While the process of diagnosing is common to all fields of medicine, this article specifically addresses missed “medical” diagnoses in patients who present to or are being followed by psychiatrists and other mental health clinicians.
SPECIAL REPORT: NEUROPSYCHIATRY
A complete, accurate, and timely diagnosis is the cornerstone around which effective treatment is constructed. Yet, the complex skills involved in the art and science of diagnosing have only been the focus of investigation for less than a decade.
While the process of diagnosing is common to all fields of medicine, this article specifically addresses missed “medical” diagnoses in patients who present to or are being followed by psychiatrists and other mental health clinicians. It is hoped that by learning about vulnerabilities within the diagnostic process, readers will have the opportunities to reduce diagnostic error rates in their own practices and institutions.
Diagnostic errors and delays are so widespread that most individuals will experience a diagnostic error during their lifetime.1 And surely, every physician has had the experience of missing a diagnosis or coming to the correct diagnosis only after too long a delay. Memories of missed opportunities for earlier intervention weigh heavily upon us.
Recent studies have found that any individual with a “psychiatric history” or with “presenting symptoms in the sphere of mental functioning and behavior” is especially prone to experiencing a diagnostic error.2 This should alert all psychiatrists, since these are the very patients we are called upon to see. Rare diseases might be misdiagnosed, but, more often, common diseases are missed, especially if they present atypically.
In 2015, an Institute of Medicine report-Improving Diagnosis in Healthcare-concluded that, “. . . improving the diagnostic process is not only possible, but presents a moral, professional, and public health imperative.”1
The diagnostic process
How do doctors make a diagnosis? As you read the following vignette of a referral, try to be aware of your diagnostic hypotheses.
CASE VIGNETTE: Psychosomatic Concerns
According to his primary care physician (PCP) who referred him, Mr O has a long history of psychosomatic complaints. Mr O is a 63-year-old accountant who recently retired. He’s been sleeping restlessly and waking up depressed each morning. He jogs a few miles a day, most days, but says that lately he’s been feeling stiff and somewhat unsteady on his feet. Also, he says he’s “plagued by” constipation. The patient told the PCP that he can’t tell whether these physical problems are “all in my head” or whether there’s really something to worry about.
When you get a referral, you are presented with a great deal of information, and it isn’t clear which information will be relevant to your eventual understanding of the patient’s health problems. Some of the material that is contained in the referral will be explicit, some implied, details may be inaccurate, and surely, much will be left out.
Now consider that once you greet a patient in the waiting room, you will have even more information. You will notice how the patient is dressed and what he’s been doing while waiting; you will observe his facial expression, level of social comfort, his gestures and gait. And, before you even hear the patient’s voice or ask a single question, all of these experiences will affect you and have an influence on how you are inclined to think about the diagnosis.
Diagnostic problem solving is extremely engaging for physicians, but it is not the defined type of problem solving one encounters with something like a crossword or Sudoku puzzle in which there are 26 letters or 9 numbers that are manipulated to form one unique solution. Making a diagnosis is complex problem solving in an uncertain domain. Any particular information or observation might be relevant; and any particular information or observation might be irrelevant (noise). The patient might leave out crucial parts of the illness story, perhaps not realizing that they are important. Or the physician–problem-solver may have insufficient knowledge, perhaps never having heard of the diagnosis that actually explains the patient’s symptoms. There also could be more than one diagnosis; and these diagnoses might be interacting with one another.
System 1 and/or system 2 cognition
The reigning theory about diagnostic problem solving is that it involves two interacting cognitive systems-system 1 and system 2.3
System 1 cognition is illustrated by the experience of seeing a patient for the first time and “knowing” immediately that the patient has, for example, schizophrenia. System 1 is based on mental pattern-recognition; it is intuitive, but honed and refined by experience; it is fast, effortless, and operates outside of awareness. While this system is extraordinarily efficient, processing large amounts of information very rapidly, it is also highly error-prone and open to biases (predispositions to think and act a certain way) of which the clinician is generally not aware.
System 2 cognition is exemplified by the generation of a differential diagnosis list and examining the evidence for and against each diagnosis. System 2 utilizes explicit rules of logic; it is slow, effortful, entirely conscious, and analytical. Clinicians naturally favor the use of system 1 cognition because it is effortless. But the most effective clinical reasoning involves going back and forth iteratively between system 1 and system 2 cognition.
