A DSM for Psychiatrists?

 

 

No matter what version - the DSM is clearly a flash point for criticism by psychiatrists and non-psychiatrists alike. There are too many diagnoses.  People don’t like certain diagnoses or complain when some categories are eliminated. There are endless debates about diagnostic criteria, reliability, and validity. Categories are a wrong approach and we need dimensions. Philosophers have a field day imagining what the DSM is and making suggestions.  In an early post on this blog, I responded to the philosophical suggestion that the DSM was supposed to be a blueprint for living. Antipsychiatrists have no problem rejecting the entire volume of course because they are stuck in the 1970s with Szasz and maintain that there are no mental illnesses. The more flexible antipsychiatrists reframe this into everyday problems in living another decades old formulation that did not stand the test of time. Others suggest that the DSM exists to make diagnoses that lead to pharmaceutical treatment and make profits for drug companies.  The more legitimate criticism from psychiatrists is focused on the criteria and whether any diagnostic categories exist. Some of that criticism comes full circle back to why a classification system was needed in the first place. Clinical psychiatrists tend to use a fraction of the available diagnoses and in most practices can recall the diagnostic codes without looking them up. In fact, most psychiatrists use the DSM as a reference, pulling it off the shelf for rarely encountered diagnoses and then typically to look up a diagnostic code for coding and billing purposes. 

The title Diagnostic and Statistical Manual – is the first clue about the original intent of the manual and it antedates the psychiatric profession and the APA in the United States by several decades. The abbreviated history is available on the APA web site and several other Internet sites.  Initially it was to determine numbers of people by diagnosis both in the varied mental illness facilities across the country and later in military service. This function was described as administrative but there was also a consensus building aspect in the early 20^th century as diagnoses shifted from a unitary psychosis model to more nuanced.  The advent of the DSM-III was a turning point because it provided atheoretical definitions of disorders that were subsequently adopted by the ICD-9. Subsequent revisions in the DSM-IV and DSM 5 included revisions based on professionals and professional organizations, assigned work groups and their research, and eventually the general public. The original goal of classification and statistics has remained but it is used for various reasons by non-psychiatrists.

There are many examples of non-psychiatric use.  In the legal and political sphere, most states have rationed services for people with severe mental illnesses who are at high risk for hospitalization and other morbidities. Qualifying for those benefits depends on a  DSM diagnosis.  The same is true for state sponsored services for autism and developmental disabilities. In forensic settings experts are called upon to give diagnoses in an adversarial setting.  Disability, veteran’s benefits, and worker’s compensation are all linked to diagnoses.  All medical billing to insurance companies and government payers depend on DSM equivalent diagnostic codes in the ICD-11. Managed care companies ration care based on many of these codes by refusing to cover them. None of these functions were designed as an original intent for the diagnostic manual.

Heterogeneity – either explicit or implicit is another frequent criticism of the manual. Human biology and the biology of diseases and disorders teaches us that the etiopathogenesis of illnesses is diverse. There are many possible underlying biological and nonbiological causes.  Many genes and lesions can often lead to the same apparent presentation or phenotype.  That lead to the idea of intermediate phenotypes or endophenotypes to get a more consistent population to study but that has only been partially successful. The DSM was never designed to biologically classify mental illnesses, but DSM diagnoses are used for studies of biology and pharmacology. Other systems have been suggested for that purpose – most notably the RDoC system, but so far it has not exhibited any widespread success.  There is no reason to think that a verbally based system will accurately describe biologically based illness whether those descriptions are in the DSM or RDoC.

Apart from classification for statistical, administrative, and planning purposes what good is the DSM to psychiatrists? I recently saw it criticized for not including enough psychopathology. The criticism was bitter and partisan but apart from some very basic definitions the DSM is not a course in psychopathology.  All psychiatric residents need to be taught psychopathology to the point that they are experts in it. That will never happen from reading the DSM. It also doesn’t happen from reading a psychopathology text or taking a college course in psychopathology.  It happens from seminars, reading, and clinical experience – discussing psychopathology with colleagues, supervisors, and instructors.  It happens from learning in treatment relationships with people who have psychopathology not just a list or criteria but experiencing firsthand the interpersonal aspects. The DSM explicitly states that it is for use by trained professionals and that it can be used to facilitate communication between trained professionals.  

