A Trip To The Dermatologist

 

 

Pattern matching is an important skill for all physicians.  It is rarely discussed these days despite all the continuous hype about artificial intelligence replacing doctors by reading x-rays and other lab tests.  I taught a course in diagnostic reasoning for about 12 years and the examples of pattern matching I used were from dermatology and ophthalmology.   The dermatology experiment was a straightforward comparison of dermatologists to primary care physicians looking at the same slide set of rashes and skin lesions (1).  The dermatologists were correct more often, faster at diagnosing, and able to correctly diagnosis equivocal cases compared with the primary care physicians.

I want to be clear that does not mean that primary care doctors don’t do a good job.  Some are so good that dermatologists know that they need to attend to the diagnosis and treatment of some physicians who refer them significant numbers of patients with melanoma and other types of cancer.  All these factors were probably in my subconscious when I decide to call to see a dermatologist.

It was not easy.  The first appointment was a teleconference and I would call it a swing and a miss. I was given a very expensive prescription for ocular rosacea that did nothing. When I called again to be seen in person, I was given an elaborate algorithm based on how many problems I wanted to be seen about. The more problems – the longer the wait.  I decided to go outside of my usual healthcare providers to a private clinic closer to my home. 

I have noticed a gradual accumulation of dermatology problems with age. I have made every effort to avoid direct sunlight.  If I must be outside at any time when my shadow is shorter than my height, I am wearing a high SPF shirt and sunscreen, a baseball cap, and wrap around polarized sunglasses. I have probably been sunburned twice in my life and tanned once. About 2 years ago I noticed a ring-shaped red lesion on my right forearm.  Every now and then it seemed to burn but generally it was static.  I saw my primary care MD and he did a scraping and potassium hydroxide preparation to see if it was ringworm (tinea corporis). It was not, so he told me to apply the betamethasone ointment that I typically use for eczema to the area.  I did for a couple of weeks and there was no effect.

At about the same time, I happened to notice a blue spot on the lateral aspect of my left ankle.  That is a difficult area to see.  I went into see a primary care MD who used an ophthalmoscope for magnification and concluded it was a collection of pigmented cells that did not look like a melanoma.  She said she would describe it in my chart including recording the diameter so it could be followed along by primary care. 

I described all these problems to the Dermatologist's assistant before he walked in the room.  I had photos of all the dermatology products I had been using and what had been tried in the past.  Even though the pattern matching diagnosis in Dermatology is good, like other areas of medicine – the treatments seem to be hit or miss and even then the response seems to vary over time.  I made a note to myself that I should look for papers claiming that these are placebo treatments or it is just all regression to the mean.  But I doubt that there are any anti-Dermatologists out there complaining about that and too many diagnoses and too many medications.

The intake form that I completed was just 2 pages long and there was an occupation section probably to consider environmental exposures.  When the Dermatologist came in he was very cordial and talkative.  He established that we both went to the same medical school (27 years apart), lived in the same neighborhood while we attended, and knew some of the same professors.  He took the history and clarified the technical points to his assistant who had now become his scribe.  He used a dermatoscope to inspect the lesions and make rapid diagnoses on the right forearm (actinic keratosis), left ankle (fibroma secondary to trauma) and left malar area (actinic keratosis).  He recommended freezing the malar area and forearm with liquid nitrogen and said the fibroma was just a skin reaction to some trauma that did not require treatment. At that point we went into a more detailed discussion of the rosacea and ocular rosacea and failed treatments with doxycycline and tacrolimus.  He recommended a compounded product of azelaic acid, metronidazole, and ivermectin, advised me of the cost, and has his assistant set that up. It was a very efficient process – the diagnoses, freezing treatments, and discussion took about 20 minutes.  At the end all of the follow up, prescriptions, and documentation was done and he was moving on to the next person.

There are times when it pays to see an expert and this is an illustration of one of those times.  I had been looking at these lesions for 2 years and trying to take the next steps.  There are as many barriers to seeing a Dermatologist as there are to seeing a psychiatrist.  I knew enough to monitor these lesions and they did not seem to get worse, but they were also not improving. After 2 years I got the definitive diagnoses and treatment I had been looking for as well as reassurance that the ankle lesion was not a melanoma.

This is an impressive result compared with most physician visits.  Even considering that there were a couple of things that did not fit.  Sun exposure for one.  I am what is referred to as a white fish in upper Midwest vernacular.  That means apart from my blue veins and the redness of rosacea – my skin is generally as white as the background of this page.  I had some early exposure to people with skin cancer and have been very diligent about keeping my skin and retinal exposure to direct sunlight at a minimum.  I suppose there are other factors at play such as age and know there are senile keratoses – but this did not resemble typical lesions in my dermatology texts or online. The Dermatologist predicted that the freezing treatment would cause these lesions to slough off and be replaced by normal smooth skin.  I have a follow up in 3 months to see if that happens and if the compounded topical rosacea medication works.

I am currently studying high prevalence polygenic diseases and have included eczema on that list.  Some estimates say that 20% of the population may have it.  There is the association with asthma but in my case as my asthma improved with age, I developed eczema and then worsening eczema.   I expect there will be many parallels with psychiatric disorders and diseases when my comparison is done. 

In the meantime, a Dermatologist in the right setting is a good consultant to have in your corner.

 

George Dawson, MD, DFAPA

 

References:

1:  Norman, G.R., Brooks, L.R., Rosenthal, D., Allen, S.W., & Muzzin, L.J. (1989). The development of expertise in dermatology. Archives of Dermatology, 125, 1063–1068

 This is the original reference I used in my course on the diagnostic process and how not to make a mistakes.  The first author has written significant papers about this.  

Graphic:

I mapped the dermatology conditions onto the body outline. If someone has a better body outline or one that they use on a standardized form and you want to send it my way - please do.  I can make a much better graphic if the outline is a separate shape.  The actinic keratoses areas on the map are probably both only 2 cm in diameter.  The rosacea can happen anywhere on the face and most annoyingly on the eyelids.  The eczema is a whole body condition that started out subtly as intense pruritis on the extremities and eventually spread to the abdominal area, chest, and back.  Pruritis is the most significant symptom with occasional lesions that looks like abrasions.  It can be exacerbated by skin contact with allergens like ECG electrodes. 

How Physicians Think

One of the more interesting aspects of my career has been contemplating how physicians make decisions on both the diagnostic and therapeutic side. Early in my career there was an explosion of activity in this area. Much of it had to do with internal medicine. There were computerized programs that were designed to assist physician decision-making. There were also entire courses taught at the CME level by experts in the field. At the time those experts included Jerome Kassirer, Stephen Pauker, Harold Sox, Richard Kopelman, Alvan Feinstein, and others.  The New England Journal of Medicine has a long-standing feature entitled Case Records of the Massachusetts General Hospital that showcases both diagnostic reasoning and the associated clinicopathological correlates. They added additional articles and a long standing feature on diagnostic decision making. After studying the subject area for about 10 years, I started to teach my own version to 3^rd and 4^th year medical students. It was focused on not mistaking a medical disorder for a psychiatric one.  It included a complete review of cognitive errors in that setting and how to prevent them. I taught that course for about 10 years.

There are a lot of ideas about psychiatrists and how they may or may not diagnose and treat medical disorders. Systematic biases affect the administrative and environmental systems where psychiatrists work.  Many psychiatrists are very comfortable at the interface of internal medicine or neurology and psychiatry. The most common bias about psychiatrists is that other medical conditions need to be “ruled out” before the patient is referred to a psychiatrist. From a psychiatric perspective the real day-to-day problems include inadequate assessment due to an inability to communicate with the patient and considerable medical comorbidity. Psychiatrists who work in those problem areas need to be competent in recognizing new medical diagnoses and making sure that their prescribed treatment does not adversely affect a person with pre-existing medical disorder.

