What are the most common causes for stochastic IC failure aside from those caused by stress?

On rare occasions, ICs sometimes fail in an application in spite of having ratings well beyond that application's worst-case operating point (2 or 3 times more). Re-running the IC through a full standardized test/evaluation process purveys no awry symptoms. Most of the other ICs survive throughout their intended life-cycle without any stark contrasting characteristics from the failed ones. So what other plausible underlying reasons are there for seldomly failing ICs?

Remember heat is also “stress”.  Stress can involve many parameters and it is easy to overlook one. Are they the same ICs that fail?
Surges on the supply may exceed stress values momentarily.ESD can degrade or destroy performance and it can degrade further over time, to the pint of failure. The ESD can happen in handling during manufacture, or via fingers touching during operation.PCB flexing can cause SMD ceramic capacitors to crack and fail. Perhaps this could cause a knock on effect on the ICs.‘Fake’ components may have entered your supply stream.Blame it on the software- we are hardware guys after all!

I think Aubrey covered most of it but a couple additional  points come to mind.If it's always the same part on the same board that's failing, you should have a look at PCB layout. Where the part is physically located and how the traces are routed can have an effect on failures. For instance, is it close to power supply or not, are there high temperature components nearby, close to a board edge or I/O connection. If the traces to the part are long (or if it connects via cable to the outside or another board), you may have transients on the data lines that exceed the manufacturers limits.

Thank you very much for your reply.Yes, I was pertaining to the pervading concept of stress, it being the most common culprit.I wanted to know if there were other reasons I wasn't seeing.Surges and ESD count as electrical stress.PCB flexing counts as mechanical stress.
Fake components does seem like a good culprit. It is sometimes tempting to neglect that. Thanks!
Best regards,Justin

Thank you very much for the additional points.That is indeed an insightful way of looking at it without accounting for stress.PCB layout isn't just something one can do off-handedly. It should be well-thought of in order to avoid those hazards.

use regulated ic based power supplies.   never  exeed the max  voltage rating as per the manufacturer  for example if the min - max voltage is from 3to 5 never apply 5v  better to apply only a min  of3v to a max of 4.8 v for the safety of the ic  .  give max possible pcb  track distance,  give thicker tracks for tracks  carring more current,  +ive and - ive  tracks should be always thicker and shorter in length .cover all the un used places and tracks to ground with out affecting the circuit  to creat a shield effect to avoid all type of interferences, use tvs diodes inthe power supply  stages.  connect  only low value cap on the out put stages of the ic .provide   reverse protection diodes for the ic in put supply. if required  provide heat sink  for  ic.