This well-done retrospective cross-sectional study delivers just what we want from such a study design: a clear path forward to develop a prospective trial to answer a specific question.
Source: Feuer AJ, Rhai A, Demmer RT, et al. Association of stimulant medication use with bone mass in children and adolescents with attention-deficit/hyperactivity disorder. JAMA Pediatr. 2016;170(12):e162804; doi:10.1001/jamapediatrics.2016.2804. See AAP Grand Rounds commentary by Dr. Daniel Doherty (subscription required).
Investigators looked at data from 6489 children and adolescents in the NHANES database who completed questionnaires and also underwent dual-energy x-ray absorptiometry (DXA) scans. Of those, 159 participants reported use of stimulant medication, presumably for treatment of attention-deficit/hyperactivity disorder (ADHD). They found lower bone density measurements in the stimulant users.
The study points out a number of important caveats when interpreting such studies. First, NHANES is a self-reported questionnaire, so of course the information reported has not been verified independently. Second, the primary outcome, bone density, is a surrogate outcome, also termed a biomarker. Nobody really cares what his or her bone density measurement is; what really matters is whether someone has a fracture, so the utility of DXA is entirely in how well it predicts whether an individual is at significantly increased risk for fracture or other debilitating bone disease. Even after many years of DXA use, it's still tough to determine the best criteria that predict long-term outcomes. Third, such cross-sectional survey studies have difficulty sorting through the many differences among groups that also may influence group assignment and outcome, in this case whether a child receives stimulant medication, or whether the families even complete the survey or obtain the DXA scan. So, it's not surprising that in the Table 1 of this article, describing baseline characteristics of the study participants, there are a lot of statistically significant differences. For example, stimulant users were more likely to be male and non-Hispanic white.
Dr. Doherty's discussion brings up additional key points. Other studies have pointed out long-term benefits from stimulant use in ADHD, such as a decrease in criminal behavior, trauma, and accidental injury including fractures. So, it's not a simple matter to decline stimulant use based just on DXA results.
This brings us back to my initial premise: rather than providing us with immediately useful information to use in the clinic, a good cross-sectional study instead points the way towards future prospective studies. Unfortunately, the outcomes of interest regarding risks and benefits of stimulant medication use in children with ADHD are long-term factors, so we may not have a more definitive answer until cohorts are followed prospectively for many years.