Grantee Research Project Results
2005 Progress Report: Examining Epidemiologic and Environmental Factors Associated with Microbial Risks from Drinking Water
EPA Grant Number: R831727Title: Examining Epidemiologic and Environmental Factors Associated with Microbial Risks from Drinking Water
Investigators: S. Eisenberg, Joseph N. , Moe, Christine L. , Uber, Jim
Institution: University of Michigan , Emory University , University of Cincinnati
Current Institution: University of California - Berkeley , Emory University , University of Cincinnati
EPA Project Officer: Packard, Benjamin H
Project Period: December 23, 2004 through December 27, 2007 (Extended to December 27, 2009)
Project Period Covered by this Report: December 23, 2004 through December 27, 2005
Project Amount: $589,806
RFA: Microbial Risk in Drinking Water (2003) RFA Text | Recipients Lists
Research Category: Nanotechnology , Drinking Water , Human Health , Water
Objective:
Results from drinking water intervention trials have provided a wide range of outcomes, ranging from no evidence of risk to attributable risk estimates as high as 35-40 percent. This range of risk estimates is problematic for regulators. One reason for this variation is because of the differing environmental conditions in each of these studies, such as source water concentrations, treatment barriers, and distribution systems. Risk models can provide insight to these epidemiologic data by interpreting the variability observed across studies. These insights then can be used to help design future studies. The objectives of this research project are to: (1) develop a population-based dynamic model that can be used to characterize drinking water risks to communities and apply this risk model to human calicivirus (HuCV), an important pathogen on the U.S. Environmental Protection Agency (EPA) Candidate Contaminant List; (2) develop an exposure model that describes the pathogen fate and transport from source water through to the distribution system for distribution systems representative of those in urban areas of the United States that will incorporate factors that have a potential role in determining human exposure; and (3) combine the models developed in Objectives 1 and 2 and conduct sensitivity studies to categorize those factors with respect to their relative importance in determining risk.
Progress Summary:
Interim Results
Progress during the Year 1 of this project centered around three areas of study: (1) literature reviews of both outbreaks caused by norovirus as well as the occurrence and persistence of norovirus in the environment; (2) reanalysis of secondary data associated with the norovirus biology; and (3) model development for both the health effects and environmental fate and transport components of a risk assessment model.
In September 2005 the three co-investigators (J. Eisenberg, C. Moe, and J. Uber) met in Bilthoven, Netherlands, at the National Institute of Public Health and the Environment (RIVM), for a project kick-off meeting. This was a 1-day meeting to map out a specific research agenda and time line and to meet with RIVM scientists who have expertise in norovirus research (Dr. Marion Koopmans, Dr. Yvonne van Duynhoven, Dr. Ana Maria de Roda Husman, and Dr. Peter Teunis). Although the grant had a start date of January 2005, a delay in the University of California (UC) at Berkeley receiving the funds and a subsequent delay in executing the subcontracts resulted in a later start date. Furthermore, the principal investigator has moved the grant from UC Berkeley to the University of Michigan, resulting in additional delays. Based on the expenditures during this first year, we have requested and received permission to shift unused portions of the budget to fund Year 4 of the project.
The following provides a summary of the interim progress in our three areas of study:
Literature Reviews
An extensive literature review was conducted in Medline using the following key words:
- Human Calicivirus (HuCV) & Outbreak & NOT animal studies
- “Disease Outbreaks”[MeSH] AND “Gastroenteritis/virology”[MeSH] AND Norovirus[MeSH]
- Norwalk virus[MeSH] AND “Disease Outbreaks”[MeSH]
- Gastroenteritis/virology[MeSH] AND “Disease Outbreaks”[MeSH]
- Caliciviridae Infections[MeSH]AND “Virus Diseases/transmission”[MeSH] AND (“Disease Transmission, Vertical”[MeSH] OR “Disease Transmission, Horizontal”[MeSH])
A total of 343 articles were obtained. Of these, a total of 257 articles were found to describe 205 distinct outbreaks. Table 1 lists the distribution of outbreak types.
Table 1. Distribution of Outbreak Types
A number of drinking water and foodborne outbreaks were identified as having sufficient data to allow for subsequent analysis. Currently, we are categorizing these outbreaks by genotype and cluster (outcome variable) and by a number of covariates (outbreak type, location, date, primary attack rate, secondary attack rate, season, and age group) to examine the association of genotype/cluster with characteristics of the outbreak.
