I saw a picture today of a duplex bearing from another helicopter TR pitch change mechanism where the inner race and the ball bearings had separated longitudinally from the outer race which is an interference fit within the housing - manufacturing fault??????
Maybe but rare. Or the application of a large load in the axial direction, in which bearings have a relatively low structural strength, if not specific end-load bearing. Other alternatives are a large misalignment of the shaft combined with a large axial load (not really possible in normal service), jumping/climbing out due to partial seizing from insufficient lubrication, or an enormous amount of wear in the bearing - not something you'd expect to go unchecked on an aircraft.