Best quote from AMIA-2008

"It doesn't mean that our systems are bad, it is just that the combination of systems and practice doesn't work."

No comments.

Defying the Statistics

- What is the chance that you meet a dragon on the street?
- Fifty per cent.
- ???
- Well, either I meet him, or I don't!
A joke

Statistics plays a large part in psychiatry, as in the rest of medicine, because it helps clinicians to make decisions about what is likely to be beneficial or harmful to their patients. Most of the guidelines for physicians are based on the statistically significant conclusions from the clinical trials. The argument goes like this: It has been found that, say 70% of the patients with a certain diagnosis and a particular set of other characteristics (age, race, height, weight, etc.) get better on a medication. Therefore, if you have a patient Ivan Johnson who fits all these characteristics, his chance of benefiting from this medication is seventy per cent, which means he should take it.

Here at the American Medical Informatics Association conference I constantly witness conclusions like this being drawn. They have become basis of medical expert systems and much of medical informatics in general. Yet, I am going to show with one simple example, that this logic is erroneous. It is a case of scientific confusion between the probability of an event observed in many previous experiments, and the ability to correctly predict the result of one next experiment.

Let's not go as far as medical predictions. Let us take the simplest model of probability - a tossed coin. The probability of the heads is 50% - everybody knows that. Indeed, if we toss a coin a billion times, we will probably observe the heads in something very close to 50% of the cases. But that is all it means! You can't go any further in your conclusions! In your ability to predict what the coin will fall next, in your one-billion-and-first experiment, you are as helpless as you were when you'd thrown it for the first time. You simply don't know what it will be - all you can say is that it will be either one or the other.

It is not because the probability is 50% that you don't know. Take an experiment with the chance of outcome A being 90% and the outcome B being 10%. What is the probability of the outcome A in the next experiment? Nine out of ten, very well. But what outcome will the next experiment have? You can't ask this. You don't know. It will be either one or the other.

So what does this nuance mean for the system of psychiatry and health care in general? Not much, because the system operates on statistics, it deals with patients en masse, and therefore the results of multiple experiments correlate well with the predictions based on probability. But it means the world for an individual patient; for you and me.

Even if we forget for a minute that the experiments in which the probability is calculated were not actually done on you and me, but on some other people; even if we forget that the medical outcomes are not binary but multidimensional and unpredictable by their very nature - even then predicting an outcome for a particular patient will be impossible because of the effect I've just described. If 99.99% of the patients like you die in a month without an operation, it doesn't mean that you will also die in a month. The probability of it is high; whether it happens is not known.

I admit of course that with a chance like that you might want to strongly consider the operation; I would probably do the same. However, the majority of psychiatric statistics operates with numbers quite remote from the extremes. People talk about reducing risks from twenty-five to fifteen per cent, of improving the outcome in sixty per cent of the patients versus forty. For the hospital and the insurance company these numbers are big business; for you and me, they are simply meaningless.

Ads removed

I decided to remove the ads from this blog. Due to their contextual nature, they were mostly about promoting various psychiatrists' services, which created a sort of irony that I tolerated for a while. But then it's ridiculous after all.

Now I can proudly say that I have no financial interests whatsoever in discussing my views about psychiatry. :) If you like this blog, however, please donate a comment or two.

Psychiatry needs a user's manual

Psychiatry may be a nice tool, but only if you know how to use it, and whether it is the right tool for your tasks. Unfortunately, modern psychiatry is usually sold to the customers without a user's manual; the patients have to figure everything out by themselves.

As with any powerful tool, such approach is bound to be dangerous. Many a patient have suffered a great deal from psychiatric enterprise, only because they didn't know what to expect from it. Everybody knows how to behave if you're stopped by the police, but very few people seem to have any idea what they ought to do in a psychiatrist's office or a mental hospital.

Will the gene of schizophrenia help you?

So why am I not being too enthusiastic about psychiatric genetics? Why am I not excited by the prospect of uncovering the genetic basis of mental disorders, despite the proponents' claims that it should make diagnosis more reliable and eventually provide avenues for effective treatment?

Because these promises do not sound realistic to me. I have to admit that the question of psychiatric genetics is very interesting academically, and it is also readily funded by both drug companies seeking profit and the government seeking some kind of solution to the overwhelming quantities of mentally disturbed citizens. Still, I find it difficult to buy that genetic advances will benefit the real-world patients, and here is why.

First of all, the genetics of mental disorders is subtle and multifactorial. Therefore, even if some genetic predisposition could be statistically proven, in any concrete individual the environment plays a dominant role in determining whether the disorder will appear. In other words, regardless of whether a particular person possesses the wrong genes or not, his being sick or normal will largely depend on his life situation (which is exactly what we've known for a hundred years).

Let's take an example from oncology. There is a gene called BRCA1 that, if mutant, has been recognized as carrying much higher risk of breast cancer development. Now, in oncology the diagnosis can be made much more accurately than in psychiatry; the genetic association with BRCA1 gene is also simple and straightforward. Nonetheless, oncologists do not usually make predictions about any particular patient; even though the risk may be significantly increased, we still cannot say to the carrier or the mutant gene: "You will have breast cancer", or to the carrier of the normal gene: "You're cool, go home."

