The idea that number of accidentals in a key signature is directly proportional to difficulty of reading is understandable, but it isn't always true. In fact, several very common chromatic harmonies are relatively ugly in D♭ but not too bad in C#. The Neapolitan chord—a major triad with a root on the lowered second scale degree—would be an E♭♭ major chord in the key of D♭ but just a D major chord in C#. All three of the standard augmented sixth chords would have B♭♭ in the key of D♭ but A in the key of C# (although, admittedly they'd all have Fx in C# but just G in D♭).
My point is not that C# is unequivocally easier to read, just that the idea of readability is not quite as simple as just counting the number of sharps or flats in the signature. More importantly, most experienced performers—although they may have once found C# difficult—no longer find reading any key to be that big of a deal at all.