Science is an ongoing process where new discoveries change old theories. For example, when I was in school, we learned that atoms were the building blocks of matter, and atoms could only be broken down into three smaller particles: electrons, protons, and neutrons. Now we know that there are many types of elementary particles much smaller than these three particles.
Another example is the 1992 Food Pyramid which was held up as the gold standard for nutrition. Much of the nutritional advice given based on the Pyramid has since been debunked; some foods that the Pyramid vilified have now been shown to be good for you.
Environmental Risk Assessment
Environmental Risk Assessment, an important section of toxicology, the science of poisons, evaluates the human population's risk from low levels of environmental toxins. Since the 1980s, scientists have used two models in environmental risk assessment.
Traditional Models
The Threshold Model: This model is used for chemicals that have not been shown to cause cancer and 
assumes a threshold dose below which there is no risk of harm.
What does the chart describe? In simple terms, the blue-angled line shows us that as the amount or “dose” of a chemical increases, the adverse health effects also increase. The dotted horizontal line demonstrates a control subject's adverse health effects, not receiving any dose of a chemical but living in the same environment. The point on the line just before an adverse effect on the treated individual, where the blue angled line meets the dotted horizontal line, is what risk assessors call the “no-effect-level” – the highest dose where no adverse health effects are seen. Every dose above this causes adverse health effects.
Most of the time, the studies using this model are done with laboratory animals, usually rats or mice. If rats or mice were the same as people, the dose chosen as the “no-effect-level” would be used as the “safe” level in people. But laboratory animals are not people, and scientists cannot quantify the differences between the two. Risk assessors divide the observed “no-effect” level by an “uncertainty factor” that results in the lowering of the “safe” level (often referred to as taking a conservative approach). Traditionally, the “uncertainty factor” used for an animal study is 100 or 1,000. The uncertainty factor is not science we can measure; it is a value judgment that lowering the number is beneficial to public health.
The Linear Non-Threshold Model: This model is used for chemicals that have been shown to cause cancer 
and assumes no safe level of exposure at which there is no risk of harm.
What does this chart demonstrate? It shows that every increasing dose of the chemical results in an adverse health effect. There is no threshold, no “no-effect” level. The data that forms the line again comes from animal studies that investigated cancer. Because the doses used in animal studies are often much greater than those in the environment, the model's amounts would be on the far-right side of the angled blue line. They then draw a straight line from these doses to the origin and look at where lower doses fall on the line.
This model is used to estimate the level of a chemical that results in an “acceptable” increase in cancer cases in the population. The slope of the line can be converted into excess cancer deaths in populations of varying sizes. While risk assessors do not use an “uncertainty factor” as in the Threshold model, they introduce uncertainty about what is, in fact, a “safe” level in humans in two ways.
- By what they believe “acceptable” means.
- By how they draw that straight-line path, their extrapolation.
How do we know that chemicals follow this straight-line path? We don’t. This is the primary source of uncertainty in this model. The Linear Non-Threshold model is used because it results in lower “safe” doses than other models – like the Non-Threshold model, a choice based on a value judgment that one should take a conservative approach in determining “safe” levels of these chemicals.
Disruptive Thought – 3rd Model
In the 1980s, Dr. Ed Calabrese, a toxicologist at the University of Massachusetts, first described the phenomenon of hormesis, which he observed in animal studies. Hormesis (from the Greek word ‘to excite’) describes the phenomenon where low doses of a chemical have adaptive and sometimes beneficial effects as opposed to high doses that have toxic effects. Dr. Calabrese developed the hormetic model of chemical behavior, also known as the biphasic or J or U-shaped model. Unlike the first two models, a chemical's effect is not good or bad, but good and bad. The Hormetic model seeks to preserve the good and avoid the bad.
An example of hormesis is the ingestion of salt. The body needs a small amount of salt to contract and relax muscles, conduct nerves, and maintain the proper balance of water and minerals. However, too much salt can lead to high blood pressure, heart disease, and stroke. Another example of hormesis is alcohol consumption. Many studies have shown that light-to-moderate alcohol drinkers have decreased risks of dementia, diabetes, osteoporosis, and cardiovascular disease compared to abstainers. Of course, heavy alcohol consumption results in effects on the liver, brain, muscles, and many other organ systems.
How is this model used in risk assessment?
It isn’t.
Although this model is routinely used in drug development, medicine, and other scientific fields, it is not used in environmental risk assessment. If you talk to EPA scientists, the reason they will give is that there isn’t enough scientific evidence behind this model to warrant its use. However, I believe that this model is not in use because of scientist’s fear that it will predict higher “safe” levels of chemicals than the other two models – leading to the loosening of environmental standards. This goes against some scientist’s long-held belief that any level of a toxin is detrimental to human health.
It is also important to realize that the definition of “very low levels” of chemicals is constantly changing as our analytical capabilities increase. For example, two decades ago, very low levels of chemicals were considered to be in the parts-per-million (ppm) range; today, we are talking about parts-per-billion (ppb) and parts-per-trillion (ppt). The use of the two accepted models will keep predicting lower and lower levels of “safety” as our analytical instruments improve. The levels considered “safe” 20 years ago would be regarded as “unsafe” today, not based on changes in our understanding of how chemicals act, but simply based on our ability to measure lower and lower levels.
Conclusions
The traditional models used in risk assessment make use of empirically derived scientific data, but they come with really significant, unproven assumptions. The Threshold Model uses “uncertainty factors” to account for the difference in response between animals and people (a substantial unknown). The Linear Non-Threshold Model assumes that chemicals act in a linear fashion, which is not backed by current scientific thought. Like all models, the hormetic model also has assumptions built into it, but it has a scientific basis that equals or exceeds the other models. Scientists should be open to all evidence-based theories and not reject them out of fear of where their usage may lead. To deny its use is simply “cancel culture,” which has no place in today's scientific world.
