A Computerized Model for Evaluating New Rainfall Closure Rules for Conditionally Approved Shellfish Growing Areas

Abstract

Agencies infrequently evaluate shellfish harvest closure rules for conditionally approved growing areas because of the logistics associated with analyzing large databases of accumulative rainfall and fecal coliform samples. The task is further compounded when the rules are composed using multiple variables, which require additional statistical analysis if a single variable is changed. We developed a computerized model capable of analyzing large datasets and performing statistical analyses for evaluating new rainfall closure rules for conditionally approved shellfish growing areas based on samples taken during critical sampling periods, and performing these tasks according to specifications of the U.S. National Shellfish Sanitation Program (NSSP). This model uses real rainfall and fecal coliform databases to calculate the upper limits of the geometric mean and the estimated 90th percentile at the 0.05 significance level for three periods: the open period under the current rainfall closure rule, the open period under the proposed new rainfall closure rule, and the critical period when the growing area is closed under the existing closure rule but would be open under the proposed new closure rule. The simulation portion of the model uses field-collected fecal coliform samples and rainfall datasets to evaluate the proposed new rainfall closure rule according to the NSSP standards. The second, statistical portion of the model generates a series of outputs that either approve or reject the statistical validity of the proposed new rule. The model has the option of extending the NSSP standards by including not only the estimated 90th percentile and geometric mean but also their upper limits at a significance level of 0.05. This extended method incorporates the effect of sample size and introduces greater precision by using these parameters, thus enhancing the decision making process and reducing the risks to public health. The model also requires a modification to the sampling procedure described by the NSSP to collect additional samples during a defined critical period. For a growing area in Arcata Bay, California, the upper limit of the geometric mean at the 0.05 significance level for the critical period is 4.0 MPN per 100 mL, which is less than 14. The upper limit of the estimated 90th percentile at the 0.05 significance level for the critical period is 14.8 MPN per 100 mL, which is less than 43. Therefore, the dataset for the new closure rule complies with the NSSP 14/43 standard at the 0.05 significance level, and the responsible agency could choose to approve the proposed new closure rule at the 95% confidence level. The model performed these tasks rapidly and accurately and was field tested by a public health agency.