Another possible reason given for short valve life is cathode stripping or more accurately cathode poisoning....opinions are divided on this and it's been furiously debated on many forums in the past but the theory goes along something like this:
If the HT is applied in advance of the heaters there is a possibility of ions bombarding the cathode reducing it's emissive capability and so shortening life...with the heaters properly warmed up the electron cloud surrounding the cathode repels the ions.
The old WAD amps with solid state rectification applied HT for approx 30s or more before the heaters had time to reach full temp....if you power cycle such and amp often enough the cathode is subjected to these ion attacks...as I said that's the theory...
The counter argument is that the potential difference between HT on the plate and earthed (effectively) cathode needs to be about 10Kv for this to take place
and so at typical domestic valve amp HT of 300~500v it's hardly an issue...
One interesting bit I dug up was that of 'cathode sleepiness' where leaving a valve in an effectively cut off state by removing the heater voltage whilst still maintaining HT for extend periods of time turns the plate into a getter: The gas impurities present in the valve envelope (it's never a perfect vacuum) get attracted to the plate and stop it conducting...