June 17, 2010

A Diagnosis Of Schizophrenia

map of hobbit.jpg
should be accurate to 10 dolichoi

When I read psychiatric articles, I wonder if the authors think that because the stars in the sky are small, a microscope would be the best tool.

Nassir Ghaemi and Frederick Goodwin make the case that diagnostic divisions (between bipolar, schizoaffective, and schizophrenia, in this case) are still very important.

They write,

[The] overly broad use of the term schizoaffective was illustrated in a patient referred to one of us as schizoaffective. However, his manic and depressive episodes, both of which included periods of florid psychosis, responded prohylactically to a combination of lithium and valproate... When the lithium dose fell below 75 mg/day, psychotic symptoms recurred;  when lithium was reestablished at only 75 mg, the psychotic symptoms disappeared. Clearly, this patient is not schizoaffective but rather has severe psychotic symptoms integral to bipolar illness.

Slow down. 


A 22 year old Omani male with no prior psych history, but with a family history of psychosis, presents with change in personality.  He says that he hears "spirits" that tell him other people because they have done bad things.  He says his food has been poisoned by spirits.  His father has been possessed by the devil.  He is not interested in doing anything, and does not want to do anything but sit in his room. 


within traditional Omani society, abrupt personality changes or altered states of consciousness are commonly attributed to spirit possession.  The belief in possession is embedded in social- cultural teaching, in which invisible spirits are deemed to inhabit the earth and influence humans by appearing in the form of an anthropomorphic being.
Father confirms this is part of their belief system.

He is given Lamictal 100/d and Risperdal 2/d, and is cured.


If you're confident he doesn't have a demon, applaud yourself for your advanced scientific insights.

One of the main tenets of a biologically based system of psychiatry is that an Omani man with schizophrenia may have different cultural manifestations of the disease, but it's schizophrenia nonetheless.  This looks like schizophrenia, not bipolar disorder.  The Lamictal was probably useless.

However, the neurologists found that even though he had no seizure activity on EEG, on SPECT scan he had low perfusion in the left temporal lobe while psychotic, and an improvement in perfusion when he recovered.

From a biomedical perspective, the condition of the current patient would suggest symptoms of chronic schizophrenia, a diagnosis that is supported by a family history of psychosis. In the parlance of modern psychiatry, the patient met criteria for schizophrenia and responded to risperidone, a known treatment for psychosis. [However] the possibility remains that lamotrigine may have ameliorated the patient's psychotic symptoms by controlling 'non-convulsive seizures'.

If you think he was merely misdiagnosed, you have missed the point.   There is nothing short of a SPECT scan that would have made this diagnosis.  Prior to these tests, he had schizophrenia-- not appeared, but had it-- formally, according to phenomenology, course of symptoms, family history, and degree of impairment.

No amount of evidenced based medicine will protect you from this.  A thorough SCID, family history, and full panel of labs would diagnose him with schizophrenia.  Under that diagnosis-- which he reliably has (several people using the same diagnostic techniques diagnose him the same)  there would be no "evidence" for the use of Lamictal. 

Worse, under Ghaemi's plan, improvement on Lamictal would be suggestive of bipolar disorder.  Which he didn't have at all, i.e. not based on either biology or phenomenology.


An example.  You don't need to know anything about the internal anatomy or biochemistry of a rhinoceros to know one when he gouges you in the face.  Purely on phenomenology alone, you get it right 100% of the time.   This is important: 100% of the time.  It could be bigger, smaller, whiter, blacker-- none of this will confuse you.  Ever.  In the same way, there's no theoretical reason we would need to know the biology of schizophrenia to diagnose it accurately. 

However, the simple reason we get the rhino right is because we've all seen a picture of a real rhino. No one tries to identify a rhino based on compiling a list of shared characteristics of rhino-ness. We do it with a split second comparison to an already agreed upon sign.

rhino.JPGi don't like it when people try to impose labels on me  

Even the word "rhinoceros" is a sign.  You don't phonics out the letters "r-h-i-n-o-c-e-r-o-s" in order to know what it says.  The word is actually an image.

Schizophrenia isn't like that at all.  There's no "ideal" schizophrenic to match it to.  Even when you think you're sure, it could turn out to be... non-convulsive seizures.  Or bipolar.  Importantly, this isn't a case of "schizophrenia is wrong; he actually has non-convulsive seizures."  By the most rigid definitions, he has both disorders, in the same way a kid with pesticide toxicity also has ADHD.  Which, of course, is nonsense.

So how would you know?  You wouldn't, so what Ghaemi proposes is to take medication response as informative.  That's even less reliable. 

If a depressed guy responded to Seroquel in 1999, did he thus have bipolar?  The answer is thus yes, but of course not.


