The original adventure was written assuming levels 1-3, with only the stuff at the very end being a challenge for a 3rd level party. (And that's a third level "OD&D" party, not the durable heroes of 5E, although using modern stats for the monsters mitigates that some.) For a modern game, The Keep On the Borderlands should probably have been done assuming levels 1-5, with a lot more of the midrange stuff being factored for 3rd level groups, and the tough stuff assuming 4th or 5th. And really, looking at the math, I probably should have realized that just based on the encounter XP compared to the XP required to level up, I just didn't take the time to figure it out.
So, oops. ¬.¬
At the same time, this was always intended to be a "disposable" adventure, to show Seifer how it's done, so I'm not sure it warrants doing a lot of refactoring work. I put my own spin on things, turning the "Cult of Evil Chaos" into a cult of Hastur specifically and using that to spin the personalities, goals, and methodology of the various factions involved, but I have no plans for it beyond what's in the module and no real notion of a followup. If there's enough interest from the players, we might carry on a campaign, but we'd have to figure out what it would entail. If nothing else, I could just string modules together– I've got most of the "classics" from 1E through 3.x and ten years of Dungeon magazine to pull from.
I do know that after this, it'll be a while before I want to run low-level adventures again. The Silver Coast game started at 1st level because it was a new edition and I used the Starter Set as a kickoff, but the group had just hit 5th? 6th? when it imploded. If I was starting a new campaign with an experienced group, I'd probably launch the game at 3rd or 5th right out of the gate. I'd like to see what 5E looks like on a higher tier, given that SlyFlourish says it still feels like D&D at high level in a way 3.x and 4E didn't.
Anyway, we'll see where it goes. One of my DMing strategies is "never prepare more than a few sessions in advance," and certainly that holds true here. Tonight's session will probably be the deciding factor on what happens with this particular game. If they go the direction I expect them to, they'll pretty much "break" the Caves of Chaos (or get broken themselves in the attempt), at which point I'll have to refactor it anyway because they will have thrown a major spanner into the works of the monster factions' balance of power.
And if I have to basically overhaul the whole thing, it becomes time to decide whether it's worth moving forward, and how we might want to do so, anyway.