Friday, 10 August 2012

DSM, diagnose thyself

There’s an eye-popping synapse-tingling connectome-reconfiguring passage in The Psychopath Test, the new book from Jon Ronson (the man who wrote The Men Who Stare at Goats), when he interviews Robert Spitzer, the architect of the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM III), and  Allen Frances, chief editor of DSM IV.

DSM-III, Ronson explains, was in part a reaction to the audacious Rosenhan experiment  when eight healthy American volunteers  presented themselves to psychiatrists throughout the United States. The volunteers gave false names and jobs, but otherwise followed instructions to answer every question the psychiatrists posed truthfully except in one respect: they were to report that they could hear an indistinct disembodied voice saying something that sounded like “hollow” or “empty” or “thud”. All eight were admitted to psychiatric wards where they told staff that the one abnormal symptom, the hallucinated voice, had stopped. All eight were diagnosed insane, given powerful drugs,  and released 7 to 52 days later, and only when they admitted insanity.   Reassuringly, when the findings were published in the journal Science,  the American psychiatric profession experienced a collective breakdown and was sucked into a vortex of self-doubt. Spitzer, a progressive psychiatrist  who had argued successfully for the removal of homosexuality as a mental disorder, began the healing process by reforming  the DSM, replacing old discredited clinical judgement with new rigorous repeatable road-tested diagnostic checklists. The enterprise seems to have come naturally - perhaps too naturally - to Spitzer who classes  himself as one of nature’s categorizers. Unfortunately, by all accounts the undertaking didn’t quite reach the heights of clinical excellence a layperson might naively hope for: “the psychiatrists Spitzer invited [to DSM-III editorial meetings] would yell over each other. The person with the loudest voice tended to get taken the most seriously.... 'Of course we didn’t take minutes’ says Spitzer, 'We barely had a typewriter'”.

Allen Frances, Spitzer’s successor, is delightfully candid about the results: “It’s very easy to set off a false epidemic in psychiatry and we inadvertently contributed to three that are ongoing now: autism, attention deficit and childhood bipolar. With autism it was mostly adding Asperger’s...the rates of autistic disorder in children went from less than one in two thousand to more than one in one hundred. Many kids who would have been called eccentric , different were suddenly labelled autistic”.
In a recent op-ed piece in the New York Times about the forthcoming DSM 5 his criticism is even sharper: “ DSM 5 promises to be a disaster... it will introduce many new and unproven diagnoses that will medicalize normality and result in a glut of unnecessary and harmful drug prescription.”

Can we, should we do better than the DSM?  Part of the problem with the DSM’s diagnostic checklists is surely, as Allen Frances suggests, that unusual behaviours, which are in the tails of the bell-curve of normal behaviour, are labelled as disorders. When the behaviour to which the label is attached (idiosyncratic language, preoccupations unusual in intensity or focus, whatever) is not itself a cause of suffering, this seems hugely wrong-headed. Another problem is that diverse conditions that may have nothing in common biologically and which have quite distinct  prognoses, and possibly respond very differently to interventions, are being lumped together as the same condition. What’s more, completely different diagnostic criteria may have overlapping symptoms, calling into question the value of the distinct diagnoses.

There’s an intriguing article in Neuron  by Joshua Buckholtz and Andreas Meyer-Lindenberg that reckons that by using some deliciously clever ideas from network theory we might be able to do better than current practice.  The paper argues that, as much as it must irk the natural born categorizer,  many mental disorders probably don’t divide neatly into discrete types in the way that  butterflies and beetles do. What’s more, mistakenly grouping them together could be actively harmful: “Many in the field believe that the notion of discrete, categorical mental disorders, originally articulated by the Research Diagnostic Criteria and reified in the DSM-III and DSM-IV, is so far removed from biological reality that it actually impedes clinically useful scientific discovery. ...Though originally intended to be ‘merely’ reliable operationalizations of clinical phenomena, over time, these categorical classifications came to be treated as though they were natural kinds — inherently meaningful, ontologically (i.e., biologically) valid taxons. This has produced the assumption that each DSM-defined disorder is ‘‘real’’—a distinct, independent entity with a unique set of causal factors and pathophysiological processes.”

The authors argue that this view that all mental disorders fall into distinct biologically meaningfully categories  is not supported by evidence: “comorbidity between mental disorders is the rule rather than the exception...covariation among psychiatric diagnoses is so prevalent, and so extensive, that it alone belies the artificial nature of phenomenologically based categorical classification.”

