Clinical reasoning

A High Conners Score Is Not a Diagnosis: 4 Things No ADHD Rating Scale Can Tell You

Faheem Ahmed··11 min read
WHAT NG87 1.3.3 ASKS FOR THE SCALE GIVES YOU Symptoms, counted Structured, multi-informant, repeatable. Genuinely useful. 1 requirement of 4. It cannot establish… Onset — before age 12, trait not event Pervasiveness — two or more settings Primary or secondary to something else Freedom from its own built-in bias All three of the first come from the history.
NG87 1.3.3 requires four things. A rating scale can deliver one of them — and carries a bias in the items themselves.

A GP refers a 14-year-old girl. The parent-completed Conners 3 comes back with an inattention T-score of 79 — well into the "very elevated" band — and the referral letter asks whether she has ADHD and should start medication. Most of the diagnostic work has apparently already been done. Almost none of it has.

The short version

What does NG87 actually require?

Three things, and a rating scale is not among them. NG87 1.3.1 asks for a full clinical and psychosocial assessment; a full developmental and psychiatric history; and observer reports plus a mental-state assessment. Scales appear in the next recommendation, 1.3.2, and they appear there as "helpful added tools" — in the same sentence that says a diagnosis should not be made solely on rating-scale or observational data.

Then the threshold, at 1.3.3: symptoms meeting DSM-5 or ICD-11 criteria, causing at least moderate impairment on interview or direct observation, occurring often, in two or more important settings.

Put those two recommendations side by side and the hierarchy is unambiguous. The interview is the assessment. Everything else supports it.

The four things no rating scale can tell you

RequirementWhy the scale can't reach it
OnsetWhether the pattern was there before age 12, and whether it is trait-like rather than a change from a previously sound baseline. That lives in the developmental history — school reports, early years, milestones.
PervasivenessA parent form describes home. Presence in two or more settings is a separate evidential task, and it is the requirement most often waved through.
Primary or secondaryAnxiety, depression, sleep restriction and trauma all degrade attention and generate elevated inattention items. The instrument cannot tell you which came first.
Its own biasClosed items shaped by norms, culture and gender. They encode an externalising, disruptive presentation that skews male — one reason ADHD is under-detected in girls.
Informant disagreement is information, not noise.

When a parent rates inattention as very elevated and a teacher rates it as average, the instinct is to average the two. Don't. Ask what is different about those two settings — structure, demand, one-to-one support, anxiety under assessment. That question is usually where the diagnosis actually lives.

None of which makes scales useless. They cast a wide net, they structure multi-informant input, and they give a repeatable baseline for measuring change on treatment — the last being their strongest and most neglected use. They were designed to quantify reported symptoms in a structured way. That is a genuinely useful thing. It is a different thing from diagnosing.

Why the symptom count is overrated

DSM-5-TR asks for six or more symptoms in a domain (five from age 17). It is easy to treat that as a measurement. It isn't, because the items are not independent of each other.

If a child genuinely cannot sustain attention, then they will also make careless mistakes, fail to finish work, look disorganised, lose things and be easily distracted. Those are not six separate findings; they are one difficulty seen from six angles. Once the underlying problem is real, clearing the threshold follows almost automatically.

The count earns its place as a structure — it stops a clinician skipping a domain because a parent was talkative about a different one. What actually converts the items into criteria are the qualifiers everybody skips: inconsistent with developmental level, causing impairment, present six months, several symptoms before age 12, in two or more settings, and not better explained by something else.

The six-month clause is routinely misread. It does not mean the problem started six months ago — in real assessments you rarely get an onset that clean. It means the symptoms must be currently present. Someone can have had mild ADHD that improved on its own; if there have been no symptoms for the past year, you cannot diagnose them now. You might record a lifetime history not currently meeting criteria — which is a different statement.

Read the symptom. Don't tick it.

This is the part that separates an assessment from a checklist, and it is the easiest thing in the world to get wrong.

A child is fidgety. Before that becomes an ADHD symptom, ask what the fidgeting is doing. They may be fidgety because they are under-aroused and need to move — that is an ADHD symptom. They may be fidgety because they are anxious, and we all fidget when we are anxious — that is not.

A child has difficulty waiting their turn. That may be impulsivity. Or it may be that they lack the social skills to read when it is their turn, how long to wait, and whether it is acceptable to interrupt — which looks identical across a table and points somewhere quite different.

Be suspicious of "hyperactive type" in a teenager.

If a young person is hyperactive enough to be impaired by it, you would expect inattentive symptoms too. The predominantly hyperactive label makes sense in a five- or six-year-old who isn't yet expected to sustain academic focus. In a fifteen-year-old it deserves a second look.