A literature review was initiated to describe the environmental occurrence of norovirus, the effects of water treatment processes including removal and inactivation, and the relationships between different measurement techniques observed in the literature. Environmental occurrence information will inform probabilistic models that predict the time series of occurrence of norovirus at the treatment source, whereas treatment data will inform similar models to estimate removal and inactivation of norovirus under typical coagulation/sedimentation/filtration processes and disinfection with chemical oxidants. Information about measurement techniques is necessary for establishing a common base unit of measurement for norovirus obtained from various studies, which in turn will require a basis for converting between measurement approaches. Following is a brief summary of information obtained from the literature review to date; the literature review is ongoing
The concentration of norovirus in sewage samples was as high as 107 RNA-containing particles per liter, and was observed to be relatively high independent of outbreaks. A different study measured norovirus concentrations of approximately 105 PCR detectable units (PDUs) per liter in raw sewage versus 103 PDU/L in treated sewage.
Recombination of norovirus strains may occur because of the different strains present in treated sewage, with potential exposure to more virulent and pathogenic strains. In Germany, norovirus average genome load was found to be 9.7 x 10-5 genome equivalence (gen. equ.) per liter in a raw sewage and 8 x 10-4 gen. equ./L was found in the effluent of the treatment plant. The receiving surface water had 1.8 x 10-4 gen. equ./L because of dilution.
Enteric viruses are more resistant than bacterial pathogens to sewage treatment processes, including chlorination. Yet approximately 2 log10 units of virus removal from sewage can be achieved by activated sludge and aeration. Volunteer studies concluded that a residual chlorine concentration of 5-6 mg/L is needed to inactivate norovirus completely (1-2 mg/L greater than allowed for drinking water residual in water distribution networks). Seroconversion was considered as the detection level of norovirus. Norovirus could be entrained in floc particles during the water treatment coagulation/sedimentation processes, providing some protection from the inactivation process for norovirus particles in floc that escape physical sedimentation/filtration processes. A 2 mg/L dose of preformed monochloramine achieved a 1 log10 reduction of norovirus after 3 hours of contact time. Many enteric viruses have been successfully inactivated by ozone with CT99 values much less than 1 mg/L-min, and norovirus was reduced by 3 log10 units after a 0.37 mg/L ozone dose and 10 seconds contact time.
Two decades ago, electron microscopy and serologic tests were the only available detection methods and norovirus could only be identified in 20 percent (by electron microscopy) to 40 percent (by serological tests) of outbreak cases. The more recent development and use of reverse transcription-polymerase chain reaction (RT-PCR) has helped to understand that norovirus is implicated in a majority of nonbacterial gastroenteritis outbreaks.
Real-time RT-PCR gives more reliable results regarding inactivation of norovirus with different doses of disinfectant (ethanol, sodium hypochlorite) compared to conventional RT-PCR, which uses agarose gel for amplification of RNA. Conventional RT-PCR detects the presence/absence of virus, and real time RT-PCR involves a degree of quantification in terms of cycle threshold delay. In inactivation studies with norovirus, coliphage MS-2 and poliovirus 1 were also considered and their concentrations after inactivation were compared for infectivity assay technique and RT-PCR. It was found that RT-PCR underestimated the inactivation of virus.
Data Gathering and Analysis
Most of the work in this section has focused on the collection and reanalysis of data from Christine Moe’s laboratory that will be incorporated into the risk model. To date, these data include: dose-response, latency period, shedding duration and magnitude, innate and conferred immunity, and environmental persistence on surfaces. The initial goal in this project component is to model dose-response and virus shedding and to incorporate these submodels into the risk model. Specific activities in this area included:
- Development of quantitative real-time RT-PCR techniques to quantify the virus titer (in terms of number of genome copies) in the Norwalk virus and Snow Mountain virus inocula used in the human challenge experiments conducted by Dr. Moe’s laboratory. A preliminary step was the development of cDNA Norwalk virus and Snow Mountain virus standards to include in the quantitative real-time RT-PCR assay. We compared these new data to the previous endpoint titration RT-PCR data for the inocula titer. These data will be used to refine our characterization of the dose-response relationship.