Against that backdrop I decided to read 2 relatively new books. Both of them have the same title “How Doctors Think”. One book was written by Jerome Groopman, MD hematologist-oncologist by clinical specialty. The other book is written by Kathyrn Montgomery, PhD – a professor of Bioethics, Humanities, and Medicine. As might be expected from the writers’ qualifications Groopman is writing more from the standard perspective of a physician with an intense interest in medical decision making and Montgomery is describing the clinical process and analyzing it from the unique perspective of philosophy and the humanities. It follows that even though the titles are the same these are two very different books.

Groopman’s approach is to use a case-based style of looking at medical decision-making from the perspective of several clinicians-including his own work. The mistakes that occur are teaching moments and are explained from the perspective of heuristics or common cognitive biases. It is the approach I used in my course on preventing cognitive errors associated with psychiatric diagnoses. To cite one example, he describes an athletic forest ranger in his forties. The kind of a guy an internist might say: “I am not worried about his heart – he does his own stress test every day.”  He noticed increasing chest discomfort for a few days without any associated cardiopulmonary symptoms. He presented for an assessment on a day when the pain did not go away. He was seen and thoroughly examined.  There were no physical symptoms, exam findings, or laboratory finding to suggest a cardiac problem and he was released from the emergency department.  He returned a few days later with a myocardial infarction.  Discussions with the attending physician indicate that there were two issues associated with the missed diagnosis of cardiac chest pain – the generally healthy appearance of the patient and a lack of any positive tests indicating coronary artery disease.  Groopman discusses it from the perspective of representativeness bias (p 44) or being affected by a prototype – in this case the patient’s apparent level of fitness and attributing the chest pain to musculoskeletal pain rather than pain of cardiac origin. 

This case also allowed for a discussion of attribution errors especially if the patient fits a negative stereotype.  In the next case, a 70 yr old patient with alcohol use presents with and enlarged nodular liver on exam.  The presumptive diagnosis is alcoholic cirrhosis and the team’s plan was to discharge him back home as soon as possible. Closer examination confirmed that the patient was not drinking that much and searching for other causes of liver disease resulted in a diagnosis of Wilson’s disease.  For most of the book, Groopman uses this technique to illustrate substantial errors, the kind of cognitive bias that it reflects, and corrective action. The reality of “making mistakes on living people” comes though.

He recognized the importance of pattern matching and pattern recognition in clinical practice. There is an initial conversation with a physician that collapses pattern recognition to stereotypes and their associated shortcomings.  He elaborates on the concept and quotes a cognitive scientist to illustrate that pattern recognition may not require any conscious reasoning at all.  An expert can arrive at a diagnosis in about 20 seconds that may take a medical student or resident 30 minutes. Experts begin collecting information about the patient on contact and are immediately considering diagnostic possibilities. I have personally had this experience many times, typically for acute neurological syndromes (strokes, cerebral edema, encephalitis, meningitis) in patients who were referred for me to see in a hospital setting. Pattern matching clearly occurs in the diagnostic process, but it is more difficult to write about and discuss than verbal reasoning.

A major strength of the book is a fairly detailed look at uncertainty in medicine. The diagnoses are not etched in stone and no outcomes are guaranteed based on the accuracy of the diagnosis or not. He introduces a pediatric cardiologist who advances the argument that most of his cases are novel and that there are no set guidelines for what he treats. Even more complicated is that fact that what may appear to be sound science-based treatments like closing an atrial septal defect with a 2:1 shunt in kids it can be an illusion.  Many of those children do well without the surgery and many have had unnecessary surgery. The cardiologist also points out that study of this kind of problem is impossible because of the length of time it would take to do a randomized study.

Another major strength is advice to patients about how to keep the doctor they are seeing thinking about their case.  Numerous examples are given ranging from seeing large number of healthy patients where abnormalities are rare to seeing patients with real problems who have been stereotyped for one reason or another. Groopman is very specific in coaching prospective patients in how to overcome some of the associated biases.  This advice centers on the fact that biological systems are complex and don’t necessarily support logical deductions.  The astute doctor needs to be systematic, evaluate the data for themselves including the elicitation or more history, and question their first impressions. The patient aware of these limitations can ask the correct questions along the way to assist their physician in staying on track. He advises the patient to express their concern about the worst-case scenario to get that out there for discussion and to keep their doctor focused.  The patient is informed of how their history, review of systems and exam may need to be repeated along with some tests that have been previously done. The physician may have to ignore common aphorisms or maxims that are designed to focus on common problems and consider the complex – like more than one diagnosis being suggested. Business management of the medical encounter is seen to impair and obstruct this interactive process.

Groopman’s book is very good both as a guide to patients and a review for physicians who have been educated in diagnostic thinking. In the body of the book technical jargon is avoided and the case scenarios thoroughly explained. There is an excellent list of references and annotations for each chapter at the end of the book. 

How Doctors Think by Kathryn Montgomery takes the unexpected form of a philosophical argument against medicine as a science. She qualifies her criticism by being very clear that she is considering Newtonian or positivist science and not biological science. She recognizes several features of biological science that make it an integral part of medicine, but also not at all like the criteria for science that she sets as the premise for her argument. This is problematic at two levels. First, deterministic and reductionist physicists like Sabine Hossenfelder are very clear that everything is reducible to known subatomic particles and that particles in a brain are deterministic.

“Biology can be reduced to chemistry, chemistry can be reduced to atomic physics, and atoms are made of elementary particles like electrons, quarks, and gluons.” (5)

So for at least some scientists – reductionism is not a problem and the boundaries are not very clear between physical science, biology, and medicine.  Second, it is now known that biological organisms have a wide array of stochastic mechanisms that by virtue of their own nature produce apparently random results. With that range of possibilities, it is not very clear if the standards of physical science are that much different than the biological science necessary for medicine.

Montgomery makes the argument about science and the damage that the idea of medicine as science does to both medicine and its practitioners at several levels.  First, she describes science in medical training. Medical students encounter the basic science curriculum in the first two years of medical school. It is not physical science but biological sciences relevant to understanding pathophysiology, pharmacology, and epidemiology/evidence-based medicine.  She suggests this exposure to science is less relevant as the student transitions to a clinician with adequate clinical judgment – almost to the point that the basic science is an afterthought. This aspect of training is also used to point out that medical students are not being trained as scientists and the remainder of their formal education is spent learning clinical judgement.  At places she describes the preclinical years as fairly bleak period of memorization peripherally related to clinical development.  Second, the uncertainty of biology and medicine is part of her argument.  She extends the argument from the patient side to the side of the doctor. Patients want and need certainty and therefore they want doctors who are schooled in the best possible science who can provide it. Patients want an answer and all they get is statistics. Third, she suggests that the moral and habitual practice of medicine although dependent on human biology and the associated technical advances is not really science.  Physicians are taught to practice medicine and the don’t question “the status of its knowledge” (p. 191). She describes medical practice as a set of rational procedures that are shared with many other professions in the humanities and social sciences.  Fourth, the notion of medicine as a science is “clinically useful” in that it reassures the patients that physicians are engaged in a rational process like they were taught in science classes rather than a contextual, interpretive, narrative process used by non-scientists.  She cites numerous examples of maxims and aphorisms used in medicine to guide this process like Peabody’s famous: “The secret of the care of the patient is in caring for the patient.” 

