One sided confidence interval in stability plots

[#1]

Working on stability of pharmaceuticals, I was wondering whether the stability script (Analyze -> Reliability and Survival -> Degradation -> Stability Test) allows estimation of shelf life based on a one sided confidence interval. The default script, as it is set up now, is exclusively based on the intersection between a user defined acceptance limit and the two-sided 95% confidence interval. However, ICH Q1E suggests the use of a one-sided confidence interval for attributes known in- or decrease over time.

Since I can't script (yet) I wondered whether:

- there is a quick fix,

- there exists a justification to not use one-sided confidence intervals for attributes that either in- or decrease

- SAS / JMP could modify the stability script to allow the user to select one sided confidence intervals (both upper & lower).

I am sure that this is a significant issue in the pharmaceutical industry.

[#2]

If it is a model, it is an approximation. If it is a distribution model, it is an approximation.

The normal distribution of the response errors is assumed. The model distribution of the parameter estimates is assumed. That distribution is not used for the intervals. It allows the Student t distribution to be used to obtain the quantile that is used with the standard error for the confidence interval or the prediction interval.

[#3]

Hi Ressel, 

Over a year later, I am looking at these same issues. It looks like if you plot in Fit Y by X OR Stability Analysis, you need to set alpha = 0.10 to plot the correct confidence curve line crossing the spec limit...

So when you say "uses a two-sided (alpha = 0.10) confidence interval as one-sided (alpha = 0.05) confidence interval", I assume this means:

• JMP will PLOT a two-sided (alpha = 0.10) confidence interval
• ...while JMP will CALCULATE a one-sided (alpha = 0.05) confidence interval

Can anyone confirm that this is correct thinking?

[#4]

You are correct and can easily confirm this using the Fit Y by X platform or the graph builder; i.e. try emulating a stability test using alpha = 0.10 in one of the platforms. Makes total sense. Sorry for confusing everyone. Sometimes I am just shamefully slow in my head.