- Application of quantitative real-time RT-PCR to quantify Norwalk virus and Snow Mountain virus in archived stool samples from infected subjects in the human challenge experiments conducted by Dr. Moe’s laboratory. We compared these new data to endpoint titration RT-PCR data for selected stool specimens. These data, along with information on the number of virus-positive stools and the weights of the virus-positive stools, allows us to calculate the total amount of norovirus shed by an infected individual. We expect to be collecting better virus titer data on fresh stool specimens from norovirus-infected subjects in a new U.S> Environmental Protection Agency (EPA)-sponsored Norwalk virus challenge study starting in May 2006 to examine Norwalk virus persistence and infectivity in groundwater (EPA Cooperative Agreement82911601-0).
- Dr. Moe and Dr. Teunis completed two manuscripts describing the Norwalk virus human challenge experiments (previously funded by EPA Science To Achieve Results (STAR) Grant R826139) and the Norwalk virus dose-response relationship. These manuscripts are currently under review by the coauthors and then will be submitted to the Journal of Infectious Diseases and Proceedings of the National Academy of Sciences, respectively. These dose-response relationships will be subsequently used in the risk models.
- Dr. Teunis, Dr. de Roda Husman, and Dr. Moe conducted a preliminary risk assessment for waterborne noroviruses and presented this work at the American Water Works Association International Symposium on Waterborne Pathogens in Atlanta, March 2006. This risk assessment was a preliminary exercise in examining ways to use raw data on virus occurrence, virus removal, drinking water ingestions, and the dose response data.
- Dr. Moe’s laboratory conducted studies of Norwalk virus persistence on surfaces (formica, ceramic tile, and stainless steel). Viral persistence was estimated as number of genome copies detected by quantitative real-time RT-PCR at various time points postinoculation. These studies were conducted at room temperature under ambient humidity conditions. Additional experiments will examine Snow Mountain virus persistence on the same surfaces and will also examine the effects of temperature and humidity on persistence of viral RNA. This information will be used to develop a fomite-mediated transmission submodel.
Model Development
The primary goal in this component is to develop and integrate norovirus exposure and disease transmission models. To date these two models have been developed separately.
Disease Transmission Model. We have developed an event driven stochastic modeling framework analogous to models currently being published in the literature for studying disease processes such as smallpox, influenza, and severe acute respiratory syndrome. The difference with our model of norovirus transmission is the existence of an explicit environmental component (i.e., our model accounts for the fact that norovirus can survive for long periods outside the human host). The theme of our study is to examine the ways in which heterogeneity will impact transmission. Heterogeneity is accounted for in the model in various ways. First, transmission occurs at different rates in different settings. We account for this feature by including social structure (i.e., in our initial model design we have included three settings: households, schools, and workplaces). Second, the infectiousness of individuals varies. Infectiousness is determined by both the rate of pathogens that are shed and the length of time that an individual is infectious. We currently account for this process by assuming that each individual has a super-spreading status. Third, environmental contamination varies in time and space. The environmental model described in the next section will account for this.
In brief, the model can be described by individual-, household-, and community-level attributes as summarized in Table 2.
Table 2. Disease Transmission Model Individual-, Household-, and Community-Level Attributes
Individual Attributes |
Household Attributes |
Community Attributes |
Household ID |
Node ID # |
# of households |
Age |
|
# of schools |
Work or school location |
|
# of workplaces |
Super-spreading status |
|
|
Every individual belongs to a household and depending on their age has a designated school or work place. Each individual also has a super-spreading status. Households are assigned a node number that spatially locates it within the drinking water distribution system. Communities are specified by the number of households, number of schools, and number of workplaces. A specific realization of the model is defined by the user by defining the following distributions: the number of kids per household, the workplace size, the school size, the number of super long shedders, the number of super high shedders, and the number of households along a given node. To run the model requires identifying specific parameter values. The parameter values of the model include: recovery rates, person-person transmission rates within the different social settings (home, work, and school), and transmission rates from water to humans. Ultimately, recovery rate parameters will be obtained from Christine Moe’s shedding data, person-person transmission rates will be obtained from outbreak data, and transmission rates from water to humans will be obtained from both rates of drinking tap water and from Christine Moe’s dose-response data.