Thinking about Mr O
Now let’s return to the case vignette. The patient’s story is in your mind when you go into the waiting room. Depending upon your fund of knowledge and your experience, you might have what turns out to be an accurate diagnostic impression from the referral information. But perhaps you don’t. To get to the correct diagnosis, you might have to overcome the “framing effect” in which your cognition has been biased by the way the patient’s problem was first presented to you.2 In the case of Mr O, your intuition may have been biased toward thinking that the patient is depressed with somatic complaints precipitated by his recent retirement.
Now you escort the patient from the waiting room and by the time you and the patient have settled into the chairs in your office you have new information to include in your diagnostic formulation. Your system 1 cognition immediately registered that Mr O had parkinsonian features. His affect was somewhat flat and he walked slowly, with a diminished arm-swing.
Now you use system 2 logic in an effort to piece together the information you have so far. Does the narrative presentation fit with your observations? You interview the patient, carefully framing your questions to better understand the patient’s symptoms and the arc of his illness over time.
The diagnosis of Parkinson disease
Knowledge and experience are both crucial in making the diagnosis of Parkinson disease in Mr O’s case. In addition to the blunting of facial expression and “feeling stiff and somewhat unsteady on his feet,” you learn that Mr O has been worried about falling, especially when he walks down a staircase. This is consistent with a shift in the center of gravity and forward propulsion associated with Parkinson disease.
Mr O also has had depression, prominent constipation, and fatigue-these symptoms are often prodromal in Parkinson disease, Lewy-body dementia, multiple systems atrophy and other, less common parkinsonian syndromes. If you know the symptoms for Parkinson disease, you ask Mr O whether his handwriting has gotten smaller. It has. Then you inquire about his sleep. He has had restless sleep for several years; on a number of occasions he has fallen out of bed. In addition, he has hit his wife in the middle of the night more than once, although he doesn’t remember doing this.
You might recognize that this is the classic presentation of rapid eye movement (REM) sleep behavior disorder, which is strongly associated with the subsequent development of Parkinson disease and related syndromes. In REM sleep behavior disorder, there is a failure of the usual inhibition of muscle tone that accompanies REM sleep. This allows the sleeper to move about during REM sleep and to act on dream content, including “fending off attackers” by lashing out.
Inevitable cognitive bias
To get to a correct diagnosis, clinicians must navigate a wide variety of inherent cognitive biases. For example, with Mr O, if you hadn’t realized from the referral that he might have had Parkinson disease, you had to overcome the impact of “anchoring,” the tendency to stick with your first diagnostic impression. You also had to overcome “availability bias,” the tendency to think that a diagnosis that readily comes to mind is likely to be the correct diagnosis.
For psychiatrists, major depression or dysthymia with somatic concerns is a diagnostic formulation that is commonly seen and readily comes to mind. Parkinson disease comes less readily to mind.
Cognitive biases are ubiquitous; they reflect the flow of nonconscious predictions we make, based on past experiences, as we move through the world. The influence of cognitive biases on diagnostic reasoning has been well described by Croskerry2 who has studied diagnostic errors in emergency department settings where physicians have to make quick decisions with little information about the patient and are forced to rely heavily on error-prone systems.1
In the case of Mr O, it would have been a cognitive error to believe that the etiology of his symptoms was the difficulty he was having adjusting to retirement, along with his tendency to focus on somatic concerns. This would have been an “attribution error” in which you held the patient responsible for his or her disease. All physicians are prone to making attribution errors, especially when making a diagnosis in patients who have behavioral symptoms.
Another common bias is a tendency to look for confirming evidence-“confirmation bias”-rather than disconfirming evidence, although disconfirming evidence is much more powerful. It turns out that Mr O’s symptoms of restless sleep, constipation, fatigue, and depression actually began before he retired, not after.
Furthermore, it is helpful for clinicians to be aware of the possibility of “premature closure,” the tendency to stop investigating once you have come upon a diagnostic hypothesis but before that hypothesis has been fully proven. This is an especially easy error to make when a medical condition has a long prodromal period without symptoms that are specific enough to make a clear medical diagnosis. As psychiatrists, we often encounter diseases of this type; Parkinson disease is but one example. Others include: multiple sclerosis, Wilson disease, Alzheimer disease, Cushing disease, fronto-temporal dementia, Huntington disease, and more. Periodically reviewing a patient’s condition while maintaining diagnostic uncertainty forces the clinician to rely more on system 2 cognition.
