by Cooper RStudies in History and Philosophy of Biological and Biomedical Sciences. 2015 51:1-10.The latest edition of the Diagnostic and Statistical Manual of Mental Disorders, the DSM-5, was published in May 2013. In the lead up to publication, radical changes to the classification were anticipated; there was widespread dissatisfaction with the previous edition and it was accepted that a "paradigm shift" might be required. In the end, however, and despite huge efforts at revision, the published DSM-5 differs far less than originally envisaged from its predecessor. This paper considers why it is that revising the DSM has become so difficult. The DSM is such an important classification that this question is worth asking in its own right. The case of the DSM can also serve as a study for considering stasis in classification more broadly; why and how can classifications become resistant to change? I suggest that classifications like the DSM can be thought of as forming part of the infrastructure of science, and have much in common with material infrastructure. In particular, as with material technologies, it is possible for "path dependent" development to cause a sub-optimal classification to become "locked in" and hard to replace.
"The DSM provides a common language for mental health research, policy and care. Almost all papers published in psychiatric journals refer to the DSM; the use of DSM categories to select subject populations for study is near universal. Worldwide, textbooks for mental health professionals, and treatment guidelines, tend to be structured around DSM categories. In the US, the DSM also plays an important bureaucratic and economic role; in particular, the DSM contains the codes that insurers commonly require before paying for mental health treatment."
It is unlikely that many of us have subscriptions to Studies in History and Philosophy of Biological and Biomedical Sciences, or will run across it in a waiting room somewhere, but I think it adds a perspective to the endless deliberations about the DSMs that would be helpful to anyone interested in the topic. I hope that it will find its way to the Internet in full at some point. It’s one of the few things I’ve run across that actually made me feel some sympathy for the dilemmas faced by the Task Force that was asked to revise it. No, I haven’t gone over to the dark side. It didn’t make me feel that sympathetic. But it did help me at least understand how the DSM has developed such a curious inertia, keeping it from iterating into something more useful…
Without spending $41.95 for the article, there are plenty of reasons for “stasis”. I recently posted on delirium and how the classification and codes proceeded to DSM-III in 1980, and it has been the same ever since [Delirium Reinvented]. The other 2 relevant dimensions is that the DSM will soon be a guidebook to the ICD-whatever simply because the ICD classification has minimal diagnostic descriptors and most clinicians use only a handful of diagnostic codes. In a large Danish study of 1.2 million diagnoses, 75% of all diagnoses consisted of 49 codes.
The logical reason for stasis with a delirium code is that the optimal verbal description has been reached although it may not be far from the optimal neuroscience description, since it is a metabolic final common pathway for several hundred conditions.
link to George’s post added by 1bom
The American Psychiatric Association made a Faustian bargain with DSM-III through DSM-5, opting for a mostly illusory reliability at the expense of validity. Then the field more widely bought in to the orthodoxy of the DSMs for a host of pragmatic reasons that displaced the primary scientific thrust of the classifications. These included turf wars, business models, and access to research funding. The locked-in result was what I have called methodologic imperialism. In my view, the greatest harm has been the sabotage of effective research progress because of rigid adherence to flawed DSM categories as the independent variables. We should not expect to move the ball down the field until that is fixed – and RDoc won’t be the answer.
Dr. Rachel Cooper provides a preprint of her article here.
i understand everyone’s arguments about why the DSM sucks.
but the bigger issue of ‘why’ is not just because of codes or whatever. it is the lack of definitive medical studies linking brain function with physical location. this is obviously not trivial, but such studies are necessary in order to add life (pictures) to manual.
fMRI data has the promise of being able to provide those pictures, but the ‘raw’ signal is still noisy. the electrical activity causes changes in the magnetic susceptibility (paramagnetic deoxyhemoglobin) at that respective spatial location.
any analyses proposed for fMRI for dichotomizing disease must be homogenous (i.e. no parameter-fiddling or anything. same methods and parameters, the only thing that changes is the subject whose data is input) for it to be admissible in the DSM. it must also implicitly deal with the problem mentioned in the previous paragraph.
furthermore, any such analyses procuring brain-behaviour linkages must generalize across subject groups. i.e. same experiment (say, the sternberg item recognition paradigm [SIRP] http://www.psych.upenn.edu/~saul/hss.pdf) on matched suspected-psychosis subjects and healthy controls. such an experiment could be tested across the world.
in the experiment being administered, we have a priori beliefs about what is being tested (hence the experiment). using SIRP seems popular because it tests the information processing, which is suspected to be impaired in schizophrenics (even if their performance is similar to controls). thus, fixing the administered stimuli and WHEN they’re administered SHOULD allow detection of the deficiency in *each* subject’s scan (move to the respective ‘time point’ where the stimulus was administered).
note: i’m not saying schizophrenia has a trivial etiology. as dr friston’s work in the early 90s showed (with PET), schizophrenia can be further refined. the diversity of the observed behaviours should be reflected in the etiologies (revealed by the fMRI scan while they’re performing the mental task).
it should be reasonable to assume that giving schizophrenics a task (like SIRP), should reveal similar etiologies when they’re administered the identical stimulus in a homogenized environment.
i believe this is how we would have to define the mental disorder (dichotomizing mental disease based on responses to a ‘core’ set of stimuli).
assuming existence of such ‘trigger stimuli’ (that detect the disorder, and thus enable dichotomization) the manual could then say, ‘during information processing tasks, can see differences in medial temporal lobe’ and show a picture (see ingvar franzen: https://www.dropbox.com/s/5ldmpp9jjhx5k79/ingvarfranzen.pdf?dl=0).
i see the DSM as the ‘bible’ that collects the results and maybe adds some consensus commentary.
we have a long way to go. i think it’s possible given the commoditization of MRI. this technology is a culmination of some of the greatest minds in science over the past few centuries.
it’s about time for us to step into the next century as well.
we need to start cataloging. i cannot stress this enough. nature is so diverse we have to start SOON.
sorry for this long post but i just figured that i wanted to share all manuals suffer from the same problem, and i believe the way forward is something like the above. we need to have confidence that someone with schizophrenia is going to have similar etiology to others when administered a stimulus (where the latter may take some theory/testing to discover!).
sorry for the rant. i didn’t want to write it but i figured this would be a good time.