You are here

Australian regulatory guidelines for sunscreens (ARGS)

Version 1.2

30 August 2019

Book pagination

5. Reproducibility of SPF test results

This guideline is currently under review.

The in vivo testing of the SPF of a sunscreen product according to the procedure in ISO 24444:2010 and AS/NZS 2604:2012 produces a reasonable estimate, but not a highly accurate and precise measure, of the true SPF of the product applied to the skin at a rate of 2 mg/cm2. The test data exhibit a considerable inherent variance which needs to be taken into account when interpreting the test results and labelling of the product, and also needs to be taken into account when interpreting the results of any subsequent retesting of the product.

The test procedure in ISO 24444:2010 and referenced by AS/NZS 2604:2012 requires the product to be tested on a minimum of 10 subjects and for the arithmetic mean, standard deviation and 95% confidence interval (95% CI) for the mean to be calculated using the formula 95% CI = m ± t.s/√n, where

  • 'n' is the number of individual SPF data,
  • 'm' is the arithmetic mean of those data,
  • 't' is the value of Student's t for n-1 degrees of freedom and p=0.05 (double sided), and
  • 's' is the standard deviation of the test data.

There is a 95% probability that the true SPF of the product lies somewhere within the 95% CI.

For the test to be considered valid, the 95% CI must fit within ± 17% of the mean and, if not, the product must be tested on further subjects (up to a maximum of 20) until the 95% CI based on the data for all subjects does fit within ± 17% of the mean. If testing on 20 subjects does not bring the 95% CI within ± 17% of the mean the whole test must be rejected. In practice, use of more than 10 subjects would be necessary only if the coefficient of variation (CV = s/m) is greater than 24%, and testing on 20 subjects would only fail if the CV was greater than 37%.

Statistical analysis of the SPF test data submitted to the TGA over recent years in support of SPF claims made for a large range of sunscreen products has shown that the data typically exhibit a relative standard deviation (RSD) or coefficient of variation (CV) in the range 5-20%. Only rarely is the CV less than 5% or greater than 20%. Thus, in the majority of cases, testing on 10 subjects would yield a 95% CI well within the ±17% limits and testing on additional subjects would not be required.

Subsequent retesting of a sunscreen is likely to yield a mean SPF anywhere within the 95% CI from the original testing of the product or even a few SPF units beyond either end of that 95% CI. Consequently, if the original test result was close to the lower limit for a particular SPF claim allowed by the Standard, the retest result could be lower than that lower limit and appear to cast doubt on the validity of the labelled SPF claim. However, it would be necessary to retest the product several times and obtain consistently low mean results before any conclusion could be drawn about the labelled SPF being unjustified.

Book pagination