Thursday, July 18, 2013

How to Improve the Accuracy of Psychiatric Diagnoses - My Take

Allen Frances, MD has just blogged his ideas about how to improve the accuracy of psychiatric diagnoses.  His ideas basically come down to "be extremely alert to severe mental disorders and extremely cautious and patient before diagnosing mild ones."  He suggests a posture of "watchful waiting" of mild conditions to avoid attributing a treatment effect to a medication when in fact it is a placebo response.  He suggests erring on the side of underdiagnosis rather than over diagnosis.  I suppose that is all well and good but I have a few ideas on my own:

1.  Be a physician first - anyone coming for an evaluation needs to be assessed from a triage perspective.  Most American Psychiatric Association (APA) guidelines emphasize the need to assess the psychiatric parameters of acuity such as suicidal ideation and aggressive ideation and the risk of those behaviors, but there is very little medical guidance.   Psychiatrists need to be able to rapidly recognize both acute medical illness and medical illness that is causing the psychiatric presentation. They need to be able to rapidly assess medical problems that may interfere with the treatment of the psychiatric disorder.  The best way to have those  skills is to have adequate exposure to the full range of medical problems that can be encountered, especially from a pattern matching and pattern completion perspective.  That occurs only from treating many people with variations on the problem.  That starts in Medical School where every prospective psychiatrist should be focused on those experiences.

2.  Interpret your own studies - that means actually taking a look at actual brain scans, ECGs, lab tests, and other reports relevant to the care of your patients.  Psychiatrists need to be actively involved in the medical aspects of the care their patients, especially when they know more about the problems than the other physicians on the scene.  A few examples would be in the area of drug interactions, movement disorders, toxic syndromes like neuroleptic malignant syndrome and serotonin syndrome, the evaluation of delirium, electrocardiogram effects of psychiatric medications, and drug intoxication and withdrawal syndromes.

3.  Communicate well with the patient and their family.  Psychiatrists are trained and observed extensively in interviewing techniques.  They should understand the limits of specific interview situations and they should have well developed therapeutic neutrality that other physicians do not necessarily have.  In that environment they should be able to have the most productive dialogue with the patient and their family.  Psychiatrists should be experts in a diagnostic process that includes information from multiple sources.  Psychiatrists are also schooled in the concept of a therapeutic alliance and the implications of that orientation in treatment.

4.  Recognize the importance of psychotherapy.  Many diagnostic sessions require that psychotherapeutic interventions to be woven into that interview to support the patient, alleviate acute anxiety and to allow for a more thorough diagnosis.  Careful approaches to the diagnosis and treatment of patients requires recognition of the fact that some people will not tolerate any medications and psychotherapy may be the only available modality.  I do not hesitate to tell patients after an assessment that psychotherapy may be the best approach to the problem as well as discuss non medical approaches that have documented efficacy.

5.  Perform an actual psychiatric diagnosis.  This task is critical in the training of psychiatrists there is a lack of understanding about what making a diagnosis actually means.  Contrary practically everything that you read in the media, checking off criteria in the DSM 5 is not a psychiatric diagnosis.  Rating scales are also not psychiatric diagnoses and they are not quantitative measures.  It is very common these days for a psychiatrist to see a patient who carries 4 or 5 misdiagnoses like Bipolar Disorder/Major Depression + Attention Deficit-Hyperactivity Disorder + Intermittent Explosive Disorder + Asperger's Syndrome.  These folks are frequently on medications that are supposed to address the various disorders and they may not have ANY of the disorders.  In some cases they may not require medical treatment.  There are many people out there making complicated psychiatric diagnoses and initiating treatment in a 20 minute visit who are not qualified to make these diagnoses.  The other line of demarcation is the impact that a disorder has on the patient.  People who are functioning well in all spheres of their lives, by DSM definition - do not have a psychiatric disorder.  Many people are relieved to hear that they do not have a diagnosis or if they have had a diagnosis in the past that they no longer require treatment.

That diagnosis should be more comprehensive than a list of diagnoses.  There should be a formulation that describes the phenomenology and potential etiologies of the current disorder(s).  A narrative that makes sense to the psychiatrist and the patient.  At the end of my diagnostic session with the patient, I will frequently state it out loud in order to let the person know what I am thinking and get their feedback on my formulation.  I think that there is inherent flexibility in these formulations because the psychological etiologies can still vary based on the model that seems most applicable or the model that the psychiatrist prefers to use.  As an example it could be psychodynamic, behavioral, interpersonal, or existential.  It may employ a more recent model like one based on third generation behavior therapy or be a more supportive model focused on bolstering the patient's defenses.  The formulation is part history but also a discussion of etiologies (biological, social, psychological), dynamics, and defensive patterns.  The formulation can provide convergent validation for the diagnoses.  It provides both a pathway to understand the patient and guide psychological interventions.  The bulk of the material for this assessment occurs in parallel with the discussion of symptoms.

