Page 14: Guidelines for Canadian Drinking Water Quality: Guideline Technical Document - Enteric Protozoa: Giardia and Cryptosporidium

Appendix D: QMRA model

Mathematical models have been developed as a means to quantitatively assess the potential microbiological risks associated with a drinking water system, including the potential risks associated with bacterial, protozoan and viral pathogens. These models have been developed by international organizations (Smeets et al., 2008; Teunis et al., 2009), as well as by groups within Canada (Jaidi et al., 2009). QMRA models have also been used to estimate the potential health risks through other routes of exposure (Mara et al., 2007; Armstrong and Haas, 2008; Diallo et al., 2008). Although some of the assumptions vary between models (i.e., the choice of reference pathogen or selection of dose-response variables), all are based on the accepted principles of QMRA--that is, hazard identification, exposure assessment, dose-response assessment and risk characterization.

A QMRA model was developed by Health Canada as part of the risk assessment process for enteric pathogens in drinking water. This probabilistic model explores the potential disease burden, with associated uncertainty, for user-defined scenarios for a drinking water system. The model includes user inputs for the protozoal quality of the raw water source and the specific treatment train (defined in terms of filtration and disinfection approaches). Cryptosporidium and Giardia are used as reference protozoans. For drinking water systems where data are lacking for the above parameters, the model includes values from published literature and from expert opinion as a starting point for the assessment. For source water quality, the model provides users with the choice of four categories. These source water quality estimates were developed only to be used within the context of the QMRA model for evaluating the impacts of variations in source water quality on the overall microbiological risks. It should be noted that although a source may be a particular category for enteric protozoans, it may have a different source water quality category for bacterial or viral pathogens. For treatment processes, the model uses a range of literature values to more accurately represent the range of effectiveness of treatment methodologies.

The QMRA model uses this exposure information, along with the dose-response model and the DALY calculations for Cryptosporidium and Giardia, to estimate the potential disease burden (in DALYs/person per year) for the site-specific scenario information. The quality of the outputs from the QMRA model are dependent on the quality of the information that is input into the model. Measurements, as opposed to estimates, for exposure levels will result in a higher-quality risk assessment output. Even with high-quality exposure data, the QMRA model requires numerous assumptions that introduce uncertainty into the assessment:

  • It is assumed that the distribution of (oo)cysts in water is random (Poisson). However, it is likely that the (oo)cysts are not randomly distributed but rather occur in clusters, either loosely associated with each other or tightly bound to or within particles (Gale, 1996). Such clustering means that most consumers will not be exposed, but a small portion will be exposed to 1 or more (oo)cysts. This model does not account for clustering and will therefore underestimate the probability of exposure and infection.
  • Treatment efficiency is modelled based on data in published literature for various treatment processes, which may be an underestimate or overestimate of the performance at a specific site. Also, treatment efficiency data are derived using laboratory-adapted strains of Cryptosporidium and Giardia, which may not respond to treatment processes in exactly the same manner as the Cryptosporidium and Giardia present in source water.
  • It is assumed that all (oo)cysts are viable. The current risk assessment model assumes that errors caused by overestimating viability are at least partially counterbalanced by poor (oo)cyst recoveries during the detection procedure.
  • All (oo)cysts are assumed to be equally infectious. Dose-response experiments have shown that there can be significant differences in infectivity among different strains of Giardia and Cryptosporidium. Until routine, practical methods to identify infectious Giardia and Cryptosporidium (oo)cysts are available, it is desirable, from a health protection perspective, to assume that all (oo)cysts detected in source waters are infectious to humans, unless evidence to the contrary exists.
  • The model assumes a daily consumption of unboiled tap water of 1.0 L/person. In a population, there will be a distribution of tap water consumption that is not represented by this point estimate.

The model is based on risks associated with source water contamination and does not consider potential contamination within the distribution system.

Page details

Date modified: