Estimating risk assessment exposure point concentrations when the data are not normal or lognormal

Area of Science:

• Environmental Science
• Risk Assessment
• Geospatial Analysis

Background:

• The U.S. Environmental Protection Agency (EPA) recommends a specific method for calculating the exposure point concentration (EPC) in Superfund risk assessments.
• This method, based on the 95% upper confidence limit of the mean assuming normal or lognormal distribution, can lead to overestimation of contaminant levels.
• Overestimation can result in unnecessary cleanup actions at hazardous waste sites.

Purpose of the Study:

• To investigate and compare alternative methods for calculating the exposure point concentration (EPC).
• To address limitations of the EPA's recommended approach when contaminant data do not fit normal or lognormal distributions.
• To identify suitable EPC calculation methods for diverse soil concentration datasets.

Main Methods:

• Comparison of several alternative EPC calculation methods.
• Utilized soil data from three hazardous waste sites in Montana, Utah, and Colorado.
• Evaluated methods including Chebychev inequality, Wong's method, studentized bootstrap-t, Hall's bootstrap-t transformation, and parametric bootstrap.

Main Results:

• The EPA's method performs adequately only when data are well-fit by a lognormal distribution.
• For lognormal data, Chebychev inequality or the EPA method may be appropriate.
• Wong's method is suitable for gamma-distributed soil concentration data.
• Bootstrap methods (studentized bootstrap-t, Hall's bootstrap-t, parametric bootstrap) are recommended for poor distribution fits.

Conclusions:

• The choice of EPC calculation method significantly impacts risk assessment outcomes.
• Alternative methods offer more accurate EPC estimations when data deviate from standard distributions.
• Recommended methods include Chebychev, Wong's, and various bootstrap techniques based on data distribution fit.