May 22, 2008

What's Wrong With Research In Psychiatry?

Apart from the high fives, bravado, and binge alcoholism.


There are no independent psychiatric researchers.  Young academics are mentored by older academics; this isn't optional, for either person.   In virtually no circumstance do they study something entirely of their own choosing, it is either an outgrowth of the mentors' research, or is the mentors' research.

Distant from reality:

Young academics almost never work outside of the university.  Theirs is all selection bias.  The only patients they see are the ones the university gives them: either Medicaid/indigent on the inpatient unit, or patients of the disposition to want to be involved in clinical trials.  Academics are like government economists: "we haven't had two consecutive quarters of declining GDP, so we're not in a recession."  Regular psychiatrists are the management at Wal Mart: "I'm not sure what this is called, but no one is buying anything."


Academic psychiatrists are nearly all on the same page, and refer to one another as if they have a relationship, even when they've never met.  ("Chuck Nemeroff is doing some good work on...")  It's pointless to list the other characteristics of groupthink here, except to highlight one: the purpose of groupthink is not to promote an ideology, but self-preservation, and this is unconscious.  They don't realize that their lives are devoted to preserving the group, yet young researchers are brought on who connect with the group; peer reviewers-- and journal editors-- come from the group; grant reviewers, and NIMH people themselves came from, and support the group.

An example of groupthink preservation is the referencing of studies.  Academics support their propositions with previous studies; however, no one checks the accuracy of these studies.  No one has the time, and the group necessarily must trust the work of others in the group.  Even if an error were to be found, it would be described as an isolated error.  A cursory stroll through this site alone suggests just how "isolated" such errors really are. 

Financially isolated:

Medicine is a market.  Buyers and sellers.

Academics make a salary, but their survival at university depends on the grants they can bring in.  That means their market, their customers, are funding agencies, not patients.  It doesn't mean they don't care about patients, it means that the service they provide is  nuanced and directed towards Pharma or the NIH. 

If the funding agencies are stacked with people who like antiepileptics for bipolar; if the grant goes to Pfizer who is looking to create a bipolar indication, etc, etc, that's the research that can be expected.  I'm not even worried that the results will be... predestined.  I'm worried that such pressures direct what kind of research, what kind of questions get asked, at all.

Too much data

We're busy talking about bias and hidden results and skewed statistics and nonsense.  So we call for more studies, as if they will somehow be better studies, despite no other structural changes being made.  The reality is that we have information inflation: new studies have less value because they get lost; and old studies completely disappear, as if somehow their validity is temporal.

There are a quadrillion studies already conducted in psychiatry. There is plenty of data that can be analyzed, meta-analyzed, pooled, parsed.  If all current research ground to an immediate halt, and researchers just looked back at what we already have, we would save billions of dollars in future research and future bad treatments, and we would learn so much.

Outcomes Research Is Purposefully Avoided, or Ignored:

You might think in a field with nothing but outcome studies (e.g. Prozac vs. placebo) I might not be able to make this claim, but I do.

Most studies are short term.  The few long term studies that exist (e.g. Depakote for maintenance) are either equivocal (e.g. Depakote for maintenance) or show no efficacy (e.g. Depakote for maintenance.)  And they are ignored.

But these outcomes are distractions.  The question isn't is Depakote good for maintenance bipolar.  The question is, is there any value to the diagnosis of bipolar?  In other words, if you called it anxiety, or personality disorder, or anything else, and then treated them ad lib, would the outcome be different?  Is there value to the DSM?  You might argue the diagnosis leads us to the treatment, but in most cases, meds are used across all diagnoses, and more often than not a diagnosis is created to justify the medication.

Are hospitals valuable?  You would think that by now we'd have a clear answer to this, the most expensive of maneuvers.   I can say, however, that reducing the length of stay from several months to 5-7 days has not affected the suicide rate.  I'm not saying they are or are not valuable, I am saying that I don't know-- and that's the problem.  It is 2008 and there are more studies on restless leg syndrome then there are on hospital vs. placebo.  You know why?  See above.

Are one hour sessions associated with better outcomes than 2 minute med checks?  I know 2 minute med checks sound bad, what I want to know is if they are actually bad.  Higher suicide rates?  More days absent from work?  More divorce?  More sadness?

The system is completely ad hoc, with each party yelling loudly to protect their fiefdom. It allows everyone to declare themselves an expert without having to prove it. Tell a Depakote academic you're suspicious about the utility of the drug, and he won't tell you you're wrong, he will tell you you don't understand.  Try it.  He will evade the existing data ("not enough people," "studies are difficult to conduct," "we know from clinical experience," "more work is needed") and rely on appeal to authority.  Appeal to authority is the signal you're being bullshitted.

Outcomes research will never be conducted in psychiatry because its existence depends on not knowing the answers.  It will eventually be conducted on psychiatry.  You can't tell you're an idiot, someone has to tell you.