Part of the solution, they argue, may be to reverse engineer the wiring map (or connectome) of the brain, and look for anomalies, similar patterns of dodgy wiring that could underly may different mental disorders. This is possible, it turns out, by photographing blood oxygen levels in different parts of the brain over time in an MRI scanner, and analysing which bits light up and in what order. With this information and some heavy-duty computational statistics, it’s possible to deduce something about the strength of the connections between different brain regions.  Two strongly connected brain areas, for example, might be expected to have a tendency to light up during the same tasks, one just a fraction of a second after the other. Such approaches have led to the identification of a number of brain circuits and have also shown that precise properties of the brain circuits or networks vary between people. These variations can, to some extent, be inherited, and seem to be related to the risk of some mental disorders.

Network view of mental disorder: disruption to one brain 
circuit can affect multiple cognitive processes leading to a 
variety symptoms. While some symptoms are  specific to 
certain diagnostic categories (disorders A & B) others might 
be shared across different diagnoses. 
Figure adapted from Buckholtz & Meyer-Lindenberg, Neuron 2012.
This is interesting for me professionally as it draws connections with a growing literature in network theory (which is also important in epidemiology, my own field of research), and could open gates to interesting avenues for research (gates that may have been held shut by the epistemic barriers of the modern DSM system). It is interesting for me  personally too as the father of a child diagnosed with autism, an area where  the DSM  seems particularly baffling. For example, what sense does it make to group  under the same diagnostic heading kids who have very strong  linguistic skills but little interest in engaging in social interaction, with other kids who are very interested in social interaction but who lack the linguistic skills to do so? It seems about as sensible as grouping double foot amputees with the blind drunk on the basis that both exhibit ambulatory impairments. To my knowledge, in the case of the so-called autistic spectrum disorders (so-called as there seems to be no evidence of a one-dimensional continuum that "spectrum" implies)  there are no statistical or mechanistic reasons for suspecting common aetiology or common prognoses (if there are, I would love to hear them). It seems more plausible to me that the bizarre grouping of such heterogeneous conditions is a chance upshot of the chaotic DSM III meetings, rather than the result of rational deliberation. What’s more, I don’t want to read about what interventions are helpful for kids who have received a label that reflects such a broad and heterogeneous condition. I want to read about what might help for children like my son. It’s hard to see how DSM can be a helpful tool for this, and the candour of Frances Allen in drawing attention to its deficiencies is admirable. While a mechanistic understanding of mental disorder has to be worth striving for, and the network ideas outlined in the Neuron paper are intriguing, it’s probably going to be a long time before such approaches gain a foothold in routine clinical use. It seems strange to me that in the short term the field doesn’t make more use of statistical approaches to identify commonalities among conditions (this type of thing) rather then rely on old discredited clinical judgement to define the disorders. Meanwhile  the tendency to label disorders as a substitute for understanding, and to mistake a label as understanding, is going to be  a hard habit to break. This is a shame as the stigma of the diagnosis of mental illness is  probably as prevalent and lamentable in most parts of the world now as it was in 1973 when Rosenhan was writing.  His article - lucid, humane,  and trenchantly honest - is worth revisiting:

“we  tend  to  invent  "knowledge"  and  assume  that  we  understand  more  than  we  actually  do.  We  seem  unable to  acknowledge  that  we  simply  don't  know.  The  needs  for  diagnosis  and  remediation  of  behavioral  and  emotional  problems  are  enormous.  But  rather  than  acknowledge  that  we  are just  embarking  on  understanding,  we continue  to  label  patients  "schizophrenic,"  "manic-depressive,"  and  "insane,"  as  if  in  those  words  we  had captured  the  essence  of  understanding. The  facts  of  the  matter  are  that  we  have  known  for  a  long time  that  diagnoses  are  often  not  useful  or  reliable, but  we  have  nevertheless  continued  to  use  them.”


[1] Ronson J (2011). The Psychopath Test.
[2] Rosenhan DL (1973). On being sane in insane places. Science 179; 250-8.
[3] Frances A (2012). Diagnosing the D.S.M. N Y Times, May 12, A19.
[4] Buckholtz JW, Meyer-Lindenberg A (2012). Psychopathology and the human connectome: toward a transdiagnostic model of risk for mental illness. Neuron 74; 990-1004.

No comments:

Post a Comment