Monday, January 23, 2012

The New York Times' DSM 5 Autism Coverage

The New York Times apparently thinks that life is too calm for people in the autism community.  The paper has decided to shake things up a little by publishing a story suggesting that many people who have a diagnosis of autism now will lose their diagnosis under the proposed DSM 5 autism criteria.  The story was published on Thursday night and was quickly picked up by news outlets around the web.

The story, which is supposedly based on a "new study", is filled with hyperbolic statements like -
Proposed changes in the definition of autism would sharply reduce the skyrocketing rate at which the disorder is diagnosed and might make it harder for many people who would no longer meet the criteria to get health, educational and social services" 
The changes would narrow the diagnosis so much that it could effectively end the autism surge, said Dr. Fred R. Volkmar, director of the Child Study Center at the Yale School of Medicine and an author of the new analysis of the proposal. “We would nip it in the bud.”
But after you get past the chicken little routine and start looking for the real data behind these provocative statements what you will find is ... nothing.  The story isn't based on a newly published study but rather is based on a presentation given at a meeting of the Icelandic Medical Association by one Dr. Fred Volkmar just the day before the story was published.

The actual study won't be published for another month or two.

To make matters worse, the members of the public aren't the only ones that are being kept in the dark about the contents of this study.  It seems that the American Psychiatric Association (APA), the group that is responsible for the DSM 5 autism criteria, was also taken by surprise by this story.  Members of this organization read about how their new criteria are going to end the autism epidemic in the NY Times just like the rest of us, I guess it was news to them too.

One of the DSM 5 autism committee members, Dr. Catherine Lord, had this to say about the release -
We’re really just hearing about this today,” says Cathy Lord, the director of the new Institute for Brain Development and a member of the DSM committee working on the diagnostic changes (Volkmar at one point was a member of the committee, but resigned). “It was not sent to the committee, it was not brought up to the committee, and it’s embargoed, so technically they’re not even supposed to be talking about it. You’re supposed to not even release the information until it comes out in the journal.
The APA also issued a press release defending the new autism criteria the same day that he story was originally published.

If I were in a cynical mood, I might have to wonder why Dr. Volkmar decided to present embargoed data to the Icelandic Medical Association and why he participated in the NY Times article.  Because it looks suspiciously like Dr. Volkmar decided to take his issues with the DSM 5 autism criteria to the court of public opinion rather than to the scientific literature were they belong.

It is one thing for non-scientific people (like me) to talk about these things on blogs and to theorize about what the new DSM 5 criteria might mean.  It is quite another when a scientific expert does the same in an apparent end run around the other members of his field.

Regardless, even though the study has not been published and the real data is not available, I think enough information was published that we can get an idea of what the final study will look like.  The following is based on what I read and my own speculation about what the data is, take it for whatever it is (or isn't) worth.

From the NY Times article, we know that the original data came from a 1993 (some sources say 1994) field trial of the DSM IV and included data on 372 children and adults who were "among the highest functioning".  The press release on Yale's web site tells us that these children and adults were selected because they did not have intellectual disabilities.  From other versions of the story (sorry, can't find the link), we know that the original trial had roughly 1,000 participants.

Based on above, I think that the original data came from here.  This was a field trial of the DSM IV criteria in 1994 that involved Dr. Volkmar and had 977 participants.

The breakdown of the participants from this paper is follows -

Disorder Number IQ > 70 Mute
Autistic Disorder 454 118 245
Other PDD 240 122 84
PDD-NOS/Atypical 163
Asperger 48
Rett 13
CDD 16
Related Disorders 83 130 93
Mental Retardation 132
Language Disorders 88
Childhood Schizophrenia 9
Other Disorders 54
Total 977 370 422

The 370 number with an IQ > 70 (no ID) very closely matches the 372 cited in the NY Times but I have to wonder if this new result includes the "Related Disorders" group.  It doesn't seem like it would make sense to do that but without adding in that group you can't get to the 372 number.  Maybe the related disorders group was excluded and the the cutoff wasn't strictly at an IQ of 70.

Going back to the current result, the researchers applied the new DSM 5 criteria to the group of 372 and found that about 45% - roughly 167 - would no longer qualify for an autism label.  The breakdown that is given is about 25% of those with "classic autism" (autistic disorder), 75% of those with Asperger, and 85% of those with PDD-NOS.

Please remember when you consider these figures that they are based exclusively on the "high-functioning" group.  So before you can get excited about 85% of children and adults with PDD-NOS not qualifying for an autism label, you have to ask yourself what part of the entire PDD-NOS group are they talking about.

There are many potential problems with this result.

