The RAADS-14 Screen: A Three-Minute Autism Screen for Psychiatric Settings
Fourteen questions cut from an eighty-question instrument. It catches 97 percent of autistic adults. It also cannot reliably tell autism from ADHD.
The RAADS-14 Screen was designed for a specific problem. A psychiatrist has forty minutes with a patient who has depression, anxiety, a personality disorder diagnosis and a long history of not quite fitting anywhere. Autism might be underneath it. There is no time for an eighty-item questionnaire, and the ones that exist were validated against healthy controls rather than against the other patients in the waiting room.
That is the gap this instrument was built to fill, and it is the reason it is more interesting than its length suggests.
What is the RAADS-14 Screen?
The RAADS-14 Screen was published in 2013 by Jonna Eriksson, Lina Andersen and Susanne Bejerot at the Karolinska Institute in Sweden. They started with the 80 items of the RAADS-R Test and worked out which handful carried the most discriminating power.
Fourteen survived. The response format came across unchanged from the parent instrument, and it is the thing that makes this family of questionnaires different from the AQ family. You do not rate agreement. You say when something was true of you:
- True now and when I was young
- True only now
- True only when I was younger than 16
- Never true
Each answer scores 0 to 3, so totals run from 0 to 42. One item is reverse-scored. That lifetime framing matters, because autism is developmental. An instrument that asks only about the present is answering a different question.
The three subscales
Factor analysis produced three groupings:
- Mentalizing deficits, seven items. Reading intent, understanding what people mean rather than what they say, knowing how a conversation is going.
- Social anxiety, four items. Discomfort and avoidance in social situations.
- Sensory reactivity, three items. Sound, texture, light and the rest.
Internal consistency across all 14 items was strong, with an alpha of 0.90 in a combined sample of 1,233 adults.
How the RAADS-14 Screen is scored
Scoring the RAADS-14 Screen takes a moment. Add the fourteen item scores. The published threshold is 14 out of 42.
| Score | Reading |
|---|---|
| 0 to 13 | Below the threshold. Autism is unlikely on this evidence, and the instrument is genuinely good at this. |
| 14 to 42 | At or above threshold. Further assessment is indicated. This tells you much less than the number feels like it should. |
Median totals in the validation study give useful context. Autistic participants scored 32. Participants with ADHD scored 15. Participants with other psychiatric diagnoses scored 11.
Look at the ADHD median. Fifteen. The cutoff is fourteen. The average person with ADHD and no autism diagnosis scored above the threshold for autism, and that is not a rounding problem. It is the central limitation of the instrument.
How accurate is the RAADS-14 Screen?
The RAADS-14 Screen validation enrolled 1,233 adults: 643 psychiatric patients, of whom 135 had an autism diagnosis, and 590 non-psychiatric controls. Patients came from seventeen Swedish psychiatric clinics. That is a serious sample, and considerably more relevant to real clinical work than the healthy-control designs used for most autism questionnaires.
Sensitivity was 0.97. Of the autistic participants, 97 percent scored at or above the cutoff.
Specificity is where it gets complicated, because the answer depends entirely on who you are comparing against.
| Compared against | Specificity | AUC |
|---|---|---|
| Non-psychiatric controls | 0.95 | 0.99 |
| Other psychiatric disorders | 0.64 | 0.91 |
| ADHD | 0.46 | 0.88 |
This is not a defect peculiar to this instrument. Autism and ADHD overlap heavily in the traits that self-report questionnaires can see, and the two conditions co-occur often. It is a problem for the whole field. But it is unusually visible here because the authors did the honest thing and published the comparison, which many validation studies avoid by using healthy controls only.
The missed cases
Nine of the 135 autistic participants scored below the RAADS-14 Screen cutoff. Two of them scored zero. Out of forty-two.
Think about what that means for a moment. Two people with a clinical autism diagnosis answered all fourteen questions in the least autistic way available. That is masking, or poor insight, or both, and it is a reminder that a low score on any self-report instrument is evidence rather than proof.
What the study did not report
The RAADS-14 Screen validation did not publish positive or negative predictive values. That is a meaningful gap, because those are the numbers that answer the question people actually have: given my score, how likely is it that I am autistic? Sensitivity and specificity answer the reverse question, and converting between them requires knowing the base rate in your population.
In a psychiatric outpatient clinic, where autism might be present in five or ten percent of patients, a sensitivity of 0.97 and specificity of 0.64 produce a positive predictive value in the region of fifteen to twenty-five percent. Most positive screens would be wrong. That is normal for screening instruments and it is why a screen is followed by an assessment.
Who the RAADS-14 Screen is for
The RAADS-14 Screen was built for, and validated on, adults attending psychiatric services. That is where it belongs and where it performs best.
It is a sensible choice if you are a clinician with limited time and a patient whose history does not add up. It is also reasonable for an adult who already has a psychiatric diagnosis and has begun to wonder whether something underneath it was missed, which is an extremely common route to late autism diagnosis.
