%20(1).webp)
Look at the last report you signed as a trained ADHD assessor. The coding field almost certainly says F90.0, because that is what the electronic record offers. The methods paragraph almost certainly names DIVA-5, which is built from DSM criteria. And somewhere in the formulation there is probably a sentence saying the patient met "the diagnostic criteria for ADHD" without saying whose.
That sentence reads as a single, settled threshold. There is no single, settled threshold, and for adults sitting near the boundary - which people referred in their thirties and forties routinely do - which manual you had in mind determines the outcome.
UK services nominally work to the WHO's ICD-11. Every structured instrument on the shelf is derived from the American Psychiatric Association's DSM-5-TR. Gomez, Chen and Houghton set out the divergence in the World Journal of Psychiatry in May 2023, and four differences matter in a report.
Onset, duration, cross-situational presence and impairment are functionally aligned between the two. The divergence that changes decisions is the threshold.
The five-symptom adult threshold is the only part of this that reliably changes decisions, and it is worth knowing where it came from. DSM-5 lowered the adult requirement from six per domain to five.
Matte and colleagues, in Psychological Medicine in 2014, tested the DSM-5 criteria in a population-based sample of 18- to 19-year-olds, ADHD having been absent from the official DSM-5 field trials, and found five inattention symptoms and four hyperactivity-impulsivity symptoms gave the best balance of sensitivity and specificity against functional impairment.
So an adult with four unambiguous hyperactive-impulsive symptoms and moderate occupational impairment is not subthreshold in any absolute sense. They fail one manual's count, satisfy another manual's judgement standard, and sit at the empirical optimum from the one field test that looked.
DIVA-5, CAADID and ACE+, the three semi-structured interviews UKAAN's Adult ADHD Assessment Quality Assurance Standard names as acceptable, are all structured around DSM symptom items. The ASRS was built against DSM-IV Criterion A and revised against DSM-5. CAARS 2 aligned its symptom scales to current DSM criteria. There is no equivalent for ICD-11's 22 items.
The result is a quiet mismatch almost nobody documents. The service codes to ICD. The instrument counts DSM symptoms. The report describes the result as though the two were the same exercise. Where an assessment is challenged, that gap is the first thing an informed reader finds.
NG87 is often cited as though it settles this. Recommendation 1.3.3 states that symptoms should meet the diagnostic criteria in DSM-5 or ICD-10 (hyperkinetic disorder), cause at least moderate impairment based on interview or direct observation in multiple settings, and occur in two or more important settings.
Read the manuals it names. DSM-5 and ICD-10 — not ICD-11, which took effect on 1 January 2022. So the guideline currently permits one current manual and one superseded one, and if you assess against ICD-11 6A05 you are not applying the criteria set NG87 names. It also states its own two-settings requirement rather than deferring to either manual.
NG87 leaving the choice open is a delegation, not an oversight. A delegated decision that is never recorded is indistinguishable from a decision that was never made.
The evidence does not compel a position, so here is one. It costs a paragraph and pre-empts the first question a challenging reader will ask.
Three sentences, placed in the diagnostic formulation rather than the methods section, where the next reader will find them. In the straightforward case, the paragraph names five things:
The borderline variant is the one that earns its keep. Where a patient reports four inattention symptoms at clinically significant frequency, one below the DSM-5-TR adult threshold, with clear cross-situational impairment, say that ICD-11 specifies no count and places corresponding weight on clinical judgement, and that the diagnosis recorded is therefore threshold-dependent. Nobody is required to write that. It is simply harder to challenge than silence.
It does not say. ICD-11 asks for "several" symptoms from each domain and leaves the weighting to clinical judgement, which is the substantive difference from DSM-5-TR.
No. Gomez, Chen and Houghton note the absence directly and describe it as an obstacle to both research and clinical practice.
Neither. NG87 recommendation 1.3.3 permits DSM-5 or ICD-10 with no tie-breaker, and predates ICD-11 coming into effect.
This is fixable at the level of documentation, at no cost, today. One sentence in the formulation saying which criteria set you applied tells the next clinician whether a borderline diagnosis was a judgement call or a count.
Global ADHD Network's assessor training works through criteria application and report structure in practice. Our team is at info@globaladhdnetwork.com.
