Tuesday, June 3, 2014

The Issue With Patient Management Problems

So-called patient management problems have been building up on us over the past 30 years.  I first encountered them in the old Scientific American Medicine Text.  They are currently used for CME and more importantly, Maintenance of Certification.  To nonphysicans reading this they are basically hypothetical patient encounters that claim to be able to rate your responses to fragments of the entire patient story in such a way that it is a legitimate measure of your clinical acumen.  I am skeptical of that claim at best and hope to illustrate why.

Consider a recent patient management problem for psychiatrists in the most recent issue of Focus, the continuing education journal of the American Psychiatric Association (APA).  I like Focus and consider it to be a first rate source of the usual didactic continuing medical education (CME) materials.  Read the article, recognize the concepts and take the CME test.  This edition emphasized the recognition and appropriate treatment of Bipolar II Disorder and it provided an excellent summary of recent clinical trials and treatment recommendations. The patient management problem was similarly focused.  It began with a brief descriptions of a young women with depression, low energy, and hypersomnia.  It listed some of her past treatment experience and then listed for the consideration of the reader, several possible points in the differential diagnosis including depression and bipolar disorder, but also hypersomnia-NOS, obstructive sleep apnea, disorder and a substance abuse problem.  I may not be the typical psychiatrist but after a few bits of information, I would not be speculating on a substance abuse problem and would not know what to make of a hypersomnia-NOS differential diagnosis.  I would also  not be building a tree structure of parallel differential diagnoses in my mind.  Like most experts, I have found that the best way to proceed is to move form one clump of data to the next and not go through and exhaustive checklist or series of parallel considerations.  The other property of expert diagnosticians is their pattern matching ability.  Pattern matching consists of rapid recognition of diagnostic features based on past experience and matching them to those cases, treatments and outcomes.  Pattern matching also leads to rapid rule outs based on incongruous features, like an allegedly manic patient with aphasia rather than a formal thought disorder.

 If I see a pattern that looks like it may be bipolar disorder, the feature that I immediately hone in on is whether or not the patient has ever had a manic episode.  That is true whether they tell me that they have a diagnosis of bipolar disorder or not.  I am looking for a plausible description of a manic episode and the less cued that description the better.  I have seen evaluations that in some cases say: "The patient does not meet criteria for bipolar disorder."  I don't really care whether the specific DSM-5 criteria are asked or not or whether the patient has read them.  I need to hear a pretty good description of a manic episode, before I start asking them about specific details.  I should have enough interview skills to get at that description.  The description of that manic episode should also meet actual time criteria for mania.  Not one hour or four hours but at least 4 days of a clear disturbance in mood.  I recall reading a paper by Angst, one of Europe's foremost authorities on bipolar disorder when he proposed that time criteria based on close follow up of his research patients and I have been using it ever since.  In my experience practically all substance induced episodes of hypomania never meet the time criteria for a hypomanic episode.  There is also the research observation that many depressed patient have brief episodes of hypomania, but do not meet criteria for bipolar disorder.  I am really focused on this cluster of data.

On the patient management problem, I would not get full credit for my thinking because I am only concerned about hypersomnia when I proceed to that clump of sleep related data and I am only concerned about substance use problems when I proceed to that clump of data.  The patient management problem seems more like a standardized reading comprehension test with the added element that you have to guess what the author is thinking.

The differential diagnosis points are carried forward until additional history rules them out and only bipolar II depression remains.  At that point the treatment options are considered, three for major depression (an antidepressant that had been previously tried, an antidepressant combination, electroconvulsive therapy, and quetiapine) and one for bipolar II depression.  The whole point of the previous review is that existing evidence points to the need to avoid antidepressants in acute treatment and that the existing relatively weak data favors quetiapine.  The patient in this case is described as a slender stylishly dressed young woman.  What is the likelihood that she is going to want to take a medication that increases her appetite and weight?  What happens when that point comes up in the informed consent discussion?

