discrepancies in these observations, and further research is needed to explore this issue. In particular it would seem desirable to improve exposure assessment by carrying out water sampling at individual households, or at least in the immediate neighbourhood at more frequent intervals.

1 Swan SH, Waller K, Hopkins B, et al. A prospective study of spontaneous abortion: relationto amount and source of drinking water consumed in early pregnancy. Epidemiology (1998) 9 p126-133.
2 Waller K, Swan SH, DeLorenze G and Hopkins B. Trihalomathanes in drinking water and spontaneous abortion. Epidemiology (1998) 9 p134-140.

EPA Review Finds Philadelphia Turbidity Study Seriously Flawed

The turbidity levels examined in this study were in the range of 0.14 to 0.22 NTU - considerably below the levels of many water supplies in the United States and other developed nations. Therefore the results presented in the paper had major implications for water treatment practices for water authorities around the world. Publication of the paper provoked considerable controversy in the US, not only because of the content, but also because the authors had not followed customary procedures for consultation and internal EPA peer review.

A subsequent expert review of the paper by US EPA personnel has found “a myriad of serious concerns about a number of the major elements of the article, including the research design, methodology, and analysis”. We report here on the findings of the reviewers.

The turbidity data was obtained from the Philadelphia Water Department (PWD) by an EPA employee (the second author on the paper) and the research was partly supported by EPA funds. In keeping with usual EPA procedures, the PWD expected to have an opportunity to consult with the researchers prior to publication of the study but no such consultation was undertaken by the authors. In addition, it is the common practice of the EPA to subject papers to internal peer review prior to submission to journals - again the authors did not follow this procedure.

In the light of these facts, and the perception by some readers that the paper may represent the EPA’s position, a review by EPA experts in epidemiology, turbidity measurement, drinking water treatment, public health, and statistics was undertaken. The PWD was also invited to comment on the published paper. The major findings of the review are outlined below:

Hypothesis

The stated purpose of the paper was to investigate how well turbidity serves as an indicator of microbiological integrity of water. However it is then assumed (without validity) that turbidity is a direct surrogate of microbiological content, and the authors (in the words of the reviewers) “try in myriad ways to relate this surrogate to daily hospital use for enteric diseases which may or may not have been contracted through drinking water”.

Turbidity measurements

• The mean daily turbidity measurements used in the study were in the range 0.14 to 0.22 NTU. The PWD commented that the turbidity meters used at the time of the study were calibrated at approximately 4 month intervals using standards from 0.2 to 1.0 NTU. 75% of the readings used in the analysis were below the calibration range and thus cannot be considered reliable.
• According to the manufacturer the “stray light characteristic” of this model of turbidity meter is 0.04 NTU under ideal conditions. This effectively represents the sensitivity limit of the instrument - samples with readings 0.04 NTU difference may in fact have the same turbidity.
• One of the 3 water treatment plants routinely rounded turbidity readings to the nearest 0.05 NTU, while the others reported to the nearest 0.01 NTU.
• A simple average of the 3 plants was used calculate average turbidity for the city despite the fact that the 3 plants supplied different amounts of water.
• No consideration was given to the effect of chemical corrosion inhibitors on turbidity.
• Although the authors were supplied with minimum and maximum turbidity readings as well as mean readings, these were apparently not analysed. If the basic hypothesis is correct, one would expect stronger effects to be seen with maximum readings or turbidity spikes.

Disease Classification

The authors used the well recognised International Classification of Diseases (ICD) codes (9th revision) to classify the reason for hospital attendance. These codes are used to characterise primary and secondary diagnoses.

• The definition of primary gastroenteritis cases included codes for non-infectious gastroenteritis, and codes for specific microorganisms such as Salmonella, Shigella and Vibrio which are readily killed by chlorine. The reviewers feel that these should not have been included.
• The authors also included cases with secondary diagnosis of gastroenteritis if they judged that the primary diagnosis might be associated with gastrointestinal disease. However the primary codes in this category included dermatitis, eczema, and gastrointestinal symptoms (as distinct from gastrointestinal diagnoses). Again the reviewers felt this was inappropriate.

