Psychiatry is politics, it is politics in the way that running for office is politics.  It is not a science, it is not even close to science, it is much closer to politics.

A doctor makes a diagnosis of a patient and writes it down on the chart.  If it were science, then I should be able to evaluate the patient myself and come up with the same diagnosis.  If it is a science but not an exact science, I should be able to come up with the same diagnosis most of the time, and the other times where I disagree I should be able to see why the other person thought what he thought.

But if I can guess the diagnosis without actually seeing the patient at all—but by knowing the doctor—then we do not have science, we have politics.

If you are watching the TV news with the sound turned down, and a Republican senator is talking, and the caption reads, “Tax Breaks for the Rich?” you can guess  his position.   In fact, the actual issue doesn’t matter—what matters is his party affiliation.  Everything follows from there.  Not always, certainly, but enough times that you don’t bother to turn the sound back up on the TV.

Psychiatry is the same way.  It is very easy to determine who is considered a “great” psychiatrist, or a “thought leader in psychiatry” based on who is making the evaluation, and not on any merits of the psychiatrist himself.  Down one hallway Freud is lauded; down the other he is villified; Kay Redfield Jameson is the hero.  But their value, of course, is not at all dependent on what they did—it is dependent on who you are.  Ronald Reagan was either a god or a devil depending on who you are, not who he was.  It doesn’t seem to matter that most people can’t  name one specific thing he did in office, what wars and battles he presided over, what he did or did not do to taxes.  Ronald Reagan isn’t a person, he is a sign.


It’s even possible for me to guess the medications a patient is taking based solely on who prescribed them, and not on the symptoms of the patient.  Importantly, the possible medications vary widely from doctor to doctor; it is wrong to think my predictive accuracy is based on any fundamental logic or science to medication selection that should be true across all psychiatrists.  It's just his regular, unthinking, habit.  "I like Risperdal."  Are you an idiot?  Are there internists  saying, "I like insulin?"

Let me be clear: I’m not talking about doctors having unique insights into which medication might benefit a certain patient.  (“I think Geodon could work really well here.”)  I’m talking about each doctor having a set of drugs he prescribes with such regularity that I can guess them.  

---

It stems from a lack of appreciation that mental illness is not a genetic disease, or even primarily a biological one, or even, surprisingly, a psychological one.  It is a social disruption.  On a desert island, no one can tell you are insane.

The key evidence against my position is that biology is so obviously relevant. There is a hereditary component to many mental illnesses; twins raised apart still often have higher concordance rates than non-twins.  But this misses the point of the problem entirely.  Consider diabetes: it is obviously a biological disease, with a heritable component.  Much more biological than any mental disorder, because you can point to the dysfunctional biology in diabetes, but you can’t do that in bipolar disorder.  But despite this biology, the environment is so massively important as to often overwhelm this biological component.

We can consider even further the actual relevance of genetics.  Things that we assume are simple genetic outcomes are often more complicated than they seem.  Eye color is every 7th grader’s primer for Mendelian genetics.  But—surprise—there is no gene for eye color.  There are in fact three genes for eye color, and the color is determined by the interplay of all three.  So while you can guess eye color based on the parents, you are not always right—because each parent is giving three different genes.

It may be, in fact, true, that bipolar disorder is genetic.  Perhaps overwhelmingly genetic, let’s say 40%.  We go wrong because we consider genetics a “fixed variable”—we think we can only affect the other 60% of the factors.  Right?  Wrong; genetics is not fixed.  Having a gene may be a fixed, but whether you express this gene or not is most certainly under outside control.  Consider gender; absolutely genetic, correct?  Not much one can do about it?  But lizards can alter the sex of the progeny by changing the incubation temperature of the egg.  Think about this.  Now, is it not probable that the expression of the genes for bipolar have a lot to do with how you are raised?  And we already know that environment affects gene expression, so I’m not speculating here.

