Appendix C Microbial Risk Assessment Methods
Risk may be defined as the possibility of suffering harm from a hazard. Risk analysis provides the tools by which the magnitude and likelihood of such consequences are evaluated. The objective of risk analysis is to allow a standard to be set to achieve a certain level of public health or ecological protection, or perhaps to balance the costs of achieving such a standard (for example, by application of treatment processes) with the benefits obtained.
The formal risk analysis process can be divided into four discrete components (NRC, 1983):
Hazard assessment, in which the amount of a substance discharged from various sources is estimated.
Exposure assessment, in which the distribution of a hazardous substance among individuals in the population of concern is estimated. This may involve use of transport or attenuation models.
Dose-response assessment, in which the relationship between the level (concentration, duration) of exposure and the likelihood of effects of various degrees of severity is assessed.
Risk characterization, in which the information from the previous three elements is integrated.
Most human health risk assessments have been conducted on carcinogens. Such risk assessments are based on chronic, lifetime exposure to a particular agent and almost always rely on evidence obtained from animal tests (such as rats or mice) to quantify the level of exposure below which risk is maintained at an acceptable level. The level of acceptable risk to carcinogens has generally been
set in the range of 10-6–10-4 (excess cases per lifetime) (Travis et al., 1987), although there are differing actions taken under different federal statutes (Rosenthal et al., 1992).
With infectious microorganisms, for which uniform guidelines do not exist, the process of risk assessment may be conducted in a similar manner (Haas et al., in press). There are several distinguishing aspects of quantitative microbial risk assessment, however, which bear notice:
Generally adverse outcomes (infections, illnesses) associated with infectious agents may result from single exposures, rather than from chronic long-term exposures, although the duration of exposure will influence the likelihood of the outcome.
Secondary cases of illness may occur in persons not exposed to the original contaminated material, such as by person-to-person contact.
In general, human dose-response data are available for a variety of pathogens, and it is not always necessary to rely on animal studies. In fact, the issue of potential host-specificity of certain agents may make extrapolation from animal studies a less generally applicable procedure (without extensive validation).
Because the doses of microorganisms that may provoke adverse outcomes are very small (in principle one competent organism has the potential to act in this matter), the statistics of small particles adds an element of variability that it not present with many chemicals (where even for very potent carcinogens, an average daily dose of millions or more molecules may be needed for an adverse effect to occur). This variability must be accounted for in the dose-response model and in the estimation of exposure (and is superimposed on other sources of variability, such as in the analytical method itself).
There is also the possibility (although its importance has not been conclusively demonstrated) that populations exposed to pathogens over time may change in sensitivity (either to become more or less sensitized to the particular agent).
The process is illustrated schematically in Figure C-1. From the dose-response curve, the risk corresponding to a particular pathogen concentration and water ingestion rate (volume per day of water ingested without treatment, such as boiling, which might inactivate pathogens) can be determined. If this is above an acceptable level, then the concentration of the pathogen in the water supply may be regarded as too high. In initial formulations of the Surface Water Treatment Rule, an annual risk of 1 infection per 10,000 persons was regarded as being acceptable (Regli et al., 1988) although there is some question about the appropriateness of this value (Haas, 1996).
The use of this approach is attractive in that it has the potential to impute risks far lower than might be readily detectable by epidemiological surveillance in a routine manner. Furthermore, it may be applied to a series of pathogen monitoring results to assess any trends in the implied risk.
There are a number of uncertainties that need to be considered with this approach. First, the ingestion of water is variable from person to person [in common EPA risk assessments, a default value of 2 L/person/day is generally used (Cohrssen and Covello, 1989) emanating from the NRC report Drinking Water and Health (NRC, 1977)]. It is now recognized that use of a population distribution may be more appropriate (Roseberry and Burmaster, 1992). Second, the methods for pathogen measurement (and especially determination of viability and infectivity of pathogens in environmental samples) have limitations, although it is far from clear whether particular methods may err on the positive or negative side. Third, existing methods to detect pathogenic microbes in water tend to be time consuming, and so use of this approach on a rapid or real-time basis to assess water quality is not realistic. Rather, this approach is more suited for periodic retrospective assessment of long-term trends. Finally, the dose-response curve used in this approach has typically been obtained on a defined population (generally ''normal, healthy" individuals) and may not be reflective of differential susceptibilities in other segments of the population (Gerba et al., 1996).
Risk assessment may be particularly beneficial in conjunction with an epidemiological surveillance program. If the dose-response curve is extrapolated to low levels of risk, it is possible to ascertain the level of infection implied by a particularly low level of exposure. At the 1 in 10,000 risk level, surveillance of many thousands of persons over a period of time would be necessary to ascertain a positive effect. If pathogen-monitoring data are available, the imputed risk from the dose-relationship can be compared to the bounds from surveillance to assess consistency.
With pathogens, it is possible to perform risk assessments on a frequent basis, perhaps monthly or quarterly, as new pathogen monitoring data becomes available. It is also possible to use the monitoring data in conjunction with dose-response information and perhaps other data (such as the presence of excess turbidity in a finished water) to define levels of contamination where particular public health actions must be taken. Limitations on method efficiency and rapidity may, however, limit the present ability to use microbial risk assessment in such a real-time manner.
Cohrssen, J. J., and V. T. Covello. 1989. Risk Analysis: A Guide to Principles and Methods for Analyzing Health and Environmental Risks. Springfield, VA: National Technical Information Service, U.S. Department of Commerce .
Gerba, C. P., J. B. Rose, and C. N. Haas. 1996. Sensitive Populations: Who is at the Greatest Risk? International Journal of Food Microbiology 30(1–2):113–123.
Haas, C. N. 1996. Viewpoint: Acceptable Microbial Risk. Journal of the American Water Works Association 88(12):8.
Haas, C. N., J. B. Rose, and C. P. Gerba. In press. Quantitative Microbial Risk Assessment. New York: Wiley.
National Research Council (NRC). 1977. Drinking Water and Health. Washington, DC: National Academy Press.
NRC. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: National Academy Press.
Regli, S., A. Amirtharajah, B. Borup, C. Hibler, J. Hoff, and R. Tobin. 1988. Panel Discussion on the Implications of Regulatory Changes for Water Treatment in the United States. Advances in Giardia Research:275.
Roseberry, A. M., and D. E. Burmaster. 1992. Log-Normal Distributions for Water Intake by Children and Adults. Risk Analysis 12(1):99–104.
Rosenthal, A., G. M. Gray, and J. D. Graham. 1992. Legislating Acceptable Cancer Risk from Exposure to Toxic Chemicals. Ecology Law Quarterly 19:269–363.
Travis. C. C., S. A. Richter, E. A. C. Crouch, R. Wilson, and E. D. Klema. 1987. Cancer Risk Management. Environmental Science and Technology 21(5):415–420.