6.  Know the literature on borderland syndromes.  There is a significant overlap between medical conditions that are fairly non-specific in terms of diagnosis and treatment response like chronic fatigue syndrome, fibromyalgia, and chronic pain.  There are a significant number of people who present to medical and surgical clinics with symptoms and they never receive a diagnosis or an explanation for those  symptoms.  Familiarity with these syndromes will greatly assist in the diagnosis and treatment of these individuals if they are referred for psychiatric evaluation.  Specific knowledge of these conditions will allow the psychiatrist to consider an effective approach and effective patient education.

7.  Don't compromise your process because of extraneous variables.  The largest extraneous variable these days is the intrusion of business into the practice of medicine.  Psychiatrists may find that they are subject to limitations that do not apply to other physicians.  As an example, I have been told (by a managed care company reviewer) that psychiatrists don't diagnose or treat delirium when I was the only physician capable of making the diagnosis.   If you assess the patient and believe they need further diagnostic procedures or a medication trial that may be diagnostic do not give in to a case manager or pharmacy benefit manager who refuses to authorize what you need.  Make sure you communicate what you think the best possible care is to the patient rather than what the business people think.  Don't confuse medical quality with what a managed care company is calling "value".  They are probably unrelated.

8.    In the case of children, the best diagnostic approach looks at the family process both initially and in an ongoing manner.  The family should see the psychiatrist as someone who is not only an interested observer, but someone who can offer good advice right from the start of the process and recognize that symptoms in the identified patient can be a product of family dynamics.

9.    Take enough time.  The only valid way to make a diagnosis is to see the patient and interact with them in such a way that they feel understood.  Anything that takes away from that process can negatively impact on the flow of information and the task of providing that person with the best possible diagnosis and treatment plan.  The patient in this situation should not have the same experience they would have in primary care clinic discussing their depression or anxiety symptoms and the most obvious difference should be the total time spent talking with the patient.

10.   Review your findings thoroughly with the patient and family members if they are involved.  The process of psychiatric diagnosis differs from typical medical or surgical evaluations because of the sheer amount of data involved.   As an example, it might typically involve a sleep history similar to what might be obtained in a sleep lab with an additional 200 data points to look at the major diagnostic categories.  Even at that point there may be constraints on the data in terms of accuracy or detail that require corroboration of active debate.

11.  Know your diagnostic thought process - there a number of biases in the diagnostic process that have been written about in the literature on diagnostic decision making and in some journal features like the excellent series in the New England Journal of Medicine.  If you know the heuristics involved you can prevent diagnostic errors.

12.  Consult with your colleagues - consultation with colleagues serves a couple of useful purposes.  No matter how industrious you are it is impossible to see every possible presentation of every possible illness.  When you discuss patient presentations with colleagues who are also treating patients you are in effect extending your own pattern matching capability to include what your colleagues have seen and treated.   In many cases your colleagues have diagnostic and treatment experience with very low volume illnesses that are ordinarily seen a few times in the course of a career.

These are a few ideas I wanted to post today and there are a lot more.  Many of them seem like common sense, but the diagnostic approach to mental illness as practiced in most medical settings these days is anything but common sense.  You cannot get a comprehensive evaluation and diagnosis in ten minutes and you cannot really be walking out of a clinic with multiple prescriptions for medications that are supposed to work for that diagnosis in ten minutes.    

George Dawson, MD, DFAPA


  1. This should be required reading for not only every resident, but for every residency director. How anyone can think that a symptom checklist, in which pervasiveness and duration requirements are basically ignored, is better than a comprehensive biopsychosocial evaluation - including pertinent follow up questions when there is subtle evidence the patient does not know what the doctor is getting at - is totally beyond me.

    1. Cost is one, the stupid insurance-government complex is another


  2. George, this is an excellent Dirty Dozen on psychiatric diagnosis. You're right, most of this is common sense; but common sense is not used commonly. Thanks.

    Jim Amos, MD
    The Practical Psychosomaticist

    1. Thanks Jim.

      After reading your LinkedIn thread I picked up a copy of Chisolm and Lyketsos and am pleased with what I have read so far. The irony is that whenever I am lucky enough to get comprehensive records sent to me, it never resembles their narrative format they used for Poe. It is usually a telephone book sized copy. A couple of hundred of pages generated by an EHR with surprisingly little information per page (I can generally discard 95% of the pages without losing any relevant information). An endless series of fill-in-the-blank templates that are incoherent.

      I have started an outline for my own book that will be focused more on the cognitive aspects of the evaluation and if it ever gets to the writing stage, their approach offers promise in an age where only the government and businesses are supposed to use health care information.