I was emailed a link to a 2003 article in the Psychiatric Times, which describes a maddening report out of California is so blatantly politicized that Arnold himself is embarrassed.

The report says, insanely, that there are not, and will not be in the future, enough psychiatrists to meet the needs of California.   (Actual report PDF here.)

Well, not exactly true, is it?

When you say shortage, what do you mean-- 5000 psychiatrists for one state isn't enough?  Oh, you mean that for some inexplicable reason, 63% of the entire state of California's psychiatrists work in the Bay Area or LA?  Sounds like you have plenty of shrinks, they're just not distributed very evenly.  Why would that be?

48% of all psychiatrists in California are in a solo or 2 physician practice.  Hmm. 75% were male, 65% white.  Hmm.  Perhaps the problem is that your solo psychiatrists want to work in a nice area with good pay, and not in an inner city where-- ironically or tragically, your choice-- the need is greatest but the pay is least?

The nuts filing the report continue to lament that there aren't enough child and geriatric psychiatrists.  Enough for what?  Oh-- enough for Medicaid and Medicare.  What did you expect?  After suffering through a Child psych fellowship, why would go work for peanuts in a community mental health clinic, where you have a better chance of getting stabbed than getting rich?

Their complaints are misplaced and deluded.  They do not reflect reality.  Let me give you reality: the shortage exists in community (read: Medicaid) mental health, primarily because the pay sucks.  But even there, the problem is not as dire as they make it sound.  

First, even if there are numerically more psychiatrists seeing private patients, the community mental health psychiatrists see many, many more patients in a day.  I'm going to guess the ratio is five to one. (Oh, you're upset they see them in ten minute intervals? When you give them a case load of 3000, what did you expect them to do?  Psychoanalysis?)

Second, psychiatrists aren't the only ones providing "community mental health."  Advance practice nurses (APN) and nurse practitioners (NP) also prescribe medications; in some states physician assistants can prescibe; and very soon psychologists will be able to prescribe, as they already can in New Mexico (and I think Louisiana.).  (Care to retract your asinine prophecy, "the center predicts that there may actually be too many psychologists in the future.")

Third, primary care docs handle far more psychiatry than we can imagine. They just can't bill for it.  (And so how good a job are they incentivized to do?)

The shortage is for "psychiatrists" proper (i.e. MD/DOs), not "providers of psychiatric medications." 

The question then, uniquely, is whether we need psychiatrists proper at all to do community mental health.  Are community mental health psychiatrists, as a group, better at diagnosing and treating than anyone else, for example an NP?  Sadly, the answer is currently undeniably no. No one reads anymore, no one studies, and worse, the half-learned information that still lingers is so incomplete as to be misleading.  Post residency, we get our info exclusively from drug reps and throwaway journals.  Ergo, most residents are better psychiatrists than someone in practice ten years.

Woah-- be careful.  Think long and hard before you hurl "clinical experience is more important" at me.  Make sure you want to go down this road.

I am certain that I can take anyone with a college degree in any science, and in four months make them better than an above average psychiatrist.  This is an open challenge to the APA.  I'll repeat it: I'll take any person with a B.S. and in four months make them an academic psychiatrist.

But back to our "shortage" problem, or more accurately our distribution problem.  The solution to this is elementary, but bitter.   Either raise the standards necessary to be a practicing psychiatrist-- more audits and tests, greater documentation in notes, recertification exams with consequences to failing, and outcome/performance evaluations graded against other psychiatrists-- but also raise the pay, dramatically-- you can use the prescription drug savings when you implement my other plan-- so as not to lose the smart people to internal med or neurology; or lower the requirements so that more people can be prescribers, and lower the pay so that you can afford more of them.  Either of these two will satisfy the growing "need."  Which is better for the patient is up to California to figure out.

Hi. On 9/25 I recommended 5 stocks.  Here's where we stand one month later:
 

BRK.B    3123   3333 +210 (7%)
SHLD     160   180    +20   (13%)
GOOG    404     480  +76   (19%)
USO        56      52   -4   (-7%)
COP        57      61   +4   (7%)
AAPL       76      81   +5   (7%)

For comparison, the S&P rose 4% during this month.   