Two developmental points sit alongside this. Hyperactivity doesn't vanish with age — it changes form, from the six-year-old climbing furniture to the fifteen-year-old with internal restlessness who takes frequent walks. And attention itself develops in stages: a three- or four-year-old who can't settle to one thing is displaying an age-appropriate attention span, not a disorder. Compare with same-age peers, never with adults or a quieter older sibling.

So is ADHD over-diagnosed?

Both errors are real and they land on different children — but the national picture is not the one the public argument assumes.

Who gets missedHow the opposite happens
Girls and women (NG87 1.2.2)A scale score treated as a diagnostic test
Quiet inattentive presentations that don't disrupt a classroomSymptoms taken at face value, without onset or pervasiveness
Bright children whose ability masks impairmentSecondary inattention — anxiety, low mood, lost sleep — labelled primary
Children assessed under the older DSM-IV / ICD-10 criteriaSingle-informant assessment, usually a worried parent
Mild presentations where impairment is real but undramaticWaiting-list and commercial pressure toward a quick answer

The UK data is worth stating plainly, because it cuts against the received wisdom. England recognises and treats ADHD at a lower rate than Norway, Denmark or Spain. Of children meeting strict diagnostic criteria, fewer than one third have contact with any mental health service. Only 15–25% of those meeting criteria receive pharmacological treatment. The estimated cost of unsupported ADHD exceeds £17 billion.

Population prevalence has not risen dramatically. Referrals have — driven by awareness, by the DSM-5 changes (a lower adult symptom count, autism co-diagnosis permitted from 2013, onset moved from age 7 to 12), by catch-up diagnosis in adults who were missed as children, and by the pandemic. None of those six drivers is "people making it up."

"Is ADHD over-diagnosed?" is the wrong question.

Ask instead: is this assessment good enough to tell? Most diagnostic error in either direction traces to a scale trusted too far and a history taken too thinly.

One consequence worth carrying into practice: you don't need a diagnosis to access support. A child with sub-threshold symptoms can still have the sleep, the school adjustments, the parenting support and the concentration strategies. A diagnosis is required for certain clinical decisions — medication above all — not for recognising need and acting on it. Given waits of four years or more for children, that matters enormously.

The case: Amara, 14

Back to the referral. Here is what was in the pack.

SourceFinding
Conners — parentInattention T = 79 (very elevated). Hyperactivity/impulsivity T = 54 — average.
Conners — teacherInattention T = 62, high-average only. Free-text note: "quiet, dreamy, anxious under assessment."
Conners — selfInattention endorsed — but loading heavily on worry and rumination.
SDQEmotional subscale high.
RCADSSocial and generalised anxiety elevated; depression moderate.
Sleep diaryOnset 01:00–02:00, roughly five hours, phone in bed, over two weeks.
Developmental historyPrimary school unremarkable. Milestones normal. Strong and organised to the end of Year 8.
HistoryChange over eight months. Friendship group collapsed. Avoiding PE and drama. A panic episode in a test. Maternal history of depression.

The chronology carries the answer, and most people miss it on first hearing. This is not primary ADHD. It is secondary inattention on a background of depression, anxiety and chronic sleep restriction. Onset is far too late; DSM-5-TR wants several symptoms before age 12 and the collateral shows none. The parent–teacher divergence is not an inconvenience, it is the finding. And moderate depression with five hours of sleep is a sufficient explanation sitting untreated — while everyone looks at a questionnaire.

The strongest objection in the room was the right one: girls are under-diagnosed, so could this be masked ADHD surfacing when demand rose? It deserves respect — and the answer is the shape of the curve. Compensation failing under academic demand produces gradual strain across a year or two. This is an eight-month collapse with panic, tearfulness and social withdrawal. That is an illness trajectory.

Crucially, both hypotheses converge on the same next step: treat the mood, anxiety and sleep, then reassess attention at a rested, euthymic baseline. Nothing is lost by sequencing it that way. A great deal is lost by handing a stimulant to a depressed, sleep-deprived fourteen-year-old.

When it is right to query somebody else's diagnosis

This is the part most relevant to prescribers, because in shared care you meet these children after the diagnosis is made.

If you are monitoring a young person and something doesn't fit — unexpected side effects, prominent anxiety, a response that makes no sense — it is legitimate to query the diagnosis and ask for it to be looked at again. That is not a challenge to a colleague; it is part of safe prescribing.

It matters because an incorrect diagnosis is not a neutral event. It means the wrong treatment, wrong expectations of the young person by school and family, wrong expectations of themselves, and damage to self-esteem — and it closes down the search for what was actually wrong. The consultant who taught this session was blunt that he has made this mistake himself, more than once, with secondary inattention. That is the error to be humble about.

Red flags, and when to refer

Stop and rethink on: sudden onset or loss of previously acquired skills (regression is not ADHD — investigate); symptoms in one setting only; a clearer primary explanation such as trauma, absence seizures, autism or profound sleep loss; and safeguarding concerns, which outrank everything and stop the assessment that day. Escalate urgently for severe impairment or risk. And always take a cardiac history — a family history of sudden death under 40 needs a specialist opinion and an ECG before any stimulant is considered.

