It will dramatically shorten the life expectancy of the sensor. I don't have enough hard data to tell you if it affects accuracy or not at any given moment, but in the long run it certainly isn't helping.
Burning oil generates particulate matter, and that KILLS the sensor. Think of it as running EXTREMELY rich and filling the sensor with carbon flakes.
I'd pull the sensor.
If you have to run it, get some spark plug antifoulers, and cut the guts out of one, and make a spacer out of it, so as to keep the sensor out of the direct exhaust stream. BUT, this dramatically effects it's response time.
Just in eyeball tests, I've run one conventionally and one sheilded like I mentioned, and it would seem there's like a .1 sec delay between the two.
HTH