Must a shelf-life spec be different than the release spec?

We have the luxury of detecting a potential significant negative trend for our main component assay which has the lower specification limit of “Not Less Than 90.0%”. After 24-months into the 36-month shelf-life the results were verified at 92.3% (down from 99.7% at release). Previous intervals bounce around, but a trend line could be drawn to show the latest results are consistent.

We’ve never had any significant trend for this (or any) parameters for this product commercially in the past several years, and the final 36-month interval results have routinely been around 5% above the lower limit. It’s an older product, thus the agency has not pushed us to consider tightening the spec.

The investigation is in its infancy, so we’ve not yet discovered if the result might be due to lab or process variabilities, but this product is an oldy, but goody and all the quality pundits are asking why the release specs aren’t set higher than shelf-life specs. Never mind that it was released so high it wouldn’t have made a difference—the finger pointing is quite intense.

Historical data certainly would support raising the release limits to 95.0% or even higher but why mess with something not broken? How has your company handled specifications like this?

Why mess with something that is not broken? The finger pointing sounds like a political ploy and not a quality ploy. To play along with the politics, consider running the statistics, showing that the specs could be raised and just leave it at that—no harm, no foul to stability either way, as the data seems to suggest. Please tell us your thoughts?