What if the problem was reframed such that nobody ever found out about the decision and thereby that their estimates of risk remained unchanged?

Then it's wildly and substantively different from moral decisions people actually make, and are wired to be prepared for making. A world in which you can divert information flows like that differs in many ways that are hard to immediately appreciate.

It is certainly possible that there is some underlying utilitarian rationale being used.

The reasoning I gave wasn't necessarily utilitarian -- it also invokes deontological "you should adhere to existing social norms about pushing people off trolleys". My point was that it still makes utilitarian sense.