So there you go.  And, for the record, I now own only GOOG and BRK.  USO should continue to fall (and so should consequently, COP) until either it gets really, really cold outside, or until April (it is cyclical, after all.)  But when USO goes up, GOOG should go down. 

And USO is a winning play over the next decade. I can only be sure of that with SHLD and BRK. 

In case anyone cares at all, I also own AKAM; I'll be buying back AAPL this week; and I'm looking at ATI.   These should be solid plays until Christmas, though I'm expecting some sort of massive retracement this week.

This is what a subscription to JAMA gets you:

 from the abstract:

RESULTS: Among the 201 women in the sample, 86 (43%) experienced a relapse of major depression during pregnancy. Among the 82 women who maintained their medication throughout their pregnancy, 21 (26%) relapsed compared with 44 (68%) of the 65 women who discontinued medication. Women who discontinued medication relapsed significantly more frequently over the course of their pregnancy compared with women who maintained their medication (hazard ratio, 5.0; 95% confidence interval, 2.8-9.1; P<.001). CONCLUSIONS: Pregnancy is not "protective" with respect to risk of relapse of major depression. Women with histories of depression who are euthymic in the context of ongoing antidepressant therapy should be aware of the association of depressive relapse during pregnancy with antidepressant discontinuation.

Read it again.   What's the message they are trying to communicate?

The study found that pregnancy is not protective, and stopping your meds during pregnancy raises the risk of relapse.  Any other way to interpret this abstract?  Am I putting words in their mouths?

I read the entire article, with familiar horror.  This was a naturalistic study that followed 201 women with MDD and their medication dosages and saw what happened.  That this study had nothing  to do with the "protective effect of pregnancy" is right now a secondary issue.  The real problem is that the actual study says something very different than the Conclusions:

The study did find that more people relapsed if they stopped their medications.  But it also found that more people relapsed if they increased their medications.

Exactly how were you to know this if you only read the abstract? 

 Don't you think that might have been important?  Tthe medication changes themselves are not the cause of the relapse-- how could both stopping them and raising them both be bad?-- but are logically explained as representing  something else. 

The Conclusions should have read:

Conclusions:  Taken together, these findings suggest that pregnant women who are stable (on medication) tend not to relapse, but those who are unstable (and need  med changes or who go off them) relapse at higher rates.

The authors do address, slightly, this odd finding-- on the last page.  But so what?  Only liars read the last page.  What makes this misrepresentation so egregious that it is near unforgivable in a journal of JAMA's arbitrary status is that they and we know doctors are not reading these studies from start to finish; for the most part, we skim over the abstracts. So we're going to skim over this abstract, it supports our existing prejudices so we don't give it a second thought, and go on with our deluded lives.

So to write the abstract this way is absolutely volitional, and absolutely misleading. The problem is not with the study, which was excellent, but with the presentation of the findings, which is psychiatric propaganda.

I would demand my subscription to JAMA be cancelled immediately if I had one in the first place.

---

But this isn't really the disturbing part. 

What's really sad is that I am, apparently, the only one who noticed this.  None of the three Letters To The Editor about this article complained.  One of the three letters did complain, but not about the article-- rather about the authors' ties to drug companies.   Yes, that again.  That's what passes for ccritical thought nowadays.  That's now the default moral high ground soundbite of bitter doctors, akin to "the war is just about Halliburton" or some other half-thought  deduced from two hours of the Colbert Report and the table of contents of the New York Review of Books.

That's the problem.  We're not critical of our fundamental principles.  So we attack windmills.  We doctors are conditioned (yes, conditioned) to find Pharma bias everywhere, and never to see-- so that we don't have to see-- the real bias in the way we have set up psychiatry.  It's the same reason we spend so much time on statistics.  Pharma and statistics are witches in The Crucible.  

