I believe what is being asked for is a subtle difference in units. The phase shift relative to sin(bx) would be "c" in units of radians or degrees. This describes the shift of the waveform at x = 0 The phase shift can also be described in units of x because it causes a translation of the waveform on the x-axis. This translation is a horizontal shift. Both really are the same thing, but the measurement is given from a different perspective/units.
To find the phase shift in units of x (usually x is time, t):
Acos(bx) = non-shifted waveform. It has amplitude A at x = 0.
With a phase shift, what is x at the same amplitude of A?
Acos(bx + c) = A
cos(bx + c) = 1
bx + c = 0 (after taking inverse cosine)
x = -c/b (same formula you have).
Note that c is in units of radians or degrees, so -c/b is in units of x which is typically time, but in this case no one has said what units it is...so it is in general a horizontal shift.
Typically, one would say that the cosine wave has a phase of c radians, or is out of phase by c radians. If one were asked for the phase shift in radians, it would be c (not -c/b which is not in radians, but in units of x which is normally time). If one were asked for the phase shift in units of x, they would want to know the horizontal shift or movement translation of the waveform with the phase c versus the waveform 0 phase. This is the -c/b equation which is in units of x (time). I believe this is the horizontal shift asked for.
In other words, the phase shift itself is usually described in units of phase (radians or degrees) and so is conventionally just the value c in your equation because this is the phase that has been added to Acos(bx) to result in a phase shift. In most applications, the quantity bx would be wt or 2*pi*f*t and the units would become radians or degrees so c would also be in radians or degrees.