Guiding Product Recall Strategy with Probabilistic Modeling

Product adulteration requires the application of a recall strategy as defined in 21 CFR § 7.42. The first step of this strategy is to conduct a health hazard evaluation. For allergenic substances, Congress put in place the Food Allergen Labeling and Consumer Protection Act (FALCPA) in 2004, which requires that food labels clearly identify the food source names of all ingredients that are or contain any protein derived from the eight most common food allergens—eggs, milk, fish, crustacean shellfish (e.g., crab, lobster, shrimp), tree nuts (e.g., almonds, walnuts, and pecans), peanuts, wheat, and soybeans.

The central tenet of the act is to protect consumers from ingesting products that intentionally contain one or more of the eight most common food allergens. However, inadvertent contamination or adulteration of unlabeled products with allergenic proteins does occur and poses a health risk to the allergic consumer because avoiding an eliciting allergen is the most successful way to prevent food allergic reactions.

Importantly, adulteration is defined by the analytical limits of detection that are currently available and, with the development of new and more sensitive techniques, the number of adulterated products will eventually increase, prompting more recalls. This increase will become an ever-growing burden on the food industry because eliminating all allergens from food is nearly impossible and there is no process in place to assess the health risk to consumers when allergens are detected at increasingly lower analytical limits. Thus, because limits of detection for allergens are analytically driven, the pivotal questions are how to quantify the population health risk posed by an exposure to an adulterated product and how to determine the appropriate risk management strategy for guiding a product recall.

It is now known that allergic individuals have “thresholds” or levels of exposure at or below which an adverse effect is not seen in an experimental setting (Taylor et al., 2002). Unfortunately, defining such thresholds has presented major challenges to physicians and food producers because allergen sensitivities vary greatly from person to person, and accumulating enough data to identify the lowest tolerated dose to a particular allergen with confidence has been difficult. In 2006, the Center for Food Safety and Applied Nutrition at the U.S. Food and Drug Administration (FDA) formed the Threshold Working Group to comb the data on the exposure of allergic subjects to allergens and evaluate the possible approaches to establishing food allergen thresholds (The Threshold Working Group, 2008). Although they identified four general approaches (analytical methods-based, safety assessment-based, statutorily-derived, and risk assessment-based), they concluded that that “the quantitative risk assessment approach provides the strongest, most transparent scientific analyses to establish thresholds for the major food allergens.”

Quantitative risk assessments provide numerical estimates of the chance of illness or death after exposure to a specific hazard, such as an allergen, that can be used to establish a threshold. Analytical methods-based, safety assessment-based, or statutorily-derived approaches rely either on a limit of detection, a no-observed adverse-effect level (NOAEL) and uncertainty factors, or an extrapolation from an exemption established by Congress for another purpose to establish a safe exposure limit. Quantitative risk assessments integrate dose-response relationships between allergen ingestion, the associated biological effects, and the inherent variability in the allergic population to estimate the risk of developing an adverse reaction to specific allergen. Furthermore, with the application of probabilistic modeling, quantitative risk assessments provide both the level of protection and the degree of uncertainty associated with an exposure level.

With quantitative risk assessments, it is then possible to utilize approaches adopted by the FDA for evaluating risk from exposure to substances without clearly defined thresholds, such as genotoxic and carcinogenic impurities in drug substances and products (FDA, 2008). For these impurities, the threshold for the estimate of daily exposure expected to result in an upper bound lifetime risk of cancer is less than 10^-6 (one in a million), and is considered to be a level at which the health risks to consumers are negligible.

In conclusion, quantitative risk assessments that utilize probabilistic modeling are important tools for quantifying the population health risk posed by an exposure to an adulterated food or supplement and are the preferred method for managing a company’s response to product adulteration, such as in the health hazard evaluation required as part of a firm-initiated or FDA-requested recall.

Dietrich B. Conze, Ph.D.
Contributing Editor
Senior Science Director, Spherix Consulting, Rockville, Md.
[email protected]

Claire Kruger, Ph.D.
Contributing Editor
President, Spherix Consulting, Rockville, Md.
[email protected]