Abstract of Meeting Paper

Society for Risk Analysis - Europe 1998 Annual Meeting

Epidemiologic Risk Assessment of Cryptosporidium Parvum in Watershed. Hussni O. Mohammed and Susan Wade, Section of Epidemiology, Department of Clinical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY 14853, USA, telephone (607) 253-3566, fax (607) 253-3083, e-mail hom1@cornell.edu; and O. Cerf and Moez Sanaa, Epidemiology and Animal Health management Laboratory, Alfort Veterinary School, Maisons-Alfort, France

Cryptosporidium parvum has been recognized as one of the coccidian that has a zoonotic impact affecting both animal and man. Its occurrence is worldwide. During the past decade C. parvum infection has emerged as one of the most significant waterborne pathogens in lieu of the several outbreaks reported and the number of illness encountered. The disease has been associated with severe illness and mortality in immunocompromised population.

Since 1990, we have been studying the risk that C. parvum poses to people in a defined watershed that serve a large community, New York City. We have been carrying out epidemiologic risk analysis studies that were designed to address the above-perceived risk. The epidemiologic risk analysis approach we undertook has three complementary components: risk assessment (identification), risk communication, and risk management. The risk analysis approach allow us to design a systematic strategy where the significance of the perceived risk is evaluated, this risk is communicated to the concerned, and cost-effect strategies were devised and implemented to reduce or manage this risk.

In this manuscript we will focus on the risk assessment component of our studies. We hypothesized that there are several potential sources for this organism: man, domestic animal, companion animals, and wildlife. Since these sources are present in the watershed, they have the potential to contaminate the surrounding environment from which C. parvum transported to the stream edge and enter the water supply system. A hierarchical series of risk identification models (within sources, from sources to the environment and the risk of human exposure) were developed. Data on putative risk factors affecting the risk at each source were collected and evaluated. The logistic regression statistical procedure was used to identify a risk model for each source. The risk model consisted of factors significantly associated with the presence of C. parvum in each source and in the environment. We used simulation models to complement the traditional deterministic risk models. The simulation models were intended to explicitly acknowledge the uncertainty and variability associated with the data in relation to the unmeasured putative risk factors.

We have collected and analyzed 10,862 samples from dairy animals, 4,000 samples from wildlife, and 870 samples from soil on farms. Our risk models have identified factors that were significantly associated with the likelihood of C. parvum in the source and in the environment. Some of the factors were associated with increased risk while other were associated with decreased risk.