Miscellaneous Comments

• No consideration was given to the time taken for water to travel from the treatment plant to consumers when presumed “incubation times” were examined.
• While relevant regulations required disinfection treatment adequate to achieve a 3 log reduction in viruses, the PWD in fact carried out a much higher degree of disinfection. The estimated removal of viruses ranged between 35 logs in winter and 200 logs in summer. The PWD commented that it is difficult to see how minor changes in turbidity could produce detectable changes in the rates of viral illness (postulated by the author to be responsible for the association with illness in older children), given the high removal levels.
• No information was given on residual chlorine or chloramine levels in the system.
• No reason was given for the age categories used in data analysis (0-2 and 3-15 years).

Data Analysis and Methodology

• No consideration was given to the actual movement of water through the distribution system, instead a simplistic assumption of “exposure” was made based on residence in the supply area for each plant.
• PWD commented that it was not asked to supply ZIP code information for plant service areas, and it is not clear how the authors assigned the water source to gastroenteritis cases. PWD also stated that many areas of the city receive blended waters from more than one plant, and that these areas change over time.
• The lag time from ingestion to development of symptoms is highly variable even for a single pathogen (the reviewers cite examples of 12-50 hours for enteric viruses, 2-25 days for giardia, 2-4 weeks for amoeba). Therefore even if the authors were correct in the hypothesis that increases in turbidity equate directly to increased exposure of the population to pathogenic microorganisms, it is difficult to imagine how distinct temporal peaks in illness could be produced.
• Some of the time lags considered (1 or 2 days) were biologically implausible given the time for water to travel through the distribution system plus the minimum incubation time of potential pathogens.
• The 3 plant service areas had strikingly different rates of hospital visits, which the authors attributed to the distances from the hospital. The area with most visits (the one surrounding the hospital) contributed 59% of cases but showed no association of illness rates with turbidity rises.
• The apparent susceptibility patterns found by the authors - a time lag of 4 days (typical of viral agents) for older children, and 14 days (typical of protozoa) for young children are not in agreement with current biological and clinical knowledge.
• If the associations found at 4 and 10 days were due to different classes of pathogens (as suggested by the authors) then they should be independent of each other, however from the results presented in the paper this was not the case.
• From the description of data analysis in the paper, the reviewers concluded that more than 84 models were tested on the data set. The authors made no comment on the likelihood that a statistically significant association could be generated purely chance under these conditions.
• Some findings quoted as supporting the association between turbidity changes and illness have 95% confidence intervals spanning 1.0 - which indicates they were not statistically significant.
• No explanation is given for the final selection of lag times from the many models used.

Based on their evaluation of the paper, the reviewers posed 16 questions for the authors. The EPA has also requested that the authors supply a copy of the data set and the complete analyses that were derived from it, so that the concerns of the reviewers can further investigated.

(1) Drinking water turbidity and pediatric hospital use for gastrointestinal illness in Philadelphia. Schwartz J, Levin R and Hodge K (1997) Epidemiology 8 (6) p615-620.

Comment
The findings of the reviewers closely parallel the questions raised by Health Stream in our assessment of the paper (see Issue 8). We believe that the information regarding the limitations of the turbidity measurements (supplied to the reviewers by the Philadelphia Water Department) is sufficient in itself to invalidate the study. The other major problems with study design and analysis would also be, of themselves, sufficient to call the conclusions into serious doubt.

It is disturbing that the authors did not check the precision of the turbidity data with PWD prior to embarking on their analysis, and that the serious methodological flaws in a paper with such significant implications for the water industry in the US and elsewhere were not detected by peer scrutiny before publication.

Canberra Swimming Pool Associated Cryptosporidiosis Outbreak

Introduction In early January 1998, the Communicable Diseases Control Program of the ACT Health Department was notified by a concerned private pathology laboratory of 9 faecal specimens which were positive for oocysts of Cryptosporidium parvum.

Cryptosporidiosis was not notifiable in the ACT at that time and therefore reporting was not mandatory. Telephone interview of the 9 initial cases found no common exposure. On the 28 January, the National Centre for Epidemiology and Population Health (NCEPH) was contacted and invited to assist with the investigation of the outbreak.

A review of the 41 cases reported to 30 January found that 32/41(78%) of cases were aged under 15 years, and 9/41 (22%) were aged between 15 and 40 years. Most cases were resident in a southern area of Canberra. Completed interviews were available on 36 of 41 cases. Of these, 20/36 (55%) reported swimming in a swimming pool in the two weeks prior to onset of symptoms. Two swimming pools were implicated by the report of 18 (90%) of the 20 swimmers.

