What is abnormality? This seemingly simple question is intricately linked to the history of mental illnesses, psychiatry and, indeed, medicine in general. Over the years, many different definitions have been proposed, each having a very different perspective on what constitutes normality.
The mathematical approach, well exemplified by the use of diagnostic cutoffs in laboratory tests, emphasizes a deviation from the mean; abnormality would be defined as statistical infrequency. Social psychologists have argued instead that violation of social norms is what defines abnormality.
The approach taken by the DSM-5 defines abnormal as meeting diagnostic criteria for a disorder and causing functional impairment. Rather than talking about statistical infrequency or violation of social norms, the APA chose to talk about failure to function adequately.
Any psychiatrist will tell you that no mental illness can be diagnosed without impairment of function, and that the most severe psychopathies can fail to meet diagnostic criteria if they do not have a significant impact on a patient’s life. This view also emphasized clear disease categories; bipolar disorder is different from schizophrenia, which itself is different from substance use disorder or generalized anxiety disorder.
With the advent of the Research Domain Criteria (RDoC) and a move towards thinking of mental illnesses as dimensional rather than categorical, psychiatrists and primary care physicians will need to change their practice and objectify the diagnosis of mental disorders.
ADHD and Psychosis
In what represents a widespread shift in conceptualization of mental illness, psychiatry is evolving more and more towards placing mental illness on a spectrum rather than in discrete etiological categories. The DSM-IV spoke of autistic disorder and distinguished various types of so-called pervasive developmental disorders such as Rett syndrome and Asperger’s syndrome. With the advent of the DSM-5 in 2013, the terminology changed to “autism spectrum disorders” and incorporated various former diagnoses, such as Asperger’s syndrome, under this new umbrella term. Now, Asperger’s is understood as a “mild” form of autism, a form that is closer to “normal” on the spectrum. Patients suffering from Asperger’s syndrome are, in essence, high-functioning autistics under the new terminology.
This shift in understanding of psychiatric disorders does not involve only autism. Researchers have been contesting the existence of ADHD as a disease entity for decades. The notorious lack of diagnostic reliability would make it an invalid diagnosis.
Will the next iteration of the DSM speak of “attention deficit spectrum disorder”?
Some researchers argue that the clear inheritance of the disorder does not mean that it is a genetic disease, but rather that behaviors impaired in ADHD such as attention and impulsivity/hyperactivity are genetically inherited. Several studies have failed to find a discrete categorical basis for the disorder and have argued that it should be conceptualized as a continuum of impairments with a diagnostic threshold based on impairment (Marcus et al, 2011; Haslam et al, 2006). This is consistent with clinical practice, where the intensity of symptoms and the degree of impairments is highly variable from one patient to another.
Anyone working in ADHD clinics has experienced the variability in symptoms and impairment. How often do you see the young college student wishing to get into medical school and complaining of difficulty sustaining attention to study for their MCATs? These patients have very mild, yet distressing, attentional “difficulties.” Subjectively, they suffer from attention deficit, but objectively, their attention is likely at least upper limit of normal. On the other hand, some other patients have essentially given up reading altogether because they have never been able to sustain attention, and they can present in their early 30s, struggling to keep up with their finances or respecting the deadlines at work. Subjectively, they suffer from attention deficit, but objectively, their attention is likely much more severely impaired than the above example. With this new framework in mind, will the next iteration of the DSM speak of “attention deficit spectrum disorder”?
The conceptualization of mental illness falling on a spectrum has also made its way into psychosis. Recent meta-analyses have shown that psychosis is temporally continuous across the general population. They’ve also shown that the development or persistence of psychotic disorders results from added risk factors which push symptoms past the diagnostic threshold. This has led many scientists and psychiatrists to propose a model of psychosis-spectrum disorder, ranging from mild but persistent psychotic features to severe schizophrenia (Van Os et al, 2009; Guloksuz et al, 2017).
This is also strikingly consistent with clinical experience; from the mild, well-controlled schizophrenic patient who suffers few relapses to the severely psychotic, disorganized patient who has very little response if any to the strongest antipsychotics, schizophrenia is also very much a spectrum, and the impairments vary in intensity.
Diagnostic Unreliability in Spectrum Disorders
With autism spectrum disorder and the evolving concepts of attention deficit spectrum disorder and psychosis-spectrum disorder, psychiatry is moving towards eliminating discrete categorical diagnoses and placing psychiatric illnesses on a continuum. To talk of a spectrum implies an underlying normal distribution, with individuals at either end of the bell curve. In other words, there is no single cause to schizophrenia, autism or ADHD, but rather a collection of risk factors which collectively combine to produce clinical manifestations. With this new framework, psychiatric illnesses are no longer diseases defined by clinical criteria; they represent a deviation from the mean, with a diagnostic threshold based on the severity of the phenotype. The more risk factors an individual has, the more likely they are to cross the threshold and subsequently develop a more severe phenotype. This is the classical liability threshold model.
The above discussion raises another question. If there is to be a threshold, how can one define or measure the severity of a phenotype? In other words, how do you quantify clinical impairment? If the DSM-5 model worked well in the context of diagnostically distinct disease entities, it becomes problematic when trying to reconcile it with the spectrum model, which calls for a quantitative rather than qualitative threshold.
Whereas functional impairment is currently diagnostically used as a binary decision for most psychiatric diagnoses, using it to define a diagnostic or clinical severity threshold is dangerously relativistic. To use a different example, a cashier suffering from mild-to-severe inattention could be less functionally impaired than a neurosurgeon suffering from slight inattention. Currently widely used rating scales for ADHD rely on a patient’s subjective evaluation of symptom severity, which can be highly variable and dependent on external factors, such as one’s occupation or level of education. Eliminating objective diagnostic criteria runs the risk of having “floating diagnoses,” which depend on the interaction between someone’s clinical impairments and their functional impact and which can change dramatically from one individual to another. Diagnostic reliability, therefore, far from being improved, would be all the worse.
A Possible Alternative
An alternative would be for the notion of functional impairment to remain binary and for the threshold to be defined by clear neuropsychological measures of disability. This is done in highly specialized tertiary care centers, but it is often deferred in a high-volume primary care setting.
Mental health professionals will need to leverage the extensive technological means available to physicians and objectify the measurement of disability. Every single patient diagnosed with a mental health disorder should have a documented, objective, and quantified measure of impairment across the various cognitive domains affected by their disease. This would allow not only better diagnostic reliability, it would also facilitate treatment decisions and monitoring and prognostication of patients. An endocrinologist would never feel comfortable selecting a dose of levothyroxine in a patient suffering from hypothyroidism without an objective measure of thyroid function. Why should a psychiatrist make these very same therapeutic decisions blindly? If functional neuroimaging is not yet ready to make its way into clinical practice in psychiatry, we could very well see a future where fMRI-confirmed impairments in specific cognitive domains could replace or complement neuropsychologic testing, and truly revolutionize psychiatry.
Psychiatry is at a crossroads. As the field is moving towards defining mental illness as a spectrum of impairment, the perennial dilemma of how to define abnormality becomes once more relevant and calls for a change in practice, which would drastically change patient care and research alike.
P.A. Bilodeau is a 3rd year medical student at McGill University. He is particularly interested in social studies of medicine, philosophy of science and anything brain-related.