Providing some incentive that applies even when the study is not continued would be great. But the only thing I can come up with is escrow for some sum of money until the results are sent in. But that's also an incentive for fabricating any results once it's known that the trial is going to fail.
The notion of having to register early in a study is an excellent one. In a perfect world, the FDA should interpret an unpublished clinical trial as "EVERYONE WHO TOOK THIS DRUG CAUGHT FIRE AND DIED" and factor that in to the decision to approve or deny a potential drug/device/methodology.