“…the scan which will enable doctors to diagnose autism more cheaply and quickly. The rapid test has already proven more than 90 per cent accurate in adults…”
“What the computer can do very quickly is to see that a patient has autism…even though their brain, to the naked eye, looks very normal.”
These bold statements appear on the website of the Maudsley Hospital’s national specialist services department (1). They appear to have been recycled from press releases and media interviews back in August, after publication of a study which looked at computerised pattern recognition of MRI brain scans, in adults with autism.
At the time, the “90% accurate” claim, and the suggestion that the scan could replace current clinically-based diagnoses, was heavily criticised by the head of the authoritative Oxford-based Centre for Evidence Based Medicine (2). A very different “5% accurate” was Dr Carl Heneghan’s view.
The key issue is that because the scan gives a “false positive” result in 20% of people without autism, it is unlikely ever to be useful, on its own, for diagnosis. If the prevalence of autism in the whole adult population is around 1%, then the “false positive rate” of the test would probably have to be at most 0.1% (3).
I wish the researchers (based at the Maudsley’s academic partner, the Institute of Psychiatry) the best of luck in getting their false positive rate down from 20% to this extremely low figure. I won’t be holding my breath.
What may be more feasible, and is implied by some of the comment back in August, is to combine the scan with a shorter (and cheaper) clinical assessment than the one which the Maudsley uses currently.
But I think patient groups, particularly the large and influential National Autistic Society (NAS), would want to see proper testing of such an approach.
Following the criticisms in August, the lead researcher responded: “we have clearly stated that we are not yet ready to make our approach available in the NHS just yet.” (2) (4).
So I was surprised to see the Maudsley website piece, dated November 4th, say “Adults who are interested in being scanned will need to ask their GP, consultant or health professional for a referral letter to the Behavioural Genetics Clinic. Privately funded assessments or scans are not available.” (1). There is no mention of any research showing the “Autism Scan” to have been improved.
Next week I am meeting with a board member (7) of DANDA (Developmental Adult Neuro-Diversity Association) (5), and will suggest that patient groups might want to clarify these apparently inconsistent statements (6) before recommending the “Autism Scan” to their members and supporters.
(3) That would mean about 1 false positive diagnosis in 1000 people, which I think would be the most any ordinary clinician would allow. To demonstrate such a small false positive rate would also require a much larger study than the one reported. Because the “Autism Scan” is said to be “90% Sensitive”, 9 people in 1000 with autism will be diagnosed correctly, while 1 person with autism will be “missed”.
Dr Heneghan does not mention the issue of the lack of so-called “confidence intervals” in interpreting the false-positive and false-negative figures from such a small study (20 patients with autism, and 20 controls without), but this seems important to me, especially for a test being offered to the NHS.
(4) Two “yet”’s in the original.
(5) The study also looked at whether the Scan could distinguish between autism and ADHD. It could, but even less well than between autism and no-autism. DANDA is concerned with the overlaps between autism, ADHD, dyslexia and dyspraxia, so I think this will be of interest to them.
(6) In addition to the other points, the 4th November piece (quoted in my first line) states “more than 90 per cent accurate”, but the abstract of the research paper (link from (1)) has “sensitivity…of up to 90%” [my italics]
(7) Changed from “the acting head” on 3rd December. My thanks to Erika Musselwhite of DANDA for pointing out that noone has yet taken over the national coordinator role which Mary Colley so energetically pursued until her recent much-regretted death.
Yesterday I attended the second annual meeting of UKAAN (UK Adult ADHD Network), which is led by academics and funded by a leading pharmaceutical company (1).
It was very interesting to hear how local NHS adult services, all English, were coping with increased referrals of patients. What seems clear is that many people are screened out, often by so-called “gateway workers”, who tell patients that they do not “meet the criteria”.
When the NICE guidelines came out, I wondered if patients with ADD / ADHD would often be found to not “meet the criteria” on the basis that their condition was “only” mild. However, it looks as though a lot of mild and moderate ADD / ADHD is simply being diagnosed as “no ADD / ADHD”.
I have already seen many people with moderate-to-severe disorders, who should fall within the NICE guidelines, but have been told that they should stop “medicalising their past failures” and just get on with their lives.
The President of UKAAN, Professor Phil Asherson, told the meeting that in many areas services are being closed. This means that people who have been on waiting lists for months may end up being told that they will not be seeing a specialist after all.
I will be suggesting to ADDISS and AADD-UK that they consider telling their members and supporters about the reality of all this in greater detail and that, for those who can afford it, paying to see an independent specialist may be the best option for at least the next 2-3 years (2).
(1) There appeared to be no promotional activity at all within the meeting itself.
(2) “2-3 years” is based on the wide perception of how long the most intense NHS spending restraint may last.