The bias isn't Pharma related.  It's much more fundamental.  What's at issue here is the approach, the worldview of the authors and psychiatrists everywhere.   They are seeking to support the notion that antidepressants work and prevent relapse-- not even because that's what they believe, but because that's what psychiatry is. They are not asking a theoretical question and impartially looking for the truth; they're unconsciously trying to validate their existence. So they see what they want to see, and anything that isn't obviously in support of these postulates, this paradigm, is cursorily dismissed-- or is altered to mean something else.  This is important: they're not hiding data, they just interpret it with the only paradigm they have.

Blaming Pharma is easy because it seems obvious-- money buys truth-- but also protects the blamer from needing to perform any actual critical thought, any internal audits of their prejudices.  So what if Pharma bought those doctors start to finish?  You still need to read the study and figure out how the buying altered the data, if it did.  But that would be work.

10/30/06  Addendum:  I sent a modified (i.e. nicer) version of this as a Letter to JAMA.  It was rejected in less than a day. 

Goethe's popular comic book, Sorrows of Young Werther, published in 1774, allegedly inspired two imitations; many young men decided to kill themselves, and many, many young men decided to dress in yellow pants and blue jackets. 

Well, that's all great, but I decded to investigate whether suicide rates really do increase after one is popularized-- a propos of the previous post's discussion about whether Lott's roommate's suicide pushed Lott towards it as well.  I was pretty sure it existed, but I may have been wrong. Preliminarily, here's what I found, through 4 examples:

A study in Austria found that gun suicides increased in the three weeks following a famous gun suicide (as compared to the three weeks preceding).   There are lots of this kind of study, which are correlations based on statistical anomalies.

A better kind of study actually interviewed the suicide attempters to see what had affected them.  For example, a U.S. study found that exposure to parental suicide was not associated with suicide; exposure to a friend or acquaintance's suicide was mildly protective, and media accounts were strongly protective.    However, this study wasn't about the immediate risk (e.g. in the following month), and the authors did note that this protective effect was only if the friend's suicide or media report was greater than a year in the past.  It is easy to speculate that the longer you have to think about what they did, the more likely you are to think it wasn't the best option.

A 1993 study in adolescents found that within one month of the suicide of a friend, depression and suicidal ideation increased; but actual suicide attempts did not. 

Most of the studies finding no correlation are done using the general population; how would it be different if we looked only at people with established mental illnesses?  A 2005 study found that suicides in mental illness patients were clustered in terms of place, time, and method.  Unfortunately, this study looked at the clusters and did not identify whether or not the victims were actually even known-- or whether the patients had even heard about the suicides.    (For example, they might happen at the same clinic, but that doesn't mean they knewabout each other.)

Etc, etc.  So clustering, at least in terms of lethal attempts, appears not to happen much, (and if it does it is primarily in teens.) 

As an observation, most of the articles finding Werther effects were written pre 1980, while most finding no relationship were written post 2000.  One explanation is that we are more rigorous now (HA!); the other explanation being that there is considerably less idealization of suicide now.  In fact, suicide now is unremarkable.  Consider the "medicalization" of depression and suicide, as biological diseases rather than character pathology or expressions of emotion, a communication of sorts.   Suicides then "meant" something-- something more than "I'm depressed," while suicides now are simply symptoms.  Suicide= more Wellbutrin. 

I still think they "mean" something, and I try to interpret it, but the focus nowadays is certainly not to interpret suicide as an expression of anything.  Too bad.

Would you predict suicides increase in the sunshine/summer or darkness/winter?

Obviously, if I'm asking...

A Greek study-- and Greece has one of the lowest suicide rates in the world, about 5/100,000 (U.S. is about 17) with two major findings:

1. Suicides in the northern hemisphere, across 18 very diverse countries (Europe, Eastern Europe, Japan, North America,) peak in May/June, with a relative risk 1.08-1.5.

