Drinking water uncertainty

There is uncertainty as to what levels of MTBE in drinking water cause a risk to public health.9 U.S. EPA has issued an advisory suggesting that drinking water should not contain MTBE in concentrations >20-40 pg/L, based on taste and odor concerns, but has not issued a federal maximum contaminant level (MCL) for MTBE, which will be based on the ongoing U.S. EPA studies.1... 

Tables 1-1 through 1-4 show the relationship between exposure to acrylonitrile and known health effects. Short-term and longer-term Minimal Risk Levels (MRLs) are also included in Tables 1-1 and 1- 3. These MRLs were derived from animal and human data for both short-term and long-term exposure, as described in Chapter 2 and in Tables 2-1 and 2-2. The MRLs provide a basis for comparison with levels that people might encounter either in the air or in food or drinking water. If a person is exposed to acrylonitrile at an amount below the MRL, it is not expected that harmful (noncancer) health effects will occur. Because these levels are based only on information currently available, some uncertainty is always associated with them. Also, because the method for deriving MRLs does not use any information about cancer, an MRL does not imply anything about the presence, absence, or level of risk for cancer.

Most scientists would hold that these unknowns and uncertainties in the regulatory risk-assessment model would tend to favor risk overestimation rather than underestimation or accurate prediction. While this view seems correct, it must be admitted that there is no epidemiological method available to test the hypothesis of an extra lifetime cancer risk of about 10 per 1000 000 from methylene chloride in drinking water. The same conclusion holds for most environmental carcinogens. It is also the case that more uncertainties attend the risk assessment process than we have indicated above. 

Note that some of the risk information is actuarial (based on statistical data, typically collected and organized by insurance companies), and some of it has been derived from the type of risk assessment discussed in this book (chloroform in chlorinated drinking water, afla-toxin in peanut products). While the uncertainties associated with the figures in Table 11.2 are much greater for some risks than for others (not a trivial problem in presentation of risk data), such a presentation, it would seem, is helpful to people who are trying to acquire some understanding of extremely low probability events, of the order of one-in-one million. 

The MRL was based on a NOAEL of 26 mg/kg/day in the drinking water for 4 days for hepatic effects in mice (Larson et al. 1994b). The NOAEL of 26.4 mg/kg/day was divided by an uncertainty factor of 100 (10 for extrapolation from animals to humans and 10 for human variability) to arrive at the MRL of 0.3 mg/kg/day. A study performed by Moore et al. (1982) found renal effects in CFLP Swiss mice dosed at 65.5 mg/kg/day by gavage in oil. Another study by Larson et al. (1993) found both hepatic (elevated SDH, ALT and AST, hepatocyte necrosis) and renal (proximal tubule necrosis) lesions in Fischer 344 rats and hepatic lesions only in B6C3Fj mice induced by chloroform administered at 34 mg/kg/day once by gavage in oil. Lesions in the Larson et al. (1993) study were ranked as less serious LOAELs. 

Risk characterization is thus the step in the risk assessment process where the outcome of the exposure assessment (e.g., daily intake via food and drinking water, or via inhalation of airborne substances) and the hazard (effects) assessment (e.g., NOAEL and tolerable intake) are compared. If possible, an uncertainty analysis should be carried out, which produces an estimation of the risk. Several questions should be answered before comparison of hazard and exposure is made ... 

This slide shows an example taken from an European standard for the analysis of mercury in water. For drinking water the standard states a reproducibility variation coefficient of 30% on a mercury concentration level of 0.8 pg/l. Provided that we can prove that we can perform as described in the standard our expanded measurement uncertainty (95% confidence) is estimated to 60%. 

This slide shows an example from a drinking water PT provided by the University of Stuttgart. At the same level as in the previous example we find a reproducibility variation coefficient of about 20%. So our expanded uncertainty is estimated to 40%. 

Risk assessments are also conducted to derive permissible exposure levels for acute or short-term exposures to chemicals. Health Advisories (HAs) are determined for chemicals in drinking water. They are the allowable human exposures for 1 day, 10 days, longer term, and lifetime durations. The method used to calculate HAs is similar to that of the RfD methodology using uncertainty factors. Data are being developed from toxicity studies with durations appropriate to the HA. 

The safe levels of hepatotoxins in food and drinking waters have been evaluated in terms of tolerable daily intake (TDl). This value should be ideally taken as a result of human studies, but often such studies are inadequate or nonexistent. Alternatively, they are often obtained after animal studies, although the variability in sensitivity between animals and humans that makes necessary the estabhshment of safety factors to deal with this uncertainty should be taken into account. 

---