Methods for projecting long-term dietary exposure from short-term survey data for environmental contaminants

Public health risk assessments often involve dietary exposures over long periods of time. However, most information about dietary consumption habits comes from short-term surveys that are conducted for periods of three days or less. When employed for characterizing long-term exposures, short-term surveys are likely to underestimate the number of persons consuming a particular food, while overestimating the amount consumed by each individual. Direct application of short-term data is particularly misleading for foods that are consumed infrequently. If a more accurate population estimate for chronic dietary intake is needed for a risk assessment, then two general techniques may be considered. The first method is simpler, while the second is more accurate. Both methods require information about the size of the population consuming the food over the long-term period. The simpler fractional adjustment method reduces consumption across the entire distribution by the ratio of consumer population sizes. Since this method will tend to underestimate high-end exposures and overestimate low-end exposures, it is most useful as a quick bounding exercise. Since short-term surveys are better at characterizing the behavior of frequent consumers, a second method employs an exponential function to reduce the low end of the population distribution by a greater amount than the high end. If available, additional information may be used to select the parameter values for the exponential adjustment. Otherwise, an uncertainty range may be used for the parameter values. Since the frequency-based method is more complex, it is most valuable when used as part of a chronic exposure simulation. Examples of both methods are given for the estimation of chronic wine consumption.