First, as mentioned in the NY Times article, the study from 1994 might not have the proper data to even apply the DSM 5 criteria.  For example, the 1994 study makes no mention of any data concerning sensory issues but sensory issues are one of the criteria under the DSM 5.  Since higher functioning individuals are, in my experience at least, more likely to have sensory issues than other forms of restricted or restricted behaviors, the absence of this data could make a big difference.

Second, this data set is twenty years old and was collected at a time when autism was thought to be extremely rare.   Since that time the rate of autism has exploded from about 5 in 10,000 to 91 per 10,000.  Clearly something about how autism is diagnosed and who gets an autism label has changed in the past 20 years and those changes - whatever they are - would definitely have an impact on the current results.

If nothing else, if you wanted to do an analysis of what the changing criteria meant for people currently on the spectrum, it would have made more sense to look at the bulk of the people who are currently on the spectrum.  Even a back of the envelop calculation suggests that the number of children with autism who are currently eight years old (born in 2005) is far larger than the number of people with autism born during the entire decade of the 1980s (the population in the 1994 study).

If you wanted to do a real analysis of the impact of the criteria, you should look at the people who just received a DSM IV diagnosis and see what impact the DSM 5 would had.  Going back to the dark ages of autism and looking at an ancient, incomplete data set doesn't really tell you anything.

But the biggest problem with the story is what is doesn't cover.  It doesn't tell us anything at all about the "lower functioning" side of the spectrum.  Well, I say lower, but what I really mean is the majority of people with autism.  For all this talk about what the new DSM 5 will mean for the highest functioning, very little is being said about what the changes mean for the rest of the spectrum.

Consider again the numbers in the table above from the 1994 study.  In this study, the majority - over 60% -  had an IQ that would place them in the intellectually disabled group.  Over 40% were considered to be mute, meaning that they either had extremely limited or no verbal skills.

These numbers have not changed that much over the years.

There is still a large group of children and adults, perhaps as high a 50%, who would test as intellectually disabled.  And there is still a significant group, again as much as 50%, that have substantial problems with function speech.

The exact source and breakdown of these numbers will have to wait for another day but lets take a highly conservative estimate and assume that 40% of all people with autism have issues with either intellectual disability or functional speech.  The actual number is likely quite a bit higher because these aren't necessarily the same groups - you can have ID and have functional speech and not having functional speech doesn't mean that you will test as having ID.

Now here is the problem.

The DSM 5 has removed all of the criteria deal with lack of functional speech and other types of functional communication and replaced them with criteria that deal with social communication.  So the new criteria take  something that was a substantial problem for more than 40% of people with autism - in 1994 as well as today - and throws it out the window.

And when it comes to the intellectually disabled group it gets even worse.  While the new criteria don't come out and say it explicitly, there is language in the criteria that seems to imply that anyone who would qualify as intellectually disabled will no longer qualify for a diagnosis of autism.  So the new criteria was the potential to exclude what was originally the majority of people with autism and is still today a substantial group.

Or in short, the DSM 5 is minimizing what is an extremely disabling aspect of some people's autism and has the potential to kick out a large segment because the also are intellectually disabled.  Remember that 40% I mentioned a short while ago?  That is a conservative estimate of how many "lower functioning" people the DSM 5 has the potential to impact.

Now lets put the scary 45% percent number from the NY Times in context.  That number refers to a subset of the spectrum, somewhere between 35% to 50%, depending on the exact data from the original study that was used.  So lets split the difference and say that it refers to 42% of the spectrum.  That means that the scary 45% number really would be about 20% of the entire spectrum.

So, once again, the media is dwelling on the needs of the higher functioning minority (20%) while completely ignoring what is happening with the lower functioning majority (40%).  I would say that I am surprised but I really am not.  The media is in the business of selling stories and no one really wants to hear about children on the lower functioning side of autism.  The bottom line is that both groups have serious issues and both groups deserve attention.

As for what this new result means in the end, my guess is not much.  We will have to wait and see what the actual study says instead of relying on media accounts but so far I am not impressed.


  1. If John Robison can be diagnosed as autistic when he admits he has no disability under the current criteria, I doubt the new one will change anything. However, if it resulted in Robison (and probably some others I could think of) losing their diagnosis, then it would be providing a useful service.

  2. I think your analysis and thoughts are very even-handed and important.

    I've been following the DSM-5 stories as they emerge, and have been trying to keep a close eye on all the attacks (and much rarer, praise) of the changes.

    Much of the time it really is alarmist, uninformed, reactionary rubbish, so it's great to be read pieces like this which try to calmly figure out the real facts and issues. Thanks for taking the time to write something worth reading.