Why I Have (Mostly) Given Up on Diagnosis

The Professional's Big Book of Stories?

Every year about this time I review my template file for new client notes. It has blank sections for name, presenting concern, history, plan, and a number of other categories.

This year I found myself staring at it, considering whether a revision was in order. And the category that leapt out at me was “Diagnosis.” The truth is, I seldom use it any more.

Once, that part of the template was more elaborate. I had space for all 5 diagnostic axes of DSM-IV, and sometimes convinced myself that by filling them in I was doing something useful. That didn’t last long – at least not for the majority of people I see.

I dropped the detail years ago, leaving only that “Diagnosis” heading. When DSM-5 came out it seemed that I had been prescient. The axes were gone. The new Bible revealing the truth of immutable human nature had been revealed, and the axes were not a part of it. The nature of disease had, once more, changed overnight.

Maybe new copies of the DSM should come with an “Orwell pill” that erases memories of previous editions. This might help older clinicians like me forget that things were ever different, and we would more readily leap onboard with a fervent belief in the new revelation.

It’s hard for a psychologist not to diagnose. In many jurisdictions, diagnosis is one of the few controlled acts that distinguish our profession from other counselors: We’re allowed to do it, they are not. In reality, of course, they do it all the time. We say “This individual suffers from panic disorder.” They say “This individual reports symptoms that appear to meet criteria for panic disorder.”

Nevertheless, it’s an important distinction. Somehow. If it’s your one Special Power, it’s hard not to use it.

What’s diagnosis for?

Definitions vary, but the core of diagnosis is the identification of the character and etiology of (in this context) a disorder. (Quibble if you wish.) The essential point of diagnosis is to guide action. By knowing what you are faced with, you may know what to do about it.

Diagnosis is enormously useful in most of medicine. A patient might present with a host of symptoms – perhaps including fever. Diagnosis might attempt to determine whether the person suffers from a viral or bacterial infection. If bacterial, an antibiotic might be effective. If viral, the antibiotic will be a waste of time. Diagnosis takes a host of symptoms and creates a knife-edge dichotomy: yes or no, this or that. If done correctly, the side of the line on which one falls can make a huge difference in the treatment used and the outcome expected.

Diagnosis becomes more difficult when problems are continuous rather than dichotomous. Instead of “Does she have mumps or does she not?” the question is “Are the symptoms severe enough to merit a diagnosis of depression?” In the first case, we are making an inference about the underlying reality of the disorder. In the second case, we are drawing an artificial line, much like the lines defining different time zones. We could move the line a few kilometers to the west or east, and it would be meaningless for someone to argue that we had done it incorrectly.

In striving to emulate the rest of medicine, psychiatry and the other mental health professions saw diagnosis as essential. In many ways this was true.

• In order for a study on the effective treatment of OCD in Baltimore to be relevant to the treatment of patients in Edinburgh, we had to know we were talking about the same thing. 
• In order to apportion scarce mental health resources, we had to have a systematic way of admitting patients to treatment – it isn’t fair for the primary determinant of whether one gains access to a mental health center to be which side of Oak Street one lives on. 
• And indeed, there are a few conditions – very few – that seem genuinely and sharply distinct from others. An example from the 20th century was tertiary syphilis – a condition that might, to the untrained eye, resemble other conditions but had an entirely distinct cause.

There may be others. Currently a great deal of work is looking at the genetic markers for disorders like schizophrenia. Although for the most part these seem only to be hints of the level of risk for a disorder, it is conceivable that distinct causes will be discovered.

As well, in some cases we may not have sliced the pie thinly enough. Many people suspect, for example, that one problem with Major Depression is that we are looking at a cluster of disorders rather than a single entity. If we could identify distinct types we might be able to select treatments and predict outcomes better than we can now.

In clinical settings, however, the main point of diagnosis is utility. If we take all the complexity of the human life sitting before us and boil it down to a nice clear label, this will tell us what to do.

The trouble is, it almost never does this.

I frequently get referrals of clients who already have a diagnosis. Setting aside the fact that the diagnosis is often incorrect (not surprisingly, having usually been made in the context of a five-minute consultation), this tells me little. In order to decide what course treatment should take, I need to take all the detail back out of the box to see the complexity again. What’s going on in the person’s life? When did it all start? Which of the symptoms of that disorder does the client actually have? What thoughts are they having as they exhibit those symptoms? What do they think is going on?

We are indoctrinated in the church of diagnosis so firmly, that it took an inordinately long time for me to begin questioning what I was doing. I would see the client, pull up my intake template, pull out my DSM, turn to the disorder that seemed to describe the person most closely, and puzzle through whether they had the requisite “5 of 9,” or “6 of 11,” or “A, B, and 3 of the 5 symptoms for C” to meet criteria. I’d put down the answer and that would be the last time I looked at it.

As a tool for guiding treatment, diagnosis was the ultimate null entity. A placebo, if you like.

True believers are infuriated by this. Take even a simple case, they say. Your client is too afraid to get on a plane. What are they afraid of? If it’s that the plane will crash, they have 300.29, Specific Phobia. If it’s that they will have a panic attack, and these occur in other settings as well, then it’s 300.01, Panic Disorder. This makes a difference! In the one case you’ll look at the overprediction of catastrophe; in the other you’ll do panic treatment. A classic knife-edge diagnostic distinction.

