Skip to Main Content
An official website of the United States government
Epidemiology and Genomics Research Program

Identifying Extreme Exposure Values

There are various perspectives on whether to exclude potentially unlikely exposure values. If the researcher chooses to do so, several approaches exist for identifying extreme values. We examined the plausibility of the reported frequencies for each food item in the NHANES 2009-10 DSQ and chose to identify extreme values using a method that identifies them based on their actual distribution in the sample, but also minimizes the number of values identified.

Because all items have a real probability of being consumed zero or a small number of times, we focused on the higher ends of the distribution, and examined the times per day distributions. For each food item, we identified discontinuous points of the distributions. We defined the highest frequency before the discontinuity as the maximum acceptable value. Values above this maximum were top-coded (not excluded).

Definition of Extreme Values for DSQ in NHANES 2009-10

Food Group NHANES 2009-10 (N=8541)
Maximum Acceptable Daily
Frequency Value
Number of Identified
Values
Fruit 8 4
Fruit juice 8 8
Salad 5 0
Fried potatoes 5 1
Other potatoes 3 2
Dried beans 4 0
Other vegetables 5 0
Tomato sauce 2 1
Salsa 3 2
Pizza 2 0
Soda 8 23
Fruitades/sports drinks 7 6
Cookies, cake, pie 7 0
Doughnuts 5 0
Ice cream 5 0
Sugar/honey in coffee/tea 10 8
Candy 8 4
Any milk (not soy) 10 12
Cheese 6 2
Any cereal 7 1
Whole grain bread 6 2
Cooked whole grains 4 1
Popcorn 3 1
Red meat 6 2
Processed meat 4 1

Because the NHANES is a national representative survey and this screener was administered to children aged 2 years through adults aged 69 years, we expect this guidance to be appropriate for most U.S. populations.