2. This peak is actually due to the amount of sunshine.  More sun=more death. 

#2 seems a stretch to me, so I looked it up further.  Wow. 

Same guys, find that there was no relationship between suicide and that day's sunlight; but there was a strong correlation with the past days sunlight.  There were several specific sun/day-suicide interactions, but in general for males that past 8 days and the day before, and for females that past 4 days  (but not he day before) were correlated to increased risk.

If you consider that the solar radiance in June is 26 MW/m2, and December is 6 MW/m2, then the risk of suicide increases 3% for every 1 MW/m2. 

Others have found the same.  An illustrative example is the Chile study finding the springtime peak of suicides, but this effect was absent in the north, which is closest to the equator and thus has the least seasonal variability (Chile is a strip that runs up and down the western part of South America.)   Interestingly, other equatorial regions have failed to find seasonal suicide links (e.g. Singapore); some have even identified a reverse pattern in the southern hemisphere.  And  urban areas seem to have a less pronounced or absent effect.

This is all quite interesting, but in order for it to be useful we have to show that there isn't another  obvious explanation.  Here's one: people kill themselves in June because there are more available tools.  Jumping off a building, outdoor hanging, drowning, all prefer better weather.  For example, you don't mull jumping off a building during a week of rain. 

Now you could counter that such a suicidal person would simply come up with something else (e.g. OD) but that's not what happens; suicides are very specific and personal acts.  The jumper doesn't instead use a gun.  (Consider that people with multiple suicide attempts use the same one or two methods each time.)  If two methods are similar, however, then I think such a move could happen.  But if the person is considering drowning, then an OD is probably not an option, because drowning means something, it ihas unconscious significance, and that can't be ignored.

I might even propose that non-OD and non-self-cutting suicides  are just as much about the act as about the desire to die.  When you get drunk and then stab yourself in the abdomen 45 times, you're communicating something as well as trying to die.

Following from this, it has been observed that there is no seasonal pattern to non-weather related methods: cutting, OD, gassing, (i.e. non-violent methods).  There's no seasonality (skew towards winter) to jumping in front of a German subway, which is thankfully free of sunlight's evil effects.

So if it rains for a week, instead of moving to another completely different suicide method, I believe they would simply postpone (i.e. continue to ruminate about it)  their attempt, unless a similar method is available.

BTW, this is about completed suicide, not suicidality or suicide attempts, for which I have no idea about the seasonality. 

I bring this up because of the discussions I've had, especially with the residents at my hospital, on the extent of volitional control in suicide.  I say it is a cognitive process and not necessary outcome of a disorder, and the idea that sunlight or weather can influence the timing or method of a suicide goes along with that.

Good debate going on over at Shrink Rap.

I've posted a few comments there, but in summary: 

Hunter College didn't expel her for being suicidal; they EVICTED her from her DORM for ATTEMPTING suicide.  They wouldn't be allowed to do the former; they are obligated to do the latter, for public safety.

People want to refer to the George Washington University case.  Ok, but let's get our facts straight, from the Superior Court: 

Jordan Nott's roommate jumped out the window OF HIS DORM in April 2004, while Jordan and a friend were trying to break in to save him.  In October 2004, thinking about this, Jordan himself became suicidal but went to the ER instead.

The University suspended him. Well, not exactly:

 in order to be “cleared” to return to the residence hall, the UCC and Community Living and Learning Center (“CLLC”) had to assess Jordan’s “ability to obtain recommended treatment” and ability to “live independently and responsibly.” In order to obtain clearance, Jordan had to set up an appointment with UCC within 48 hours, and develop an agreed-upon treatment plan."

That's not GWU's defense-- that's Jordan's complaint.  That he had to go through that.  Is that so  unreasonable? 

Living in the dorm is a privilege, not a right.  If the University thinks you are a risk AND it doesn't violate constitutional rights, you're gone.  It's not up to them to prove you are a risk; it's up to you to show you aren't. 

