Friday, March 1, 2013

More on DSM-5: An Irrelevant and Perhaps Incoherent Defense

To supplement the previous post on DSM-5, I thank Dr. Bernard Carroll for calling attention in his comment to psychotherapist Gary Greenberg's blog:
--which in turn refers to a paper in JAMA extolling the virtues of the soon-to-be-published DSM-5 from some of its authors:

I am not acquainted with Mr. Greenberg or his blog and would not ordinarily cite his work under those circumstances, but I have for a good while followed Dr. Carroll's excellent work and so will take a chance on relying on his endorsement of the post.

As a very minor introductory point, sticklers for accuracy might have noted that in previous posts I prevaricate between calling the new manual DSM-V and DSM-5. The JAMA piece uses "DSM-5" but also includes mention of the previous editions of the manual, which were designated by Roman numerals. Thinking they planned to be consistent, I started using the designation DSM-V a while back, and it now appears I was in error. And if using "5" instead of "V" was the only quibble about the new edition, we'd be in great shape.

Mr. Greenberg focuses in on one questionable statement in the JAMA "advertorial" as he puts it.  The American Psychiatric Association folks state, "Many of the revisions in DSM-5 will help psychiatry better resemble the rest of medicine, including the use of dimensional (eg, quantitative) approaches." Two points here. First, it's not clear at all why it should be so important that "psychiatry better resemble the rest of medicine," except for psychiatry's own inferiority complex, for which there is no doubt an expensive drug that they can take. Second, and much more important, is Mr. Greenberg's main response: "...there are no dimensional measures in the DSM-5. They tried to develop and implement them, but the effort was hurried and chaotic and poorly planned and ultimately soundly rejected by the membership of the APA. The dimensional approach was going to be the signal achievement of the DSM-5 (that is, after the first signal achievement, the tying of neuroscientific findings to DSM disorders, had to be abandoned for lack of evidence); it didn’t pan out. The APA leadership is understandably reluctant to own up to this fact."

So what's new, exciting, and innovative about DSM-5 is offering quantitative, dimensional measures to diagnose mental illness, to please nonpsychiatrists by making psychiatry better resemble the rest of medicine, except for the fact that there are no quantitative, dimensional measures in DSM-5. To which a third point could have been added. If there had been quantitative measures in DSM-5, that would not have made it so much different from older versions of the manual at least as far back as DSM-III. What made DSM-III special for its day was precisely that it demanded quantitative criteria, e.g. before you could diagnose a patient with depression, you had to have so many out of the following list of criteria for at least so many weeks. So the idea of having easily agreed-upon and measured criteria that would appeal especially to non-psychiatrists is hardly all that big a departure. Admittedly, demanding that a person have (say) 6 out of 9 criteria for a given mental disorder, when each of the 9 is described qualitatively, is not as quantitative as having something like a blood pressure cuff you could wrap around the patient's head and read a number off the dial. But it's the number-off-the-dial that (Greenberg tells us) the APA looked for and failed to discover.

Greenberg focuses on the illogic of the case as stated by the DSM-5 committee, but the larger issue may be trying to change the subject. As I have followed the issue, the main complaints about DSM-5 have not been the lack of quantitative measures or making psychiatry look different from the rest of medicine. The main recurrent concerns have been 1) the serious conflicts of interest among the panelists, and 2) the creation of new categories of illness that will vastly increase the number of patients susceptible to a diagnosis. The article in JAMA does not address the first concern at all, and addresses the second only glancingly.

To illustrate how wonderful the DSM-5 is going to be, the authors give us glimpses into the new criteria for autism spectrum disorder, binge eating disorder, disruptive mood dysregulation disorder, posttraumatic stress disorder, and the removal of the bereavement exclusion for diagnosis of major depression. They state that the new criteria for autism "is not expected to significantly alter prevalence rates." Others have looked at the new criteria and disagree, I gather, so I don't know who's right. In none of the other examples did they address how many additional people might be diagnosed. The concern about conflicts of interest, of course, is that if you're in the pay of the drug companies, you naturally want to see more people being prescribed drugs, which means they should all have psychiatric diagnoses for which drugs can be recommended. After that one comment that say nothing more about how many more patients might receive the other diagnoses.

The discussion of bereavement and depression seems especially disingenuous (see my earlier post on this, The DSM-5 folks make it sound like this is first off, a very minor issue that affects only a few patients; and second, the problem for those few patients is that they may suffer terribly under present guidelines and DSM merely fixes things for them. Specifically, while grief is common and can last a long time, it is also possible for people who have just lost a loved one to have an unrelated major depression. By the old guidelines, they say, these people would have been denied proper treatment for their major depression, and with DSM-5 removing the exclusion for diagnosing depression in the face of grief, now these people can get proper treatment.

Now, if the goal of DSM-5 is to make psychiatry more like the rest of medicine and make it easier for nonpsychiatrists to handle psychiatric diagnoses, maybe somebody should point out to these folks that if a patient has, for example, congestive heart failure and kidney disease, you don't deny them treatment for the kidney disease because you are treating them for heart failure. The possibility that a person can have two diseases at the same time is one that most physicians can figure out, and they therefore know that guidelines that tell you how to diagnose any one of the diseases must be modified when a second disease may be concomitantly present. That's called clinical judgment and guidelines cannot totally replace clinical judgment. So we can be forgiven if we think that the real reason to remove the bereavement exclusion is to make a lot of people with uncomplicated grief reactions candidates for antidepressant drug therapy--and to be very skeptical that any significant number of patients who actually have major depression rather than grief are today being denied proper therapy.

