Mental Health disorders are diagnosed today based on subjective interviews and administration of partial sets of subjective questionnaires. How can we do it better?
Diagnosing mental health disorders is a real challenge. Today, without indicators of underlying physiological dysfunction, it must be done purely based on symptoms that are largely self-reported. Every patient presents with an array of physical, behavioral, cognitive, emotional and social symptoms which are collated together and then diagnosed into a “best fit” strategy that will ultimately determine treatment. An immense amount of investment goes into descriptive diagnosis manuals such as the DSM-5 and ICD-10 and their associated clinical interviews (SCID-5, CIDI).
In addition to these manuals, there are a diverse array of clinical and research-focused questionnaires which have been developed to review the symptoms of individual disorders, whether that be the Positive and Negative Syndrome Scale (PANSS) for schizophrenia, Yale-Brown Obsessive Compulsive Scale (Y-BOCS) for OCD or the Patient Health Questionnaire 9 (PHQ-9), for depression (to name but a few).
Here we outline four fundamental challenges in the current state of mental health diagnosis:
Asking the “right” questions.
Clinical interviews are designed to map out the symptoms of the patient in order to classify them towards a particular diagnosis. But what are the right questions to ask in these interviews and questionnaires? For example, although clinical interviews such as the SCID-5 include some neutral questions, they also predominantly focus on the presence or absence of particular negative symptoms. Similarly, clinical questionnaires which focus on one disorder type, such as the PHQ-9 includes negative behaviors and symptoms, and excludes the opposing positive assessment. Other questionnaires such as the Beck Depression Inventory (BDI) and Hamilton Depression Rating Scale (HAM-D) similarly focus on negative symptoms. In each instance the most positive option is the absence of a negative symptom, rather than the presence of a positive one. For instance, high empathy, a positive symptom, may be highly correlated with symptoms of severe sadness and maybe quite different from depression arising in the absence of empathy. Are they, and should they both be treated, the same?
A sample of questions from the PHQ-9 is shown below:
In addition, many of the questions used to diagnose disorders such as autism spectrum disorder (for example the Autism Spectrum Disorder Parent Interview) or ADHD (e.g. the Conners Parent Rating Scale) interpret everyday normal behaviors as possible symptoms of a clinical disorder, depending on their severity. This, in combination with criticism of the latest DSM-5 from some individuals that it has an “increasing tendency to ‘medicalize’ patterns of behavior and mood that are not considered to be particularly extreme” highlights the difficulties in determining where “normal” ends and “disorder” begins. Behaviors considered normal in some cultures may be considered aberrant in others.
Examples of these questions are shown below:
|“Does s/he dislike teeth or hair brushing, having hair or face washed, haircuts or washing hair, fingernails or toenails cut?”
|Does your child play with objects that are not usually toys, for example, carrying around DVD cases or strings?
|Messy or disorganized at home or school.
|Not true at all (Never)
Just a little true (Occasionally)
Pretty much true (Often)
Very much true. (Very often)
|Argues with Adults.
|Only attends if it is something he/she is interested in.
Can subjective questionnaires sufficiently “do the job”?
There are a range of screening questionnaires that are used to provide subjective assessments of a patient’s behavior, either from a self-report perspective, or from the perspective of the clinician or caregiver. These are critical to providing a holistic impression about the symptomatology of the patient. However, subjective assessments of behavior only go so far and are open to bias. For example, a patient may differ in the way they talk about or describe their symptoms or level of distress depending on their cultural background. Selecting the option ‘frequently’ as an option to how often you experience feelings of sadness may mean something quite different to an American than a Korean or even to a millenial and baby boomer. Although this is taken into account to some degree in the latest DSM-5 guide, it highlights one disadvantage of only relying on purely subjective measures.
A partial views of symptoms
Furthermore, although some screening and health questionnaires target a range of symptoms across multiple disorder types (for example, the Symptom-checklist-90), the majority of questionnaires are designed to focus on one or two disorder types (or really categories of symptoms) and therefore may miss symptoms outside of this category. This further accentuates the funneling of a patient towards a single disorder type, rather than providing the “bigger picture” of a patient’s mental and emotional prognosis where relevant. Someone with attention issues may become frustrated and upset by being unable to accomplish the things expected of him and easily be diagnosed with ‘depression’ without understanding of its root.
Further, the practice of arriving at a single score from these questionnaires pose a significant problem. In the DSM-5 for instance, 128 different combinations of answers can lead to the same diagnosis. (See The Difficulty of Diagnosing Depression) Reducing a diagnosis to a single score based on a partial set of subjective questions severely undermines the diversity of symptoms with which the patient presents.
A disconnect with underlying physiological processes
Mental disorders don’t fit neatly within predefined boundaries. In fact genetic studies, such as those recently published in the journal Science by the Brainstorm Consortium, have found considerable overlap between different psychiatric disorders, leading researchers of this latest study to conclude that “current clinical boundaries do not reflect distinct underlying pathogenic processes, at least on the genetic level. This suggests a deeply interconnected nature for psychiatric disorders, in contrast to neurological disorders, and underscores the need to refine psychiatric diagnostics.” Trying to force fit patients into a specific diagnosis is therefore misaligned with the biological underpinnings of mental health disorders.
Subsection of genetic risk correlations among brain disorders and quantitative phenotypes (From Science June 2018, Brainstorm Consortium).
Where to next?
Slotting patients into a particular diagnosis is fraught with problems, and fails to reflect the interconnected and overlapping spectrum of cognitive, emotional and social changes and their relationship at the level of the brain. Furthermore, the use of subjective assessments which are influenced by individual, social and cultural norms and are overly-focused towards the negative side of human nature are important notes of caution to consider within the realm of mental health diagnosis.
Objective measures, such as task batteries (for example the Cambridge Neuropsychological Test Automated Battery (CANTAB), may help overcome the subjectivity. For example, with ADHD, patients can be assessed by the Conners’ Continuous Performance Test 3rd Edition, a computerized task which assesses sustained attention in children (see Many Ways to Measure Attention for other attention methods).
Ultimately, however, we must completely overhaul the diagnosis of mental health disorders to understand and treat them at more physiological levels. Brain imaging such as EEG has the opportunity to provide a noninvasive insight into the physiological underpinnings of symptoms (see ADHD and the Theta/Beta Ratio for example). That said, more sophisticated analytical approaches must be applied to realize its value (see Alpha, Alpha everywhere: what does it mean?). Simply searching for physiological correlates of already subjectively defined disorders may be futile.
In the meantime, rather than partial subjective categorization into disorders, it would be of greater value to understand broader symptom clusters, both negative and positive, to determine a course of treatment.