Vol. 2 No. 2 September 11, 1981

The information in this newsletter was compiled from a series of seminars given in the Environmental Toxicology Department during the spring quarter of 1981.

Risk Assessment and Predictive Toxicology

A very important area in the regulation of exposure to toxic substances is the assessment of risk due to exposure. Risk assessment is a predictive science and as such it makes judgement calls utilizing current scientific information developed in toxicological laboratories. Therefore, the prediction of the risk associated with exposure is only as good as the scientific information backing it up. Toxicology is a "risky" business and this is particularly true with respect to setting exposure limits for workers, and acceptable daily intakes of food additives and pesticides.

Just what is risk? Some of the common dictionary definitions of risk are "a chance of loss", "the chance of injury, damage, or loss", "the degree of probability of loss". The terms risk and hazard are often used synonymously. One of the key concepts of risk is the idea of the probability of occurrence of an undesirable outcome (such as birth defects, cancer, death, etc.). The opposite of risk is safety. Risk and safety are often in an inverse relationship such that greater risk implies lesser safety, and lesser risk implies greater safety.

Risk is often expressed as a probability figure such as 1 in 10, 1 in a 100, 1 in a 1000, or 1 in a million. The greater the denominator, the less the risk. Some occurrences are so rare that they are called "Act of God" occurrences. Examples of "Act of God" occurrences are: being struck by lightening, dying in a tornato, or being struck by a meteorite. The risk associated with these "Act of God" occurrences is usually less than 1 in a million. Thus, these are sometimes considered to be minimal risk occurrences and our perception of these is that they are very unlikely to happen to us. Many of our daily activities expose us to situations in which there is a risk of injury or death that far exceeds 1 in a million. Examples of such activities are riding bicycles or driving cars. We are aware of the risk associated with these activities, and we voluntarily expose ourselves to it because they are safe enough means of transportation to be acceptable to us.

How do we arrive at estimates of the risk associated with various activities? One of the principle means is through epidemiological studies. Epidemiology involves the study of the incidence and distribution of health related occurrences. The first type of epidemiological studies is prospective studies which define populations and then follow them for a period of time to discover the incidence and distribution of an occurrence within the different populations. The populations involved in the study are chosen specifically for differences in their make- up, such as genetic background, diet, geographical location, or other environmental factors that alter their life style. Prospective studies are undertaken when there is a question or evidence that a particular activity may be associated with an increased incidence of a particular disease. Some very long term prospective studies are being done currently in the area of heart disease. Much useful Information has been gained from these studies regarding the influences that diet and lifestyle have on the incidence of heart disease in humans. Thus, prospective studies look forward into the future. The second type is retrospective epidemiological studies which look back into the past to discover factors which may have influenced the incidence and distribution of a particular disease process. Retrospective studies are particularly effective in finding increased rates of rare occurrences. It is very difficult to use retrospective studies to discover a small increase in the rate of a disease that has a high incidence rate. Both of these types of studies are very necessary to the establishment of base-line risk factors associated with every day life. Indeed the entire insurance industry is based on risk analysis.

The perception of risk or perceived risk is often unrelated to actual risk. Thus, our perception of a particular risk may lead to a reaction that is disproportionate to the actual possibility of occurrence. If we are involuntarily exposed to a particular risk, our perception of that risk is often greater than if we voluntarily exposed ourselves to it. Risk perception is also related to the degree of knowledge we possess concerning the particular event involved. Thus, the less information we have in a particular area, the more likely we are to over-estimate or under-estimate the amount of risk involved.

What is safety, and what is safe? Someone once said that safety is "acceptable risk". That naturally leads to the question as to what is an acceptable risk, and who is to determine what is an acceptable risk. If safety is indeed the inverse of risk, and risk cannot ever be entirely eliminated (reduced to zero), then it follows that nothing can be absolutely safe.

Predictive Toxicology

Predictive toxicology is the science of risk assessment based on experimental data. It involves toxicology testing in experimental animals, interpretation of the data of these tests, and subsequent risk estimation based on the test data. The types of information required to adequately predict the risk of the exposure of humans to a particular chemical compound include complete acute and chronic toxicity data, data on the no- observable-effect-level (NOEL), the mechanism of action of the compound, the metabolism of the compound, the dose-response relationship of the compound for acute and chronic toxicity, and an estimate of human exposure to it.

In extrapolating animal data to human populations, a safety factor of 100 is often used in setting human exposure limits. The NOEL in experimental animals is determined, and human exposure levels set at 100-fold less. This factor of 100 is supposed to take into account a ten-fold factor for variation within the human population, and a ten-fold factor for variation between animals and humans. Depending on the slope of the dose response curve for the various toxicity parameters, this may or may not be a good estimation. This type of approach is theoretically sound for compounds which are considered to have a threshold dose. With respect to carcinogenesis, some chemicals are not considered to possess a threshold dose level below which cancer will not be produced. In such cases, the dose response curve is considered to be linear in the low dose range and the line may be extrapolated to exposure levels at which the incidence of cancer is 1 in a million or less.


The results of all the toxicity tests that the EPA requires new pesticide products to undergo are used to predict the risk (or hazard) of the pesticide to humans, domestic animals and wildlife. The prediction can be only as good as the tests, the data and the predictors judgement. The toxicity test models often undergo revision and improvement in order to enhance their predictive validity. Simplification of these tests is also sought to speed-up the toxicity screening process and decrease its cost.

Predictive toxicology is not an area that easily follows cut-and-dried rules and regulations. There are too many factors involved that have to be considered. The determination of safe levels of exposure to chemicals is a complex area involving science, philosophy and economics. Because of this, views of "acceptable risk" (safety) will always be related to each individuals subjective bias.



(From Human and Veterinary Toxicology, Vol. 23(4), August 1981).

For more than a year an SOT task force chaired by William W. Carlton has been carefully reviewing the US Food and Drug