Is the QbTest Accurate for Adult ADHD?

If you’ve been through an ADHD assessment recently – or you’re stuck waiting for one – there’s a good chance you’ve come across the QbTest or QbCheck, often advertised by clinics who also offer ADHD Diagnosis or assessments.

These computerised tests are increasingly common in both NHS and private ADHD services, and in October 2024, NICE recommended the QbTest as an option to support ADHD diagnosis in children and young people¹.

But what does the independent research actually tell us – especially if you’re an adult?

As someone who works with late-diagnosed adults who have, or suspect they need, an ADHD diagnosis, I’ve been digging through the published studies, and the picture is a lot more nuanced than the press releases suggest.

What are QbTest and QbCheck?

First, the basics. QbTest and QbCheck are continuous performance tests (CPTs) developed by a Swedish company, Qbtech².

During the 15–20 minute test, you sit at a computer, wear a headband with a reflective marker, and respond to shapes on screen by pressing a button. An infrared camera (QbTest) or a webcam (QbCheck) tracks your head movements.

What the test claims to measure

The tests claim to capture three things:

1. inattention (how consistently you respond to targets),
2. impulsivity (whether you press when you shouldn’t), and
3. hyperactivity (how much you move).

Your results are compared against a “normative database” of people without ADHD of the same age and sex.

Both are FDA-cleared, CE-marked, and available for ages 6–60. QbTest is the in-clinic version with specialised hardware; QbCheck is the remote version you can do at home on your own webcam.

What do the headlines claim?

You’ll find some impressive statistics attached to these tests. NHS press releases and Health Innovation Network materials ³cite figures like “£55.9 million in NHS savings”, 132,652 hours of healthcare capacity released, and diagnoses achieved 55 days faster than standard assessments.

It sounds remarkable – especially when you know how dire the situation is for children and adults in the UK, and in Scotland in particular, right now.

NICE’s 2024 guidance describes the QbTest as likely to be cost effective when used alongside standard clinical assessment⁴.

Qbtech’s own materials describe the tests as “clinically validated” with over 50 supporting studies.⁵ So far, so reassuring. But let’s look at what the research actually says.

Is the Qb Test accurate for diagnosing ADHD?

The most rigorous independent evaluation is a 2024 systematic review and meta-analysis ⁶ published in the Journal of Child Psychology and Psychiatry by Bellato and colleagues, drawing on the work of researchers at Southampton, Nottingham, Cardiff and King’s College London. It pooled 15 studies with over 2,000 participants.

The findings are sobering.

The QbTest’s total score showed what the researchers called “acceptable, rather than good” accuracy – sensitivity of 0.78 and specificity of 0.70. In plain terms: it correctly identified about 78% of people who have ADHD, and about 70% of those who don’t.

The individual subscales did worse. Sensitivity ran as low as 0.48 to 0.65 – meaning the test missed between roughly a third and a half of people with ADHD, depending on which symptom was being measured. The impulsivity subscale was the weakest, scoring barely above random chance.

The authors’ conclusion was blunt: used on its own, the QbTest isn’t accurate enough to tell ADHD and non-ADHD cases apart, and it should not be used as a standalone screening or diagnostic tool – or as a way of triaging who gets onto the waiting list.

Read that last part again. It should not be used to decide who gets added to the waiting list, or where they sit on it. Do we really think that never happens?
.

The Problem with ‘Healthy Controls’

Here’s where it gets genuinely worrying.

Many of the studies showing better results compared people with ADHD against healthy controls – people with no psychiatric conditions at all. That’s not the clinical reality for most people who are being assessed for ADHD. In practice, clinicians need to separate ADHD from the things that look like it: anxiety, depression, autism, sleep problems, trauma responses – conditions that frequently travel alongside ADHD OR can be mistaken for it.

When researchers tested the QbTest against that real-world picture, performance dropped sharply:

The pattern is consistent: the moment you ask the test to do the hard, real job – not “ADHD vs nobody,” but “ADHD vs everything else it could be” – it struggles because it’s not able to do a differential diagnosis. Inattention, hyperactivity or impulsivity are not enough to say whether someone’s ADHD or not.

Does the Qb Test Work For Adults?

This is where I have the most immediate concerns, because adults are who I mostly work with.

The NICE recommendation is for children and young people only.

There is no NICE guidance supporting the QbTest or QbCheck for adult ADHD diagnosis – and the adult evidence base is weaker, with poorer performance in real clinical populations.

This isn’t a fringe view: when Which? investigated private ADHD assessments, the clinicians they spoke to flagged that the QbTest is recommended for children, not adults, because the evidence simply wasn’t there for adults.¹⁰

This matters a lot, because adults seeking assessment tend to have the most complex presentations of all – they’ve got years of coping strategies, co-occurring anxiety and depression, sometimes other neurodevelopmental conditions as well. These are exactly the population groups where the test performs worst.

