A Response to Consumer Union's "Do You Know What You're Eating?"

Dr. Douglas G. Pfeiffer, 205C Price Hall
Department of Entomology, Virginia Tech

I. Background:
The current issue (March 1999) of Consumer Reports contains a brief account of the recent study performed by Consumers Union. Although this article ("How Safe is our Produce?") probably will be much more visible to the public than the full accounts, it is really impossible to assess much about it because of its brevity. The claims of the short version are that current laws do not protect children from consuming dangerous amounts of pesticides, that such consumption is fairly common, and that one pesticide in particular, methyl parathion, is a major contributor to the problem of residues danger. The shortcomings are not readily apparent until the full account, entitled "Do You Know What You're Eating?" is examined. A summary is posted on the web in html and the full report in PDF format.

This report used USDA data (these are large files, 9-39 MB) on survey of pesticide residues from 1994-1997. The authors reformatted the data, designing acute and chronic toxicity indices, and deriving an overall Toxicity Index (TI).

This interpretation of data by the authors represents an attempt to provide an objective way of assessing pesticide risk in the food supply. One main point we are left with is that whenever a complex situation is reduced to a single number claimed to be an objective, quantitative descriptor, one should be skeptical. There are often subjective judgements and assumptions that enter into the underlying calculations. The approach has several disastrous flaws, both theoretical and practical, which I will discuss. There are also some strengths that I will mention.

II. Theoretical flaws:
Confusion between chronic exposure and acute exposure: In discussing chronic exposure, the authors use the Reference Dose, RfD. This is the dose below which, if consumed every day over a lifetime, there is no effect. But the authors repeatedly confuse this with an acute overexposure, e.g., when talking of a single peach, "...a young child can ingest more than a safe dose ... of a specific pesticide by eating a single serving..." This wording misinterprets an index for chronic exposure for the purpose of eliciting alarm.

Effect of multiplication by constants on spread of data: The authors multiplied the inverse of the LD50 by 100 in order to make whole numbers and eliminate decimal fractions (this was not in fact accomplished ‚ we were merely left with decimal fractions 100 times bigger, like 0.200 instead of 0.002.) Conversely, in calculating the chronic toxicity index, the inverse of the RfD ranged from 6 to over 50,000. "To express the results on a more manageable scale, we multiplied them by 0.1." Multiplying data by a constant would not change the significance of statistical differences in a real test. However, it would affect the apparent spread of the data, and that is all that was used by the authors to define important differences. The authors state "...comparatively small differences (of 10-20 points or less on the TI scale) between scores for different foods are not very meaningful, statistically. Large differences...are not subject to this caveat." Since data were arbitrarily multiplied by different constants, it makes no sense to talk about distance of spread of data points unless accompanied by a statistical test.

Arbitrary assignment of critical levels of TI: The toxicity index provides a relative toxicity loading of each fruit in the study. TI levels in the study range from 0.01 to 5,376. The authors state that, "In our judgement, values greater than 100 on the TI scale show comparatively high pesticide contamination, and values less than 10 indicate that those foods are comparatively quite "clean". (Values in the range from 10 to 100 represent increasing degrees from "low" to "moderate" levels of pesticide contamination.)" What is the basis of saying that a score of 100 is associated with a real risk, greater than, say a score of 75?

A similar example of arbitrariness is the multiplication of chronic toxicity by a factor of 2, doubling its impact over acute toxicity. The rationale is that chronic toxicity is more relevant to dietary consumption. But this considers only consumer aspects. Farmworker safety would be better addressed by a greater emphasis on acute toxicity. The point is not whether farmworker safety is more or less important than consumer safety, but rather that a subjective decision was made in the derivation of this apparently objective quantitative assessment.

