TPS63001 Random Failure

Hi, I have used this regulator (the TPS63001) in over 25,000 boards so far. I have a problem which manifests somewhat randomly, eventually. After our assembly batch we have a failure rate of about 10%. I was fine with this, due to our high profits, but the customer is also having a 10% failure rate. After finally deciding to double test I realized that after a second test (of the good potion of the batch) there was in fact about 10% failure. So then I did a third test and another 10% of the twice good batch was bad. I realized that if I tested and retested enough times that all of the boards were bad. The fix was simply to replace the TPS63001 and it 100% would work again until inevitably it would eventually fail just like before. I am wondering if anything similar is known in the community as to what might cause this. My power requirements are only about 100mA (at 3.3 volts) and Im using the recommended layout of parts. Also I'm buying reels from digikey. In some cases I have had 100% success but in other times (most of the time) the failure rates are much worse. In all failure cases, it is the TPS63001. I have tried to intentionally destroy the TPS63001 to no avail, shorting out its output pins, or input pins etc. doesn't hurt it. Shorting out it's inductor, nothing.... it seems to work well, but eventually it ends up failing, which manifests itself as a non functional board for me, with a power draw of nearly an amp, quickly getting hot to the touch. Any advise?