Thursday, March 6, 2008

Unreliable Narrative: The Aesthetics of Depression

[This presentation was delivered at the Center for Cultural Analysis of Rutgers University, March 5, 2008, as an introduction to a dialogue with Dr. Jerome C. Wakefield]

With sincere humility—and not only in the spirit of the political season—I want to say what an honor it is to be on the platform with Dr. Jerome Wakefield before this audience. Dr. Wakefield is an original thinker and a distinguished scholar. He has advanced a complex theoretical position regarding diagnosis, one I reference respectfully in my work. I want to remind you that, in contrast, I am what we at Brown call a clinical, by which we mean an ersatz, professor of psychiatry. I am not a researcher. What theoretical contributions I make are, more often than not, presented in popular writing for general audiences.

Still, I gather that we are to have a debate.

I am to take the position at the core of my book, Against Depression, that depression is a disease. I am to say that the seriousness of depression is often underestimated and that depression can be falsely romanticized.

Dr. Wakefield is to take a contrasting view, that depression is over-diagnosed and that the signal cultural threat comes from subsuming adaptively useful sadness into the category of disorder.

I want to begin by saying that from my own perspective, I have already conducted this debate.

In Listening to Prozac, which I wrote in the early 1990s, I expressed concern over the potential for what I called “cosmetic psychopharmacology,” the use of medications to take a person from a normal but socially disfavored state to an equally normal state that is better valued or rewarded. I worried that cosmesis might take place covertly, through what I called “diagnostic bracket creep,” the expansion of categories of illness to embrace available treatments. In particular, I feared that as a culture we would lose our taste for melancholy.

Against Depression
, from 2005, can be seen as my attempt to delimit that concern—to outline the domain of non-cosmetic, legitimately medical psychotherapy and psychopharmacology. In that book, I spend time exposing shortcomings of our current diagnostic system. I find it flawed. Still, with demurrers, I locate the boundary around depression more or less where my profession, psychiatry does.

To the extent that there is matter for debate, then, Dr. Wakefield will be arguing that psychiatry has drawn that line incorrectly.

So Dr. Wakefield and I share an interest in this matter of domains. We both recognize an entity, depression; we both value the preservation of the ordinary as ordinary and take distinctions between health and illness to be culturally important. In fact—based on my reading of Dr. Wakefield’s book, The Loss of Sadness, along with the monograph that seems to form the statistical justification for the book, his reanalysis of data from the National Cormorbidty Survey or NCS—I have the impression that Dr. Wakefield and I agree on a number of points.
Let me list some points of agreement, in the manner of legal stipulations that need not later be at issue.

We agree that depression is a disorder, and a grievous one.

We agree that this disorder is imperfectly captured by current diagnostic systems. There are problems of validity although, to be fair, we don’t know as much as we might about the underlying entity, the platonic ideal of depression.

We agree that the Diagnostic and Statistical Manual, or DSM, can be misleading in the individual case, especially when symptoms are of short duration or mild severity.

We agree that most sadness is not depression, though here I would add a point that Dr. Wakefield downplays, that you can have substantial depression without sadness.

Like Dr. Wakefield, I am open to the possibility that certain epidemiologic studies may overstate the prevalence of depression.

Certainly we agree that medications used to treat depression have a wide range of indications and that they may also mitigate symptoms in people who have no specified disorder; Dr. Wakefield quotes me to this effect.

We agree that pathologizing normality has a price. We also agree that there are cases where a doctor might reasonably prescribe and a patient take medication, or recommend and participate in psychotherapy, in the absence of a diagnosis, and that even in these cases, clarity about categories is a matter of good faith.

We mistrust Big Pharma; I write about that mistrust frequently, though that part of my work tends to be overlooked.

I apologize for taking to much time on this matter of agreement. Again, there may be something in the air. But I hope that this introduction will minimize later occurrences of the straw man problem, which I encounter now and then. I want to sharpen the subsequent discussion about points of disagreement.

Because we are to debate, I will need to set aside most of what I write about in Against Depression. I hope you will read that book. In it, I say that I had hoped, after a decent interval, to return to the question of psychotherapeutic medications and the ethical dilemmas they generate. But twelve years after the publication of Listening to Prozac, what had changed was not the pharmacopoeia but doctors’ knowledge about mood disorders. Depression’s standing as a disease had been solidified. I wanted to trace the consequences of that new understanding—to see how the culture had been affected by our past inability to treat depression and to imagine how a society might be changed by a decision not to romanticize depression but to accept it simply as a disease, and an ordinary one, like diabetes or cancer. I write about Van Gogh and Picasso and Bonnard—trying to trace the origins and consequences of depression as a marker of creativity and emotional refinement.

As I say, I will set that line of thought aside, as it overlaps so little with Dr. Wakefield’s concerns. In order to engage him, I will need to discuss his ideas more than my own. To begin the debate:

I assume that you know what depression is. It is a common affliction characterized by extended hopelessness, mental paralysis, emotional pain, low self-regard, and a loss of interest in the future, along with the features the standard diagnostic manual recognizes, problems with memory and cognition, disruption of sleep and appetite, loss of energy, excessive guilt, and suicidality. If you have suffered this disorder, or if someone you care for has, you will know how painful, how overwhelming, how debilitating it can be.

