ENBIS: European Network for Business and Industrial Statistics

Towards a safe and (statistically) sound SPF testing procedure

23 September 2009, 11:55 – 12:15

Abstract

Submitted by

Stefanie Feiler

Authors

Stefanie Feiler (1), Uli Osterwalder (2), Philippe Solot (1)

Affiliation

(1) AICOS Technologies, (2) Ciba

Abstract

For determining the sun protection factor (SPF) of a sunscreen product, it has to be tested on human skin. Here, the current American and European guidelines (FDA / COLIPA) also serve as models for many other countries. These standards prescribe test procedures where several skin areas per test person are treated with sunscreen and then are exposed to different doses of UVA/UVB radiation. The exposure times until erythema (reddenings) occur is then used for determining the SPF.
For several years now, there is an ongoing discussion on changing these standards. New test methods have been proposed, which only require a single exposed site per person and supposedly generate less erythema, i.e. reduce the risk for the test persons.
However, a thourough discussion from a statistical point of view has been lacking. It turned out to be an advanced application of binomial considerations (e.g. when determining confidence intervals). The behaviour of the methods is illustrated in a simulation study. We hope that our contribution will increase transparency and result in facilitating the decision process.

View paper

Return to programme