Electron Bernstein waves (EBWs) are theorised to efficiently drive current in spherical tokamak power plants, e.g. Spherical Tokamak for Energy Production (STEP). At high temperatures ($T_e \gtrsim 4\,$keV), relativistic effects can significantly impact wave propagation. This work presents relativistic calculations of EBW wave propagation, damping, and current drive (CD) in a conceptual STEP plasma. Kramers-Kronig relations are exploited to efficiently evaluate the fully-relativistic dispersion relation for arbitrary wave-vectors, leading to a $> \! 50$x speed-up compared to previous efforts. Current drive efficiency is calculated using both linear and quasilinear codes. Thus, for the first time, large parametric scans of fully-relativistic EBW CD simulations are performed through ray-tracing. In STEP, three main classes of rays are identified. The first class propagate deep into the core ($\rho < 0.5$), but exist only if relativistic effects are accounted for. They damp strongly at the fundamental harmonic on nearly-thermal electrons and thus drive little current. A second class of rays propagate to intermediate depths ($\rho \approx 0.3 - 0.7$) before damping at the 2nd harmonic. Their CD efficiencies are significantly altered due to relativistic changes to trajectory and polarisation. The third class of rays damp strongly far off-axis ($\rho > 0.7$), predominantly at the second harmonic. These ray trajectories are sufficiently short and ``cold'' that relativistic effects are unimportant. In linear CD simulations, the optimal launch point corresponds to this third class of rays, suggesting that non-relativistic simulations are adequate. However, quasilinear calculations indicate that, at reactor relevant powers, current drive is maximised at $\rho \approx 0.6$. This quasilinear optimal point corresponds to the second class of rays, for which relativistic propagation does matter.