The model has been developed and implemented in C++. Preliminary simulations have been conducted to examine the basic model properties.
In addition to this model development, initial work was conducted on developing an analysis approach that can use available outbreak data to inform the model. The focus here is analysis of norovirus outbreak, and the parameter of interest is the proportion of cases that were caused by secondary transmission. We are currently developing a maximum likelihood approach to estimate the transmission and recovery rates. This likelihood function would then be used to estimate maximum-likelihood parameter values. These parameter values would in turn be used to conduct the following two sets of simulations to obtain an estimate for the parameter of interest (proportion of cases caused by secondary transmission). In the first set of simulations, all transmission rates are fixed at their maximum-likelihood point estimates. In the second set of simulations, all secondary transmission rates are set to zero, with the remaining transmission rates still fixed at their original maximum-likelihood point estimates. The attributable risk of secondary transmission can then be estimated by a comparison of these two sets of simulations.
Environmental Model. A general distribution system exposure assessment (EA) tool was developed that can be used, and modified, by the project team. This EA tool integrates a research version of the public domain network hydraulic and water quality model, EPANET, that allows multispecies chemical and biological transformations within the distribution system, with a Monte-Carlo simulation engine. The EPANET extensions were developed under another EPA-sponsored project (Contract No. 68-C-00-159, task order 85). The Monte-Carlo simulation capability allows flexible selection of various probability distributions assigned to model parameters representing source water concentrations, water demand rates, kinetic coefficients, and the like. Thus the EA tool allows, for example, statistical sampling from the output distribution of household pathogen concentrations, based on random process models of pathogen occurrence at the water source. The results are stored in a database, which can be processed to develop descriptive statistics and parameters of household pathogen exposure that will be used to link with the disease transmission models. The EA tool does not rely on proprietary software and is coded in the C programming language.
Future Activities:
Our plans for Year 2 are as follows:
- Complete our analysis of norovirus outbreaks.
- Complete our models of norovirus shedding and dose-response for Norwalk virus and Snow Mountain virus.
- Complete our disease transmission model and examine ways in which secondary transmission contributes to the risks associated with outbreaks. We will use selected outbreaks as case studies.
- Complete our initial environmental models of norovirus occurrence, fate, and transport in source waters and water distribution systems, considering the effects of water treatment processes and inactivation caused by the presence of free chlorine residual.
- Develop or adapt statistical models of norovirus occurrence in surface waters used for drinking water sources.
- Develop or adapt statistical models of norovirus removal and inactivation during common physical/chemical treatment processes and disinfection using free chlorine.
- Link the above statistical models with the distribution system transport models using the above described EA tool developed in Year 1. Use existing data from the distribution system in Hillsborough County, Florida, to support a case study analysis.
- Include information on Norwalk virus persistence and infectivity in groundwater (funded by EPA Cooperative Agreement R829116).
- Include information on norovirus occurrence from U.S. studies and RIVM studies.
- Develop preliminary conceptual and mathematical models that extend these initial environmental models to incorporate processes of contamination via distribution system intrusion, and incorporation/release of infectious norovirus to/from pipe biofilm. These models will be a focus on the exposure assessment modeling work during Year 3.
- Integrate our environmental and disease transmission models.
Journal Articles on this Report : 1 Displayed | Download in RIS Format
Other project views: | All 2 publications | 1 publications in selected types | All 1 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Eisenberg JNS, Hubbard A, Wade TJ, Sylvester MD, LeChevallier MW, Levy DA, Colford Jr JM. Inferences drawn from a risk assessment compared directly with a randomized trial of a home drinking water intervention. Environmental Health Perspectives 2006;114(8):1199-1204. |
R831727 (2005) R831727 (2008) |
|
Supplemental Keywords:
drinking water, risk assessment, pathogens, decisionmaking, epidemiology, mathematics, modeling,, RFA, Scientific Discipline, INTERNATIONAL COOPERATION, Water, Environmental Chemistry, Health Risk Assessment, Ecological Risk Assessment, Environmental Engineering, Drinking Water, microbial contamination, microbial risk assessment, monitoring, real time analysis, aquatic organisms, other - risk assessment, early warning, water quality, drinking water contaminants, epidemiological study, drinking water systemProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.