Thursday, June 4, 2015

Information versus Wisdom




I saw this post on another blog today and thought it was a good title.  I end up pondering this idea almost every day.  In medicine these days we are inundated by data scientists on the one hand and administrators on the other.  The data scientists tell us how they are going to revolutionize medicine through their analysis of large data sets.  The theory is that there are patterns in the data that can be detected only with advanced computational methods.  Having gone through the spreadsheet era and seen how easy it is to prove almost any theory with a large spreadsheet, I am very skeptical of Big Data.  Just dredging through the data, looking for patterns and writing it up does not seem very rigorous to me.  It strikes me more like one of the popular TV shows where the agents are in the field but solidly connected to the computer whiz back at headquarters who is capable of pulling up any document, any floor plan, and hacking into any closed circuit TV system in order to get the information that is needed.  I don't think that science works that way.

On the administrative side, it is the worst of times.  The statistical efforts of administrators are frequently laughable attempts to legitimize the next genius idea to come down the pike.  Their mistakes in healthcare are legendary ranging from the promise of the electronic health record to the RVU based management of physicians as widget producers, all exhaustively documented with numbers.  I sat in a meeting one day that showed 95% of the physicians in the department were not "producing" enough to cover their salary.  The problem was that nobody had done the multiplication on "work RVUs".  When the appropriate multiplier was added it was a different story.  Administrators also tend to collect a lot of numbers that they think will be useful for an analysis, without thinking ahead to the data analysis and statistics.  They seem to have no idea about basic statistical analysis much less more advanced analysis like how to legitimately analyze data over time to detect real differences.  There is no better example than the state of Minnesota collecting PHQ-9 scores over time from anyone trying to treat depression in the state.  They seem to think that unconnected collections of those numbers at different points in time will have some kind of meaning. Administrators also have the habit of creating studies that confirm their vision of the world and when those studies are complete - that is all of the "proof" that is necessary.  The entire concept of managed care rests on many of those studies.

On the wisdom side of things I can think of no better example than a colleague who I said goodbye to today.  He worked with me for the past 2 1/2 years.  He is an Internist who is also an Addictionologist and is ABAM (American Board of Addiction Medicine) certified.  He has been a physician since the early days of the HIV/AIDS epidemic and treating those patients was a significant part of his early practice.  He has an encyclopedic knowledge of the care of those patients and how it has evolved as well as being an excellent Internist.  He is interested in psychiatry and can talk in psychoanalytic terms.  He is also an expert in LGBT issues and can speak with authority on that subject.  I certainly did not want to see him go, but for the purposes of this post, I can think of no better example of wisdom that comes with medical practice.  He could be consulted on any number of complex problems in his areas of expertise and provide a very well thought out answer based not so much on information, but on what works and what the potential complications are.  Any physician can tell you that these are the folks you want to work with.  When I think about data mining approaches toward these areas of knowledge, I think about the 31 page document that is available online that looks at the issue of medication interactions of psychotropic and HIV medications.  It is a compulsively great document, but lacks the wisdom to help you pick the best therapy for a manic patient on tenofovir.

Granted my position is a thoroughly biased one.   I make no apologies for wanting to work with physicians who have the greatest technical expertise and know how to apply it.  I don't mean people who can recite facts or even algorithms.  I mean the people who know all of that and can look at the patient with the most complicated medical situation and still come up with a plan of action and how that patient must be closely monitored.  They also know when it is better to do nothing at all and that is a difficult skill to acquire.  Practically everyone leaves medical school and residency with a strong treatment bias.  You are taught to be "aggressive" and that most of the treatments that you do will do some good even if there is not cure.  In clinical practice, that is far from the truth.  In psychiatry for example, you have to recognize that there are certain biological predispositions, clinical patterns, boundaries, and personalities that are the warning signs of disaster with certain treatments.

When I first started in medicine in the 1980s, the wisdom based model was still the predominate model in most clinical settings.  Now it is much less frequent and there are departments that are just looking for people to fill in the gaps.  They don't necessarily want to retain you they just want to "keep the numbers up."  They also don't want you spending a lot of time on complex cases, because the payment rates rapidly decline if you are not shuffling people in and out the door.    When the administrators start recruiting bodies based on their revenue models and Hollywood accounting,  I hope that I will always end up on the side with the wisdom, rather than a heap of useless information.