The DSM is clearly not a treatment manual of any kind. That is why I have always found the charge that it is a source of prescriptions for the pharmaceutical industry ludicrous.  There are roughly six times as many prescribers of psychiatric drugs as there are psychiatrists and the only medication in that category that is more likely to be prescribed by psychiatrists is lithium. It is easy to speculate that the prescribing patterns of that larger group are not contingent about what is in the DSM.

What about the diagnostic side and what psychiatrists need? Although there was some criticism that the neo-Krapelinians have had too much influence on the manual it is time to acknowledge that verbal descriptions have come to their logical limits. It is also time to acknowledge that psychiatrists need to know a lot more about medical diagnoses in general in order to function in a medical environment. If medical conditions are in the differential diagnosis – how many medical conditions do psychiatrists need to know about and diagnose?  Every psychiatrist I know has stories about medical conditions that were referred to them as a psychiatric disorder where they made the correct medical diagnosis. They are typically conditions from neurology, endocrinology, and infectious disease but also general medical conditions like diabetes mellitus, hypertension, and atrial fibrillation. Approaches I have seen in other specialties include lists of conditions that the trainee or practitioner needs to know about.  That is a useful approach but lists like that in a DSM are likely to raise objections about medicolegal risk and that a larger recipe book is being made for what it takes to be a psychiatrist. There are also many psychiatrists in settings where medical assessments are impossible, where they are referred out, or where the practitioner may feel inadequately trained. I see all of those reasons as being an opportunity to advance the quality of psychiatric treatment.   

A related issue is the diagnostic process in psychiatry as opposed to the rest of medicine.   Nassir Ghaemi, MD had a recent commentary about this on his blog suggesting that the DSM approach prioritizes comorbidities rather than differential diagnosis like the rest of medicine.  He describes the typical pattern matching that occurs early in the process and suggests that the differential diagnosis point, the DSM encourages listing all of the comorbidities rather than going through a differential diagnosis process.  In other words there is a lack of a hierarchical process. 

That has not been my experience. Granted – I may be a more medically oriented psychiatrist than most (but then again had 20 colleagues doing the same work) – but when I see a patient the universe of diagnoses are all possible both in and outside the DSM. The number one priority was making sure that a life threatening medical condition was not misdiagnosed as a psychiatric disorder.  Every physician can recall being taught about differential diagnosis and having to write an exhaustive list for the first few Internal Medicine inpatients. That process illustrated that a lot of the “rule outs” occurred as a mental exercise and really did not need to be written down. By the end of that rotation the differential diagnosis list collapse from the low double digits to the low single digits. There was also a triage element based on the more pressing problem or diagnosis.   A DSM for psychiatrists could make this process explicit, discuss the cognitive aspects of pattern matching and completion necessary for generating hypotheses in the differential diagnosis, the differences between differential diagnosis and comorbidity, and probabilistic considerations in selecting the preferred diagnosis. It would potentially have training implications because in order to optimize the pattern matching required - adequate training experiences need to be supplied to develop those skills. 

A DSM for psychiatrists needs to be much more information intensive in terms of research on validators, psychiatric genetics, multiomics, endophenotyping, drug mechanisms of action, and biological markers for each category.  A typical response to that suggestion is "Well there are no biological markers, labs tests, etc."  I don't find that to be a compelling argument when I think about what is currently being ignored.  We are on the cusp where more of that information is becoming relevant and we are past the point where much relevant information can just be dismissed. Any concern about cost of a more extensive manual can be dealt with by placing it online for subscribers. This may seem like a significant task given the accumulating information, but it is time the APA and research leaders in psychiatry to realize that the task has changed.  Psychiatrists are different from other physicians and other mental health professionals.  Psychiatrists need the technical information to provide quality care and compete against other systems that claim to know more about psychiatry and medicine than they do. Time to adjust to that reality and have the necessary internal debates first.

