Obtaining acceptance criteria limits for potency assay

[#1]

Hi,

We are trying to set acceptance criteria limits for a potency assay. Most of the data are EC50, relative potency values and ratios. We have so far used the control charts with the moving range (MR) as limits or with levey jennings (LJ). However, the LJ limits which JMP calculates are often too wide and the MR ranges often seem unsuitable (limits widening after removal of an outlier). 

I am new to JMP so am still learning about the software, are there other charts/graphs which can help setting these limits? Or recommend LJ/MR for this type of data limit setting?

Thanks!

[#2]

JMP can perform many calculations and analyses. What is the accepted method used in your field to determine such criteria?

[#3]

Normally we work with PK and Biomarker assays and so use confidence intervals or standard PK acceptance criteria - 80-120% theoretical. However for this study we want to set limits based around the data we have generated.

[#4 · Accepted solution]

You could explore tolerance intervals as a way to set specification/acceptance limits.

JMP help: tolerance intervals 

[#5]

Thank you! this is very helpful!

[#6]

So you set specifications or acceptance criteria based on historical performance, that is to say, based on what has been observed in the past when the outcome was acceptable? The criteria are, therefore, not based on customer requirements or on failure limits. If so, then @KarenC's advice around using the Distribution platform makes a lot of sense.

You might also consider using ways that accommodate more than one source of variation. (Think 'peeling back the layers of an onion.'). See the chapters in the User Documentation about the Measurement System Analysis platform and the Variability Chart platform. Also, the general purpose regression tools in the Fit Least Squares platform with models that include terms for random effects or the Mixed Model platform in JMP Pro can analyze more complex studies and data sets.

[#7]

Thank you Mark!