The real issue is that you don't really need a physician who can pass a reading comprehension test.  By the time a person gets to medical school they have passed many reading comprehension tests.  You want a physician who has been trained to see thousands of patients in their particular specialty so they have a honed pattern matching and pattern completion capability.  You also want a physician who is an expert diagnostician and who thinks like an expert.  Experts do not read paragraphs of data and develop parallel tree structures in their mind for further analysis.  Experts do not approach vague descriptions in a diagnostic manual and act like they are anchor points for differential diagnoses.  Most of all experts do not engage in "guess what I am thinking" scenarios when they are trying to come up with diagnoses.  Their thinking is their own and they know whether it is adequately elaborated or not.

This patient management program also introduced "measurement based care".  Ratings from the Inventory of Depressive Symptomatology (IDS) were 31 or moderately depressed at baseline with improvements to a score of 6 and 4 at follow up.  Having done clinical trials in depression myself,  and having the Hamilton Depression Rating Scores correlated with my global rating score of improvement, I have my doubts about the utility of rating scale scores.  I really doubt their utility when I hear proclamations about how this is some significant advance or more incredibly how it is "malpractice" or not the "standard of care" if you don't give somebody a rating scale and record their number.  In some monitored systems it is even more of a catastrophic if the numbers are not headed in the right direction.  Rating scales of subjective symptoms remain a poor substitute for a detailed examination by an expert and I will continue to hold up the 10 point pain scale as the case in point.  The analysis of the Joint Commission 14 years ago was that this was a "quantitative" approach to pain.  We now know that is not accurate and there is no reason to expect that rating scales are any more of a quantitative approach to depression.

Those are a couple of issues with patient management problems.  The articles also highlight the need for much better pharmacological solutions to bipolar II depression and more research in that area.

George Dawson, MD, DFAPA


Cook IA.  Patient Management Exercise - Psychopharmacology.  Focus Spring 2014, Vol. XII, No. 2: 165-168.

Hsin H, Suppes T.  Psychopharmacology of Bipolar II Depression and Bipolar Depression with Mixed Features.  Focus Spring 2014, Vol. XII, No. 2:  136-145.  

7 comments:

  1. Today's post brought two shaking my head and laughing moments. Thank you.

    "The patient management problem seems more like a standardized reading comprehension test with the added element that you have to guess what the author is thinking."

    When it was slow in the pharmacy one of the pharmacists used to quiz me using the little mini-CME multiple choice tests in the medical journals. I would never be able to pass a real CME test, but sometimes with the ones he used, I would be on a roll with all right answers. It was just knowing how to take a multiple choice test by ruling out two answers and guessing which one was right out of what was left.

    "Rating scales of subjective symptoms remain a poor substitute for a detailed examination by an expert "

    My neighbor had a reoccurrence of back pain due to slipped disks, and was fretting that her doctor wouldn't take her seriously. So I said that they usually ask how much pain you are in on a scale of 1 to 10 with 1 being barely noticeable to 10 being the worst you have had or can imagine having, and that she should think about it and let him know what level it was. She thought about it for all of 10 seconds and blurted out, "I'll tell him it's a 20!". To which I replied that I was positive he wouldn't take her seriously with that answer.

    BTW, being realistically accurate with pain scales doesn't help much either. I have always taken a 10 to mean the worst possible pain you could image; think nuclear radiation poison pain. So I am in the 2-5 range generally and I have never said anything over a 5. Hence, even though I can be in very bad pain, I am dismissed pretty quickly by my healthcare team at those low levels.

    ReplyDelete
    Replies
    1. I have treated many people who were a "12" or "15" and that is the problem with calling this a quantitative measure. A quantitative measure is a gallon and it is the same everywhere. On that other end, I am sure there are plenty of Norwegian farmers and lumberjacks who don't even register pain until they are at least a "7.

      Belief system is also an important aspect of the pain scale. If you have the belief that the physician just has to increase the pain medication high enough and the pain will be gone, you will be less likely to rate small changes positively. The same thing is true of ADHD ratings.

      Maybe this is why they try to prevent science majors from getting into medical school. We know what quantitative analysis is.

      Delete
  2. Rating scales for affective disorder rountinely ignore the criteria that major mood disorder symptoms have to be pervasive and persistant, have to all take place at the same time, and have to be significantly different from the person's usual functioning. This fact makes them completely worthless in practice, and the example you give is a good illustration how Pharma bullcrap is being taken as Gospel by too many people in our profession.