 Montgomery’s writing is as sophisticated as you might expect from a bioethics professor with a doctorate in English and extensive exposure to medical training. Her critique depends a lot on verbal reasoning and the application of that model to numerous disciplines. Philosophical critiques of medicine and psychiatry that I have responded to in the past are typically presented as arguments with the premises being set by the author. As I read through these arguments being repeated across chapters there were clear points of disagreement.  Here is a short list:

1:  The argument about medicine not being a physical science – that is a good starting point if you want to be able to attack the scientific aspects of medicine, but does anyone really accept that premise? No physical science is taught in the basic science years of medicine.  The basic sciences are focused on human anatomy and physiology. An associated argument is that biological sciences have no overriding laws like physics and that is given as further evidence that medicine is not a science. There is an entire range of science within the basic science of medicine that cannot be explained by physical science but it is necessary for clinical medicine and innovation in medicine.  Finally science is a process that is subject to ongoing verification. That is as true for biological science as it is for physical sciences. While there appear to not be as many absolutes for biology progress is undeniable even within the boundaries of medicine.

2:  Uncertainty in biological systems and medicine - the author makes it seem like defining medicine as a science gives the false impression of certainty. I don’t think that certainty is misrepresented or minimized in clinical medicine.  Every physician I know experiences the uncertainty during informed consent and prognosis discussions. It is built into surgical consent forms and in situations involving medical treatment or testing – the discussions are even more complex. In a typical day, I will advise patients on side effects that occur at rates varying from 4 out of 10 patients to 1 out of 50,000 and tell them what to look for and when to call me.  I have had patients tell me after those discussions that they would prefer not to take a medication or do the recommended testing. I will also discuss life threatening problems with patients, and let them know I cannot predict outcomes but can advise them on how to reduce risk. The only way medicine can practiced is by having appropriate informed consent discussions that fully acknowledge uncertainty and the associated biological heterogeneity.  From the patient side, everyone has a friend, acquaintance, or family member who was healthy until the day there were not. The uncertainty of physical health and medical outcomes at that point are widely known by the general public.

An additional and lesser known aspect of the effect of uncertainty on physician behavior is encouraging the correct answer or treatment as soon as possible. Montgomery attributes some of this to the moral dimension of the physician-patient relationship and doing the right thing for the patient.  But a critical part of uncertainty is that physicians eventually learn to project their decisions out into the future. Those projections are all taken into account in developing the current treatment plan. The outcome of an idealized plan can be viewed as the direct result of the uncertainties involved.  

3:  Physician detachment is a likely consequence of characterizing medicine as a science – At points Montgomery makes the point that physician can emotionally protect themselves by assuming the detached rationality of science. It follows that abandoning medicine as a science would result in a more realistic emotional connection with patients. She has a detailed discussion of the physician-patient relationship being more as a friend or a neighbor.  She concludes that neighborliness has a number of virtues to recommend it as the relationship for the 21^st century. Two concepts from psychiatry are omitted from this discussion – empathy and boundaries. Empathy is a technical skill that is typically taught to physicians in their first interviewing courses in the first year of medical school.  It is a technical skill that allows for a more complete understanding of the patient’s emotional and cognitive predicament. In my experience what patients are looking for is a physician who understands them. That is generally not available from a friend or neighbor.  The basic boundary issue is that it is very difficult to provide care to a person who is emotionally involved with the physician. There are degrees of involvement, but any degree is important. A physician who is empathic, had a clear awareness of the relevant boundaries, and has a solid alliance with the patient is far from detached.  But I would not see them as neighborly or a friend.  The physicians job is the be in a position where they can provide the best possible medical advice. That can only happens from a neutral position where they can give a patient the same advice they would give anybody else.  That also does not mean that physicians are not emotionally affect when bad things happen to their patients or when their patients die.

4:  Do ancient Greek concepts still apply? – The author uses Aristotelian definitions of episteme and phronesis several times throughout the text. Episteme is scientific reasoning and phronesis is practical reasoning.  Aristotle’s view was that since there are no “fixed and invariable answers” to questions about health, every question must be considered an individual case.  In those cases, practical reasoning that considers context and additional factors or phronesis applies.  That allows the author to compare medicine to a number of social science disciplines that use the same kind of reasoning.  The question needs to be asked: “What would Aristotle conclude today?”  In ancient Greece there were basically no good medical treatments and medical theory was extremely primitive. Over the intervening centuries medicine has become a lot less imperfect. Uncertainty clearly exists, but the scientific advances are undeniable.  It is possible to say today that there are now fixed and invariable answers to large populations of people. Medicine has always been a collection of probability statements – but those probabilities in terms of successful outcomes have significantly improved.  One the corollaries of  Aristotle’s work is that there can be “no science of individuals” and yet the current goal is individualized or personalized medicine.

5:  Is science relevant to clinicians on a day-to-day basis? -  I think that it is.  I have certainly spent hours and even entire weekends researching patient related problems to find the best solution to a problem and to be absolutely sure that my recommended course of treatment would not harm the patient. All of that reading was basic or clinical science.  On the same day that I received Montgomery’s book, I got my weekly copy of the New England Journal of Medicine.  I have been a subscriber since my first year of medical school based on the recommendation of my biochemistry professor. Our biochemistry class was designed around research seminars where we read and critiqued basic science research. There was also the assumption that you were reading the text cover to cover and attending all of the lectures.  He encouraged all of us to keep up on the science of medicine by continuing to read the NEJM and in retrospect it was a great idea.  In that edition I turned to the Case records of the MGH (6): An 81-Year-Old Man with Cough, Fever, and Shortness of Breath. It was a detailed discussion by an Internist about the presentation and differential diagnosis of the problem. And there on page 2336 was a diagram of the ventilation perfusion mismatch that occurs with a pulmonary embolism and acute respiratory distress syndrome. I have seen this science at the bedside in many clinical settings.  

The clinical competency of pattern matching, pattern recognition, and pattern completion is left out of Montgomery’s description of how doctors think and it is an important omission.  It is a good example of non-verbal and unconscious reasoning that can be a critical part of the process. The answer to the question: “Is this patient critically ill?” and the triage that follows depends on it.  Pattern matching is also experience dependent with experts in their respective fields being able to more rapidly diagnose and classify problems that physicians who are not experts. Biases affecting verbal reasoning can negatively impact the diagnostic process, but so can the lack of experience in seeing patterns of illness and an inadequate number of cases in a particular specialty.

I consider both of these books to be good reads, especially if you are a physician and have had no exposure to thinking about the diagnostic process.  Both authors have their own ideas about what occurs and there is a lot of overlap. Both authors have the goal of stimulating discussion and analysis of how physicians think and educating the general public about it. Physicians will probably find Groopman a faster and more relatable text. Physicians may find the references and vocabulary used in Montgomery to be less recognizable. I would encourage any physician who is responding to initiatives to change the medical curriculum or critique it to read Montgomery’s book and work through her criticisms.  Both books have excellent references and annotations listed by the chapter for further reading. Non-physicians especially patients who are working with physicians on difficult problems may benefit from Groopman’s tips on how to keep those conversations focused and relevant.  As a psychiatrist who is sensitive to attacks (even philosophical ones) from many places – you may find my criticism of Montgomery’s work to be too rigorous. I tried to keep that criticism down to a level that could be contained in a blog post.  I encourage a reading of her book and formulating your own opinions. It is an excellent scholarly work.

