Conservative risk assessment
This page is a encyclopedia article.
The page identifier is Op_en3471 |
|---|
| Moderator:Henrik (see all) |
|
|
| Upload data
|
Conservative risk assessment: type of risk assessment which maximises the expectation of risk in order to make sure that the true risk is always below the estimate. In risk assessment one has often to act in the state of uncertainty. Because many risk assessors prefer to err in the direction of exaggerating the risk rather than in the direction of not appreciating the risk, the worst possible prediction is often taken as the basis of risk evaluation in different steps of risk assessments. An example is using 95 % upper confidence limit rather than the most probable (average) risk level as the basis of likelihood of a deleterious effect. Another example is the so called Linear extrapolation of cancer. It means that at one tenth dose of a carcinogenic chemical the number of cancers is also assumed to decrease to one tenth, at one hundredth dose to one hundredth, and so on all the way to zero level. An alternative way would be to assume a safe dose below which there is no cancer any more. Neither way of evaluation can be scientifically proved to be correct, but in conservative risk assessment the worst possibility is taken to be true. Conservative risk assessment has been criticised on several grounds. One is that crying the wolf all the time will inflate the message. The other point is an imbalance of risk evaluation, because conservative risk assessment is possible in some areas, e.g. in pesticide or dioxin risk assessment, but not in others such as air pollution or alcohol. This then may lead to wrong priorities, e.g. to overemphasising pesticide or dioxin risks while neglecting air pollution risks. Thirdly, the more uncertainty there is in the estimate, the higher the final estimate tends to be, while accurate estimates with little uncertainty tend to be lower. This results in systematic underrating of well-known risks, even if they were relatively high.
[1]