It is a poor choice in two cases. If you want a profile rather than a yes or no, fourteen items cannot give you one. And if you already know you have ADHD, the instrument’s weakest performance is precisely in your situation, so a positive result will tell you very little.
The five-item version
The authors arranged the items so the five strongest come first, allowing an even shorter screen. At a cutoff of 4 out of 15, sensitivity was 0.93 and specificity was 0.45 against ADHD, 0.49 against other psychiatric disorders and 0.90 against non-psychiatric controls. Faster. Blunter. Useful mainly when the alternative is no screening at all.
Searching for the RAADS-14 Test by another name
People look for this instrument under several labels: the RAADS-14 Screen, the RAADS-14 Test, the RAADS-14 questionnaire, or just RAADS-14. They are the same fourteen items. The word Screen is the authors’ own, and it is deliberate. A screen sorts people into worth-investigating and probably-not. A test implies something it cannot deliver.
There is also frequent confusion with the parent instrument. The RAADS-R Test has 80 items, four subscales and a 0 to 240 range, with its own widely debated threshold of 65. The short version shares its response format and its lifetime framing but nothing else. If you have taken one, you have not taken the other.
Strengths and weaknesses of the RAADS-14 Screen
Strengths
- Built and validated specifically for psychiatric populations, unlike almost every alternative
- Very fast. Three minutes.
- Asks about your whole life, not just the present
- Sensitivity of 0.97, so a negative result is informative
- Strong internal consistency at alpha 0.90
- Large validation sample of 1,233 adults across seventeen clinics
- The authors published the uncomfortable comparisons instead of hiding behind healthy controls
Weaknesses
- Specificity of 0.46 against ADHD, which is close to chance
- The median ADHD score sits above the autism cutoff
- No published positive or negative predictive values
- Fourteen items cannot produce a usable profile
- Validated against DSM-IV-TR diagnoses, before DSM-5
- Groups were not matched for gender, age or intelligence
- Diagnoses were taken from clinical records with no confirmatory assessment
Controversies around the RAADS-14 Screen
The ADHD problem
This is the main criticism of the RAADS-14 Screen. A screening instrument that cannot distinguish the two most commonly confused neurodevelopmental conditions has a narrow range of usefulness, and a large share of people being screened for autism in psychiatric settings either have ADHD or are being assessed for it. The authors were upfront about this. Many websites reproducing the instrument are not.
Whether social anxiety belongs in it
One of the three subscales is social anxiety, which is a diagnosis in its own right and extremely common in psychiatric populations. Including it raises sensitivity and almost certainly lowers specificity, since people with social anxiety and no autism will endorse those items. Whether that trade is worth making depends on what you want the screen to do.
Self-report in a psychiatric population
The validation relied on diagnoses recorded in Swedish clinical practice with no independent confirmation. The authors acknowledge this, and also note that some patients in the comparison groups may have had undiagnosed autism, which would have pushed measured specificity down. Both caveats cut in different directions and neither is resolvable from the published data.
Two people scored zero
It bears repeating because it is so rarely mentioned. Two diagnosed autistic adults scored nothing at all. Any site telling you that a low score rules autism out is overstating what fourteen questions can establish.
How the RAADS-14 Screen compares to other instruments
| Instrument | Items | Time frame | Validated against |
|---|---|---|---|
| RAADS-14 Screen | 14 | Whole life | Psychiatric outpatients |
| RAADS-R Test | 80 | Whole life | Mostly clinical and control samples |
| AQ-10 Test | 10 | The present | General population, then clinical referrals |
| AQ-50 Test | 50 | The present | Diagnosed adults vs random controls |
| CAT-Q Test | 25 | The present | Autistic and non-autistic adults |
| ADOS-2 Test | Observation | Observed now | Clinical diagnostic samples |
| ADI-R Test | 93 | Development | Clinical diagnostic samples |
| M-CHAT Screening Test | 20 | Toddler, now | 16,000 screened toddlers |
Against the AQ-10 Test, which occupies the same brief-screener niche, the RAADS-14 Screen has two real advantages: it asks about your whole life, and it was validated against other patients rather than against the general public. Those are the two things that most often go wrong with short autism screeners.
If you want depth, the RAADS-R Test gives you 80 items and a four-area profile. If you suspect masking is suppressing your scores, the CAT-Q Test will explain more than retaking any autism screener will.
And if the ADHD overlap described above is your situation, a single-condition instrument is the wrong tool. Our Neurodivergent Test screens seventeen conditions including both, spends its questions where your answers point, and reports impact separately from traits. For a plain-language autism screen of moderate length, the Free Autism Test takes about six minutes.
Frequently asked questions about the RAADS-14 Screen
What is a high score?