Whatever gene associations are found in mental disorders, they will not be enough to make predictions, because the environment plays a major role in the fate of any organism. Hence, we can only afford hunting for genes after we've done everything possible to fix the environment. Talking genetics when there are major problems in people's psychological well-being (look around!) is like cleaning the carburetor when the car's wheels are missing. It could help, but it's not the biggest concern right now.

The second reason why genetics is unlikely to benefit the patients is that psychiatry has a tendency of rushing things into practice as soon as they are discovered (if it can be profitable, of course). It is theoretically possible that genetic discoveries will lead to the creation of the effective treatment for psychiatric disorders, but most likely this simply won't have time to happen. What will happen much sooner is that some type of genetic analysis will be implemented, and people who are at risk will be recommended to take the medications preventively.

This will be catastrophic, because the drugs can make a healthy person sick; so there will be no way of disproving that they were really needed. This will reinforce the sense of scientific correctness, and promote further treatment and propaganda efforts, supported by the drug manufacturers. The tradition of giving drugs to healthy people, mandatory genetic testing of children and adults, and the arising legal battles will change the whole landscape of psychiatry into something entirely different from what we know today.

In the resulting confusion the original goal of the genetic research (to develop treatment) will be postponed, as scientists of the day will concentrate on getting out of the current crisis. And everything will repeat again - only the counter of the innocent victims of psychiatric help will advance by a few tens of millions... but who counts them?

Gene of Schizophrenia?

A recent advance in studying the genetics of schizophrenia is reported in this Nature article.

There are reasons why scientists keep looking vigorously for the genetic basis of psychiatric disorders despite decades of unsuccessful efforts. With such persistence it is likely that some kind of genetic association will finally be found, or at least claimed to be found. Given the computational complexity and sample sizes of the modern genetic studies, many years could pass before the results of any given study could be verified or disproved.

What worries me is that psychiatry has a tradition of trumpeting the first promising results of a new study and rushing them into practice before the results are verified. And the nature of the specialty is such that once something new is implemented, it confuses the picture so much that it becomes impossible to say if the new method was even helpful.

In practice, as soon as some believable hint is received that psychiatric disorders are genetic in nature, and some genetic testing procedures become available, it will make the case for preventive use of psychotropic drugs on people who are "at risk". This makes no scientific sense (I'll explain this point next time, otherwise I'll be late for my train), but this most definitely will be done.

When You See What You Like You Want It - Science Has Proved

"Strong cravings for alcohol can be sparked by the mere sight, smell and taste of a person’s favorite drink,"

- say scientists, who are about to publish their findings in Biological Psychiatry, a prestigious psychiatric journal, as reported by Science Daily article today.

This revolutionary scientific finding has opened our eyes on one of the most interesting aspects of human behavior: when we see something that we like, we start to want it. Turns out, alcoholics can abstain from drinking for a long time, but as soon as they see their favorite drink in an engaging environment (such as their neighborhood bar), their desire to consume the drink increases dramatically.

To discover this curious fact the scientists used the help of the rat behavioral model. They first trained the rats to drink alcohol in a special "bar cage". It isn't mentioned what types of drinks were served, which is one of the shortcomings of the study: after all, the rats' behavior would probably differ if they were offered vodka versus martini or beer (in which case domestic or imported could also make a difference). It was reasonably clear, however, that the rats enjoyed their drinking experience. Then the animals were put into a different cage, and deprived from alcohol. After a certain period of abstinence they were allowed to go back to the bar, and before the scientists knew it they were drinking again like there was no tomorrow.

“This effect is highly detrimental to humans who are trying to abstain from drinking,”

- says lead author Nadia Chaudhri, Ph. D., with the Ernest Gallo Clinic and Research Center at UCSF.

These humans, as her words imply, are hopelessly trapped inside the control centers of their essentially animal bodies that divert them from their chosen path and towards their favorite drink.

To help such unfortunate individuals, the scientists propose to broaden the context of the exposure-based therapies. The experiment has shown that when the drinks were no longer available even at the bar cage, the rats would gradually lose their interest to alcohol, and start to engage in other activities like mating and generally scratching around. Therefore, the alcoholics should, as part of their therapy, visit their favorite bars with the therapist, but - and here the ingenious scientific thought unveils itself - they shouldn't drink anything there.

"These contexts could be real, i.e., visiting bars or liquor stores, could be created using virtual reality techniques, or could simply be recreated by patients as they imagined visiting places that triggered their urges to drink,”

- says John H. Krystal, M.D., Editor of Biological Psychiatry and affiliated with both Yale University School of Medicine and the VA Connecticut Healthcare System.

It is easy to imagine how much this study can change the lives of alcoholics once its results are put into practice. And those of you who are not yet alcoholics (at least not the ones treated by psychiatrists) could also take a lot out of this study. If you know you want something, but shouldn't have it - you must never allow yourself to see it, smell it, or, perish the thought, taste it. Otherwise, your desire may increase. You've been warned!