Just means you have to do the same thing you do with any other ceramic, and always have needed to, really: check the C(V) curve for suitability.
Proof that I'm not taking the piss -- a salvaged "203Z" (20nF, Z5U temp code) ceramic disc, about 8mm dia.
Annealing was holding both leads at soldering temperature (~300C) for ten seconds, during which the ceramic dip coating bubbled and oozed (it's a wax-impregnated coating, which inevitably adsorbs some moisture -- this is normal). As you can see, it recovered some value, although not the full nominal value. The side-of-barn tolerance that Z5U has, means it was never outside of spec, so it technically doesn't matter, but the voltage change remains.
The main differences these days are:
1. Dielectrics are thinner than ever, meaning you're ever more likely to see a problem related to C(V) variation, even for low voltages (e.g., logic supply bypass);
2. You're probably more likely to see curves from the manufacturer describing this. Characterization is still really spotty (maybe 10% of all type 2 ceramic capacitors provide this?), but I'm guessing it was near 0% back in the day. Y'know, back in the days when this was still just as relevant anyway, because you might've used a 500V capacitor to bypass a tube radio IF strip, or something.
Tim