Intervention Both pools were closed and samples of backwash water sent for testing for the presence of Cryptosporidium. Water samples from one pool was positive for C. parvum oocysts. The pools were emptied and cleaned with high levels of chlorine. One of the pools returned a strong positive result post-cleaning. This pool was drained again, and cleaned using the methods detailed below. Subsequent water samples were negative for C. parvum.

In addition to pool cleaning, the public was advised of the outbreak via frequent media releases and warning notices erected at all swimming pools in the ACT. Despite this, serial contamination of swimming pools occurred. In all, the ACT Department of Health tested 13 public and semi-public swimming pools for the presence of Cryptosporidium oocysts in the pool water. Samples from total of six pools returned a positive result and were closed for cleaning. Overall, 364 cases were reported to 27 March 1998, with a dramatic decline in reports within 14 days of the closure of the implicated swimming pool.

Sampling and testing The sampling and testing regime employed required taking two 25-litre water samples from the swimming pool filter backwash water. The samples were consigned by overnight courier to Melbourne for testing by a private laboratory. The testing included the identification and enumeration of viable Cryptosporidium parvum oocysts in the sample.

Cleaning regime For those pools that were closed to the public, the Department recommended the following cleaning regime:
1. Drain all the water from the pool ensuring there is no residual water in pipeworks and settling tank.
2. Scrub clean the pool wall and floor surfaces and surrounds with a hard brush and hypochlorite solution. The purpose of which is to mechanically loosen and oxidise any organic matter.
3. Hose down the area with high-pressure water and follow by steam cleaning.
4. Discard and replace the filter media in the filtering system.

The cleaning regime follows best practice and has been successful. Post cleaning, all pools returned two consecutive negative results of their backwash water samples and were re-opened.

Contribution of the CRC The CRC for Water Quality and Treatment provided assistance in the form of David Cunliffe from the SA Health Commission and Brett Robinson from the Australian Water Quality Centre. David and Brett attended a meeting with pool owners/operators to assist in discussions about the nature of the organism, treatment options for swimming pools, preventative and remedial actions.

The actions taken by ACT Health in mounting an education campaign and in closing down pools that tested positive for oocysts was supported. It was explained that the key to reducing risks of future outbreaks was prevention and that education campaigns need to be continued. The message that anyone with diarrhoea should not use public pools needs to be reinforced and people diagnosed with cryptosporidiosis should not use a public pool for 7-14 days after cessation of symptoms.

In terms of the pools, the concentrations of disinfectant that can be used will not inactivate Cryptosporidium oocysts and it is unlikely that most pool filtration systems will remove all of the oocysts. However, it is important to maintain pool treatment systems; a poorly maintained system will remove fewer oocysts than a well maintained system. Particular attention should be paid to filtration systems and to replacing filter media when required to maintain performance. The operation of public pools needs to controlled by legislative regulation.

Acknowledgements Valuable experience has been gained in the management of outbreaks of cryptosporidiosis associated with swimming pools. Advice and support from Martyn Kirk of the Victorian Dept. of Human Services and CRCWQT experts, David Cunliffe and Brett Robinson is gratefully acknowledged.

(1) NCEPH, Australian National University, Canberra.
(2) Public and Environmental Health Service, South Australian Health Commission, Adelaide.
(3) Health Protection Service, ACT Health, Canberra.

News Items

The PSG considered more than 30 issues which had been raised by stakeholders since the publication of the 1996 ADWG. It was agreed that revision should be considered in 1998 for 8 substances or microorganisms, and an additional 15 issues were deferred for reappraisal in 1999.

The PSG also identified 5 issues which were not strictly part of the scope of the present ADWG which they believed require NHMRC attention. These were; direct potable reuse, health surveillance, education, water treatment chemicals, and the WHO timetable for guideline revisions. Further details of the ADWG revision process can be found on the Web site of the CRCWQT (see URL address on page 1).

Cryptosporidium found in oysters
US researchers have announced the isolation of viable and infective Cryptosporidium parvum oocysts from oyster populations in Chesapeake Bay in Maryland. In the March issue of Applied and Environmental Microbiology, Ronald Fayer and coworkers described how they found C. parvum in every batch of 30 oysters collected from 6 sites at 2 time intervals during 1997. The numbers infected varied from 5 - 18 of 30 oysters collected in spring, to 2 to 26 of 30 oysters collected in summer.

