If you added the word "Only" to the title, it would completely change the subtext.
So yeah, adding words does add subtext, but your phrasing of this seems to imply that there is existing subtext. What do you see, specifically?
"One in 5 people now think" is a positive-sounding construction which lends more statstical weight to the result than an impartial examination of the numbers, and people who scan headlines are far from impartial.
Other constructions are "4 in 5 still support...", or "80% still support", or "20% now think", which are all still mathematically neutral, but the subtext is different.
I'm struggling to figure out a neutral phrasing involving this data point. My reading of your response indicates that you see a temporal indicator ("now") or a delta indicator ("still" / "no longer") as both introducing bias. "One in five people think Brexit was the wrong decision" without anything clarifying significantly misrepresents the data.
It seems your issue is actually with the data point being in the headline at all. Maybe the headline should be a dry thing like "A Brexit survey happened, read inside for details". I won't entirely disagree. But I think this is less about word choice than you think.
I think that using "20% now think that..." would give a more accurate impression to someone scanning the headline.
People may regret their choices in elections, but they can revisit them every few years. Brexit is, if not forever, at least for a very long time.
Two mutually contradictory caveats:
1. The referendum itself was never intended to be serious. It was held because David Cameron thought it would fail, in order to shut up the anti-Europe wing of his party. He resigned over it.
2. Although the referendum was close, it was effectively affirmed by the elections in the wake of Cameron's resignation. It was still close, but they ended up with a majority for the party campaigning explicitly on a pro-Brexit platform.
This adds up to a cautionary tale on referendums. A big question was put to the people -- even bigger than an election -- and a largish majority (not a bare majority) are unhappy about the result.
In that sense, yes, 20% is a pretty big number. Not that it changes anything. It's much too late for that, and it doesn't tell us anything about the counterfactual where the referendum lost. But it's worth remembering going forward, both with respect to the EU and for national decision-making.