Fudging, whether it's die rolls or it's monster stats, is never a good idea.
You are playing the game to find out what happens. In this case, you've a situation where a die roll can end the world. Just go with the roll. Make sure everyone at the table knows what happens if the roll goes one way, and what happens if the roll goes the other way, and then let everyone see the roll as it's being made (it's a PC making the roll, right?). Be clear about the DC and mods and stuff too. Let everyone see the stakes and then roll.
Sometimes, if a die roll ends or drastically alters the world, that can be pretty cool. If the world ends you can always make a new world.
The Kang Civil War in the backstory of the Talislanta game started with a PC rolling an unexpected 20 in a session, events that became canon in future editions of the game.
The difference between fudging 1% of rolls vs 100% of rolls is (albeit definitely non-zero) smaller than the difference between fudging 1% of rolls vs zero. Try it. Sometimes frustrating or boring or disappointing things happen. But sometimes truly epic and awesome things happen.
I love zero-fudge play. That also means refraining from scalinghave ran the game without any fudging, neither die rolls nor ACs, DCs or monster HP, abilities etcsince we started playing D&D back in 2014, several hundreds of sessions (we play multiple times per week). It has created the sensation of what's happening in the game happening "for real". We have had many characters die and campaigns completely wrecked, cities burned down by bandits when the players failed to stop them, just complete trainwrecks and having to start over completely. TPKs and really unfair deaths, characters whose player spend hours working on the flyback story to then only see their characters die after fifteen minutes. It's been five long years of heartbreak but also of absolute joy and triumph when things do go right. I wouldn't trade the weight, edge and tension that non-fudging gives to the game for anything.