I have no proof (not measured that) but I believe so. (I'm speaking about battery applications - torches/flashlights, head lamps etc. - not 120V/240V lamps disconnected by a metallic wall switch).
Those battery powered apps usually have multiple modes such high/medium output, flashing modes etc., controlled by a single microswitch, hence a circuitry to "remember" what to start next (in case of OFF - STEADY - OFF - FLASHING lamps).
But even without that (always OFF - FLASHING - STEADY) they seem to consume something. This is the case of a cheapo bike headlamp (Logic) we have at home. Even if I put freshly recharged NiMH batteries such as Sanyo Eneloop, Varta Ready2Use or Sony CycleEnergy that shouldn't self discharge in a higher than neglectable amount inside, they're flat within a month even when nobody touched the lamp.

The LED drivers with "memory" are never completely off if a battery is installed,as a small amount of power is required to retain the previous mode/setting. However, simple battery operated LED circuits with a resistor can still slowly drain a battery if the negative is being switched and not the positive. I had an LED flashlight in which they put the switch across the negative, and the positive stayed connected all the time, that flashlight went through batteries when off just like youre describing. However, after I repaired it by putting the switch on the positive side, it corrected the problem

Normally the microcontroller should not consume anything, even when it is powered - it is supposed to be in a sleep mode, with every activity suspended and then the current consumption should really be in the nA range; dominant is the leakage of the power switch component, but that should be way below the 1uA as well, so practically nothing (with a 1000mAh battery that means way longer than 1Mhours discharge time, if nothing else would be there to discharge the battery). I've never seen even a cheepeese design not meeting that.
However what causes such problems is the leakage on the structures (paper insulation, not really clean switch,...), that affect all concepts, not only the microcontroller based ones.

Well, my experience is the contrary: The microcontroller based does not need high power mechanical switch (which tend to be way more problematic when speaking about OFF leakage - it needs very low resistance contact, so usually utilizes hefty amount of grease to keep it dry and that grease, when contaminated or degrading over time, tends to leak some current), so very frequently the "electronic switch" means way lower leakage, mainly when it gets older (the simple tactile switch works there well even with rather high contact resistance when slightly corroded, so the contact could be made as dry, without any humidity protection, so without anything then causing the leakage).

So I will look into the dirt and anything, what may hold any moisture or so. That means remains of the soldering flux, just paper as an insulation (both tend to absorb water and then cause the leakage), dirty grease in the switch (full of the metal particles from the contacts, silicon one absorbing water,...). Even the thin foil on many rechargeable cells could have cracks with remains of the glue, so again a bridge for the leakage (that makes the difference between "positive" vs "negative" switch position), generally any dirt may host some humidity, so cause a discharge path.