There I found the game Fejfar, Vlastimil - Chvojka, Jaroslav, CZE-Cup32 final email ICCF, 2015 which did not appear in The Database.
Success!
Not really, as you will see.
Of course, Vlastimil Fejfar is familiar to readers of this blog - see "Correspondence Play Parts 1, 2, and 3", "Climbing Sněžka" and "A Fierce Jerome Gambit Battle" for starters.
But I think there is something mixed up in the ChessBase Live Database...
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+ Ke7
The first clue that something might be amiss. The Database has 44 games with this position, out of 13,090 games starting 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+. That turns out to be about 1/3 of 1% - a very rare move, indeed!
Of course, Black might be "experimenting", too, but it seems unlikely that he would turn his "won" game after White's move to a slightly worse game after his own move.
5.Nxe5
Suspicious. White would have a comfortable "pull" after either 5.Bxg8 or 5.Bb3. I could see this move in a lightning game... maybe. In a serious correspondence game? No.
5...Nxe5 6.Qh5
Again, raising eyebrows. Why not the straight-forward 6.Bxg8 Qxg8 7.d4 when 7...Qc4 8.dxc5 Qxe4+ 9.Qe2 Qxe2+ 10.Kxe2 is clearly good.
Feeling adventurous? Then 6.Bb3 was the move, and after 6...Nc6 or 6...Bd4 or 6...Bb6 White could test Stockfish 9's contention that the first player has an edge.
6...d6
Strangeness from the other side of the board. Black is rated at 2295, and should have seen 6...Bxf2+ 7.Kxf2 Nxf7, with a better game, easily.
7.Qh3
No, I don't think so.
7...Kxf7
Missing something.
More likely, the "game" is bogus.
8.Qh4
No. Not even blindfolded.
8...Qe7
Consistent, but absurd.
9.Qxe7+ Nxe7
And Black went on to win - in whatever alternate universe the battle was fought.
10.h3 Be6 11.d3 Rhf8 12.Ke2 Kg8 13.Be3 Nc4 14.d4 Rae8 15.e5 b5 16.dxc5 b4 17.Kd3 Rb8
Very strange, indeed.
I went to ICCF website, looked up the event, and studied the crosstable: Fejfar came in 2nd to Chvojka, with 20.5 points to his opponent's 21.5. When I downloaded the PGN file of what appeared to be the game, however, it had only the outcome, not the moves.
I checked my copy of ChessBase's Big Database and could not find the game.
Another blow to the argument "But I saw it on the internet!"
(Years ago, when chess game databases began to proliferate, publishers were known to "seed" their databases with imaginary games, the better to use them as markers to show if others - publishers, players - later copied their work. I suppose that is one possibility, here.)
16.a3 Bxc1 17.axb4 Bb2 18.Rab1 Bc3 19.b5 Rxe4 
20.Qf3 Qe7 21.b6 axb6 22.Ra1 Bxa1 23.g3 Re1 24.Qd3 Qe2 25.Qxe2 Rxe2 26.Rxa1 Rxa1+ 27.Kg2 d3 28.Kf3 Raa2 29.h4 Rxf2+ 30.Ke4 d2 31.Ke3 d1Q 32.Ke4 Qa4+ 33.Kd3 Rf3 checkmate