Non-specific behavioral and mood alterations often represent the very first and, occasionally for prolonged periods of time, the one single and exclusive sign of an undetected physical illness. Flagrantly and convincingly “psychological” in nature on presentation, such masked physical conditions frequently mislead the examiner and obliterate any further medical consideration, resulting in misdiagnosis and thus, inevitably, in treatment gone astray.4
Recent studies of diagnostic error have not focused on patients who are being treated in mental health clinics or inpatient psychiatric units; however, past studies of psychiatric patients found that, important medical diagnoses were missed in these patients as often as 40% of the time, depending upon the setting.5 Moreover, these medical diagnoses were not uncommonly the sole cause or a significant contributing factor to the patients’ presentations.
What can you do?
How can you contribute to decreasing diagnostic error rates? Expert opinion emphasizes: expand your knowledge; cultivate good clinical habits of mind, including expanding your knowledge and experience, self-care, humility, persistence, and the use of feedback; and utilize the use of metacognition-thinking about thinking, and work to improve the institutions within which you work.1
Expanding your knowledge
There is much for psychiatrists to learn in the realm of neuropsychiatric disease. Often patients with medical conditions will first present to psychiatrists thinking they have a psychological problem or a mental illness. Many will have common diseases: focal seizure disorders; neurodegenerative diseases in addition to Parkinson disease, such as Alzheimer disease, vascular dementia, or Lewy-body dementia; the long-term consequences of traumatic brain injury; autoimmune diseases such as multiple sclerosis; or sleep disorders. Some patients will have diseases that are less common: anti-N-methyl-D-aspartate receptor encephalitis, Wilson disease, Huntington disease, frontotemporal dementia. Familiarize yourself with the clinical presentations of these and other neuropsychiatric conditions.6 And remember to keep this question in the front of your mind: Could this patient have a medical condition that I am missing?
Diagnosing is a high-level creative process; as such it relies on reflective practices that allow time for you to think about your reactions to patients and about your diagnostic process.7 Even with a few moments of mentally stepping back to get perspective are important. In addition, actively employ strategies that force you to use the conscious, analytic system 2 cognition. Ask yourself, “What could this be?” Make a differential diagnosis list. Work against premature closure in the face of doubt and uncertainty. Remember that keeping a diagnosis “open” forces your “problem-solving mind” to keep working on what the diagnosis might be.7
Improving health care systems in support of diagnosing
Psychiatrists rely on hospital systems to support the diagnostic endeavor. Health care institutions may provide effective communication systems that allow collaboration with the patient and his or her family, health information technology, reliable laboratory services, access to library resources, expert second opinions, collaboration across disciplines, and specialties, and a culture that values reduction of diagnostic error rates while, at the same time, avoiding shaming and blaming.
Dr Schildkrout is Assistant Professor of Psychiatry, part time, Harvard Medical School, Beth Israel Deaconess Medical Center, Boston, MA. She is the author of two books: Unmasking Psychological Symptoms: How Therapists Can Learn to Recognize the Psychological Presentation of Medical Disorders and Masquerading Symptoms: Uncovering Physical Illnesses That Present as Psychological Problems.
1. Ball J, Balogh E, Miller BT, Eds. Improving Diagnosis in Health Care. Washington, DC: National Academies Press; 2015.
2. Croskerry P. The importance of cognitive errors in diagnosis and strategies to minimize them. Acad Med. 2003;78:75-780.
3. Croskerry P. Clinical cognition and diagnostic error: applications of a dual process model of reasoning. Adv Health Sci Ed. 2009;14:27-35.
4. Yates BL, Koran LM. Epidemiology and recognition of neuropsychiatric disorders in mental health settings. Ovsiew F, Ed. Neuropsychiatry and Mental Health Services. Washington, DC: American Psychiatric Press; 1999: 23-67.
5. Koranyi EK. Morbidity and rate of undiagnosed physical illness in a psychiatric clinic population. Arch Gen Psychiatry. 1979;36:414.
6. Schildkrout B. Complexities of the diagnostic process. J Nervous Mental Dis. 2018;206:488-490.
7. Schildkrout B. Masquerading Symptoms: Uncovering Physical Illnesses That Present as Psychological Problems. New York: John Wiley & Sons; 2014.