And people are angry about GWU and Hunter because it seems that they're just out to minimze their liability?  Yeah, so what?  And it's not just the liability of  student who kills himself.  It's about his effect on public safety.   People say that an OD or jumping out a window doesn't put other students at risk.  Well, clearly Jordan's roommate's suicide had an effect on Jordan-- he admits this himself.  So there's the copycat risk.   And what if Jordan had decided to blow himself up?  Why should anyone in the dorm have to live with that risk?

Again, it's not up to the school to show she's dangerous; it's up to her to show she's not. 

The best line is this one:

If he had known [about the evictions, etc], he said, he never would have gone to the hospital.

So what are you saying?  That he would have simply killed himself? 

Hunter college caves to lawyers.

A student takes a bunch of Tylenol in an OD attempt, and after 4 days inpatient returns to her dorm to find she has been evicted.

The article makes it sound as if Hunter College kicks you out of school if you attempt suicide, which wouldn't necessarily be improper, but it's also not true.  Hunter College's housing contract says you can't live in the dorms if you have a suicide attempt.  That's a little different.

So the student sues, and Hunter decides to settle.  Her lawyer gloats:

“We’re pleased that Jane has been compensated for the college’s discriminatory treatment based on the stigma attached to a mental illness,” said David Goldfarb, one of the law firm’s attorneys representing her. “If Jane had been hospitalized for mononucleosis or pneumonia, I am confident that she would have been welcomed back to her dorm,” he added.

Well, gee, maybe a suicide attempt made volitionally, with a good chance of happening again, is a little different than pneumonia?  Stigma of mental illness?  So she was evicted for being on Zoloft?

The lawyer for the Bazelon Center for Mental Health Law , who should know better, says,

"Schools that exclude students who seek help discourage them from getting the help they need, isolate the students from friends and support at a time when support is most needed, and send students the message that they have done something wrong.”

Hmm.  I thought Hunter was a college, not a daycare?  Since when is it a school's responsibility to ensure adequate access to friends?  Is it responsible for finding them mates as well?  The problem with this statement is its logical conclusion: when can a school exclude students who seek help?  Never? Let's say the next time she tries suicide by turning the gas on, and she blows the dorm up.  Oops?

The most dangerous quote of all is from someone who really, really should know better, but obviously doesn't:

[Rachel] Glick, who is also associate chair for clinical and administrative affairs and a clinical associate professor in the Department of Psychiatry at the University of Michigan, emphasized that "universities should be open to being informed by psychiatrists and other mental health professionals about what to do to enhance the care of the students, rather than just thinking about protecting themselves from lawsuits."  

So let me get this straight: the university should ask the psychiatrist about whether the person could stay on campus or not?   Any psychiatrist out there who wants that liability football?

College:  So now that you've evaluated her, should she be allowed to return to the dorm?
Psychiatrist:  Huh?
College: Is she going to kill herself again?
Psychiatrist: How the hell would I know that?  I can't predict the future.
College: But you all told us we needed to seek your advice.
Psychiatrist: Hey man, don't try to pin this on me.  I'm going to lunch. 

Interestingly, the Psychiatric News article doesn't mention last month's case  where the parents of  Charles Mahoney sued Allegheny University because Charles was not put on mandatory leave of absence while he battled depression for two and a half years.

Everyone wants it their way; sue when you don't get your way, logical consistency be damned.  Has it now become outrageous to say that  the liability for a suicide attempt and its prevention lies  entirely with the person attempting it? 

(Addendum: many angry at my post, so I refer to the specifics of the case itself.) 

I was sent a question from a recent college grad about whether to go into psychology or psychiatry.  I don't know which is better; you should decide which you like yourself.  But I can help you succeed in either one.

Since I graduated residency, I have never-- never-- had a patient ask me where I went to college, medical school, or residency.   Whether I went to Harvard or Guatemala,  no one would know.

It's true that people assume I'm good because I work in an academic institution, so it lends me credibility. And it helps in court, tremendously, to say I work at said academic institution. 