Finally, the area of expertise in medicine and the associated clinical judgment of experts is still a current research topic.  The research has gone from basic experiments about who can properly diagnose a rash or diabetic retinopathy to a clear look at brain systems responding during that process. Those changes have occurred over the past 30 years. At the descriptive level it remains important to be aware of the possible cognitive biases and what can be done to overcome them.

 

George Dawson, MD, DFAPA

 

References:

1:  Groopman J.  How Doctors Think. Houghton Mifflin Company, New York, 2008.

2:  Montgomery K.  How Doctors Think. Oxford University Press, New York, 2006.

3:  Kassirer JP, Kopelman RI.  Learning Clinical Reasoning. Williams and Wilkens, Baltimore, 1991.

4:  Sox HC, Blat MA, Higgins MC, Marton KI.  Medical Decision Making. Butterworths, Boston, 1988.

5:  Hossenfelder S.  The End of Reductionism Could Be Nigh. Or Not.  Nautilus June 18,2021 (accessed on June 18, 2021) https://nautil.us/blog/the-end-of-reductionism-could-be-nigh-or-not

6:  Hibbert KA, Goiffon RJ, Fogerty AE. Case 18-2021: An 81-Year-Old Man with Cough, Fever, and Shortness of Breath. N Engl J Med. 2021 Jun 17;384(24):2332-2340. doi: 10.1056/NEJMcpc2100283. PMID: 34133863.

 

Layered Psychiatry

 

I had this idea about how to present the complexity of the psychiatric diagnostic and treatment process.  After putting up a couple of diagrams for comment, I went ahead with a PowerPoint. For about 15 years I taught a course in how not to mistake a medical diagnosis for a psychiatric diagnosis.  My audience at the time was 3rd and 4th year medical students.  The lecture included a discussion of the research at the time in pattern matching and pattern completion, heuristics and common biases, Bayesian considerations, and inductive reasoning. It was generally well received but really cannot be appreciated until you are a senior clinician.  Over the time since I taught that course there also seems to be a distinct bias toward considering DSM criteria to be the basis for psychiatric diagnosis and decision making – and that is clearly a mistake.

The very first time I really became aware of the importance of pattern matching occurred when I was a fourth-year medical student.  I was on an Infectious Disease rotation and my job was to get the consults for the day, go out and see the patients we would be rounding on, do my basic compulsive medical student work up and present the findings and my ideas about the case to the attending physicians. ID docs are very bright people and like most impressive rotations I contemplated becoming an ID specialist for a while.  My patient that day had spontaneous bacterial peritonitis and the question for us was: “Do you agree with the diagnosis and current antibiotic treatment?”  I met with the patient, took a complete history, did a physical exam, reviewed the hospital course and labs, and had time for a little research. At the time I was carrying a copy of Phantom Notes for Medicine – basically an outline of the major medicine text of the day. I looked up the differential diagnosis.  I was also carrying a copy of Sanford’s guide to antibacterial therapy – the 1982 version and looked up the recommended antibiotics for peritonitis.  I was all set for rounds at that point.

Both of our ID attendings were very serious physicians. There was not a lot of banter or joking.  I anticipated presenting all of the dry facts and either getting a brief agreement, some questioning until I could no longer answer, or a long discussion of the diagnosis and treatment.  In this case the attending came into the patient’s room. He was 15 feet away from the patient and he said: “What am I seeing from right here that is a potential problem?”  Our team consisting of the ID fellow, two Internal Medicine residents, and myself – stopped in our tracks.  Nobody had an answer.  Weren’t we here for peritonitis?  How can you diagnose that from across the room?

“What is wrong with the patient’s shin?” Dr. R stated looking as serious as usual.  Sure enough there was a light pink confluent rash covering about 10 square inches of the patient’s left shin area. Dr. R happened to be an expert in streptococcal infections. He rattled off the type of strep he expected and suggested that we get a culture and send it to his lab for confirmation. I completed my presentation.  The primary diagnosis and treatment by the medicine team did not change, but now there was a new diagnosis and treatment that depended on Dr. R’s ability to recognize the pattern of this rash and make a rapid diagnosis – even though he was not expecting it.  But beyond that – we all saw the rash (although we had to be prompted to see it). Dr. R not only saw it, he processed it as a unique rash, and then a rash most likely caused by a specific kind of streptococcal bacteria. And over the next several days he was proven correct by the culture result.

Pattern matching and pattern completion are critical skills acquired by clinicians over the course of their training and careers that allows for not only more rapid diagnosis and treatment but also more accuracy in classifying ambiguous cases. Some of the examples I used in my course included ophthalmologists compared with primary care physicians diagnosing diabetic retinopathy and dermatologists compared with primary care physicians across a series of rashes.  In both cases the specialists had a higher degree of accuracy and were better at diagnosing ambiguous cases.

Cognitive neuroscience encompasses a broad range of perceptual studies starting with the early studies of visual processing by Hubel and Wiesel to more recent studies that look at the encoding that occurs in perceptual systems and what level of processing occurs at the level of primary sensory and association cortices, what the higher-level cortical structures may be, and whether or not top down processing influences perception. According to Superior Pattern Processing (SPP) theory (3), both perceived and mentally constructed patterns are processed by encoding and integration and at that point can be used for decision making or transferring approximations to other individuals.  In my example, Dr. R not only sees the pattern of the rash, but it is integrated into a feature set that has a time, visuospatial, social, and emotional context that makes it more likely that he will make a correct diagnosis. Experimental data suggests that he is not seeing the rash like any other person in the room – largely as a function of top-down control of his perceptual process.  The actual transfer of this pattern to his junior colleagues is limited because they see the rash as being a universal truth – that is they just “missed it” and therefore need to memorize what this rash looks like and not let it happen again.  They are also unaware of the processes involved in pattern matching or processing or they might have asked him about it.  For example, a logical question would have been: “What features of this rash do you notice that are suggestive of strep or a specific kind of strep?”

The question of what represents a pattern is critical to the idea of pattern recognition and processing.  There is a natural tendency to associate the term with visual or auditory stimuli, but without too much imagining patterns can clearly exist in any sensory modality and often involves the integration of multiple sensory inputs.  Cortical organization generally reflects primary sensory input to the cortex with adjacent sensory association areas and further information flow to heteromodal areas in the frontal and temporal cortex where additional integration occurs. Patterns can be sensed, encoded, recognized encoded and processed across theses systems.  The resulting integration yields a very complex array of patterns that are not intuitive.  For example, Mattson suggests that pattern processing in the human brain forms the basis of human intellect including problem solving, language and abstract thought and that it includes fabricated patterns.  Those fabricated patterns allow vicarious problems solving without having to conduct real world experiments.  The recent cognitive neuroscience of pattern processing is a significant advance compared with the old diagnostic paradigms I taught 20 years ago.  Those old experiments were basically a comparison of a non-expert to an expert diagnostician focused on a relatively basic clinical problem like a pathology slide, x-ray, ECG, or physical finding and the results were not a surprise – the experts typically prevailed in both accuracy and speed.  The sheer amount of information in a clinical encounter looks at what is essentially an infinite array of patterns, including patterns that are generally not even mentioned as being clinically relevant.

In considering what kind of patterns that need to be recognized and processed by a psychiatrist – the patterns that exist in clinical practice are a starting point.  These patterns and the associated phenomenology have been grossly oversimplified by an overemphasis on nosology. I talk with far too many people who see psychiatric diagnoses as phrases on a page in the DSM. I cringe when I hear: “The patient does or does not meet criteria for (DSM diagnosis x)”.  Kendler was correct when he referred to the DSM approach as an indexing system.  It gets people into the same ballpark, but it is not be very useful for predicting response to treatment or that specific person’s response to being ill.  It is also based on a fraction of the information collected in a psychiatric evaluation. When I consider the feature sets that psychiatrists are considering in evaluations it may look something the graphic below.  Of course, these features sets are simplified for the purpose of making a useful graphic. They will vary with the individual, their experience, social context, and culture. They will also be blended across space and have their own individual levels of integration and patterning.  Let me provide a couple of examples to illustrate these points.