Well, yes. But it’s not the diagnosis that tells you this. It’s your assessment. Take away the assigning of the number, and you still have all the information on which your diagnosis was based. In fact you have more, because the whole point of diagnosis is to shave away the detail to leave you with a nice clear label. In order to treat the client you have to reclaim all the shavings and flesh out your understanding.

So you spend all that time packing things into a box, only to open up the box and take everything back out in order to decide what to do. The diagnosis provides you with almost no guidance.

What about that cutoff line? In addition to asking “this disorder or that disorder,” diagnosis attempts to declare “disorder or no disorder.” In my work, the cutoff is virtually irrelevant. If someone has OCD-ish symptoms and meets criteria, I’ll look at the details and work with them on the OCD – maybe with exposure and response prevention (ERP). If they have OCD-ish symptoms but fall short of the full diagnostic criteria I’ll do exactly the same thing. No one has ever shown that the diagnostic line has magical qualities: people who fall on one side respond to ERP; people on the other do not. In both cases I’ll point out that therapy requires a lot of work, and that it is up to them whether they wish to invest the effort in it.

Some people argue that my stance comes from having a psychotherapy practice. Medications, it is sometimes argued, have specific neurochemical effects and so identifying the underlying pathology via diagnosis is critical. (I am setting up a bit of a straw man, here, admittedly – given the last 10 years of psychopharmacology almost no one argues that diagnosis reveals specific neurochemical underpinnings anymore.)

In practice, however, sharp diagnostic lines do not often dictate prescribing practice. Antipsychotics, once given almost exclusively to people with psychotic-spectrum disorders (hence the name), are now as ubiquitous as Jelly Tots. Antidepressants are routinely prescribed to patients who do not meet diagnostic criteria for major depression (or any psychiatric disorder), despite the lack of evidence of efficacy of these medications for subclinical symptoms. (The relatively poor evidence of efficacy for appropriately diagnosed mild-to-moderate depression is another concern.)

So what? Is there a downside to diagnosis?

One problem with diagnosis is that we spend scarce clinical time performing an act that often has little value in structuring treatment. But there are others.

Diagnosis can shape clients’ self-perceptions. Sometimes this can be positive. Certain clients learn that they have an identifiable disorder and sigh with relief. “So I’m not just weird – and there are other people like me.” 

But diagnosis can have the reverse effect as well. It can draw a firm line dividing the person from the rest of humanity. “So I’m different from mentally healthy people. They have intact minds and I do not.” In my experience diagnosis more often has an alienating effect on clients than a soothing one.

If I sense that a client will feel better to know the name for their problem, I have no problem giving it. “Yes, these are all symptoms of depression.” But I make it clear that the dividing line is neither important nor indelible. “Just as many of us will have symptoms of a cold which then go away, we will work to bring your symptoms down to the point where you are not in a depression.” I attempt to intervene if I see the label becoming welded to the person’s sense of self - for example, if they begin referring to themselves as “a depressive.” The more they incorporate the symptoms into their vision of who they are, the more recovery will involve “killing” a part of the self or becoming a new and strange person.

This me-versus-them distinction might be inevitable if it was based on a clear reality: “Yes, you are infected with Hepatitis C and most people are not.” But again, most psychiatric diagnosis is defined by a consensus agreement of committee members around a boardroom table, not by a blood test. Most diagnosis involves the drawing of artificial and rather arbitrary lines. But it becomes a psychological reality in the mind of the person diagnosed. “The problem,” so a colleague once said to me, “is that people think these disorders are actually REAL.”

Well, the symptoms are real. The distress is real. In many cases the need for treatment is real. But the label itself takes on a reality that is often a barrier to improvement rather than an aid.

Furthermore, diagnoses find their way into the clinical record. Once a label is down on paper it can influence the outcome of court cases, career aspirations, and the quality and nature of medical care. Many have noted the chill that a diagnosis of “Borderline Personality Disorder” can give to subsequent providers, whether or not it is accurate. Similar unintended effects can happen with depression, anxiety disorders, or other problems.  This can be entirely appropriate: if an airline pilot is in the throes of major depression we want to know this. Capriciously given, however, or assigned as a routine part of a clinical consultation, a diagnosis can have long-lasting and damaging effects that outweigh any beneficial elements of the encounter.

Clients themselves are often unaware of the potential long-term difficulties associated with receiving a diagnosis. It is not uncommon for a university student to petition me for a diagnosis of ADD or anxiety disorder to be provided to examination staff. Except in extreme circumstances I have become very resistant to these requests. It is a service provided readily by many practitioners, but is not one I am obliged to offer as part of my work – and I have seen too many unintended negative consequences of casual diagnosis.

Some people have said to me, “Isn’t there an ethical requirement to assign a formal diagnosis prior to conducting therapy?” Actually, there isn’t. There is a requirement to conduct a proper assessment to see what’s going on. But the assigning of a label is optional.

First, do no harm. If I am doing something that runs a significant risk of causing harm to a client, I had better know that the likely benefits of such an act outweigh those risks. In the case of diagnosis the benefits have become steadily less apparent to me, and the downside steadily more visible.

Partly I have become less enthusiastic about diagnosis because I fail to see its usefulness in most circumstances. Partly it’s because I see it as a potential ethical failing.

Maybe we should be trying to understand our clients more, and oversimplify them less. Just a thought.