But if you are thinking of being a private practice clinician, it matters not  a lick where you went to school-- or fellowship  (so don't go.)

What matters is how you set up your practice. 

Following the logic above-- and you will initially doubt this, but bear with me-- it doesn't matter, financially, whether you go into psychiatry or psychology.   What matters is how you go into the practice of either.

I'll just refer to psychology right now.  And I'm going to talk about money only, not personal fulfillment or career advancement or awards-- all that is your business, and there are plenty of resources to help you.  I'm trying to tell you about the money side, which no one else seems ever to want to talk about.

The key difference in the need for psychiatrists and psychologists is the duration of follow-up.  The shortage for psychiatrists exists for long term follow-up in Medicaid/Medicare patients. The shortage for psychologists is for short term and CBT for private insurance patients.

To get rich in psychology, it is not necessary to have the best paying patients, but rather to have a steady stream of patients, whether they pay well or not.  If you have a waiting list, you win.   How to get such a stream?

If I was a good clinical psychologist-- PhD helpful but not necessary, master's is fine-- I would find two or three good psychiatrists and set up a group.  Every patient that comes through them has to have at least an initial eval with you, and vise versa.    This guarantees you volume, which is great, and many of the patients will continue on with you.  But even those that don't are still a win, because that first session can be billed at a higher rate. 

I would find the nearest academic institution with a "residents' clinic."  That's a gold mine.  There are a lot of private insurance patients there, who need short term therapy.   These academic clinics almost never have enough therapists, because the ones that are on staff are not really incentivized to see extra patients; they're on salary.  So there is a massive number of patients who could benefit from therapy, but are on a waiting list, etc, etc.  It seems unbelievable, so I'll say it again: university clinics need to refer out.   If you can get that overflow, if you can get one or two docs there to vouch for you and tell everyone to refer to you, you win.  How do you get them to vouch for you? Well, go back to my Steps: you need a specialty.  Do you do CBT?  Supportive therapy?  Grief counseling?  You can do everything if you want, but it is vital that you be known for regular therapy AND something specific.  You business card should read, "Grief Counseling and General Therapy" or something like that. 

That makes you different and better than other therapists, even if you're not.  And makes it ok for you to approach a psychiatrist to get referrals.  "Hi, I'm a therapist, send me patients" is very different than, "Hi, I specialize in Grief Counseling, short and long term, so if you have any patients..."  Again, the unbelievable truth about referrals: the referrer doesn't actually have to be certain you are an expert-- hell, they don't even have to know you are legitimate.  If your niche is sufficiently small, they simply won't have any other names available when they get cornered by a patient.  And if they have no other such "expert" in their minds, you will get the referrals.  And if just one patient reports back to the psychiatrist that you're good, you win.

Try to meet psychiatrists wherever you can, but the best place I know is through drug reps.  Go to one of the "drug dinners" and meet the psychiatrists who attend.  Find a psychiatrist-parent-- hell, any kind of doctor-- in your kid's school, meet them, let them know you're open for business.  Meet the guidance counselor, tell them you specialize in adolescent issues. (Obviously, make sure you actually do specialize in adolescent issues.)  Or Family Systems model.  Or divorcing parents.  Etc.

Remember: it's not "why refer to me?" It's, "who else are they going to refer to?"  A doctor who has any sort of emotional connection to you (i.e. met you once) will more likely refer to you than anyone else. 

If you want to work the "best paying patients" angle, then you are looking to work with cash only, reasonably affluent patients.  Ok-- why would they pick you?  Because you went to Harvard?  Because you have a PhD.?  No.  Either a) you in their insurance network so they can get reimbursed; b) they were referred to you.  And so we are back to the beginning: you need a niche, an area of expertise.  I am sure there are patients who would prefer to go to someone who did "Psychodynamic Therapy and General Psychotherapy" even if they simply needed grief counseling, because they'll assume you're better.

I'll write more about this later. 

You may as well find out what you're up to.