Is there a threshold of ADHD severity below which drug treatment may do more harm than good?

The short version: no trial in this retrieval tested whether mild ADHD is better left untreated, so the threshold question has no direct answer.

The evidence here is almost entirely about whether stimulants work and are tolerated, not about who should get them. A 2026 systematic review found methylphenidate improves core ADHD symptoms in the short term, with common side effects like poor appetite and insomnia [2]. A 2011 study of 21 drug-naïve children found symptom improvement with no significant blood pressure or heart rate changes [3]. These studies enrolled children already diagnosed with ADHD, not a mild, borderline group.

The closest thing to a harm signal is a 2017 case report: a 16-year-old with mild intellectual disability developed mania-like symptoms five days after starting methylphenidate, requiring two additional psychiatric drugs to stabilize [1]. That is one case, not a rate. But it shows the harms are not zero, and the systematic review itself says long-term outcomes and rare serious risks remain uncertain [2].

What this retrieval cannot see is the key comparison: mild ADHD treated versus mild ADHD untreated, measured over years. The 2018 study comparing stimulant-treated to non-treated groups found no long-term differences in symptoms, social functioning, or cognition over six years [4], which cuts against the idea that drugs change the trajectory for milder cases. No record here defines a severity threshold, and none compares drug harms against the real-world cost of untreated mild ADHD.

My call: the evidence does not establish a threshold, but it also does not justify medicating mild cases. The burden of proof sits with the prescriber, and no retrieved study meets it. Confidence: low.