Fourteen or above is the published threshold. Autistic participants in the validation scored a median of 32 out of 42. A score in the thirties is high. A score of 15 is above the cutoff but also the median for people with ADHD and no autism.
I scored above 14. Am I autistic?
No. You have crossed a screening threshold, which means further assessment is indicated. In a psychiatric population most positive screens turn out not to be autism, because the instrument is tuned to catch nearly everyone at the cost of flagging many people who do not have it.
I have ADHD. Is this instrument useful for me?
Less than for anyone else. Specificity against ADHD was 0.46 and the median ADHD score was above the cutoff. A positive result in your case is close to uninformative. You would be better served by something that assesses both.
I scored below 14. Does that rule autism out?
It is reasonable evidence against, and ruling out is what this instrument does best. But nine of 135 diagnosed autistic participants scored below the cutoff, and two scored zero. If your own experience strongly contradicts the result, the result is not the final word.
What is the difference between the RAADS-14 and the RAADS-R?
Length and purpose. The short version has 14 items, a 0 to 42 range and a cutoff of 14. The RAADS-R Test has 80 items, four subscales, a 0 to 240 range and a cutoff of 65. They share a response format and a lifetime framing. The long version gives you a profile. The short version gives you a decision.
Is the RAADS-14 Test the same thing?
Yes. The same fourteen items appear under both names, along with RAADS-14 questionnaire and plain RAADS-14. The authors called it a screen deliberately, because it sorts rather than concludes.
Why does it ask when something was true rather than whether I agree?
Because autism is developmental and begins in childhood. A trait that appeared at thirty means something different from one that was there at six. Agreement-format questionnaires like the AQ-50 Test cannot make that distinction, which is a real advantage for this family of instruments.
Can I take it for someone else?
No. It is self-report and the items ask about internal experience that an observer cannot access. For informant-based assessment, the relevant instrument is the ADI-R Test, administered by a clinician.
Is it free?
The items were published in an open access journal and are freely available for clinical and research use. You should not be charged to take it.
How long does it take?
Three minutes, give or take. Five if you think carefully about the childhood framing, which is worth doing, since that is where the instrument’s value lies.
Can I use my score to get an assessment?
It helps, particularly if you are already under psychiatric care, because this instrument was designed for exactly that setting and clinicians recognise it. Bring concrete examples as well. A score opens a conversation. Specifics move a referral.
Does it work for women?
The validation did not match groups for gender, and around ten percent of participants did not report it, so the published data cannot answer this well. The general finding that self-report instruments underperform for autistic women has no reason not to apply here, and with only fourteen items there is little room for it to average out.
What about the five-item version?
The first five items can be used alone with a cutoff of 4 out of 15, giving sensitivity of 0.93. Specificity drops further, to 0.45 against ADHD. It is a triage tool for situations where even three minutes is too long.
What to do with your result
If you scored above the threshold and you are already under psychiatric care, raise it there. This instrument was built for that conversation, and a clinician who recognises it will take it more seriously than a generic online quiz result.
If you scored above the threshold and you also have ADHD, or suspect you might, do not read much into it. Get something that assesses both.
If you scored below and you are satisfied, that is a reasonable place to stop. Ruling out is the thing this instrument genuinely does well.
If you scored below and you are not satisfied, trust that. Two people with diagnoses scored zero on this questionnaire. Try a longer instrument, look at masking, and write down the specific situations where things go wrong before you speak to anyone.
Sources
- Eriksson JM, Andersen LMJ, Bejerot S. RAADS-14 Screen: validity of a screening tool for autism spectrum disorder in an adult psychiatric population. Molecular Autism, 2013;4:49. DOI · PubMed. The development and validation paper, open access. Every figure on this page comes from it.
- Ritvo RA, Ritvo ER, Guthrie D, et al. The Ritvo Autism Asperger Diagnostic Scale-Revised (RAADS-R): a scale to assist the diagnosis of autism spectrum disorder in adults. Journal of Autism and Developmental Disorders, 2011;41(8):1076-1089. DOI. The parent instrument.
- Ashwood KL, Gillan N, Horder J, et al. Predicting the diagnosis of autism in adults using the Autism-Spectrum Quotient (AQ) questionnaire. Psychological Medicine, 2016;46(12):2595-2604. DOI. On how brief autism screeners perform against real diagnostic outcomes.
- Harrison AG, Edwards MJ. The ability of self-report methods to accurately diagnose attention deficit hyperactivity disorder: a systematic review. Journal of Attention Disorders, 2023;27(12):1343-1359. DOI. Relevant to the autism and ADHD overlap discussed above.
Search the literature yourself
- PubMed: RAADS-14 autism screening and autism and ADHD differential diagnosis in adults.
- Google Scholar: RAADS-14 validation and specificity.
- Molecular Autism, the open access journal that published the validation.
- Karolinska Institute, where the instrument was developed.
- NICE CG142: Autism spectrum disorder in adults. UK guidance on recognition and diagnosis.