It's easier than that, order a review of longitudinal studies and re-do the math so that drop-outs from treatment are counted as negative outcomes. *POOF* there goes the troonery into the made-up ether.