There is a lot of that going around these days.  



George Dawson, MD, DFAPA      



Supplementary 1:  I was going to jam in a section to comment on emotional and moral reasoning in view of the expected backlash to the Rosenbaum articles in the NEJM, but decided to add it here instead.  It would have strained the above essay.  It has been an interesting (and fully expected) exercise in political rhetoric.  Predictably the critical articles mischaracterize her position and ironically are at least as guilty of the fallacies that they accuse her of using.  In one case, a new fallacy is pretty much invented.  I think it is instructive to note that in these matters, logic goes out the window.  There is no pathway to a sound judgment.  It basically involves rallying the troops to see who can shout the loudest.  My self proclaimed bias above is part of the reason I am firmly on her side (but will refrain from the shouting).  For anyone who thinks like me, there is no convincing me that the appearance of conflict of interest is the same as actual conflict of interest.  There is no convincing me that free pizza and donuts will cause me to blindly prescribe a medication - probably because I have not eaten Big Pharma food since the early 1990s.  In fact, if I think of a more plausible thought experiment about how much cold hard cash it would take to pay me off to prescribe a drug, I can't come up with a figure but it would have to be absolutely stratospheric compared with the usual speaker fees that people are listed in the Sunshine Act database for.  All of that is based on the fact that I work for a living and treat real patients.  I am accountable to those patients.  If a medication does not produce real results or it causes too many side effects (like my early experience with paroxetine) it is off the Dawson formulary and I don't prescribe it again.

This is of course like arguing with Democrats and Republicans.  I know that some pro-appearance of COI=COI will strongly dislike my experience and the way my thoughts on this matter are anchored in the way I practice medicine.  That is the nature of arguing about emotional and moral reasoning in what the Institute of Medicine (IOM) describes as an ethical vacuum.  The recent editorials certainly don't prove a thing.

The usual focus of these debates also leaves out the big picture that many entire University departments (math, science, engineering) actively collaborate with industries and in many cases actively invite industry participation in order to advance those fields.  The notion that physicians are not able to do that because they have a sacred trust to patients and would be somehow compromised remains implausible to me, particularly when nearly all of the major decisions that physicians make in this country have been seriously tampered with if not controlled by managed care companies and pharmaceutical benefit managers for nearly 30 years.

That is a massive conflict of interest that nobody talks about and it affects 80% of all of the healthcare in this country.

Supplementary 2: The graphic at the top of the post is from Shutterstock.







5 comments:

  1. Good Morning,


    Could you please reference this 31-page document about treating psychiatric disorders in the setting of HIV.


    Thank you,
    Resident

    ReplyDelete
    Replies
    1. Sure - I have the text only version - a more compact color coded version can be found by Googling "Psychiatric Medications and HIV Antiretrovirals"

      Delete
  2. Careerism, ideological zealtory (which was the impetus for recovered memories), publish or perish (with bias toward positive results) and academic logrolling and Elsevier proliferation of useless journals are as much to blame for bad science as pharma or, as you pointed out, other industries that collaborate with academic departments. I'm not saying pharma money is innocuous but a coffee cup is not the same as a 200K honorarium and a ghostwritten article.

    Pharma COI is but one of many flavors of pernicious influences on the science. Pharma money also produces amazing treatments such as HIV meds. I do not want pharma money to go away, the net result would be awful. Biases cannot disappear in the real world with human beings, but should be disclosed. I'm for an ethical ban on academic ghostwriting.

    ReplyDelete
    Replies
    1. There are those ugly little facts that drug and generic drug production worldwide is done by private industry and those companies can replicate only 11-25% of the preclinical work that is often cited as the reason why they can produce new drugs.

      I don't wonder where the other 75-89% comes from.

      Delete
    2. Thanks for that stat. I was aware the pure science replication rates were pretty awful but that really puts it in perspective. The root causes are far more serious than pharma money.

      Delete