It depends on two factors: how well the researchers are organized and funding. I would say less than half of the research groups I know in Computer Science manage to follow their future work plans. The ones that do manage to follow their plans, have clear areas in which they are active and write their projects after half the work was done :)
Funding has repercussions on what you do at work: most of the time you will have to do the tasks related to your funding as opposed to what you would like to do (future work of your favorite article). The problem is funding stops at some point and then you take another project without getting to the future work from your previous article.
If you had worked well in a field, had some articles, there is no reason to abandon that work just because the funding ended, but you will have to follow it between the funded tasks and in your free time.
There is also the alternative that you write a follow-up research project and get to do a part of this future work in the second project.
I have gone through all these phases:
- I have future work that was never done
- I have future work that was done in the follow-up project (precisely a sequel project)
- I have future work that was done in projects that were not related
- I have also seen future work from some of our articles implemented as commercial project at a start-up by some of the co-authors of the papers
If you work around the clock and if you like the field you're working in, you will definitely find a way to integrate your long-term research goals with the current project goals or with the next project goals.
At least in CS there is no survey/study/article about such a thing, as far as I know. There are though articles that look at how often a certain technology gets into a new company.