Looking for advice from experienced miners

Hey, I have a question regarding overclocking and "healthiness" of GPU's. When overclocking, one needs to find the stable overclocks on each card, to see where it is able to perform without crashing or producing incorrect/invalid shares. On different cards of the same model this number might vary – silicon lottery etc. On trial and error one can find this number. My question is this: does pushing the card all the way to its limit make the card "unhealthy" and thus shorten its lifespan? I would make sure the temperatures are healthy all the time in this case. Say for example that his number is 2600 mem clock in HiveOS – 2650 being a number where it starts to give incorrect shares, fluctuate hashrate/give unstability, and/or crash. 2600 would keep everything nice and stable.

Would 2600 shorten its lifespan compared to a notch further down from the limit, for example 2400 to give it a bigger margin from the "unstable" number of 2650?

I want to mine with my cards for as long as possible. I'm thinking long term, and I want to change the fans and keep on mining until they are obsolete or just completely give out – which should be anywhere from 3-5 years (if I'm not wrong).

Hopefully my questions make sense, and thanks in advance.