There are some too complicated theorems in mathematics which have very complicated proofs in hundreds of pages. There are few mathematicians who are aware of the entire proof of such theorems in full details.
Now consider that a typical researcher wants to use the final result of such a complicated theorem ($T$) in a part of his research for proving a particular statement ($P$) but he is not aware of full details of its long proof. There is always a chance that what he wants to prove is used somewhere in the proof of that complicated theorem essentially (i.e. $\exists Q~~~~(Q\nRightarrow T)~\wedge~(P\wedge Q\Rightarrow T$)) and so his proof of $P$ from $T$ will be as trivial as $P\wedge Q\Longrightarrow P$.
Question: How to use a very complicated theorem (without knowing its entire proof in full details) for proving simpler statements without falling into a loop? Should we leave this to somebody else (like a journal referee or an informed person) to check or cease using these complicated and important theorems in ordinary math research?