How Many Innings till you drop an underperformer?

In one of my 20XX leagues I am scoring as one of the top 3 teams in the league, but through 21 games my team ERA is 6.14 with a 1.51 WHIP (obviously last in the league).

I've got pitchers with the following stats lines:
IP ERA WHIP Price (approx)
35.1 6.37 1.53 4.00
19 13.26 2.42 3.50
15.2 9.19 2.17 3.00
17.1 5.71 1.39 2.00
18.2 5.78 1.34 5.00

I've read many people say that in a one season sample size you should ride it out as it will even out, but how long is long enough to let my pitchers take me to a losing record when my hitting is doing well?

To me, whenever you have a bad start you need to reevaluate what youre doing. The whole notion of not making changes because it reduces your salary and numbers will even out is predicated on the notion that your team does not have systemic weaknesses being revealed by gameplay. So you can look to see if youre defense is weak or your SPs are not good fits for the park. Should you be going to quick hook and bringing in better relief earlier because starters are not very good?

I will put it bluntly: If you know your pitcher has a good card (i.e., you're not playing in a mystery league) and you drop him, you are a fool. Just because he has had bad luck doesn't make his card any worse than the day you drafted it. To the contrary, (while there's no guarantee) you can figure his performance should improve in the long run.

Now if you're talking about a mystery league, that's completely different; pitchers are really difficult to figure out.