On the pathway itself: refer directly to secondary care where impairment is severe. Where the impact is adverse but less severe, NG87 allows watchful waiting of up to ten weeks — not a mandatory queue, and paired with an active offer of group-based parent support that does not wait for a diagnosis. No universal screening in schools. And primary care does not make the initial diagnosis or start medication in under-18s.

Have a lower threshold for referral in children born preterm, looked-after children, those with anxiety or depression, other neurodevelopmental conditions, epilepsy, acquired brain injury, substance misuse, youth justice contact, or a close family member with ADHD. And especially in girls — not because you personally hold a bias, but because the criteria themselves are shaped in a way that makes ADHD easier to miss in them.

A referral that arrives with school collateral attached is the single most useful thing most prescribers can change this week.

This article is educational and written for UK clinicians. It is not individual clinical advice, and nothing here makes anyone competent to assess or diagnose ADHD — diagnosis in children and young people is a specialist act (NICE NG87 1.3.1). Work from the current full text of NG87.

The follow-on question — what happens once a diagnosis is confirmed and you have to decide whether, what, and what has to happen before the first dose — is covered in Tics Don't Block ADHD Medication: What NG87 Requires Before the First Dose.

Frequently asked questions

Does a very elevated Conners score mean a child has ADHD?

No. A Conners 3 T-score of 70 or above is "very elevated", but that raises the question rather than settling it. NG87 1.3.2 states that a diagnosis should not be made solely on rating-scale or observational data, and names the Conners scales and the SDQ as helpful added tools. The scale quantifies reported symptoms; it cannot establish onset before age 12, pervasiveness across two or more settings, or whether the inattention is primary rather than secondary — and all three are required by NG87 1.3.3.

What are the four things an ADHD rating scale cannot establish?

Onset — whether the pattern was present before age 12 and is trait-like rather than a change from baseline. Pervasiveness — a parent form describes home only. Primary versus secondary — anxiety, depression, sleep restriction and trauma all degrade attention and produce elevated inattention items. And freedom from bias — the items encode a stereotyped externalising presentation that skews male, one reason ADHD is under-detected in girls.

Why is the DSM-5 symptom count considered overrated?

Because the symptoms are not independent. A child who genuinely cannot sustain attention will also make careless mistakes, fail to finish work, appear disorganised and be easily distracted — so six items follow almost automatically once the underlying difficulty is real. The count is useful as a structure that stops a clinician skipping a domain, not as a measurement. The qualifiers do the diagnostic work: inconsistent with developmental level, causing impairment, six months, several symptoms before age 12, two or more settings, not better explained by another disorder.

Is ADHD over-diagnosed in the UK?

Both under- and over-diagnosis happen, to different children — but the national picture points to under-recognition. England recognises and treats ADHD at a lower rate than Norway, Denmark or Spain; fewer than one third of children meeting strict criteria have contact with any mental health service; and only 15–25% receive pharmacological treatment. Population prevalence has not risen substantially — referrals have, driven by awareness, the DSM-5 criteria changes, catch-up diagnosis in adults, and the pandemic.

Can I question an ADHD diagnosis that a specialist has already made?

Yes. If you are prescribing or monitoring under shared care and something does not fit — unexpected side effects, prominent anxiety, a response that makes no sense — it is legitimate to query the diagnosis and ask for it to be reviewed. An incorrect diagnosis is not neutral: it means the wrong treatment, wrong expectations, and it closes down the search for what was actually wrong. Raising a doubt is part of safe prescribing.

When should I refer a child for ADHD assessment, and to whom?

Where impairment is severe, refer directly to secondary care with no watchful waiting. Where the impact is adverse but less severe, NG87 allows watchful waiting of up to ten weeks — "up to", not a mandatory queue — paired with an active offer of group-based ADHD-focused parent support that does not wait for a diagnosis. Have a lower threshold in children born preterm, looked-after children, those with anxiety or depression, other neurodevelopmental conditions, epilepsy, acquired brain injury, substance misuse, youth justice contact, or a close family member with ADHD — and especially in girls (NG87 1.2.2). Send school collateral with the referral.

Faheem Ahmed

Educator, author and consultant across healthcare and education — and the voice behind The Pharmacy Guy. He supports clinicians, teams and organisations through teaching, training and consultancy.

This article draws on Session 1 of the MEDLRN Paediatric ADHD Series, delivered by Dr Chhitij Srivastava, Consultant Child & Adolescent Psychiatrist, on 15 August 2026.

Connect on LinkedIn →

Further reading

Comments

Seen a referral that rested entirely on one parent-rated scale? Or want the medication side in a follow-up? Leave a comment below — I read them all.