Linear extrapolation

From Testiwiki
Jump to: navigation, search

Linear extrapolation: a straight-line projection of an effect to smaller doses (see also Extrapolation). Linear extrapolation is a simple way of quantifying a toxic effect (usually cancer) at low doses, which cannot be tested reliably. Cancer rate is measured at a high dose, and found to be e.g. 10 % (a cancer in every tenth animal). An assumption is made that the rate decreases towards low-dose range linearly at the same rate as the dose. In other words, one hundredth dose gives the number of tumours that is one hundredth of the measured number (one hundredth of 10 % is then one out of 1000), and so forth. Linear extrapolation means that there is no safe dose ("one molecule can cause a cancer"). While intuitively attractive, this theory is probably false thanks to our defence mechanisms, and carcinogenicity is not different from any other form of toxicity. The actual risk is probably lower than the estimate achieved by linear extrapolation, hence this method is conservative. In theory also one tuberculosis bacillus can cause the disease, but experience has taught us that the dose of bacteria is highly relevant, and in real life one bacillus is easily destroyed by the body and cannot cause tuberculosis. [1]


  1. Jouko Tuomisto, Terttu Vartiainen and Jouni T. Tuomisto: Dioxin synopsis. Report. National Institute for Health and Welfare (THL), ISSN 1798-0089 ; 14/2011 [1]