Arsenic in Rice: Part 10 - Everyone loves Morales et al.

In my last post, I wrote that eating one serving of rice containing 9.6 μg of inorganic arsenic - every day - for 70 years - would present a risk of 7 in 1,000 excess bladder/lung deaths for females.  This, I wrote,  was 3 excess bladder/lung cancers more than what the "most protective level" of 5 μg/L would theoretically produce.

I want to be clear here and make sure that my point is not being misconstrued.  I am not making the point that 5 μg is not the "most protective level" and that it should be lower, say 0.003 μg/L.  On the contrary, what I am trying to get across is that 5 μg/L is not an appropriate number to establish a threshold.  Not because it is too high, but because it does not imply "safe" and "unsafe" like Consumer Reports and their experts are wanting us to conclude.

That number -  5 μg/L - is just a number to support getting the arsenic concentration down as low as we can to the stated goal of "zero."  If we understand environmental causes of cancer correctly, then anything above zero is a risk.  This does not have to paint us into a corner, but it does.

Why?  Because Consumer Reports implies that a serving of rice exceeding 5 μg/L is "troubling," "worrisome," "cause for concern," or "potentially harmful."  They assume that 5 ppb is "safe" and that above that is "unsafe."

"Troubling," "worrisome," "cause for concern," or "potentially harmful," was used by Consumer Reports to convey "unsafe" for two reasons in my opinion:

1. They do not understand what New Jersey's 5 μg/L arsenic in drinking water represents, and;
2. They are pushing a zero threshold agenda

Again, so that my message is not misconstrued:

If Consumer Reports considers New Jersey's arsenic standard of 5 μg/L the "safe" threshold for rice....and;

Rice that has up to 5 μg/L of inorganic arsenic per serving would not make it "troubling," "worrisome," "cause for concern," or "potentially harmful."....and;

New Jersey considers 5 μg/L to be the "most protective level"...and;

Based on the EPA Oral Slope Factor derived from data produced by Morales et al,  5 μg/L would result in a risk of 3.6 excess bladder/lung cancers in 1,000....

Then Consumer Reports agrees that a risk of 3.6 out of 1,000 excess bladder/lung cancers is most protective of the population that consumes one serving of rice per day for 70 year.

You can see why I take issue with the conclusions presented by Consumer Reports and the experts.  They imply that on this side it is "safe" and on that side it is "unsafe."  It is neither.  Which is why using an MCL should never be used to present to the public a comparison of the ranges where the MCL is exceeded.

Consumer Reports is wrong, wrong, wrong, on this.  It does not provide a "yardstick" - it provides the public nothing but confusion and concern.  Which is exactly the point Dr. Honeycutt with the TCEQ was driving at:

Proceeding with this SFo [Oral Cancer Slope Factor from Morales et al.] will unnecessarily alarm the public by giving a greater perception of harm and risk than is actually taking place.

By using red ink in their tables, as well as using the terms "troubling," "worrisome," "cause for concern," or "potentially harmful," Consumer Reports has alarmed the public by supporting a greater perception of harm and risk than is actually taking place.  This is inexcusable for a magazine of this caliber and extremely unsettling that their toxicological experts were not knowledgeable of this fact.

So let's get back to it shall we.  Is inorganic arsenic in a serving of rice that is greater than 5-ppb but less than 10 ppb "troubling," "worrisome," "cause for concern," or "potentially harmful?"

Only is you think an increase of risk from 3.6 in 1,000 to 7 in 1,000 is substantial enough.  And that increase is also dependent on consuming a serving of rice each day for 70 years with 9.6-ppb inorganic arsenic (we will ignore for now the fact that the average concentration of arsenic found in all the samples was below 5 ppb and assume that each day the rice contains the maximum level found - 9.6 ppb).  Oh, and another thing we need to consider is that the risk, from 3.6 to 7, is dependent on the Oral Slope Factor that was developed using data from Morales et al. (2000).

Let's look at that Morales et al. data, shall we?  First off, I am not trying to make a case that the data is - or is not - valid.  I will assume it is.  What I want to show is how much uncertainty there is in the final Oral Slope Factor that was calculated.  This uncertainty makes the increase in risk from 5-ppb to 10-ppb impossible to calculate with any degree of accuracy.  Yet, there it is, used by Consumer Reports to draw a line in the sand, a threshold held up as sacrosanct.