There’s also a revealing finding from one longitudinal study: adults retested about four years apart tended to improve, with fewer scoring in the clinical range the second time.

Those who’d taken their medication on test day were less likely to score above threshold than those who’d skipped it¹¹ which raises an obvious question for the real world – where plenty of us forget our meds unless we’ve built bombproof routines with backups.

For late-diagnosed adults – particularly women who’ve spent decades masking and compensating – a 20-minute test in a quiet, controlled room may not capture any of the real life difficulties that drove them to seek help in the first place.

If a private Qb Test costs you around £300, it’s fair to ask what you’re actually getting.

What does it mean if your QbTest is normal but you still have symptoms?

This might be the most important question I had when I started looking at the testing – and it’s one barely anyone is answering bluntly.

If you sat the test, scored in the “normal” range, and walked away being told you probably don’t have ADHD – but you know something doesn’t add up -the evidence says: a normal QbTest does not rule out ADHD.

Remember the numbers from before. The test misses up to half of people who genuinely have ADHD on some measures, and performs close to random chance for inattention and impulsivity in adult clinical populations. (See Bellato et al, and Brunkhorst-Kanaan et al.)

If your ADHD shows up mainly as inattention rather than visible hyperactivity – which is far more common in women and in adults who’ve learned to sit still and white-knuckle through – you are precisely the person a “normal” result is most likely to miss.

A quiet room, a single screen, no distractions, and your full attention summoned for 20 minutes is, frankly, one of the easier environments for an ADHD brain. It’s nothing like real life that’s actually falling through the cracks: unanswered emails, abandoned projects, the effort of trying to keep it all together.

A test that rewards masking with a clean result hasn’t ruled out your ADHD because honestly, it probably feels more like a wee break.

This is the gap I work in – and it’s why I’d politely push back on anyone treating one normal score as the final word.

Can a QbTest actually rule out ADHD?

The AQUA Trial ¹²– the research used to underpin the wider NHS rollout – has been reported as showing that clinicians with access to QbTest reports “doubled the likelihood of excluding ADHD.” (I’m sure the popular press and politicians will be delighted).

On the surface that sounds efficient: they quickly rule out ADHD, redirect people to other support. This sounds very different when we consider everything I’ve shared so far. If the test misses a third to a half of people with ADHD, then “doubling exclusions” raises an very uncomfortable question: how many of those exclusions were correct – and how many were people with ADHD whose test simply came back ‘normal’?

The trial didn’t follow up to find out. We don’t know how many of those ‘ruled out’ went on to be diagnosed elsewhere, or just kept on struggling without the right support.

For a late-diagnosed adult, this isn’t an abstract risk. A normal result used to exclude ADHD misses out the people whose ADHD doesn’t show up on this particular 20-minute task.

Where do the “£55.9 million NHS savings” figures come from then?

Let’s trace those juicy headline numbers – and look at what the trial actually tested.
The key piece of research is the AQUA Trial, a randomised controlled trial led by Professor Chris Hollis at the NIHR MindTech centre, University of Nottingham.¹³

There’s a crucial detail that isn’t highlighted : every participant had a QbTest. All of them.

The randomisation wasn’t “QbTest vs no QbTest” – it was between clinicians who could see the report (the “QbOpen” group) and clinicians where it was withheld until the study ended (the “QbBlind” group).

So the trial didn’t measure whether the QbTest improves diagnosis. It measured whether knowing a QbTest result exists changes clinician behaviour.

And it did. Clinicians with the report made faster decisions (76% reached a decision within six months, versus 50% without), appointment length dropped 15%, and confidence rose.

But perhaps the most important finding was this: there was no difference in diagnostic accuracy between the two groups.

The diagnoses weren’t better with the report. They were faster, and made with more confidence. A concrete number to point to – “the test shows X” – gave clinicians permission to decide faster and more confidently but that is NOT the same as deciding correctly.

From “cost neutral” to “£55.9 million”

Now the costs. The AQUA Trial’s own health-economic analysis found the savings so small that the impact was best described as “cost neutral.”

Cost neutral.

Yet the implementation materials cite £55.9 million in savings and 132,652 hours released. Those are modelled projections – what might happen if the findings were scaled across the NHS, assuming faster decisions translate straight into savings and that the faster decisions are correct. They are not independently verified, real-world savings.

It’s also worth noting the AQUA Trial received support from Qbtech in the form of test systems and data handling. While this doesn’t invalidate the work. it is relevant context when the same findings reappear, scaled up, in the company’s marketing.

So what can the QbTest actually do – and not do?

To be clear, I’m not saying the tests are worthless – not at all. The research DOES support some real uses.