Arbitrary weighting of certain factors: In addition to chronic toxicity, roles of pesticides as endocrine disrupters and carcinogens are considered. If a product was suspected as an endocrine disrupter in a popular book on the subject, it is given a weighting factor of 3. This is a heavier weight than is given to carcinogenicity. The authors state that "Endocrine disruption is responsible for some of the most devastating documented effects of pesticides on wildlife... In our judgement, potential endocrine disruption is a more important aspect of a chemical's toxicity than even potential carcinogenicity, and our scoring scheme therefore gives it great weight." The concept of endocrine disruption has been more controversial lately, especially after authors retracted some early studies in this field. Is it justified to give this phenomenon such great weight, much greater than carcinogenicity? This was a value judgement by the authors.

III. Practical flaws:
Confusing use of terms for more than one meaning: It is difficult to use the data presented in tables to reconstruct TI's. This is especially troubling because of possible errors exist in initial assumptions (see below). The authors translate the LD50 into an Acute Toxicity Index by multiplying the inverse of LD50 by 100. (ATI = (1/ LD50)X100).

The authors then multiply the ATI by the mean residue, then by the percent positive results to obtain a statistic also called the Acute Toxicity Index (ATI). Similar terminology problems exist with the Chronic Toxicity Indices (CTI).

Methyl parathion: Methyl parathion is considered to be a risk driver on many of the crops considered, accounting for a large portion of total risk. But there is an incorrect starting point in risk calculations based on incorrect LD50. (LD50 is a measure of toxicity, expressed in the amount of active ingredient it takes to kill 50% of a laboratory population, usually rats or mice, and expressed in milligrams per kilogram of body weight). According to Extoxnet, the LD50 for methyl parathion is 6-50 mg/kg for rats, 14.5-19.5 for mice, 420 for rabbits, and 1270 for guinea pigs. The LD50 for ethyl parathion is 2-30 mg/kg rats, 5-25 for mice, 10 for rabbits and 8-32 for guinea pigs.

In Table 2 of the report, the LD50 is given as 14 mg/kg for BOTH ethyl and methyl parathion, with a resulting Acute Toxicity Index of 7.14 for both. Methyl parathion is less toxic than ethyl parathion; this is why ethyl parathion was removed from the market, leaving methyl parathion available for use. If a LD50 level only somewhat closer to the upper end of the published range had been chosen, say 50 mg/kg for rats, the Acute Toxicity Index would have been 2.0, rather than the authors' 7.14. This change alone would have reduced the overall chronic toxicity index by more than two-thirds.

When the effect of formulation is considered, the situation is even more complex. Methyl parathion is normally applied as Penncap M, a microencapsulated formulation. This greatly enhances safety to farmworkers and nontarget species. The oral LD50 of Penncap is more than 2000 mg/kg, about 40 times safer than the technical material. (It should be remembered however, that this encapsulation would also act to extend life of residues on fruit. This fact is already accounted for in the report through the data on percent positive findings, and mean residue levels).

IV. Strengths of report: Despite its fatal flaws, there are some merits to this report. Relative estimates of toxicity or environmental impact have a place in discussions on pesticide policy. The discussion of imported versus domestic produce helps to dispel some notions that American food is much cleaner than imported produce. Spray patterns leaving relatively greater residues could be singled out for further examination. However, the benefits of the discussion on relative toxicity would have a more appropriate place in helping farmers modify their practices, rather than in an uncritical presentation to the public, instilling alarm rather than encouraging a smooth transition of practices.

V. Summary: The Consumers Union study contains theoretical and practical flaws that greatly limit its use in evaluating spray practices. The severity of residues is reduced to a single numeric descriptor, giving the impression of an objective measure, while masking subjective judgements of the authors. Many of the flaws could have been limited, and some of the benefits would have been greater, if this work had gone through the refereed process common in the scientific literature, rather than simple dissemination to the public. Peer review serves not only to keep poor science from being published, but also save authors from embarrassing mistakes that would have been easy to correct.

Dr. Jeff Bloomquist, a neurotoxicologist in the Department of Entomology here at Virginia Tech, has made me aware of a response to this report to the EPA from the Society of Toxicology.

Back to Apple page
Back to Mid-Atlantic Regional Fruit Loop
Send comments by e-mail to: Douglas G. Pfeiffer