Here is what the EPA has to say about the data they used from Morales et al.:

The calculation of cancer risks from the Taiwanese epidemiological data was performed using Excel workbook files. The files contained the input data for the dose-response models and spreadsheets to accept user-specified inputs, perform calculations, and summarize outputs from the assessment. Input data included male and female lung and bladder cancer mortality and person-years at risk (PYR) data for arsenic-exposed populations from 42 villages obtained from Morales et al. (2000), village water arsenic concentrations (minimum, median, and maximum data sets), and southwest Taiwan and all Taiwan reference population mortality and PYR data.

You can download those Excel files here.  Digging a bit deeper, we get this from the EPA in the IRIS draft report:

The Morales et al. (2000) ecological investigation re-analyzed data originally reported by Chen et al. (1988a, 1992) and Wu et al. (1989) from 42 villages in the arseniasis-endemic region of southwestern Taiwan by considering the number of liver, lung, and bladder cancer deaths.  Morales et al. (2000) used a generalized linear model (i.e., Poisson distribution) and the multistage-Weibull models to determine lifetime cancer risk estimates. Liver, lung, and bladder cancer mortality data were collected from death certificates of residents in 42 villages during 1973 through 1986.

Okay, they collected data from death certificates...

Drinking water samples had been collected from wells in the 42 villages between 1964 and 1966.

...and Morales et al. used sample data from 1964 through 1965.  Well...okay...

SMRs [standard mortality ratio] were used to summarize the observed patterns of mortality in the collected data. Morales et al. (2000) selected two comparison populations (the Taiwanese population as a whole and a population from a southwestern region of Taiwan) to account for urban versus non-urban populations differences.

...okay...standard toxicological stuff...

Although a non-significant trend was observed in the combined cancer analyses with respect to age, there was no observed tendency in liver, lung, or bladder SMRs with respect to age. This suggests that there is no age dependency on the risk ratio.

...okay, that takes out one confounding issue...

The Morales et al. (2000) investigation results showed that exposure-response assessments were highly dependent on the choice of the analysis model and whether or not a comparison population is used in the analysis.

And the reason?....

One possible explanation for this observation is the inherent uncertainty associated with the limitations of an ecological study design. Depending on the model used and the comparison population used in the analysis, the effective dose at the 1% level (ED01) estimates ranged from 21 to 633 ppb for male bladder cancer, and from 17 to 365 ppb for female bladder cancer. The lung cancer risk for males was found to be slightly higher than the bladder cancer risk, with ED01 estimates ranging from 10 to 364 ppb. The risk for female cancer tended to be higher than that of males for each cancer type. For lung cancer, female ED01 estimates ranged from 8 to 396 ppb.

Wait...the number used to calculate the Slope Factor is based on the 1% effective dose (ED)...and that number ranges from 21 to 633 ppb, 17 to 365 ppb, 10 to 364 ppb, and 8 to 396 ppb for the two cancers in men and woman?

And they calculated a Slope Factor that was used to support a standard of 5 ppb for arsenic in drinking water.  And Consumer Reports wants me to believe that exceeding 5 ppb up to a maximum of 10 ppb in a serving of rice is "troubling," "worrisome," "cause for concern," or "potentially harmful?"

Really...are you serious Consumer Reports?  You are basing it on a threshold for arsenic in drinking water that uses data with this in play:

Weaknesses include the ecological study design (i.e., there were no individual monitoring data and individual exposures were not available) and the fact that potential confounders such as smoking, dietary arsenic, and the use of bottled water (U.S. population) were not controlled for in the analysis.

And in spite of these weaknesses, the EPA and New Jersey, and the NRC marched forward with a Slope Factor.  Which is exactly why Dr. Honeycutt with the TCEQ made this statement:

Proceeding with this SFo will unnecessarily alarm the public by giving a greater perception of harm and risk than is actually taking place.

So how "off" can this excess cancer risk be?  Glad you asked that, now I can move on to the 11th post.


Next post:  Arsenic in Rice: Part 11 - Dose-response is the cornerstone of toxicology