You think you can explain why bipolar disorder isn't schizophrenia? 

To show how hard this really is, try it with animals:  explain why a unicorn is not a rhinoceros.

Your immediate reflex will be to call up some other agreed upon sign, and compare it to that: "A unicorn is like a horse..."

Ok, horse-- but is it a type of horse?  Is it more of a horse than a rhino?  If it is-- if unicornness is closer to horseness than rhinoness, then why did Marco Polo think a rhino was a unicorn but the horse he rode on wasn't? 

What's a triceratops, then?

These difficulties exist with animals that everyone knows on sight.  Now, imagine trying to identify an animal without some common ideal type, just based on the reaction of the animal to something.


We do that already, and we do it badly.  If I tell you "scaly cold blooded quadruped climbing a tree"  you'll at best say,  "umm, it's a lizard."  But if I tell you that the animal in question changes colors-- boom, "it's a chameleon!"

And now you are able to make numerous predictions about it, without ever seeing it: (e.g. changes color for camouflage.)

But it could be a gecko, right?  So at what level of taxonomy did you make the error?

  • Kingdom: Animal
  • Phylum: Chordata (vertebrate)
  • Class: Reptilia
  • Order: Squamata (lizards, snakes, worm-lizards; not crocodiles (crocodilia) or  turtles (testudines)
  • Family: ?
Forget species, forget genus.  You're stumped at the family level.  That's how wrong you are.   Gekkonidae vs.  Chamaeleonidae.  In order to do better than this, without resorting to an ideal image of the animal, you need not just more information, but exponentially more information. 

That's reflected in their common names: "gecko" and "chameleon" are derivations of the family name.  This is for animals which exist, that everyone "knows when you see it."

Psychiatric diagnoses suffer from the same exponential information function.  When you call someone a schizophrenic, all you're sure about is that he's either a snake or some kind of  lizard, but not a crocodile.



komodo dragon.jpg
don't lump me in with these freaks


It seems like lunacy for someone to criticize evidence collection as a basis for an empirical science, but here we go.

If you gave Aristotle ten thousand unplugged computers of different makes and models, no matter how systematically he analyzed them he'd not only be wrong, he'd be misleadingly wrong.  He would find that they were related by shape-- rectangles/squares; by color-- black, white, or tan.   Size/weight; material.

Aristotle was smart, but there is nothing he could ever learn about computers from his investigations.  His science is all wrong for what he was doing.   But Aristotle would think he knew a terrible amount about computers from his studies.  In fact, he'd probably be considered an expert.  "To fix this computer, we need to make it more rectangular.  Get chopping, malaka."

That's where we are now.   But  modern science is so "advanced,"  surely it can come closer to the truth?  No.  But surely the amount of data we have on psychiatric diseases must amount to something?   No.

The last time they did this kind of taxonomy, they built a brontosaurus and told us it was real.   

And they, at least, had real bones to work with.


I understand the temptation to refine a paradigm that's worked ok so far.

And I understand that there's a feeling that we're on the right track.  "No matter what you say, I know a schizophrenic when I see one."

I know you do, that's not my argument.  My argument is that what you think you know is lessening your knowledge, not increasing it.  When you say he has schizophrenia, you may know what you mean by that, but I don't know if it isn't a seizure.  

You have it backwards.  You think saying "schizophrenia" is some kind of detection, a whittling down of possibilities, informative.  Similarly: "I've screened him for ADHD, I think he has it."  But those diagnoses don't exclude any other possibilities at all.  Do they mean he doesn't have a seizure, depression, pesticide poisoning?  Your ghost term doesn't exclude any real things. 

You think you're telling me "he has a cluster of symptoms and behaviors that generally resemble X."   But every time we make a diagnosis, the world pauses: oh, so that's what's wrong with him.

How long would the Omani man have had "treatment resistant schizophrenia" without an uncalled for Lamictal prescription or a waste of money SPECT scan?  Forever.  "We just need to improve our diagnostic skills, tests."  You're not listening: the diagnosis of schizophrenia was the specific reason he wasn't given a SPECT scan.

"We need to improve the accuracy of the diagnosis."  No.  The diagnosis isn't going to get more precise the more we know; the diagnosis is going to disappear, replaced by thirty other more specific diagnoses.

Sure, in the meantime I'm happy to go with "schizophrenia" and chuck dopamine blockers at everyone, because that's the best we can do today.   But why the pretense that we need to refine the diagnosis, or the whole DSM, when we don't have the tools to do it?  Shifting around consonants and vowels, substituting one fairy tale for another doesn't make it more accurate.  Schizophrenia vs. bipolar isn't a distinction, it's a distraction.  "I don't think he has a demon, I think he has too much black bile."  Then it's settled.  Send the priest home and bring out the leeches.