Adult diagnosis: do doctors really know what they are doing?

Post Reply New Topic

That's probably the way it happens a lot of the time, but it seems rather glib of them to just dismiss the patient's concerns like that. Wouldn't it be better to at least give them the AQ test?

The AQ tells little though, many people score high and are not autistic.
Scoring above a cutoff is like having the personality traits on the list.

Well, I think any reasonable screening test is better than just sending them away. If I remember right, when I was seeking a DX, more than one diagnostician I contacted had the AQ test as the first step. Though you could be right that it's not the best one.

I'm wondering how you know this... that "many people" score above the cutoff but aren't on the spectrum? Are there some studies of the tests? Do some people score below the cutoff, and yet are on the spectrum?

Over the years I have had quite a bit of exposure to psychiatrists and psychologists. They seem more like astrologers, Tarot Card Readers, chiropractors, or homeopaths, to me, than actual doctors... healers that are using science. They all speak a special language (usually one they learned at university), and slowly make you think that you understand that language too.

I've wondered for years what would happen if you sent several Aspies to a dozen different "helping professionals" for a dx... Aspies with different behaviors? (... but wait! No one would agree on who our test Aspies are!)

That would be an interesting study. I have read summary accounts of work by review panels attempting to assess the accuracy of ASD diagnoses by giving more rigorous assessments to people who were already diagnosed. This would be a very interesting alternative route to the same information. I suspect there would be reluctance to do it because it would be so directed at the specific doctors.

There was a famous study about cognitive biases in the assessment of institutionalized mental patients that placed "pseudo patients" in psych hospitals and looked at how ordinary personality traits were reinterpreted by staff psychiatrists as confirmation of bogus diagnoses.

There might be ethical barriers to doing it. If you could design a study that worked, it would be fascinating.

Missed the edit window.
The experiment I was thinking of was the Rosenhan Experiment:
https://en.wikipedia.org/wiki/Rosenhan_experiment

I am thinking this would be different in that you would be sending people who the researchers had done their best to identify as definitely autistic to a variety of settings for assessment. The ethical challenge is that they would have to withhold those parts of the patient's medical records that revealed the prior autism diagnosis and they would be actively deceiving the assessing doctor about past diagnosis. Perhaps this is reason enough that this could never happen. Maybe if the team conducting the research also did the initial diagnosis, they could then withhold the results until the experiment was complete.

I've had two psychologists tell me I'm on the spectrum and one tell me I'm not. So just based on my anecdotal data, the percent of doctors who don't know what they're doing is >=33%.

There are still some clinicians who will not consider a person autistic if they are able to hold a conversation AT ALL.

All tests give false positives and/or false negatives.

For the current AQ cutoff of 32, the sensitivity is 76.71%. That is, ~77% of people on the spectrum test positive on the test, and 23% who are on the spectrum test below the cutoff.

The specificity is 74.07%. That is, 74% who are not on the spectrum test below the cutoff, and 26% who are not on the spectrum test above it.

It should be noted that the cutoff score used to be 26. That had a sensitivity of 94% but a specificity of only 52% (ie 48% of NTs still tested positive on the test). The cutoff score was revised up to 32 to avoid having so many false positives.

By contrast the RAADS-R has a clinical sensitivity of 97% and specificity of 100%. It's a far better test for screening for autism, especially in adults since it makes the distinction between behaviors as a child and behaviors as an adult.

I believe the above-stated tests have some validity--but I wouldn't use these scores alone within a comprehensive diagnostic situation.

I really believe a clinician, in order to accurately diagnose an ASD, has to be around that person for at least a day or so--observe this person in his/her daily activities.

Obviously, this is not possible--so (some) do the best they can.

I would agree. I'm somewhat skeptical of the specificity/sensitivity numbers for the RAADS-R. If it was actually as high as the research paper claims, one would expect it to be widely used for diagnosis purposes, yet it's still only used for initial screening by diagnosticians. I do still think it's a better test than the AQ though. I've seen 90/93% sensitivity/specificity numbers cited for it, but I couldn't find the source for that.

It should also be noted that the way these numbers are interpreted can be somewhat counter-intuitive. If you score a positive result on a test with 90% sensitivity, that does not mean there's a 90% likelihood you're on the spectrum. If you assume 2% of people are actually on the spectrum then scoring positive on a test with 90% sensitivity and specificity only means you have a 15% chance you're on the spectrum. See Bayes Theorum for an explanation why.

In my opinion, the above tests are really not comprehensive enough to be useful in a situation beyond a "screening."

It's similar to a (US) indictment, in my opinion. When one is indicted for a crime, it is based on an assessment of a person's guilt, by a grand jury, based upon a "preponderance of the evidence." T

his is not enough for a conviction, though. In order to obtain a conviction, a jury must be convinced, "beyond a reasonable doubt" that the person is guilty. "Beyond a reasonable doubt" carries more of a burden than "preponderance of the evidence.

The screening is the "indictment."

The diagnosis is the "verdict."

Minor mathematical quibble:

I would think the above numbers only work if you assume that 2% of those taking the test are on the spectrum. Presumably the actual probability of autism in those who are referred for or seek diagnosis is higher than the rate in the general population (~1-2%). A large portion of the 98-99% of non-autistic people will never take the test and can never produce false positives, and so on. Or have I misunderstood the application of Bayes Theorem?

That's true. However, without being able to quantify how much more likely someone seeking a test is going to be on the spectrum, I don't think you can really attach any meaningful statistics to how the impacts the test results. The 15% likelihood would probably be a lower bound, but you can't say the test is any more certain with any kind of statistical validity. If you could quantify it, you could easily incorporate it into the calculation.

My point is that people who want assessment need to identify concrete impairments caused by autism, e.g. can't hold job due to social ineptitude and getting fired, can't hold job/go to school due to sensory issues, can't hold job/go to school due to slow processing speed, can't get job due to social ineptitude, can't have manual job due to motor deficits, can't work in customer service due to social deficits, problems communicating in daily interactions, etc.

Without concrete impairments, psycholigists or GPs may be unwilling to put their name on a referral for autism assessment for someone who behaves within normal variation and doesn't come across as obviously autism. With concrete impairments, the doctor has concrete support for signing their name on a referral. It's evidence in support of the doctor's decision to refer, just like the autism specialist would put concrete evidence of behaviors for each diagnostic criterion or say that criterion is not applicable.

I wondered about this myself, especially before being diagnosed. How could someone tell the difference, if the traits are so similar? I was a skeptic before being diagnosed. And after. I knew I had some of the “symptoms”. But who knew if they a result of having Asperger’s? Or was it simply something I became (in response to my environment/upbringing)?

Nowadays, I don’t care that much about whether I have a personality disorder or Asperger’s (what I was diagnosed with). I just know I am on the outside looking in.