Consider the above diagram as representing the possible features that must be recognized in order to assess a patient presenting to a psychiatrist and formulating and optimal diagnostic and treatment plan. My overriding concern in the first few minutes of the evaluation is whether this person really has a psychiatric disorder or a misdiagnosed medical problem and as a corollary - are they medically stable? That sounds like a basic consideration but prioritizing it is not listed anywhere in the DSM or any medical text that I know about. It does involve rapid recognition of patterns of acute medical illness particularly the most likely patterns to be misdiagnosed as psychiatric disorders and what I am seeing in real time.  It also involves pattern recognition of the thousands of psychiatric presentations that I have see that were really medical disorders.  Real life examples have included an almost immediate recognition that the patient had a stroke (many cases), seizures (many cases), meningitis, encephalitis, cerebral edema, serotonin syndrome, and neuroleptic malignant syndrome.  These rapid diagnoses were all predicated on experience-based pattern recognition rather than written criteria and these diagnoses had nothing to do with the DSM at the time.

A more cross-cutting feature in the diagram would be transference issues and defenses that can arise as soon as the initial evaluation or be indirectly evident by the patients historical description of their relationships with important people in their life.  These patterns will involve several layers in the above diagram and most importantly may suggest a psychotherapeutic intervention that can be implemented as early as the original assessment.  A similar process occurs if the patient is describing features of a major medication responsive illness.  In that situation, features from multiple layers result in a pattern that may be recognizable to the psychiatrist in terms of specific medical treatments or the urgency of those treatments.

And finally - what might the graphical representations of these pattern matching processes be?  Here are a few examples.  In the case of psychotherapeutic examples, it will depend on the exposure to specific therapies in training and practice. Each therapy has a specific pattern or series of patterns that the therapy depends up as well as patterns more specific to the conduct of therapy.  These graphics contain critical books from my library shelves with those elements.  In the case of the diagnostic and treatment process - the school of therapy and potential application are important patterns to recognize in the initial assessment.

All of these books contain symbolic representations of clinical patterns in the form of vignettes designed to assist the student of psychotherapy in learning techniques. They also contain information about the patterns of intervention that are relevant for a specific therapy and in some cases the common factors required in all successful therapies. I have graphically represented what happens in pattern processing once a theme is noted in the clinical assessment of the patient.  Clinical teaching of this process is often problem identification followed by an algorithm of features that might predict a successful course of therapy or limitations in therapy based on the students knowledge level at the time. As is true for most pattern matching and processing, the more extensive a physician's previous pattern exposure - the more likely they are to match the optimal intervention to the problem. 

I will resist making this first post of the New Year too long and wrap it up at this point with a diagram that I think pulls it all together (see below).  Each layer of this diagram consists of patterns and all of the associated pattern processing that leads to psychiatric diagnosis, formulation and treatment.  A few of the key features include the fact that diagnosis and treatment are interchangeable processes.  There will be times even during the initial information gathering that a verbal treatment intervention needs to occur and the entire interview occurs in the context of empathy and what Ghaemi, et al (4) have described as an existential psychotherapy based encounter – even if the administrative focus is on pharmacology. A second feature is that the information exchange is necessarily large if the psychiatrist and the patient are capable of it. There has been no research that I am aware of on the optimal amount of information that is required, but there are many limitations.  The advent of the electronic health record for example has led to the universal use of templates that are very restrictive in terms of information, typically dichotomous responses. A third implicit feature is the concept of patterns, what they imply for diagnosis and decision making and how there is almost a complete lack of discussion about this process in an era where diagnoses seem to have collapsed to a brief list of bullet points.  Cognitive neuroscience is a critical area of research focused these processes that I first became aware of when reading Kandel’s book “The Age of Insight” (5).  It is an area that does not typically get a lot of attention from psychiatrists, but it is a logical extension of the work done by behavioral neurologists from 20 years ago.  If we really want to focus on how psychiatrists think about diagnosis and treatment – we need to study this field, especially as the experiments get more complex.

I will wrap up this post at this point with the hope that 2021 is a much better year and that mankind is able to put this pandemic virus behind us by the summer and approach future pandemics with more science and wisdom.

 

Happy New Year!

George Dawson, MD, DFAPA

 

References:

1:  Constantine-Paton M. Pioneers of cortical plasticity: six classic papers by Wiesel and Hubel. J Neurophysiol. 2008 Jun;99(6):2741-4. doi: 10.1152/jn.00061.2008. Epub 2008 Jan 23. PMID: 18216235.

2: Poirier CC, De Volder AG, Tranduy D, Scheiber C. Neural changes in the ventral and dorsal visual streams during pattern recognition learning. Neurobiol Learn Mem. 2006 Jan;85(1):36-43. doi: 10.1016/j.nlm.2005.08.006. Epub 2005 Sep 22. PMID: 16183306.

3:  Mattson MP. Superior pattern processing is the essence of the evolved human brain. Front Neurosci. 2014 Aug 22;8:265. doi: 10.3389/fnins.2014.00265. PMID: 25202234; PMCID: PMC4141622.

4:  Ghaemi SN, Glick ID, Ellison JM. A Commentary on Existential Psychopharmacologic Clinical Practice: Advocating a Humanistic Approach to the "Med Check". J Clin Psychiatry. 2018 Apr 24;79(4):18ac12177. doi: 10.4088/JCP.18ac12177. PMID: 29701934.

5:  Kandel ER.  The Age of Insight. Random House, New York, 2012.

Graphics:

All generated by me for a PowerPoint presentation by the same name.  The photo at the top are two pamphlets that I carried as a med student along with a copy of Phantom Notes.  I was carrying them when I was in the room with Dr. R as he made the diagnosis described above.  I would not trade my medical school experience for anything. 

Telepsychiatry - Day One

These are strange times.

For the past several years I have attended seminars on telepsychiatry. In Minnesota, we have an expert who has been doing it for a long time. He talks about the advantages of being an independent practitioner and using your own equipment rather than being a subcontractor. He has a definite method that includes seeing all of his patients in person at least once a year. His practice covers a large area that would otherwise probably not have a lot of psychiatric services. Over the years that I have been going to the seminars, I have thought about private practice and Telepsychiatry. I even looked at a storefront building at a shopping mall and fantasized about starting it up. But I am too close to retirement and there is a thing called tail coverage. That means if you carry malpractice insurance and retire you need to still pay the premium for two or three years into retirement in the event that you are sued.  That was a major deterrent and it seemed like I would just carry on in my current position until I decided to quit. And then the coronavirus and social isolation hit.

The transition to Telepsychiatry rapidly happened last Thursday. I was going about my day when my younger colleague told me that she was switching to Telepsychiatry this week. She encouraged me to get on board. Several people were critical to the effort and I was up to speed on the system by this weekend. I had to confirm that I had the computer power, bandwidth, and dropped frame rate consistent with software. I pulled up my schedule this morning and the main difference I was sitting at home looking at it on my big Mac Pro. I tested the camera and microphone. It produced a good image of me sitting in my home office and I was ready to go. What followed was a big glitch and some realizations about the visual aspects of psychiatry.

