Application Series 20

SIR WORKS SWIMMINGLY WITH BIOMETRICS

Sydney's water purity authority, Sydney Water, posed the question of what effect the quality of Sydney's beach water has on swimmers' health. Cameron Kirton, a Sydney based statistical consultant used SIR to find the answer.

Background

Prior to the opening of Sydney's deepwater ocean sewage outfalls, Sydney residents were concerned at the effect Sydney water was having on their health. So much so that Sydney Water organised for a series of annual surveys to be made of the health of swimmers at ocean beaches. These swimmers might be exposed to increased risks of illness due to the levels of microbiological organisms then present in the beach water.

Cameron Kirton, the Sydney based statistician charged with the responsibility of measuring this effect described the care taken to ensure that this analysis would be a success. 'In 1990, we did a pilot survey of the health of about 500 people who were recruited at or near each of Palm Beach, Dee Why, Manly, Bondi, Bronte and Maroubra beaches'. Details were recorded at recruitment of each individual's perceived frequency of swimming and/or boardriding, together with the location of these activities. Each half month during February and March the recruits (initially healthy and aged between 14 and 40) submitted responses to questionnaires on their overall swimming and health experiences for the previous half month.

'This led to a more detailed analysis during 1991-93' Cameron explained, 'a diarised form of questionnaire was used by about 2000 recruits each year for the period January to March'. Again, 2 half months per month were separately filled in, but on a daily basis giving such detailed information as where and for how long each swim occurred. Cameron described the method of collecting illness symptoms 'The data were self-diagnosed and self-reported. The recruits ticked cells in a table to indicate when various symptoms occurred'.

The 1991-93 data recording gave scope for much more detailed analysis of the impact of swimming on subsequent health. The respondents had the opportunity to nominate categories of other swimming locations (other beaches, private pool, public pool, harbour pool, rock pool, river, lake or dam), 'so additional information on the health hazards of the use of these locations was available and this proved very useful' Cameron said.

'To enhance the analysis we had microbiological data on a range of organisms from water samples at "target" beaches and some local rainfall recordings as well'.

Database Design

Seven different record types were defined as follows:-

(1) Recruitment:

Recruit Identification, self-assessed usual beach swimming frequency, age at recruitment, beach of recruitment and gender.

(2) Reporting:

Recruit Identification, year, month, half month for data returns.

(3) Target Swims:

Recruit Identification, location date and duration of swims.

(4) Other Swims:

Recruit Identification, location, date and duration of swims.

(5) Illness:

Recruit Identification, illness group, dates 'from' and 'to' of illness.

(6) Microbiology:

Date and location of sample, data values for organisms.

(7) Meteorology:

Date and location of data, data values for climatic variables.

'Some individuals were recruited in more than one of the four years of survey recruitment' Cameron pointed out 'and we were able to track them through their recruitment number, which was the primary key on most record types',

The fifth record type was generated from the "symptom" data so that groups of symptoms were allocated appropriately. 'Cold and sore throat occurring concurrently as symptoms were classified as "Upper Respiratory" illness. We did this with a PQL program to validate the raw data, then ascribe one or more appropriate illness categories and write details to the "Illness" records' explained Cameron.

The recruits were quite active swimmers. Over 130,000 swims were made by the 4,500 responding recruits during the 1991-93 surveys. About 85,000 of these were made at the target beaches, so a lot of data existed with varying intervals between swims for the different individuals.

'This has made it possible to do a detailed analysis of the relationship between swimmers' health and water quality of Sydney beaches and provide Sydney Water with the sort of information it needs.' Cameron said, and went on to describe the technicalities of his analysis.

'Extraction of data for graphical and tabular display and subsequent statistical analysis was done using PQL programs. Other PQL programs were written for performing miscellaneous statistical calculations such as "raw" relative risk estimation together with their confidence intervals.

'The management of data and its proper statistical analysis was able to be accomplished within 20Mb partition of a PC using SIR's database to manage and manipulate the data and perform the necessary pre-analysis and validation routines'.

For more information on the use of SIR for Health after Swimming analysis contact:

Cameron Kirton
Statistical Consultant
5 Carbeen Road
Westleigh NSW AUSTRALIA

Back Sir Home