The viability and infective potential of the oocysts were tested by inoculating three mice with oocysts washed from the gills of a batch collected near the shoreline close to faecal pollution sources (a cattle farm and houses with septic tank systems). Successful infections were established in two of the animals.

This is the first report of C. parvum isolation from shellfish growing an a natural environment, although a number of researchers have previously demonstrated uptake of oocysts under experimental conditions. To date there have been no reports of human Cryptosporidium infections from oysters or other shellfish, but given the fact that these seafoods are often eaten raw the possibility of infection must now be considered.

US water activists use new legal strategy
Environmental activists in the US are beginning to use a previously overlooked provision of the 1972 Clean Water Act to force states to set overall pollution limits on watersheds, according to a report in the 1st March edition of the New York Times. Until now the implementation of the CWA has focussed on the control of point source pollution by specific industries by means of a system of discharge permits. However the CWA also included a direction to states to carry out broad measures of water quality and impose overall limits on pollution of watersheds including non-point sources, if clean water standards are not reached.

Activists are now exploiting this section of the CWA to bring actions against about two dozen states in US federal courts. Their strategy involves asking the court to force the EPA to take control in states were governments have failed to clean up polluted watersheds. The EPA in turn is seeking to keep the issue a state responsibility by requiring all states to submit lists of polluted waters and their plans for remediation by April this year.

HIV+ individuals not at higher risk for Cryptosporidium infection
A paper published in the December 1997 issue of the Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology has confirmed that HIV+ individuals are not at increased risk of becoming infected during Cryptosporidium outbreaks compared to individuals with normal immune systems. By analysing the reported rates of symptoms among HIV+ and HIV- people during the Milwaukee outbreak, Holly Frisby and coauthors established that there was no significant difference in the rates of diarrhoea suffered by the two groups.

Among HIV+ individuals, the risk of becoming infected did not vary according to CD4+ cell count (a marker of progressive decline in immune function during HIV infection). However those with low CD4+ counts (severely immuno-depressed) who reported watery diarrhoea were more likely to suffer prolonged illness and to require medical treatment. This indicates that the state of the immune system has little influence on the initial infective process, but does affect the person’s ability to combat and eliminate Cryptosporidium infection.

Lunar Prospector finds water
NASA officials called a press conference on 5th March to announce that the Lunar Prospector probe had detected evidence of water on the Moon’s surface. The mission’s principal investigator, Dr Alan Binder, described the find as “ a tremendous resource that we can use in the exploration of space”. Data transmitted by the probe during the first few weeks of its lunar orbit indicate the presence of water at both the north and south polar regions.

While scientists are positive about the presence of large amounts of water (estimated at 10 to 300 million metric tons), the density of the deposits has not yet been determined. The Lunar Prospector is currently carrying out long range mapping from a high altitude orbit, and more detailed mapping from a lower orbit 10 km above the surface will not commence until the end of the year.

NASA scientists are reported to be overjoyed at the success of the probe in providing new data on the moon’s structure, and have been quoted as frequently mentioning the $40 million mission is only “a third of the cost of a Hollywood movie” (a reference to the recent blockbuster Titanic).

Judge rules water pre-payment illegal
A British judge has ruled that an electronic pre-payment scheme used by water companies is illegal because customers using this scheme were more likely to have their water disconnected then others. According to a report in the Electronic Telegraph of 21st February, the case was brought to court by six local authorities and opposed by two water companies and the industry regulator.

The councils gave evidence of their concerns over health risks arising from disconnection of water supplies, and the judge agreed that the the system did not comply with the industry code of practice to protect economically disadvantaged customers.

Ancient ice to hit the market
Drinkers will soon be offered the choice of buying “designer ice” made from 100,000 year old Greenland glaciers, following the decision of the Newfoundland government to grant the first commercial license to harvest icebergs. In addition to the claim of purity based on its ancient origin, the ice will spontaneously crackle and fizz as it melts due to the presence of trapped air pockets.

The harvesting company will use modified tugs to tow the icebergs to converted fish processing factories in Newfoundland. The ancient ice will also be used to make vodka to be sold under the name of Vodka Borealis. A representative of the Scottish Whisky Association was quoted as saying the 100,000 year age seemed “quite young” compared to water which has been “flowing over our Highland rocks for 800 million years”.

From the Literature

The printed version of Health Stream is available free of charge - to be added to our mailing list please contact Pam Lightbody (email above or fax + 61 3 9903 0576).