It is somewhat common to not be sure what is best after phase I/II and so try two different doses. Though normally you try to avoid that because it is more expensive. In this case Oxford thought the two dose was more promising, but the other had enough promise that they start it too just in case. That the second seems better is a surprise, if they expected this result they would have tested differently.
My guess is the lighter dose is really slightly less effective, but the same size is so small that statistical issues make it seem better than it really is.