Does Saccharin Cause Cancer? Do Pigs Fly? Ask Congressman Delaney.

The Delaney Amendment guarantees the production and acceptance of false theorems by the FDA.


Over the years, members of the scientific community have amused the public with systematic dissections of the claims of astrology to scientific status. There is little evidence that individuals actually allow their substantive personal decisions to be influenced by the wanderings of the planets or by those who claim to interpret them. Nonetheless, there is a great deal to be said for demonstrating that some arguments are sadly fallacious.


This concern for the purity of public discussion raises a question. Why has not the claim of Congress, through the Delaney amendment, to make decisions for individuals been subjected to the same analytical scorn as the claim of Omar the Weird to make such decisions? What should bother scientists as scientists is not that the Delaney amendment treats adults as if they are witless children. Nor that there is no attempt to estimate the human costs of substituting for or doing without particular additives. What should shock a scientific conscience is that the Delaney amendment guarantees the production and acceptance of false theorems.

The amendment instructs the FDA to ban any food additive that can be shown, via "tests which are appropriate," to cause cancer in either humans or experimental animals. But Congress neglected the elementary logic of the problems that must be faced in proofs that one state of affairs causes another.

To prove that x causes y, it is not sufficient to show simply that after x has been added to the scene, y results. To think otherwise is to fall victim to one of the hardier of the fallacies of pseudoscience. If I may borrow an example from one of the century's premier logicians, Lord Bertrand Russell, let us suppose we wish to prove the hypothesis that pigs have wings. Since pigs are somewhat unsavory, it would be desirable to test this thesis without actually examining a pig. To do so, we might infer from the hypothesis a testable proposition—for example: some warm-blooded animals fly. Looking around, we discover, lo and behold, that there are indeed some warm-blooded animals that fly. From this can we conclude that pigs have wings? We have diligently tested the proposition implied by the hypothesis. It seems true. Still, there is something troubling about affirming the hypothesis on this basis.

The fallacy is easy to spot, here, because we know something about pigs. They simply do not have wings. The birds and bats which fly are not pigs. How can we prove a theorem about pigs by looking at birds or bats? Of course, we cannot (with apologies to Nelson Goodman). We cannot, anyway, without first proving that what birds and bats can do, pigs can do too.


Now let's apply this example to the Delaney amendment. We wish to prove the hypothesis that some food additive causes cancer in humans. Again, it is desirable to test, not the hypothesis itself, but a proposition inferred from it—not because humans are unsavory, as is the case with pigs, but because it is undesirable to set out to cause cancer in humans. So we test a derivable proposition: the food additive causes cancer in some animals. One group of test animals is fed the additive, and the cancer rate in that group is compared to the rate in a control group of (the same kind of) animals that were not fed the additive. Suppose we discover that the cancer rate is indeed higher in the additive group. Have we proven a hypothesis about humans by experimenting with other animals? We cannot do so unless we have proven that what happens in other animals also happens in humans—just as we cannot prove a theorem about pigs by looking at birds or bats without first proving that what birds and bats do, pigs do too.

Thus does the Delaney amendment fall into the fallacy so easily perceivable in the flying pigs "proof." As evidenced by its instructions to ban any food additive found to cause cancer in humans or animals, it assumes that what happens in some animals will happen also—or is very likely to happen—in humans. Yet this goes unproven, not only in general, but in the particular case of the happening of cancer.

Specifically, note the problems with this assumption in the recent Canadian study of saccharin in rats, which has raised such a public furor. According to the study, ingestion of this food additive increased the incidence of bladder tumors in the rats in the experiment. Science magazine (June 10, 1977) points out some scientists' reservations; "when it comes to bladders," they note, "rats are a special breed." Because of rats' distinctive metabolism, chemicals tend to remain in their bladders for a relatively long period before being excreted. This, in turn, leads to frequent development of bladder stones, believed by some experts to cause tumors via physical irritation. In addition, rat bladders are commonly hosts to parasites associated with tumors. While these factors would be constant (statistically) across all rats, they call into question the relevance to humans of finding saccharin-induced cancer in rats. In assuming the general relevance of other animals to humans, the Delaney amendment assures the fallacious acceptance of hypotheses.


And there are other logical problems, having to do with even proving that one state of affairs causes another. A sad fact of life in science is that we cannot observe a causal relation—not in the same intuitive sense that we can look at a pig to see whether it has wings. Even if cancer shows up when testing the derivable proposition, the difficulty is in determining the cause of the cancer.

The problem is to test whether a food additive, per se, causes cancer. How do we discriminate between this possibility and the possibility that the food additive when combined with something in the test caused the cancer? The Delaney amendment does not require that the FDA worry about the possibility that the food fed the animals in conjunction with the food additive caused the cancer. Nor that these, in conjunction with the genetic strain of the animals, caused the cancer.

Indeed, it is now known that saccharin, per se, does not invariably cause cancer in experimental animals. The Science article cites ongoing studies of hamsters and monkeys which show no increase in cancer after long periods of saccharin ingestion. And there are problems with pinpointing saccharin as the cause in the Canadian rat experiment. It now appears, again according to Science, that the saccharin used was impure. Pure saccharin has not been found to produce mutations in a new and sensitive test-tube method of screening for carcinogenic substances. Perhaps even more damning, not all pathologists viewing the slides of the Canadian rat bladders conclude that they are viewing cancerous tissue! Similar issues would apply to the even more recently released study directly linking the use of artificial sweeteners and bladder cancer in males. In addition, until its procedures are publicized, it is impossible to assess the adequacy of the control group involved. These are some of the problems that plague attempts to prove that any x causes some y.


Those commenting on the FDA proposal to ban saccharin are of the opinion that the experimental evidence is mixed. Assuming that all the reported experiments can be replicated at will, what can we conclude from the fact that some experiments produced higher cancer incidence in saccharin-fed rats and some did not? It is vital to remember that each of the experiments has certain "background" features: the genetic strain of the animals, the food upon which they were fed, etc. Then we know that there is some set of background features and saccharin that does not result in higher cancer. We also know that there is some set of background features and saccharin that does result in higher cancer.

From this we cannot logically conclude that saccharin causes cancer. Indeed, we can just as well conclude that saccharin does not cause cancer.

To compound the intellectual disaster, the Delaney amendment does not require that the FDA estimate or base its decision on the functional relation between the additive and an increase in cancer. Does an extra one percent of saccharin bring about an extra one percent of cancer? An extra one millionth of one percent?


Consider two hypothetical additives A and B and their effect on cancer as charted below. With no consumption of either A or B, there is a unit amount of cancer. Clearly, there is a range over which A does not result in more cancer. Nonetheless the Delaney amendment would allow the FDA to accept and act upon the simple theorem that food additive A causes cancer.

The scientific methodology inherent in the Delaney amendment is appallingly shoddy. Perhaps that should be expected from science a la Congress. But it is sad that with all the generous expenditures on basic scientific research over the last decades, there is no widespread concern in the scientific community to protect the general public from the workings of a pseudoscientific process.

David Levy works for the National Planning Association in Washington, D.C. He advises that the views expressed here do not necessarily represent the views of anyone else.