An initial series of emails let me know that the visual feed was not working. That essentially took out the software and as a replacement I was supposed to do telephone interviews. Hoping that they could get it up and running I moved the first patient new evaluation to the last slot in the daytime. There was some suggestion that only follow-ups should be seen as telephone interviews. I was concerned that patients would have to hold the telephone receiver for 45 to 60 minutes but was reassured that it would all happen over speakerphone. The locations were all secure and managed by our clinical administrator. When it became apparent that the visual feed would not occur I started doing new assessments and follow-ups strictly on the telephone.

In retrospect I found myself myself in an ironic position. For years I studied telephone switching both as a high-tech investor and as electronics hobbyist. I eventually got involved in communications theory. The engineering version of communications theory is highly technical and interesting but I have never been able to apply it to the clinical interview. The clinical interview is an exchange of information. There is always a certain noise level that varies significantly from person to person. That noise can occur strictly on the information being exchanged or various emotional levels that can add or subtract from the overall noise level. A good example would be a person who brings a lot of biases into the interview. As an example, I have had people slow the interview down or bring it to a halt just based on my physical appearance and how it was interpreted. Some of those people would be very explicit in telling me they could only work with a psychiatrist who had a certain religion, philosophical bent, or political affiliation. There was often speculation, that I did not meet the preferred categories. Interviews done without the visual channel, removed those factors.

I dictate all my valuations and follow-ups and have done that most of my career. Critical parts of what has become known as the Mental Status Exam are dependent visual assessment. A few examples of common bullet points include:

Appearance: I comment on whether the person appears to be alert, interactive, their overall grooming and hygiene, their eye contact and social demeanor. Where it applies I also comment on whether they appear to be intoxicated, distracted, potentially delirious, and in some situations whether they realize I am in the room with them.

Psychomotor: Hyperkinetic and hypokinetic movements and possible movement disorders need to be described. Psychomotor agitation and retardation as well as motor restlessness also need to be commented on. It is about a 40 foot walk to my office and the person’s gait also needs to be described.  Gait analysis is useful because of the association with dementias, neurological disorders, and medication side effects. It is also useful in assessing chronic pain patients. The commonest acute pain disorder I notice is gout due to its high prevalence in men of all ages. Specific movements require additional examination in some cases rating scales. For example if tardive dyskinesia is noted and AIMS (Abnormal Involuntary Movement Scale) can be done to determine a baseline score. There are additional rating scales for Parkinson’s, akathisia, tics, and dystonia.

Affect: Psychopathologists like Sims have pointed out the subtle differences between affect and mood. In his text for example he describes affect as “differentiated specific feelings directed toward objects”. Mood is described as “a more prolonged prevailing state or disposition”. He comments that both terms are used “more or less interchangeably”. Modern use is much more basic and it has to do with direct observation of the patient’s emotional expression, the specific context, and whether or not it may be consistent with an underlying phasic mood disturbance.  A common error I notice in many descriptions is that the time domain is omitted - people never seem to comment on the affective state over the course of the interview or the fact that the patient's affect appears to be completely normal - despite the assessment being done for a mood or anxiety disorder.

If you are interviewing people by telephone rather than Telepsychiatry, you don’t have access to any of those three critical domains as well as other parameters that might be important. For example, vital signs, focal physical examination, and the overall determination about whether or not a patient may be physically ill or critically ill just based on their appearance.  There is also a pattern matching aspect to psychiatric diagnosis. After psychiatrist has evaluated hundreds or thousands of patients, certain patterns are evident that can facilitate diagnosis. The most obvious one is delirium. It has always been a mystery to me why that diagnosis is so difficult for a lot of people to make. Once you have seen a few delirious people, the pattern seems obvious. Other findings are much more subtle. An example might be a patient appears to be in pain but also does not want to disclose the source of that pain. It could be a self-inflicted injury or injury from intimate partner violence. Those findings would be very difficult to pick up over a telephone interview.

A couple of examples come to mind when I think of critically ill patients who did not come to see me because they were critically ill. The first was a patient who looked the whitest I had ever seen a person. I asked him if he was physically ill and he denied it. I asked him about possible causes of blood loss and that was also denied. He did eventually allow me to order a complete blood count. I got the results back his hemoglobin was extremely low and when I called him - he did acknowledged some symptoms of G.I. blood loss and agreed to go to the emergency department. He was subsequently found to have a gastric ulcer. In another case I was talking with the patient appeared to be physically ill. He seemed to have some abdominal distress. He allowed for a limited exam of his abdomen and appeared to have right upper quadrant pain and tenderness. He was also referred to the emergency department and had acute cholecystitis and required surgery. Both of these scenarios depend on how the patient actually looks to the psychiatrist and that is why the visual presentation is so important.

Many people think that psychiatry is an exchange of words. A common myth these days is that these words allow people to be grouped into diagnoses based on other sentences and phrases. A discussion between two people is always much more than that. When a psychiatrist is in the room the discussion is between two people one of whom has memories of tens of thousands of important patterns and findings that mean something. A significant number of those patterns are visual rather than strictly verbal.

I have lost count about how many times a rapid visual diagnosis played a critical part in the diagnostic process. When I see a patient with serotonin syndrome or neuroleptic malignant syndrome or malignant catatonia - I am not running down the diagnostic criteria in my head. I am thinking that they are critically ill probably have a specific diagnosis - but I have to get them somewhere fast where they can receive the necessary supportive care while that diagnosis is clarified and treated. Most of that is a visual process based on what I have seen in the past. In most cases, the diagnosis occurs in seconds to minutes.

I thought the telephone interviews went well. My notetaking was as intense as ever. I am looking at an average of about six pages of handwritten notes that I base my dictations on. But I know the process can be much better. Telepsychiatry is superior to telephone psychiatry, and I hope to find out how close it is to a face-to-face interview.

Hopefully that visual feed will be there tomorrow.

George Dawson, MD, DFAPA

They Don't Even Know What They Are Seeing.......

I was walking back from a meeting with a psychiatric colleague the other day.  There was the usual grousing about the practice environment and miscommunication and she made the following observation about why physicians and psychiatrists don't get the information they need.  She pointed out that in many cases the nonphysician  observers: "Don't even know what they are seeing."  If you are counting on people for observational data and that is true - that is a setup up for suboptimal care at the minimum and a catastrophe at the worst.

Take the case of a very basic measurement - blood pressure and pulse.  Anyone taking those measurements should be aware of the guidelines and whether or not the patient has a baseline abnormality, condition that can affect either, or medication effect that leads to changes in the vital signs.  They should also be aware of the limitations of measurement.  All of the automatic blood pressure machines in the world will not be able to assess and treat the patient unless the operators know what the numbers mean.  They also need to know that one of the problems with single operator and strictly machine operated approaches is that arrhythmias are problematic even if the blood pressure is fine.  There have been situations where I had to put together a continuing education course on blood pressure and pulse and the correct assessment of both.  That was a long time before the recent article on common mistakes made by medical students in these measurements.

If measurements that are considered routine and done hundreds of times a day are problematic what about observations that occur on the other end of the spectrum.  A common health care myth today is: "If I have a checklist and check off all of the boxes on that list that will lead me to some kind of diagnosis."   That is probably a minimization of the myth.  In the case of psychiatry, the myth is more: "If I convert a standard psychiatric assessment into a form (or a checklist) - the ultimate product of going through that list will basically be a psychiatric evaluation and diagnosis."  Systems of care who use this approach can deny these myths as much as they want but I see this happening every day. Organized psychiatry and the DSM approach to diagnostic criteria is partially responsible, although the manual does say that it can't be used by anybody.  It doesn't say who specifically should use it and it does not suggest (like Kendler) that it is an indexing approach.

