Freeman and Cowling comment in the Journal of Infectious Diseases on a paper last year by Yang and colleagues (I made a note about that one here). They also re-analysed one of the conclusion and found that, for this dataset at least, having H7N9 antibodies did not afford a reduced risk of death.
Freeman and Cowling conclude that this doesn't negate using convalescent sera (the bit of blood, minus the cells and the clotting factors, that contain proteins, water and the antibodies we make against an infection we've had) as a treatment option. But from the data in Yang's paper, the pre-infection existence of higher levels of antibodies that react with H7N9, did not improve chances for survival.
More study is needed.