Objectives Widening disease definitions is a major driver of overdiagnosis. In response, the Preventing Overdiagnosis working group of the Guidelines International Network developed a checklist to provide guidance on issues to consider when modifying disease definitions. This checklist recommends panels outline definition changes and the trigger for change, and examine research informing the potential changes to prevalence, the prognostic ability and precision of the disease definition, the potential benefits and harms of the disease definition and balance between them.

Using this checklist as a framework, we examined the documented considerations of the panel responsible for modifying the ADHD diagnostic criteria, focusing on the age of onset (AOC) criterion, which widened the definition by changing the requirement that symptoms causing impairment need to be present before the age of 7 (DSM-IV), to the presence of symptoms before age 12 (DSM-5).

Method For the checklist items requiring panels consult research studies (e.g. prevalence, prognosis, precision, benefit and harm), we examined the research considered by the panel modifying the DSM-IV ADHD AOC. We recorded the research studies identified by the panel and described how these informed their conclusions. We appraised these studies for risk of bias (ROB), and on their ability to address the checklist item (e.g. limitations in design and generalisability). We conducted searches to identify research that would have been available to the panel at the time of the modification and compared it to the research considered by the panel. We also identified studies related to checklist items published since the modification, assessed ROB, extracted data and compared the consistency of findings of these studies to the conclusions reached by the panel.

Results DSM-5 panel documentation cited two studies: a ‘systematic literature review’ of studies related to the AOC conducted by panel members (the methods of this study were not documented), and a longitudinal cohort study assessing prevalence. Cited within the review, were studies we assessed as related to some checklist items (precision, prognosis and benefit) however no reference to these constructs were reported in the document and no systematic appraisal of this research or comment on the strength of the evidence was provided. The cohort study reported ‘negligible’ change in prevalence. We appraised the ROB in the cohort study to be low, however, study design precluded confidence in the prevalence estimate, and subsequent research reported larger prevalence increases with AOC changes. We found overlap in the studies identified by the panel and the studies we identified as being available at the time which we assessed for ROB and strength of evidence.

Conclusions Minimal documentation of the considerations and decisions of the panel limits transparency and makes it impossible to judge the rigor of the process behind the modifications to ADHD diagnostic criteria. The information available suggests that rigorous consideration of important issues identified by the checklist did not occur, although this may be a problem of reporting. Panels modifying DSM ADHD diagnostic criteria comprise clinical and research leaders. Critical thinking and rigorous methods are their forte. Future changes to DSM diagnostic criteria should ensure all process are documented clearly and rigorous appraisal of research used to support any further changes. Use of the checklist for modifying disease definitions would ensure a more thorough and transparent assessment of important issues prior to recommending changes, and that these changes can be more robustly supported.