Good question. I don't know.
My mind is drawn however, to an experiment I tried in my youth. I drilled a hole into a pine tree trunk (the tree happened to be an invasive weed, in case you think I was being pointlessly vandalistic) and inserted a tubular commercial cardboard sweet box with a bottom and top made of light, thin, more or less rigid, sheet metal. The tube I filled with a hot explosive. The experiment proved educational in various respects, but in particular I was interested to find both the top and the bottom metal components of the container lying close together some 10 m away from the trunk of the tree, and on the same side as the hole.
So what?
So I interpreted the event as follows. The very hot gases had formed a rapidly-expanding bubble, big-bang-style, on one side of the tree and the expansion had continued until the bubble formed a reasonably good, cool vacuum. It then collapsed, imploded if you like, sucking all loose objects towards its centre. The most durable light objects remaining from the explosion were the two discs of metal, and that was where they landed, instead of having been blown far away, as one might have expected naïvely.
Now, what strikes me about the two situations, is the resemblance between the two trains of events. First there was a violent expansion, then a somewhat more drawn-out implosion. This was pretty obvious in the case of the tree, but I am inferring it, somewhat arbitrarily, in the case of the storm.
the storm first pulled off the roof, removing a lot of the area from the interior, plus in fact a lot more air from the space above where the roof had been, but without collapsing the walls. We are left with a huge mass of air above a low-pressure bubble, which immediately begins to collapse. It produces a longer lasting down-current of a far larger mass of air than had exited through the open roof from the inside of the house, and accordingly could do a far better job on the fan blades. I speculate that in houses with damage of that type, one would find a lot of the ceiling board on the floor, rather than outside with the roof, and also a lot of the house windows blown out rather than in.
All this is however the most vacuous of speculation. I have never seen damage of the kind. Simpler, less elaborate principles might have dominated for all I know. For instance, did anyone inspect the structure of the fan blades to decide whether they simply were easier to bend down than upwards?
As I said, I simply do not know!
