That's because Sanford's deep unpopularity with the electorate as a whole makes the race in this dark red district very competitive. Comparing his general election head-to-heads with Bostic's is instructive:
Sanford (R): 45
Colbert Busch (D): 43
Bostic (R): 43
Bostic's favorability rating is quite poor, at 30-42, but Sanford's is downright abysmal, 34-58. His notorious history makes him more than just polarizing: He drives 15 percent of GOP voters directly into the hands of Colbert Busch, who sports favorables of 45-31. Against Bostic, by contrast, she only takes 10 percent of Republicans. But even Bostic's toplines are pretty soft for a seat like this, which went for Mitt Romney 58-40 in last November's presidential election.
Still, there are far more undecideds in the Colbert Busch-Bostic matchup, and since they lean heavily Republican, they're very likely to come home to Bostic in the end. Sanford, by contrast, would have a harder time winning over a sufficient number of voters who haven't yet made up their minds simply because there are fewer of them. It's definitely still very much in reach for him, though, just given the demographics of the district, which is why a Colbert Busch win would be such an upset, even against some as disliked as Sanford.
The real tell about the competitiveness of this race will come if and when we see any serious outside spending here, particularly from the major party committees (the DCCC and the NRCC). It's not always a sure indicator that a race is close, though: Last cycle, Democrats spent heavily to protect a blue seat in an Oregon special election, while the GOP did the same in Nevada. Both contests wound up being quite lopsided, so sometimes third-party money comes in simply as a matter of insurance.
There are also always the caveats about polling special elections, which can be extremely unpredictable, and special election runoffs, which can be even moreso. But PPP has a good track record in recent years surveying oddball races, and their respondents say they supported Romney over Obama 56-40 last year, which is quite close to the actual figures. That doesn't mean PPP is necessarily right—the electorate could certainly wind up being either redder or bluer than usual—but it does mean they're in the plausibility ballpark.
But no matter what, it once again looks like the GOP is at risk of making what should be a safely Republican seat far more competitive than it ought to be by tapping a weird and unpopular candidate as their nominee. If Bostic's smart, he'll use these numbers to make an electability argument ahead of the runoff. But those kinds of appeals seldom seem to work with conservatives, and if Sanford can hang on, we could be in for a very interesting showdown on May 7.