Arsenic in Apple Juice

An Abbreviated History of Arsenic in Food

Arsenic has always been a major contaminant of interest at the FDA and for its predecessor the Bureau of Foods.  It was discovered that fish has unusually high levels of arsenic about 100 years ago, and it was determined soon thereafter that the arsenic in fish is far less toxic than a form well known to be toxic (arsenic trioxide; Coulson et al, 1935).   It was later determined that the main form of arsenic in fish is an organic form that is far less toxic than the inorganic arsenic that occurs in drinking water and other foods (WHO, 2011).  A survey conducted by the FDA in 1977 identified many other sources of arsenic in food, with rice standing out as having relatively high concentrations (Jelinek and Corneliussen, 1977).  However, before the 2001 Arsenic Drinking Water Rule, the main source of exposure to inorganic arsenic in the United States was drinking water, which mitigated concern for lower exposures coming from food.  Drinking water still is the dominant source of exposure for individuals with relatively high levels of arsenic (e.g. in unregulated well water or systems that are near the upper limit of 10 ppb).   But, on the average, exposure to inorganic arsenic from food is greater.

Doctor Oz Squeaks

Since rice is the major source of dietary exposure to inorganic arsenic, that was likely to become an issue.  But, in September 2011, the food supply wheel squeaked elsewhere when Dr. Oz ran a segment on arsenic in apple juice on his popular television show.   The focus on apple juice arose for several reasons: 

• A Consumer Report survey determined that some apple juice is above the EPA standard for drinking water.  The survey was fine, but using the tap water standard for apple juice is not.  Since the EPA level was determined with a cost-benefit analysis, it would be more appropriate to use the same policy-making methodology for apple juice. 
• Apple Juice is predominantly consumed by children.  This is true.
• Since they consume more food relative to their body weight, children are more highly exposed.  This is also true – for arsenic and just about everything in food.
• Children are more susceptible to arsenic than adults.  There is some evidence that children are affected at lower doses than adults, but there is also evidence to the contrary (i.e. both children and adults are both affected more or less equally). 

But, when the wheel squeaks, the soundness of the reasoning behind it matters little.  Something must be done.  However, there are different strategies that can be employed.  For example, it could be explained to the public that regulating contaminants is hard and that the agency has limited authority to regulate risks that are largely theoretical.  Or, the commissioner could promise an action level that would protect the public from arsenic in apple juice.  A risk assessment of some sort would be useful for either approach.  The FDA went with Plan B. 

  

The Risk Assessment

Even though they are very different decision making processes, the safety assessment paradigm and the risk assessment paradigm are both often referred to as risk assessment.  Thanks, EPA.    So, asking for a “risk assessment” without carefully specifying exactly what is expected can yield something that isn’t what is expected.  But, since the agency was already using a safety assessment-based methodology for arsenic in apple juice that identified a tolerable level of 23 ppb, the only other thing to do was a quantitative risk assessment.   The risk assessment that was produced (Carrington et al, 2013) was about as uncontroversial as possible, or at least that was the intent.  In addition to the dose-response model used for the EPA Arsenic Drinking Water Rule, a second model was developed that used newer data, dose estimates from a WHO (2011) evaluation, and a probability tree to represent the uncertainty associated with extrapolation.   The results using either model were quite similar, but with greater uncertainty acknowledge with the FDA model.

The Right Handed Shell Game

Almost as soon after the Arsenic Drinking Water Rule was instituted, a draft EPA cancer dose-response model that predicted roughly 17 time greater risks than the previous model released in 1988 and about 6 time greater that the unofficial model used in the drinking water rule and the model developed by the FDA for the apple juice assessment.   Although the draft model was never finalized and official abandoned in 2011, it has been widely quoted by consumer advocacy groups.   Therefore, the FDA risk assessment was criticized for underestimating the risk.

However, the far stiffer opposition came from the juice and rice industry, who had strangely allied themselves with something called the “Arsenic Science Task Force” that also represented the interests of arsenic-based wood preservatives.  That was dumb.   If I were a food industry advocate, I would try to argue that the arsenic in our product a) occurs naturally, and b) we would be delighted to get rid of it if only we knew how.  Standing with an industry that is selling arsenic doesn’t send that message at all.

In any case, when Office of Management and Budget (OMB) circulated the assessment for interagency review, the Small Business Administration (SBA) represented the forces of arsenic by playing the right-handed version of the shell game, which is pretty much played the same way as the left handed version:  Political arguments are met with scientific arguments, while scientific arguments are met with political ones.  The same litany of false dichotomies are recited.  The objective of preventing the release of intelligible information is also pretty much the same.  The only real difference is the status quo that is to be preserved.  Instead of starting with the presumption that it is necessary to prove safety, the right handed game starts with the presumption that it is necessary to show harm.

Since it purveyed As-Likely-As-Not as the central message, along with confidence intervals, the FDA risk assessment was impervious to the shell game, no matter which side it is played from.  But that didn’t stop the FDA and the SBA from talking past one another for half a year or so while OMB listened.  Part of the problem was that the more or less canned criticism from the SBA presumed that the risk assessment followed the 2005 EPA guidelines, when in fact it didn’t.  In particular, the FDA dose-response model included a probability tree that acknowledged the possibility of both “linear” and “nonlinear” dose response models.  But, the SBA didn’t understand that, probably because they didn’t want to.   Eventually, OMB got tired of listening and let the risk assessment go.  

References

Carrington CD, Murray C, and Tao, S. (2013). A Quantitative Assessment of Inorganic Arsenic in Apple Juice. 

Coulson EJ, Remington RE, and Lynch KM (1935).  Metabolism in the rat of the naturally occurring arsenic of shrimp as compared with arsenic trioxide.  J. Nutr. 10: 255-270.

Jelinek CF and Corneliussen PE (1977).  Levels of arsenic in the United States food supply.  Environ Health Perspect. 19: 83–87.

World Health Organisation (2011).  Safety evaluation of certain contaminants in food. Arsenic (addendum).  WHO Food Additives Series 63: pp. 153-316.

Official Post Soundtrack

Ocean Blue (1989).  Between Something and Nothing.  In: The Ocean Blue, Track 1.

Post Notes

Post Thesis #36.  This is part one of two risk management theses concerned with AJ. Technical details concerning the risk assessment will be discussed in separate post on a separate thread (or read the 2013 assessment).