Canoeist Pathogenic Illness Guide
Chapter 3 - Epidemiological Research (Continued)
3.2 - Large Scale Sea Bathing Epidemiological Investigations
3.2 a) UK Studies at Langland Bay and Moreton
3.2 b) Swimming at Sydney Beaches - Australia
3.2 c) Ontario Beach Study - Canada
Forward to Chapter 3.3 - Watersports Epidemiological Studies
When carrying out large scale epidemiological studies of the effects of bathing in sewage contaminated water, the main problem is coping with the very large number of variables which can affect the rates of illness. Not only does the concentration of pathogen and indicator organism in the water vary with time and position, the degree of exposure of each individual will also vary depending on the activity they undertake, such as a short wade on the beach edge, or a prolonged and vigourous swim. In addition there are many other non-bathing risk factors such as the type of food consumed by each individual before and after the bathing event. If valid results are to be obtained then the statistical methods used to analyse the results will need to be very powerful and very carefully thought out.
There can be two ways of coping with this problem. If the activity and water monitoring programme is very tightly controlled, perhaps by specifying exactly what activity each volunteer undertakes, and in what location in relation to the water sampling point, then the variability is reduced, and it may be possible to obtain statistically valid results from a relatively small sample of participants. Alternatively if variability in bathing activity
is allowed (to simulate real life) then a very large number of participants will be required to produce statistically valid results. Phillipp (1991) reports one study which called for 4, 200 non swimmers and 4,200 swimmers to detect an excess illness incidence of 30% from amongst a background illness rate of 5%.
Fleischer, J. M. et al (1993). Kay, D. (1994)
Work in the UK at Langland Bay and Moreton between 1989 and 1993 was a series of carefully designed, controlled cohort studies which attempted to limit or exclude most of the non-bathing related variables involved in this issue. The water at Langland Bay and Moreton passed the EC Bathing Water Directives mandatory level, but failed the advisory level (see Appendix D).
Extensive advertising in the media weeks before the study day recruited cohorts of approximately 600 per day. After careful interviewing to exclude unsuitable candidates, and to record a history of non water related risk factors (see table 3), the cohort was split randomly. Half the cohort consumed a picnic on the beach (provided by the research team), whilst being supervised to ensure that they did not enter or contact the water. The other half undertook a supervised swim in a controlled location close to a sampling point. The bathers were required to swim for at least ten minutes, and then immerse their head three times. They then consumed a similar picnic provided by the research team.
Table 3. Non water related risk factors for Gastroenteritis (Fleischer J M 1993)
A few days after the trial day, each participant was given a post exposure face-to-face interview by a researcher to check for short term symptoms, and a mailed follow up questionnaire after four weeks.
Water samples were taken at twenty minute intervals at three separate depths, testing for faecal streptococci and faecal coliforms:
For the statistical analysis, the set of samples closest in time and location to each participant was used to determine the exposure of that swimmer. Six permutations were therefore available, testing reported rates of illness symptoms against each of the three sample depths for each of the two organisms.
Of these permutations, only the test for faecal streptococci showed a statistically significant association with gastroenteritis rates amongst bathers, and then only for the water samples taken at chest depth.
|Exposure Status||Faecal Streptococci exposure per 100 ml||Number of Bathers||Symptom Rate per 100||Number of Bathers||Symptom Rate per 100|
|Bather||more than 70||24||37.5||8||50|
Table 4 (above) shows one analysis of the results extracted from the reference. The faecal streptococci exposure for each bather was placed in one of four categories:
In table 4 the rates of reported symptoms can be seen to be higher amongst bathers than non bathers. It should be noted however that in waters containing up to twice the USEPA faecal streptococci standard, the rate is not significantly above the normal background rates shown by non bathers. Only when the water quality standard is exceeded by more than a factor of two does the symptom reporting rate rise dramatically.
Both references contain more detailed analyses of this data, especially in an attempt to allow for the effects of the consumption of hamburgers, mayonnaise and cold meat pies by the swimmers before and after the trial day. All these analyses tended to confirm the basic result, which was that within this very tightly controlled cohort study, faecal streptococci levels could be used to indicate levels of gastrointestinal risk, but at levels up to double the USEPA standards, the actual risk from bathing approximated to the risks from other non water related risk factors such as fast food.
One other significant aspect of the study was that the rates of both faecal streptococci and faecal coliforms showed significant spatial and temporal variations.
|Geometric Mean||Range||Geometric Mean||Range|
|Faecal Coliforms||31/100 ml||0-1310||145/100 ml||0-556|
|Faecal Streptococci||42/100 ml||0-196||29/100 ml||0-159|
This emphasises the fact that very little weight can be given to any single sample of an indicator organism. Statistical methods must be used to analyse large numbers of samples or to note how sample levels change with changing conditions. (see also HMSO (1982) page 8).
Fleischer reported that his work:
questions the appropriateness of current marine water quality standards, as well as the notions of "acceptable risk" upon which these standards are based
Corbett, S. J. et al (1993)
In 1989 a survey of bathing water quality showed that one third of water samples tested at many popular Sydney beaches failed to meet existing water quality standards. This caused a wave of public disquiet and protest. Subsequently doctors in local suburbs reported increased incidence of gastrointestinal, respiratory and other viral illnesses. The desertion of the public beaches following these reports had major commercial implications for Sydney's tourist economy. Against this background a Beach User's Study was commissioned to investigate the relationship between ocean swimming and illness.
The research followed the by-now standard method for these studies, recruiting cohorts of swimmers and non-swimmers, questioning about non-swimming risk factors such as fast food, and then looking for relationships between swimming and symptoms of illness. In this study subjects were recruited by approaching beachgoers on a number of beaches, and then interviewing them about their swimming and non swimming activities. This method meant that the swimming activities were less tightly controlled than in the Langland Bay study. Whilst this added variability in activity might make the statistics more difficult, the study was based more closely on "real" activities. A total of 9600 person were approached.
The results of the study are similar to the results of other studies.
It was found that those who reported swimming at least one additional time between the initial study day and the follow up interview (who could be assumed to be regular swimmers) reported fewer illness symptoms. This could be because regular swimmers had a higher immunity to infection, although it could also be because those with an infection might have wanted to bathe, or normally be frequent swimmers, but could not because of illness.
Unlike the Langland and USA studies, it was found that faecal coliforms were better predictors of reported symptoms than faecal streptococci. It was suggested that the differences may be due to the different morbidities of these two organisms in the warmer Australian waters, and the common practise in America of chlorinating sewage effluent.
Seyfreid, P. L., et al (1985)
In a similar study to the Sydney Beach project, 8,000 people were interviewed on selected Ontario beaches in the summer of 1990. Sampling of indicator organisms showed that the beaches were inside the accepted guide-lines.
The rate of reported illness symptoms in swimmers (191 in 2,743 = 7.0%) were slightly over double the rate of symptoms in non swimmers (53 in 1,794 = 3.0%). This again is a similar rate to other studies where the bathing water is inside current water quality guide-lines. Eye, ear, nose and throat ailments made up over 50% of symptoms, with gastrointestinal symptoms being much less common.
In this study the best indicator to predict illness rates was a count of total staphylococci. This produced a better fit than either faecal coliforms or faecal streptococci (although both of these organisms showed a weak relationship with illness rates).
During the study microorganisms were recovered from beach sediment, as well as from the water. The levels were found to be typically at least ten times higher in the sediment. This may be of relevance in busy beaches, where extensive paddling and wading may alter the distribution of microorganisms within the total beach environment. Weather conditions such as heavy surf could also have an effect.