This was a major lesson that was learned during the early tube computer era. The best approach was NOT simply swapping out tubes after so many hours "to prevent in-service failures", but periodically running diagnostics checking pulse levels, etc.to identify tubes that were actually starting to slump off.
The failure rate vs life curve for most components (tubes included) has a high initial failure rate (so-called "infant mortality"), followed by a long period of low failure rates, which eventually trends upward at an increasing rate at end-of life. This produces a curve with a flat bottom and 2 peaks at the ends, like a cross-section of a bathtub.
By swapping out tubes before they hit the end of life, you push the entire tube complement in the equipment over toward the "infant mortality" end of the curve, actually INCREASING the failure rate over careful monitoring and replacing only those tubes that are actually starting to fail. All that tube swapping also results in increased failures through the increased handling of the glass tubes (breakage and seal leaks), wear on the sockets from pulling and inserting tubes, etc. The highest equipment uptime was achieved by not actually replacing tubes on a fixed schedule, but by overall system checks to identify and replace individual failing tubes BEFORE they progressed to the point of total failure.
Experience with electronic installations containing tens of thousands of tubes produced a huge amount of statistical data on component reliability, laying the foundations for modern reliability models and MTBF calculations.
A good read from 1960, when all this was being figured out is "Getting the Most out of Vacuum Tubes" by Bud Tomer, available on Archive.org:
https://archive.org/details/GettingTheMostOutOfVacuumTubes_105