Actually that is a misconception, most of the failures of silicon devices (this mostly affects power devices, for obvious reasons) are not due to the temperature in itself, but to thermal cycling, as they are caused by mechanical failure of the connections between the various materials in the package, in this case it would be the bonding between leads and the die itself.
The main cause of failure is mechanical fatigue due to differing CTEs of the materials (coefficients of thermal expansion), this is not due by temperature in itself but by thermal cycling, that is going from hot to cold to hot again.
in this case if the zener is kept at a constant temperature throughout the design life i wouldn't worry.
if you want proof of this you have to look no further than ovenized quartz oscillators, despite working at a quite toasty 75 degrees C, the ones inside test equipment that are kept constantly at the same temperature can last for twenty years or more