Your best bet are probably follow-up studies. These studies begin with a standard RCT to determine treatment efficacy, and are then followed up by long-term monitoring of participants after completion of treatment. In this way, participants on the active medication can be compared to controls on post-treatment outcomes (eg, after discontinuing medication).
Realistically, modern long-term stimulant efficacy trials are difficult to control from a practical and ethical perspective. Parents of participants who don't respond to medication may drop out, subjects whose symptoms subside are encouraged to discontinue, and parents of unmedicated participants are likely to seek medication if symptoms persist, causing significant selection bias. Thus, this is one of those rare cases where the best studies are older - at a time when treatments were not so readily available as they are now.
Literature reviews are a good place to find lists of such trials, and there are several good reviews of (particularly older) follow-up studies in which both arms are likely to have remained similarly unmedicated at follow-up after the end of the controlled phase: Barkley (1977), Gadow (1983), Hechtman (1985), Jacobvitz et al (1990), Thorley (1988). The studies reviewed measure a wide variety of clinical, cognitive, social, medical, and general life outcomes.
There are 3 typical findings from these reviews:
- In the long-run, the majority of outcome measures are not affected by medication. In line with this finding, the benefits of medication also do not last - eg, Hechtman & Greenfield (2003):
Most authors who have examined stimulant treatment in childhood as a
predictor of hyperactive adolescent outcome have found no effect – as
if the initial benefits of medication cited by many studies do not
somehow carry over into positive long-term outcome.
- Though evidence is inconsistent, some reviews do find persistent benefits in a few outcome measures - such as Wilens et al (2003):
Our results suggest that stimulant therapy in childhood is associated
with a reduction in the risk for subsequent drug and alcohol use
- One outcome measure that may remain negatively affected after cessation of stimulants treatment is height; evidence for this is also inconsistent - eg, Roche et al (1979):
This temporary effect on growth is present during the first few years
of treatment and seems related to drug dosage and to the presence or
absence of drug holidays.
- These findings remain broadly consistent with conclusions of modern long-term follow-up studies.
- 3-year RCTs are non-existent for reasons including the ones mentioned above, and because many official clinical guidelines recommend stimulant treatment be limited to a maximum of 2 years, usually with psychotherapy and training helping to reduce reliance on medication. The closest is probably Hechtman, Weiss, & Perlman (1984), a non-RCT follow-up study that looked at adult outcomes of children medicated for 3-5 years (and then stopped medication) relative to matched untreated controls, and finding some (minor) persisting long-term benefits.
- "Levels of motivation" is an unusual outcome variable for an intervention study on ADHD. However, long-term follow-up studies looking at depression do exist - finding no effect (Jensen et al, 2007; Lam et al, 2019) or favouring stimulants (Daviss et al, 2008; Biederman et al, 2009; Rasmussen et al, 2015; Chang et al, 2016).