What it can do

• Provide additional objective data as part of a comprehensive assessment – the key words being “part of.”
• Help clinicians feel more confident, and help communication – some people find a visual report easier to grasp than clinical observation alone.
• Support treatment monitoring – tracking whether medication is shifting attention and activity over time.
• Add efficiency to a pathway when used appropriately, potentially reducing the number of appointments needed.

What it can’t do

• Reliably confirm or rule out ADHD as a standalone tool.
• Serve as a screening or triage tool for who gets access to assessment.
• Reliably distinguish ADHD from conditions with overlapping presentations.
• Provide sufficient accuracy for adult diagnosis, especially in complex cases.
• Recognise the difference between inattentive, hyperactive and combined presentations.

Questions to ask if you’re offered a QbTest

If a QbTest or QbCheck is offered as part of your assessment, it’s reasonable to ask:

And one last point for you:

if you’ve previously been told you don’t have ADHD partly on the basis of a QbTest, is it worth a second opinion?

That’s your call – but I’d suggest revisiting the assessment, and perhaps taking this information back to whoever ran it.

If you’re being offered a QbCheck as the primary or only assessment tool – particularly as an adult – that should raise real concern because the evidence doesn’t support it.

The bigger picture

The rapid spread of the QbTest has happened against a backdrop of overwhelming demand and waiting lists stretching to years. The pressure to find efficiencies and find support for people is very understandable.

However, we can’t allow efficiency to come at the cost of accuracy. A test that misses a substantial share of people with ADHD, or can’t reliably separate ADHD from other conditions, isn’t solving the problem of long waiting lists, it’s just creating NEW problems.

So many of us have already been missed, whether that’s by assessments that weren’t built to recognise ADHD in girls and women, in high achievers, in people who learned early to compensate, in children with high IQs. The very last thing any of us needs is another tool that rewards masking with a “normal” result.

The research is clear. The QbTest and QbCheck can play a supporting role in ADHD assessment. They are not – and should not be treated as – a shortcut to diagnosis.

If a test told you you’re fine – and you know you’re not

If a 20-minute test said you’re fine, and every other part of your life says otherwise, the problem isn’t you. No short, computerised, test webcam monitoring of you doing a single task in a quiet room can see how your brain actually works most days.

That’s the whole question “Why Am I Like This?” (WAILT) was built to answer. It doesn’t use a database or clinical-range, but is an in depth, proper look at how your ADHD actually shows up: Lightbulb ADHD’s unique assessments, followed by a ninety minute, live conversation, and a clear picture of all the things that a 20-minute automated test cannot catch.

If you’ve been dismissed by a test and you’re still sure something’s going on, start here →

For those who want ongoing support turning that understanding into a life that actually works, 1:1 coaching is the next step – but understanding yourself comes first.

FAQs about the QbTest and QbCheck

1. NICE Guidance: QbTest for the assessment of attention deficit hyperactivity disorder (October 2024) – https://www.nice.org.uk/guidance/mtg90 ↩︎
2. Qbtech – QbTest product information – https://www.qbtech.com/adhd-tests/qbtest/ ↩︎
3. NIHR MindTech / Health Innovation East Midlands implementation figures – https://www.mindtech.org.uk/innovators/case-studies/qbtest ↩︎
4. See NICE guidance 1. ↩︎
5. See Qbtech information 2. ↩︎
6. Bellato et al. (2024) “Practitioner Review: Clinical utility of the QbTest…” Journal of Child Psychology and Psychiatry – https://acamh.onlinelibrary.wiley.com/doi/10.1111/jcpp.13901 ↩︎
7. Hult et al. (2015) “ADHD and the QbTest: Diagnostic Validity of QbTest” Journal of Attention Disorders – https://pubmed.ncbi.nlm.nih.gov/26224575/ ↩︎
8. Brunkhorst-Kanaan et al. (2020) “The Quantified Behavioral Test – A Confirmatory Test in the Diagnostic Process of Adult ADHD?” Frontiers in Psychiatry – https://pmc.ncbi.nlm.nih.gov/articles/PMC7100366/ ↩︎
9. Johansson et al. (2021) “The Quantified Behavioral Test Failed to Differentiate ADHD in Adolescents With Neurodevelopmental Problems” Journal of Attention Disorders – https://journals.sagepub.com/doi/abs/10.1177/1087054718787034 ↩︎
10. Which? – “Is it worth getting a private ADHD assessment?” – https://www.which.co.uk/reviews/private-healthcare/article/is-it-worth-getting-a-private-adhd-assessment-aExlr3U9RisV ↩︎
11. Hollis et al. (2018) “The impact of a computerised test of attention and activity (QbTest)…” Journal of Child Psychology and Psychiatry – https://pmc.ncbi.nlm.nih.gov/articles/PMC6124643/ ↩︎
12. AQUA Trial Protocol – Hall et al. (2014) BMJ Open – https://pmc.ncbi.nlm.nih.gov/articles/PMC4256543/ ↩︎
13. See Hollis et al. ↩︎