That concludes my suggestion for a DSM for psychiatrists, but I am open to more suggestions.  And for the record I am suggesting two different publications instead of a general manual full of qualifiers about expertise.  We need a manual for experts and another one like the current version - for everybody else.

 

George Dawson, MD, DFAPA

References:

1:  Horwitz, A.V. (2014). DSM - I and DSM - II . In The Encyclopedia of Clinical Psychology (eds R.L. Cautin and S.O. Lilienfeld). https://doi.org/10.1002/9781118625392.wbecp012

2:  Kim YK, Park SC. Classification of Psychiatric Disorders. Adv Exp Med Biol. 2019;1192:17-25. doi: 10.1007/978-981-32-9721-0_2. PMID: 31705488.

3:  Cooper R, Blashfield RK. Re-evaluating DSM-I. Psychol Med. 2016 Feb;46(3):449-56. doi: 10.1017/S0033291715002093. Epub 2015 Oct 16. PMID: 26470724.

4:  Shorter E. The history of nosology and the rise of the Diagnostic and Statistical Manual of Mental Disorders. Dialogues Clin Neurosci. 2015 Mar;17(1):59-67. doi: 10.31887/DCNS.2015.17.1/eshorter. PMID: 25987864; PMCID: PMC4421901.

5:  Blashfield RK, Keeley JW, Flanagan EH, Miles SR. The cycle of classification: DSM-I through DSM-5. Annu Rev Clin Psychol. 2014;10:25-51. doi: 10.1146/annurev-clinpsy-032813-153639. PMID: 24679178.

6:  Grob GN. Origins of DSM-I: a study in appearance and reality. Am J Psychiatry. 1991 Apr;148(4):421-31. doi: 10.1176/ajp.148.4.421. PMID: 2006685.

Supplementary:

It has been suggested that a hierarchical approach informs the usual differential diagnosis exercise but it may be the application of the parsimony principle. To me there is an open question about how well parsimony works for complex biological systems.

Photo Credit:  Eduardo Colon, MD

 

 

Components of Patient Outcome

It is a good idea to go back and take a look at some basic components that predict patient outcome in medical (including surgical) treatment. There is a  skewed representation of practically all medical information that is portrayed in the press.  A big part of that misrepresentation comes back to missing what the real components are that determine the outcome.  I have listed a few of what I consider to be the major components in the above diagram and will take a look at each one in order.

Technical expertise of the physician is obvious in surgical and other invasive procedures.  Henry Marsh covered it in a book on his career as a neurosurgeon (1).  He discussed the process of surgery but also the unpredictable results.  An operation may have seemed flawless but the patient ends up with a postoperative complication that had to be due to the surgery and dies or is permanently disabled.  I have discussed the issue with neurosurgeons myself and directly observed them in action.  There are clear differences in skill level and operative outcomes that appear to be based strictly on manual dexterity and coordination. There are differences based on the number of procedures that are done.  Expertise is easy to conceptualize at a manual level.

Psychiatrists on the other had usually balk when I refer to the technical expertise required in the field.  The clearest example in training is learning about inpatient psychiatry before proceeding to outpatient clinics.  Most psychiatric trainees just don't know enough as first year residents to proceed to an outpatient clinic and start to treat anyone who walks through the door.  The inpatient setting provides the training necessary to address emergencies first.  Outpatient skills can be built on that foundation. An associated issue is how much you need to know. If you are immersed in medical settings like hospitals, you probably need to know about a broad range of medical and surgical conditions and how they can lead to or complicate treatment of psychiatric disorders. That knowledge base includes a broad range of basic sciences and the updates relevant for the field. Expertise in psychiatry also encompases the interpersonal dimension. Psychiatrists should be the medical specialists that have the highest level of competency in the interpersonal dimension and how their personal feelings toward patients can complicate care.  That requires a significant level of training and expertise.

