Reply to comments on "Derivation of numerical values for the World Health Organization guidelines for recreational waters".

The contribution addressed reveals an optimistic design philosophy likely to systematically underestimate risk in epidemiologic studies into the health effects of bathing water exposures. The authors seem to recommend that data on the 'exposure' measure (i.e. water quality) in such studies should be acquired in a similar manner to that used for regulatory sampling. This approach may compromise the quality of the epidemiologic investigations undertaken. It may result in imprecise estimates of exposure because it ignores the fact that regulatory timescales and spatial resolution (even if artificially compressed to a bathing day) can mask large spatial and temporal variability in water quality. If this variability is ignored by taking some mean value and attributing that to all of those exposed in a period at a study location, many bathers may be misclassified and the studies may be biased to a 'no-effect' conclusion. A more appropriate approach is to maximise the precision of the epidemiologic investigations by measurement of individual exposure (or water quality) at the place and time of the exposure, as has been done in randomised volunteer studies in the UK and Germany. The precise epidemiologic relationships linking 'exposure' with 'illness' can then be related to the probability of exposure to particular water quality by a 'normal bather' using the known probability distribution of the exposure variable (i.e. faecal indicator concentration) in the regulated bathing waters. We suggest that any research protocol where poor sampling design for water quality assessment is justified because regulatory monitoring is equally imprecise may be fundamentally flawed. The rationale for this assessment is that the epidemiology is the starting point and evidence-base for 'standards'. If precision is not maximised at this stage in the process it compromises the credibility of the standards design process. The negative effects of the approach advocated in this 'comment' are illustrated using published research findings used to derive the figures illustrated in Wymer et al. [2005. Comment on derivation of numerical values for the World Health Organization guidelines for recreational waters. Water Research 39, 2774-2777].

Derivation of numerical values for the World Health Organization guidelines for recreational waters.

In April 2001, draft 'Guidelines' for safe recreational water environments were developed at a World Health Organization (WHO) expert consultation. Later the same month, these were presented and discussed at the 'Green Week' in Brussels alongside the on-going revision of the European Union Bathing Water Directive 76/160/EEC. The WHO Guidelines cover general aspects of recreational water management as well as define water quality criteria for various hazards. For faecal pollution, these include faecal indicator organism concentrations and an assessment of vulnerability to faecal contamination. Central to the approach set out in the WHO Guidelines are: (i) the concept of beach profiling to produce a 'sanitary inspection category' which implies a priori hazard assessment as a core management tool and (ii) the prediction of poor water quality to assist in real time risk assessment and public health protection. These management approaches reflect a harmonized approach towards the assessment and management of risk for water-related infectious disease being applied by WHO. Numerical microbiological criteria for intestinal enterococci are proposed in the new Guidelines. These were developed using a novel approach to disease burden assessment, which has been applied to both recreational waters and urban air quality. This paper explains the scientific rationale and mathematical basis of the new approach, which is not presented in the WHO Guidelines for recreational waters.