The 217Plus results have temperature and environment factors applied per its model. So what you would be doing is applying them again, only you’d be applying them using a completely different model (the MIL-HDBK-217 models). I don’t see how one could claim that that would be a valid method.

It seems like the real problems is: You don’t like the results MIL-HDBK-217 produces, so you are trying to make them “better”. If that is the case, here are several valid solutions:
1) Switch to another method entirely, such as 217Plus.
2) Use the VITA51.1 approach to MIL-HDBK-217. This follows the MIL-HDBK-217 method, but provides guideline as to how to adapt the Quality factors such that they better represent today’s commercial components. It also provides guidelines for using modern devices (e.g. larger microprocessors; higher density memories, etc.) with MIL-HDBK-217. The VITA 51.1 method was developed in concert with noteworthy reliability organizations. Using Vita 51.1 it will often improve MIL-HDBK-217 predictions by orders of magnitude, without having to resort to questionable data manipulations that may not be defensible. You would then have results that are fully traceable to published standards. The VITA 51.1 guideline is fairly inexpensive, (Google “Vita 51.1”) and easy to understand.