Regressing Relative Standard Deviation vs. Time

Six Sigma – iSixSigma Forums Old Forums Healthcare Regressing Relative Standard Deviation vs. Time

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
  • #25758

    Stan Alekman

    It is the practice in the pharmaceutical industry to establish the expiration period of drug products by regressing the characteristic of interest with time to find the time where the characteristic no longer falls within specification.
    One characteristic of interest for drug products where the active ingredient is not fixed in position as in a tablet or capsule, but is mobile as in an ointment or creme where diffusion is possible.
    When the active ingredient diffuses, its homogeneous distribution throughout the mass is altered and assays of product segments taken from locations throughout the mass begin to substantially differ from one another. Homogeneity is measured by Relative Standard Deviation of the assays from various locations in the mass. And there is an upper limit to RSD beyond which the product is considered failed.
    I have done this and find very good empirical fits to data by various indicators including normal probability plots of residuals, lack-of-fit test, and other indicators.
    I have no theoretical bases for these fits.
    I wonder if anyone has assessed RSD vs time in situations like these.
    All comments are welcome.
    Thank you.


    Robert Butler

      I’m aware of  RSD and the use you mentioned.  I can’t point to any study of the type you have done but, as a biostatistician, I can’t think of any reason why what you have done would be wrong. 
      I did some work on gels and ointments for the cosmetic industry many years ago and while we didn’t use RSD we did use other aspects of the gels to make predictions concerning product shelf life
     In light of what I did at that time it would seem to me there would be a possibility for additional predictive accuracy if you were able to take into account the physical/chemical characteristics of the ointment or creme since, in addition to simple main effects, there should be ample opportunity for various composition x drug interactions which could have an effect on RSD.


    Stan Alekman

    Thanks for your reply.
    RSD is just one of the product characteristics studied in the stability trials. Drug assay, degradant formation and viscosity are others.
    Stan Alekman

Viewing 3 posts - 1 through 3 (of 3 total)

The forum ‘Healthcare’ is closed to new topics and replies.