Biological variability is poorly understood by nonphysicians.  The best examples are allergic reactions to medications and severe idiosyncratic reactions like liver failure or kidney failure in response to a medication.  But they can also be more common on a population wide basis and mediated though a number of mechanisms.  The variability has to do with both the way the medication is absorbed and metabolized and also specific effects that occur at the tissue level.  In some cases this leads to clinical trials where the results do not seem that impressive (antidepressants, radiofrequency ablation for atrial fibrillation) but where there is a clear consensus that the intervention works.

 Every procedure in medicine whether it is a prescription, a surgical procedure or a medical device is characterized by a study.  The quality of those studies varies in terms of design and statistical analysis.  The usual goals of early studies is to gain FDA approval for release into the market in the US and elsewhere.  In the process of the safety and efficacy studies, quire a lot is learned about the drug characteristics applied over a larger population.  That data is applied toward describing that population based effect to the physicians who will eventually be prescribing the medical or procedure.  In some cases there are considerable political cross currents that can affect the straightforward statistical data as in the case of coronary artery bypass grafting or the use of beta blockers for hypertension or preventatively post acute MI.  There is currently an active debate about prostate screening and the burdens and harm done by screening, but at the same time there are clinical trials that show prostate cancer survival is improved by radical prostatectomy (RP) - but that there is no difference between standard RP versus robotic surgery.  It is  doubtful that the statistical facts of the procedure will ever be far from the expertise domain.

The final critical outcome parameter is the patient's ability to consent. It is not an easy task to hear about current medications or interventions with imperfect results and decide what the best course of action is.  Apart from the cognitive analysis, there is an emotional component of having a significant illness.  There is also the requisite ability to tolerate risk. Practically no medical interventions are risk free but some have much higher risks that others. The risk ranges from death or disability to an allergic reaction or long list of possible side effects.  The risk of not taking the medication or having the procedure done can also be significant including death.         

One of the best examples in the literature about the importance of the patient related component of these outcome variables was Kurt Gödel.  He was a renowned mathematician and logician who suffered from severe medical problems including gastric ulcers, prostatic hypertrophy with severe lower urinary tract obstruction who either postponed or refused treatment to the point where he was near death or in severe distress. He had a severe urinary tract obstruction and refused surgery to correct the obstruction. He elected to remain catheterized for a period of 4 years until the time of his death.  He eventually died of starvation in a hospital refusing to eat because he believed his food had been poisoned and at the time of his death weighed 65 pounds (he was 5'6" in height). Gödel was one of the geniuses of the 20th century who experienced problems that consistently and adversely affected his health.  Shortly before his own death and Gödel's death his friend economist Oskar Morgenstern made this observation in his diary (from reference 2; p 251) after a phone call from Gödel:

"It is hard to describe what such a conversation means for me; here is one of the most brilliant men of our century, greatly attached to me ..... [who] is clearly mentally disturbed, suffering from some kind of paranoia, expecting help from me, and I am unable to extend it to him. Even when I was mobile and tried to help him I was unable to accomplish anything. [Now,] by clinging to me - and he has nobody else, that is quite clear - he adds to the burden I am carrying." - July 10, 1977.

Morgenstern died 16 days later of metastatic prostate cancer.

 The main takeaway message from this brief commentary of variables affecting patient outcome is that it is a complicated process. There needs to be open communication about the uncertainty at every step of the way. In the informed consent discussion (that should always occur) there should be enough information exchanged so that the patient/decision maker has a clear idea of the risk involved.

Even then, the most intelligent decision makers can fail.


George Dawson, MD, DFAPA



References:

1:  Henry Marsh.  Do No Harm. Thomas Dunne Books, St. Martin's Press; 2014.

2:  John W. Dawson.  Logical Dilemmas: The Life and Work of Kurt Gödel. Natick, MA: AK Peters, LTD; 1997: 229-253.