Bottom line-- this paean of praise for DSM-5 in JAMA appears to reinforce rather than remove concerns about the new edition. My advice in the previous post, that other specialties should boycott this product, stands.


Bernard Carroll said...

There is a new defense of removing the bereavement exclusion from DSM-5, written by one of the panel who made the decision, Dr. Sid Zisook from UCSD. This new defense has just appeared in Scientific American – see this link

Dr. Zisook has been talking up a study of his in which he treated bereaved and depressed people with buproprion (Wellbutrin). He says it worked. This study is laughable. It was sponsored by the drug’s manufacturer, for whom Dr. Zisook also gave paid speeches. It was uncontrolled, so much of the reported improvement would be expected had there been a placebo treated comparison group. There were no self report depression measures, so we don’t know whether the patients also thought they had improved in terms of major depression. Their self reported grief scores improved, but once again we have no comparison group, and the natural course is for improvement.

But most of all, it was a tiny study in a skewed population – they enrolled fewer than 1% of the subjects they invited to participate (22 out of 3998, to be exact, or 0.55%). It is impermissible to generalize from this minuscule sample in order to make broad policy for DSM-5. Can we imagine that Kenneth Kendler, another member of the panel, would pontificate about the genetics of depression based on such a sample?

Dr. Zisook glossed over the fact that many people do indeed meet the nominal criteria for major depression, as Paula Clayton taught us decades ago, even while we acknowledge that their grieving is a normal process. In other words, there are many false positive diagnoses of major depression among the bereaved. Dr. Zisook doesn’t explain how he identifies the ones who need more than a hug.

Suzy Chapman said...

In March 2010, APA issued a news release setting out its rationale for dropping the use of Roman numerals for its forthcoming edition of DSM and for subsequent editions, copy here:

According to this 2010 news release:

"Following the publication of the DSM-5, ongoing review groups will be established to coordinate and oversee periodic assessments of advancements...Incremental updates will be identified with decimals, i.e. DSM-5.1, DSM-5.2, etc., until a new edition is required."

APA has more recently said there will be opportunities to reassess and revise DSM-5's new disorders, post publication, and that it intends to start work on a 'DSM-5.1' release.

It's unclear how these "ongoing review groups" will be assembled, whether there will be opportunities for stakeholder input or whether this will be a public process, as revisions to ICD-9-CM and ICD-10-CM have been.

On February 27, Susan Donaldson James reported for ABC News on concerns for the implications for all illness groups of the controversial new DSM-5 category 'Somatic Symptom Disorder,' that subsumes and replaces four of DSM-IV's 'Somatoform Disorders':

Report here:

Joel E Dimsdale, MD, who chairs the 'Somatic Symptom Disorder' work group, told ABC News:

"...People talk a lot about the DSM being the Bible of psychiatry. I see don't see it as that at all. I see it as a useful working guide to help doctors diagnose and treat patients. If it doesn't work, we'll fix it in the DSM-5.1 or DSM-6."

Patient groups, advocates and professionals are not reassured by APA's 'publish first – patch later' approach to science.

Suzy Chapman said...

In March 2010, DSM-5 Task Force Vice-Chair, Darrel Regier, attended a public meeting of the ICD-9-CM Coordination and Maintenance Committee, transcript here:

Dr Regier had attended the meeting to lobby on behalf of APA for postponement of the proposed partial code freeze for ICD-9-CM and ICD-10-CM, in order to allow APA more time in which to catch up on its DSM-5 development process, by this time slipping its targets.

Extract from official transcript:

"Darrel Regier: Good morning. I’m Darrel Regier from the American Psychiatric Association. And we are in the middle of a major revision of the diagnostic and statistical manual of mental disorders.

"We have just released draft criteria on a website on February 10th at And we’ll be having a field trial starting in July of this year. We’ll then have another revision based on field trial results going into a second revision or second field trial in July of 2011.

[SC: Note, the second wave of field trials never took place.]

"As a result, we will not have our final recommendations for the DSM-V probably until early 2011. So our clear recommendation would be to have the final freeze of the major classification for mental disorders, the chapter 5 in this case for October 1st, 2012.

"The importance of this for us is that we had a complete conversion table (inaudible) as you will for DSM-IV and ICD-10 that was prepared back in about 1995 or so. So it’s been sitting, waiting, ready to go for ICD-10 for quite some time. Our expectation is that we will be working with the central office at WHO on the mental health division throughout this time. They, of course, are working on ICD-11, which they hope to implement or approve in 2014.

[SC: ICD-11 currently scheduled to complete in 2015/16.]

"Our expectation though is that ICD-10-CM will be the procedure – will be the diagnostic code for this country probably for the next 20 years. Maybe not as long as 36 years as ICD-9-CM has been, but our plan is to really have concordance between the proposed ICD-11 major categories and disease names in agreement and harmonization with the DSM-V by about October of 2012.

"So, that’s what we’re working toward. From the ICD standpoint, this would give them really a wonderful field trial for their ICD-11 if we introduced the mental health codes into the ICD-10-CM that essentially will be going into ICD-11...

"...What would be remarkably helpful is if we could basically hold on the firm freeze of the ICD-10-CM so that we could have this synchronization with DSM-V and then we would have a system that would be supportive of mental health diagnosis coding for probably a couple of decades."

In the event, APA were unsuccessful in their appeal and CMS implemented the partial code freeze on ICD-10-CM, as planned, with no concessions towards APA's slipping timelines.

So, "retrofit" DSM-5 codes into ICD-10-CM with the suggestion that these could be used as an ICD-11 "field trial" and if a disorder "doesn't work," release DSM-5.1, DSM-5.2, DSM-5.3 patches.

Is this science or is this Windows 8?