IMHO, I think you are talking about some sort of "implicit management", creating "screen with childs" to achieve layer compositing. So, why don't make layered the architecture of the game screens?

You can have several layers in the game screen (i.e.: background, main, ui, notifications), each having its own FSM (Finite State Machine) and being independent. Then, you can update them sync/async as long as you render them bottom-up. In the same way, you can also send "user events" (click, touch, drag,..) up-bottom in the layer stack.

I think that it can be "implicit" or "explicit" but there's always "management".