    Hamilton himself railed against the use of his own rating scale for anything but screening people to determine who should have a more comprehensvie evaluation, because he knew that the symptoms rated on the scale are non-specific and appear widely in anxiety disorders. And don't get me started on Bipolar II. I've had several personal conversations with Hagop Akiskal.

    ReplyDelete
    Replies
    1. No arguments from me David.

      I trained in the heyday of the "biological psychiatrists". We had to do HAM-A and HAM-D ratings on all of our clinic patients. Depending on the inpatient service some people ordered daily Beck depression inventories. That is another reason I cringe when I hear the buzzword "measurement based psychiatry" and the associated political rhetoric.

      I don't see Big Pharma as the main proponent here. I think managed care systems, especially those who employ their own doctors, want to use all of these measurements as leverage against their doctors. Set arbitrary and meaningless numbers - more or less like they do with productivity numbers right now. What better fantasy can a manager have than a clinic fill of psychiatrists seeing 20 people a day and producing 20 rating scale numbers for his/her viewing pleasure.

      To your point about a lack of rating scale specificity, in training it was well known to all of us that patients with borderline personality disorder produced the most consistent "high" score that were resistant to any treatment intervention. They also maxed out the suicide scale and it often did not come down for months to years.

      The other faction who are trying to benefit from measure based care are the politicians who have to keep telling their constituents that they are "reforming" medicine and keeping doctors accountable. What better way to divert attention from their complete lack of leadership than to produce a massive volume of meaningless numbers and call it "pay-for-performance".

      Delete
  3. These patient management problems assess a level of competence that ranks below expertise. Similar to the medical student write-up of a new patient, where completeness — making sure all the categories are considered — is the standard of excellence. At this level, intuition, i.e., expert heuristics guided by experience-based pattern-matching, is actively discouraged. After all, a novice is apt to be led astray by hunches.

    I agree this is the wrong test for an expert. The widespread, burgeoning use of such tests reflects a denial of medical expertise and the commodification of medical practice. The push to standardize medical practice with externally defined quality and performance measures helps to raise the practices of those at the bottom, while denying and blunting the brilliance of those at the top. No non-medical overseer of health care wants you to use your hard-earned expertise at pattern-matching. Simply follow the flow-chart of parallel differential diagnoses in the manner of any good medical student, and you will be a competent "provider," an interchangeable cog in the gears.

    It's interesting to ponder, by the way, what would constitute a good test for an expert, other than such "measures" as reputation among peers, word-of-mouth referrals, and the like.

    ReplyDelete
  4. Well put Steven. I agree with you on both the intellectual/scientitic and political bases. I can recall the exhaustive medical student differential diagnoses and how by the time you wrote it down - you knew it was an irrelevant exercise.

    Your idea about an adequate test for experts is intriguing. I would imagine it is possible to have a real time interaction with a machine that has hidden diagnostic features and to document the heuristics used by the expert, If there isn't there should be and it would also be a good experiment to do an exhaustive examination of patterns to the correct diagnosis.

    The most important aspect of your post that I try to repeatedly emphasize here is the move to dumb down medicine and psychiatry to make it seem like experts can be replaced by checklists and that they are somehow equivalent.

    That is a clueless argument and unfortunately one that many physicians have fallen for and continue to do so. It is after all the reason why you practice medicine.

    ReplyDelete
  5. Well put Steven. I agree with you on both the intellectual/scientitic and political bases. I can recall the exhaustive medical student differential diagnoses and how by the time you wrote it down - you knew it was an irrelevant exercise.

    Your idea about an adequate test for experts is intriguing. I would imagine it is possible to have a real time interaction with a machine that has hidden diagnostic features and to document the heuristics used by the expert, If there isn't there should be and it would also be a good experiment to do an exhaustive examination of patterns to the correct diagnosis.

    The most important aspect of your post that I try to repeatedly emphasize here is the move to dumb down medicine and psychiatry to make it seem like experts can be replaced by checklists and that they are somehow equivalent.

    That is a clueless argument and unfortunately one that many physicians have fallen for and continue to do so. It is after all the reason why you practice medicine.

    ReplyDelete