Looking at the graphic at the top of the page illustrates why a form or a checklist does not suffice.  The observer/psychiatrist in the drawing is doing more than asking the subject a series of yes or no questions.  The psychiatrist is looking for patterns in symptoms (medical and psychiatric), what is happening in relationships with the person (including the relationship to the psychiatrist), and the person's conscious state - specifically whether there has been a departure from baseline.  There is often a balance between historical detail, phenomenology, the person's ability to describe what has happened and a plausible scenario based on probability estimates from the psychiatrist's previous experience.  Any psychiatrist who has been trained in many presentations of complex psychiatric illness is more likely to see those patterns than somebody who has not been.

To illustrate some of these concepts I will describe several cases that are all what non-psychiatrists (nonphysicians and other physicians) called hysteria. Hysteria is an old word that dies hard.  The DSM equivalent is histrionic personality disorder.  The generic use of the term suggests a person who is overly emotional, dramatic and attention seeking but there are 8 diagnostic criteria that are unchanged between DSM-IV and DSM-5.  Many clinicians opt for the term Cluster B - a DSM-IV originated term that grouped personality disorders in groups according to some common diagnostic features.  The Cluster B group included individuals that often appear dramatic, emotional, or erratic.  Those personality disorder diagnoses include antisocial, histrionic, narcissistic, and borderline.

The rule-in criteria (significant impact on life circumstances and onset when you expect a personality disorder to occur) and the rule-out criteria (not due to another mental or physical disorder) are predictable for any causal reader of a DSM and could be included on any checklist or form.  How does all of that play out?  Well here are a few examples:

Hysterical patient #1:   A 30 year old woman presents for a therapy intake.  She is mumbling and laughing.  The therapist describes her as "odd and having an odd affect."  She alludes to some suicidal behavior in the past but is smiling and joking about it.  The therapist has the impression that she is manipulative and overly dramatic.  He contacts the clinic psychiatrist and says that she is histrionic but he is concerned about her suicide potential.  The psychiatrist sees her that day and makes a diagnosis of bipolar disorder-mixed type with psychotic features.  The patient is eventually stabilized on lithium and an atypical antipsychotic.

Hysterical patient #2:  A 25 year old woman is being treated on a general medicine ward for dehydration from a respiratory infection.  She suddenly gets tearful and agitated.  Family members visiting have to physically restrain her when when she tries to get out of bed.  She starts to make very loud high pitched vocalizations.  A psychiatrist is called to go in to assess hysteria and possibly sedate the patient.  The psychiatrist sees an agitated young woman who is not able to respond coherently to any examination questions.  Brief neurological examination suggests increased intracranial pressure is the problem and the patient requires immediate transfer to a neurological intensive care unit. 

Hysterical patient #3:  A 58 year old man is referred acutely from a therapist for acute panic attacks and "probable Cluster B" personality traits.  He has recently retired due to osteoarthritis of the knees.  He had no earlier history of panic attacks but the therapist thought that he was overly dramatic at the initial session 2 days earlier when he was unable to relax and breathe normally with behavioral techniques that are usually effective.  The psychiatrist gets a history of the patient needing to abort an exercise stress test two weeks earlier due to the arthritis and having  a prolonged period of immobility at home due to sore knees. During that time he developed acute shortness of breath.  The episodes of anxiety that he described were secondary to shortness of breath and not panic attacks.  The psychiatrist sends the patient to the emergency department where an acute pulmonary embolism is diagnosed and he is admitted to the ICU.     

These are just a few examples restricted to one collection of psychiatric symptoms that illustrates what my colleague was referring to.  The value of psychiatric training goes far beyond what is in the DSM and what checklists and templates can be extracted from it.  I have never really met a psychiatrist who was focused on the DSM probably because it is implicitly evident to us that it is an index more than a diagnostic manual. We are focused on what is not in the DSM and as far as I know that is not well documented in many places.  Those are the patterns associated with clinical practice and that should have been gleaned along the way with medical training.  The DSM doesn't tell you how a pulmonary embolism presents. It is possible that you night have never seen one. But in medical training I can guarantee that it was discussed somewhere along the line in the differential diagnosis of dyspnea.  I can guarantee that one of those attendings discussed the phenomenon of the healthy young adult immobilized by air travel who gets off at their destination and suddenly has an acute pulmonary embolism. All of those features and urgencies should be in a physicians conscious state when they are seeing the whole patient and not some DSM/checklist version of a patient.

This brief post also illustrates the biasing effects of language.  What  does "Cluster B" really mean?  Aren't people who are acutely medically (or psychiatrically) ill dramatic, emotional, or erratic?  Hysteria is an extremely biasing term that over the centuries has been applied selectively to women rather than men.   The examples above illustrate that point.  If you are seeing the world through DSM language and that is your only lens - you are by definition not seeing the whole patient.  The list of possible errors in that landscape is very large.

There are a number of constraints that will get  in the way of a trained psychiatrist trying to see the whole patient.  Inadequate time is one, but time frames vary significantly.  Diagnosing a life threatening medical problem upon seeing a patient may take a matter of minutes and is clearly the most important diagnosis.  Seeing a long series of new patients briefly to prescribe treatment will necessarily mean that certain features in the above diagram will be missed.  So-called measurement based care depending on a large number of checklists to "quantitate" affects or other psychiatric states makes the same mistake.  Collaborative care where a psychiatrist looks at these rating scales and recommends treatments makes the same mistake.

The best assurance that the critical aspects of care will not be missed is to be sitting across the room from someone who has been taught all of the critical aspects of care.  That process is complex and as far as I know has never been adequately described.  A first approximation is whether that person knows what they are seeing and how to respond.

George Dawson, MD, DFAPA     

Physicians Replaced By Computers - Lessons From A Roomba

My Memorial Day project was purchasing a Roomba and getting it up and running.  I am a big believer that robots will make all of our lives easier at some point and decided now is the time to start walking the walk.  For those not familiar with the Roomba, it is designed as a robotic vacuum cleaner.  Once you have set up the rooms and programmed it, it is basically supposed to vacuum your floors automatically and then park itself in a docking station for charging.  The machine itself is about a 14 inch diameter disk that rises to a height of about 3 1/2 inches off the floor.  It is a light 8.4 pounds.  It is able to accommodate sharp angles with a secondary brush that spins on an arm that extends from under the main disk.  This combination of the main disk spinning and the extended spinning brush cleans the corners of a room.  I purchased the latest model, a Roomba 880 after consulting with friends and relatives who had earlier models.

One of the considerations in buying the Roomba was whether it would help turn my home into an even cleaner environment than it currently is.  That is a tough act.  One of my friends who is a physician gave his opinion that my home is "museum-like".  My office is probably the only problematic room with stacks of books and journals piled everywhere.  Disarray certainly but minimal dust.   My entire first level is hardwood flooring that is typically vacuumed with a built in system.  To its credit the Roomba contains all of the debris in the machine until it is emptied and all of the exhausted air is HEPA filtered to avoid exhausting any dust particles.  The main cleaning mechanism consists of two debris extractors that are rubberized bars that spin at a high rate of speed across the floor surface to capture dust, hair and larger particles.  But the most interesting aspect of the Roomba was going to be its observed behavior.  It has two modes when vacuuming.  It can start in a spot and spin increasing circles in an outward direction until gets to about a 3 foot diameter and then it spirals back in to the center spot.  In the more typical mode it heads to the room perimeter and then "automatically calculates the room size and cleaning time."  The most valuable tip in the manual was to take measures to restrict it to one room at a time and it comes with two Virtual Wall^®Lighthouse^TM devices that allow for easy demarcation of the work area.

