Part of the risk analysis of the 4 engine argument is a conservative assessment of the unknown. Zero risk is unattainable. On your remaining engines you still have wear out, infant mortality and random failure mode. When you mentioned lies and stratistics HDRW, I laughed when I also recall one of the models most commonly used in deciding on chances of failure of 4, 3, 2, 1 is the Monte Carlo Simulation (named after the gambling resort).
Perhaps the 'per flight' risk in the subject of this discussion was calculated as being low and that is why they decided to 'push on'. The data and engineering judgement so far (EK this incident of push on and SQ return) indicates there is not yet a consistent set of ground rules.