But it can't be said for sure that use isn't a factor even if it's a very small change.
Yes and no. We obviously don't have data on all IEMs and what when how something turns out to be audibly different. But we do have several people who tried to find (as in, measure and test under some modicum of control) changes over time. My longest running test has been over 2 years with an ER4SR, where I get changes that are relevant, but once I change the filters for new ones and use the old triflange I only keep aside to use for measuring an Ety, I get results close enough to wonder if the differences that remain aren't entirely caused by slight placements and ambient noise differences. (it's hell to try and insert an IEM into a coupler twice with the absolute exact same insertion and tip deformation. The trebles at least are super hard to keep identical, for example. So a test need something big enough of with a profile odd enough to appear beyond the typical small differences we get no matter what.
I went half mad trying to do a proper test with a pair of headphones. The pads would keep changing the sound with more time spent clamped on my mini DSP E.A.R.S. So I tried to remove them every time then put them on, count 5mn and measure to hopefully remain with a stable amount of pad compression and driver distance. It was a mess and placement ruined everything I tried to measure. Again with the difficulty of placing something twice at the very same place, an issue that exist on our heads too. But somehow we usually don't notice those very real, measurable (so apparently bigger than burn in changes in my cases) and often technically audible(at least if the differences were directly A/B). It's just that people don't think about that, don't focus on it, while some are a little bit obsessed with the concept of burn in.
Anyway, After 2 hours of messing up with false good ideas for the headphone burn in test (real play time of the headphone was maybe under 5mn in total as I used short sweeps for measurements, and a square wave to help rapidly adjust the placements), I decided to remove the pads entirely and clamp the headphones onto the rig with actual clamps, just so it wouldn't move at all over time. I also remove the silicone ears to make sure those wouldn't give in over time. I didn't even keep those tests, as they revealed nothing to me. I did get changes, but I always do if I look at magnitudes small enough. It's the kind and magnitudes I also get when measuring just about anything several times at different moments of the day(ambient noise, changes in the electrical grid, time that my ADC spent turned on, Gremlins messing with my tests, etc).
It's the kind of stuff I actually wish to find, Because Of course I expect and know that a moving part will change over time. Same with may high-res stuff and night and day cable whatever. Except I did find cable impacts, usually they're simply highlighting an extremely unstable device or some IEM with ludicrous impedance curve reaching single digit impedance values somewhere. Those impacts I found, heard (under somewhat controlled conditions) and measured. And they're actually predicted by good old Ohm's law for the most part. More can happen, but Impedance does explain very many cases.
With burn in, I just got disappointed after 3 IEMs and one headphone, and I lost interest too much to really look for better equipment and testing methods. I'm at the point where I think it's possible for something to change significantly (as in, so audible, someone does notice over days), but if that actually happens, I would suggest getting rid of the IEM. I value stability, it's the first step toward fidelity.
In my example with headphones, we could argue that the few minutes of playtime to set things up were enough to have the main changes already gone. Or we could wonder if the maker didn't have them play a bunch of hours before mounting them onto headphones, because if there is a defect or a failure, it tends to happen at an early stage and testing them would probably greatly reduce the number of returns for defect/failure. We could also wonder if I suck badly at testing stuff and if the important variables were even on my radar(an idea brought up many times, even by people way smarter and better informed than me). Or we could wonder if my orientation in audio makes me go for the tested stable, gears with good objective background, so they're less likely to exhibit odd behaviors like stuff made out of chinesium, or products from a small brand with one guy tuning and making stuff by ear.
Again, just consider how many more possibilities probably exist beyond what feels like a plausible answer in your mind. The fewer the controls, the more you can probably find.
Back to Ety, I got so comfy with the idea that they were super stable, that I often use them with a known voltage into them to calibrate REW before testing something else. In principle they do change, metal has elasticity, but it's not perfect, IEMs will receive shocks(falls or whatever, I dropped my old SE215 on the ground just yesterday and whined about it for half an hour). Magnetic stuff tend to lose some of their force over time(veeeery sloooowly. And ear wax will be vaporised/atomised or whatever the correct word is to say that the vibrations from music will turn some of it into a sort of mist that is likely to go beyond a damper and into the driver itself. Temperature, atmospheric pressure and hygrometry are relevant enough for some measurements about audio having to take them into account(I don't because I'm a noob amateur with crap tools).
Things cannot remain static. And yet, it's so damn hard to demonstrate the kind of changes that would explain the legions of people, dead certain it's the transducer that changed and not them, their use habits, their excitement about new toys, or just flawed and evolving memories.