The most fascinating aspect of getting started with the Roomba was going to be setting it up and watching how it went about the task of vacuuming.  I did some very minor room preparation, charged it up, and turned it loose.  As expected it headed straight for a wall and then attempted to establish the perimeter.  I remembered this as standard rodent behavior.  If you have ever confronted a mouse in an open area of your floor, their first move is to dash to the baseboards and run parallel to them to escape.  That strategy works well in the wild because the maneuver is associated with more cover and makes them less susceptible to predators.  It works much less well when confronted by a human who knows that it is their first move.  And yes, scientists have bred mice that do not exhibit this behavior.  My guess is that they would not fare well if they made it outside the lab.  The Roomba's behavior is less rigid than a typical mouse with some exceptions.  In the hour and 20 minutes it took to vacuum the adjacent kitchen and great room - it circled a kitchen island perfectly at least 10 times, but at the wall perimeters it was much less predictable.  At times the Roomba would peel off and take off across the room in a single pass or rarely return and continue along the original wall.  Sometimes it would head off the wall at a 45 degree angle and at other times 90 degrees.  There were never the usual adjacent passes that a human would make using a standard vacuum cleaner.

According to the literature,  the Roomba is supposed to "crisscross" the room in order to clean the floor.   I placed two small pieces of popcorn in the middle of a large section for flooring to use as markers of cleaning efficiency.  In the course of an hour, the Roomba passed these markers many times, sometimes very closely without vacuuming them up.  During that time it was very difficult to detect how much crisscrossing had occurred since mouse-like it spent the majority of the time in the periphery, bumping and spinning around walls and furniture.  It eventually did break free from the walls and set off on a 45 degree path picking up one popcorn fragment at about the one hour mark and the other at about one hour and ten minutes.  The old adage about pictures is true and I happened across this 30 minute time lapse photo of the Roomba working a room (with permission from the SIGNALTHEORIST web site).  It correlates well with my description of the actual paths.

As I surveyed the job afterwards, the floor was definitely clean and the warning light on the machine was saying that the dustbin was full.  When emptied, it contained an impressive amount of debris and dust relative to what seemed visible to the naked eye.  Another win for the robots?  Well, not really.  It is an interesting tool that I will continue to use and study, but in comparison with humans it is not efficient and at this point certainly not autonomous.  Despite all of the guidance in the manual the Roomba can still encounter unpredictable surfaces and get itself into trouble.  In my case it was the pedestal of a recliner.  The wood at the edge was about 3/4 of an inch high.  For some reason, the machine did not recognize it as an obstacle and continued to run up the base and get hung up.  A loud spoken error message would sound advising the human in the room to pick it up and start it in a new place.   The other concern is efficiency.  It spent far too much time in the perimeter and a low percentage of time covering the main floor areas.  That was tremendously inefficient.  It took at least 5 times longer to vacuum the main rooms than I would have if I was pushing a vacuum cleaner.  Even though it allows me to do other things, it says something about current state of available and affordable domestic robots.  They can't match the performance of humans on a fairly basic task.  This is an important concept.

Before any of the futurists out there jump on me for being a Luddite, let me disclose a few details.  I am a member of the IEEE and have been for the past 18 years.  I am currently a member of the IEEE Robotics and Automation Society.  I am not an electrical engineer and I have not designed or built any robots, but hope to start doing this when I retire from psychiatry.  I consider myself to be an expert in the human brain and the advantages it confers on humans over other animals and machines.  The Roomba is a basic case in point.  It cannot sense and adapt to novel conditions quickly enough to match a human doing the same task.  Even more striking is that although it is designed to vacuum homes and I have a fairly typical home with a better than average floor surface, it still encounters situations that exceed its response capacity.  In those situations it needs a human assist.  What is it about the human brain that leads to that kind of an advantage?  First and foremost, it is a rate of pattern matching and pattern completion capacity that allows us to recognize vacuuming problems, anticipate them and correct them by developing novel solutions even before the problem leads to a stop in action.  Some of this happens when a human goes around the room to set it up for the first time for the Roomba.  That human has made some assessment of the machines capacities and limitations and is problem solving for the machine before it is turned on.

Observing the limitation of the Roomba leads me to a point where I can address both the idea of computers replacing doctors and how that fits into the common anti-physician narrative in this country.  Is there a connection between the two?  My experience tells me that there is.  For nearly 30 years there has been a constant stream of antiphysician rhetoric.  The sources have been expected.  One of them is the key opinion leaders (KOLs) of the managed care industry.  I can recall reading one of the the first books written by one of them, a non-physician who was widely acclaimed as being an expert in managed care.  His early theory was that the high cost of health care was due to the decisions that physicians make.  But in the middle of the book he wrote what he thought of physician salaries and only grudgingly acknowledged that they should probably be paid a good wage due to their education.   I have posted here many times my experience at a managed care conference in the 1990s.  The speakers at that conference were very clear that the explicit agenda of their industry was to replace all of the specialists with primary care physicians.  The examples given were orthopedic surgeons and psychiatrists.  When a psychiatrist in the crowd pointed out the shortcomings of that philosophy - he was called a "whiner" by a Governor who was an anointed KOL in the industry.  Then the KOLs from the financial services industry started weighing in.  You could find glimpses of it while reading the investment literature.  People who were investors with no particular degree started saying that some day, physicians would get what they deserve - with the implication being that whatever that was - it was not good.  Any physician has experienced this prejudice.  The comments about how physicians are "expensive" as a rationalization for working them to death by not hiring any additional help.  Replacing physicians with computers seems like a logical extension of this rhetoric.  Googling this topic returns a number of provocative articles written from a point of view that is generally consistent with who the author is.

I know that some of those authors know the difference between a robot or a computer and a doctor, but it is also clear that some do not.  They certainly don't seem to understand that the real processing power of a human diagnostician's brain is in the area of pattern matching.  In order to duplicate that property with current technology, takes a massive computer and it is one of the reasons why my new $700 Roomba, although well designed - can easily be beaten by a human with a standard vacuum cleaner.  But the human advantage goes far beyond that.  Human diagnosticians do far more than match simple patterns.  They are able to complete fragments of patterns and anticipate what the whole pattern should be.  For example, is it likely that a depressed person is in this current state as the result of an inherited form of depression, their current state of detox from an opioid and/or benzodiazepine, current stressors or interpersonal conflicts, brain trauma, an undiagnosed medical condition, childhood adversity, psychological trauma as an adult, or defects in reasoning at either the emotional or cognitive levels.  Then there is the matter of acquiring all of the data to make the determinations.  Patterns upon patterns of data.  The Roomba-like approach would be to give the person a checklist of depressive symptoms and pretend that is all that needs to be known.  Checklists are already being administered by a computer and may be administered by robots someday.  

Yet it takes the pattern recognition, and several layers of it, as well as human experience dependent learning in order to make a real medical or psychiatric diagnosis.
        

George Dawson, MD, DFAPA






Supplementary 1:  The graphic at the top of this post is a photo that I shot of the inside of the box that my Roomba came in.

Supplementary 2:  I don't want to give the false impression that I do a lot of vacuuming.  My wife does practically all of it, but I am trying to do more especially if there is a high tech twist to it.  Some of the first robotics I hope to work on will be human controlled arms and hands designed to do yard work and move heavy objects around in the house.  I can't believe this is an area that has been ignored.