In one guise or another, depression has been understood as a disease for as long as there has been writing about disease; alternative viewpoints have an equally distinguished lineage.

DSM says that if you have five signal symptoms, including low mood or anhedonia, for two weeks and if they are of sufficient severity, then you are depressed. Here is how the manual defines the necessary level of impact: “The symptoms must cause clinically significant distress or impairment in social, occupational, or other important areas of functioning.”

As I point out in Against Depression, this mechanical diagnosis is more prognostic than descriptive. I write: “In twenty years of practice, I have never seen a new patient who complains of just two weeks of moderate depressive symptoms. If such a patient were to consult me, I might tread water—extend the period of evaluation. If the despair lifted, my thought might be that the episode was self-limiting. More likely, I would tell myself,
He was never really depressed.” But the diagnosis has predictive validity. If you have two weeks of the syndrome, you will likely go on to many months and then the career of recurrent depression.

This flawed working definition has been revolutionary for research.

It has defined a sufficiently homogeneous group to allow scientists to identify marked differences associated with or inherent to depression: differences in the brain (most recently the focus has been on the hippocampus and prefrontal cortex), and in the hormonal, circulatory, and skeletal systems, and elsewhere.

The definition has allowed for public health research that finds excess death across the life cycle, much of it from cardiac disease.

The definition has allowed for social scientific inquiries showing associations between depression and under-education, underemployment, and poor social supports.

And it has led to thought-provoking findings in genetics, including some we may discuss here, regarding the so-called “Woody Allen gene.” Early in life, this gene, for a serotonin transporter, seems to modulate the effect of stress on mood disorder. Young people with the protective version of the gene, even if they are otherwise prone to depression, will remain very largely immune to adversity; they may get “uncaused” but not “caused” depression.

Dr. Wakefield disputes whether this scientific progress is as substantial as I suppose or as probative regarding diagnosis. He challenges the method of diagnosis as containing a category error. He thinks that when the relevant syndrome is not severe and when it arises from a clear cause, it should be deemed mere sadness. Here, he points to the example of bereavement.

I have so far given an incomplete account of the
DSM algorithm. You are not diagnosed depressed if your cluster of symptoms arises from your response to the death of a loved one . . . unless your condition includes suicidality, morbid preoccupations with worthlessness, marked functional impairment, frank psychosis, or a type of general mind-and-body slowing called psychomotor retardation—or unless the standard syndrome lingers for two months.

The exception is narrow: between two and eight weeks, if you are bereaved, symptoms causing clinically significant impairment but not marked impairment do not qualify for a diagnosis.

Dr. Wakefield asks, why only the loss of a loved one? Why not the loss of a job or a marriage? Unless the syndrome is severe, aren’t we better off also calling those “caused” brief depressions ordinary sadness? Aren’t we robbing the culture of sadness when, regarding a person who has reason to be distressed, we call a month of anhedonia and loss of mental acuity and the rest depression?

I am sympathetic with this concern. But we should begin by saying that in operational terms it is limited. How often do doctors treat caused, moderate depressions? Rarely, I should think. We will need to know whether the new sort of definition improves on the sort that has proved useful over decades. And we will want to ask how this change looks in historical context.

Let's begin at the height of Freudianism, at mid-century. Freud's followers believed that virtually all psychological symptoms were caused.

Freud wrote about melancholia, which corresponds to the serious end of the depressive spectrum. He discerned psychological causation, related to loss of an “ambivalently held object”—a loved one (and secretly a hated one) who aroused strong mixed feelings. Meanwhile, neurotic depression was a bread-and-butter target of analysis. Personally, Freud was sophisticated about biological underpinnings for a variety of conditions. But it would overstate the case only slightly to say that fifty or sixty years back, in American psychiatry, which was dominated by psychoanalysis, there were no uncaused depressions, only insufficiently curious patients and doctors.

Of course, in analytic theory, which held sway for most of the twentieth century, the cause of a depression might extend back past the current adversity; that stressor might stand in for a prior loss which itself echoes inner conflict over unacceptable impulses from infancy. But for psychoanalysis, spontaneous mental affliction was something like an oxymoron, suggesting an error in epistemology. So it is not at all clear that it is culturally intuitive for depression to be uncaused.

In contrast, today's highly biological psychiatry is fond of multiplying etiologies , or categories of causes, for depression. Small, otherwise inapparent strokes, medications like interferon, hormonal shifts or deficits, and a variety of psychic stressors, not all related to loss, can trigger mood disorders. The most common variety, what I would call wear-and-tear depression, seems related to genetic liability, early life trauma, and then later adversity. Behavioral genetics, and here I am thinking particularly of the work of my friend Ken Kendler, has traced the course of this chronic, recurring and relapsing condition. Generally, early episodes of depression have identifiable social or psychological triggers. As the disease progresses, later episodes are likelier to be spontaneous. Otherwise, in a life history of depression, caused and uncaused episodes generally proceed or alternate without evident pattern. A mixture of major and minor episodes is common.

The data are not perfect, but depressions, whether from stroke or interferon or wear and tear, look similar. The same parts of the brain are implicated. The same social supports afford protection. The same treatments avail. The notion that uncaused episodes are pathognomonic, or characteristic, runs counter to the model.

The opposition of caused and uncaused is closest to the theories that arose in an intermediate period, as psychoanalysis was losing ground to medical models. Researchers—Dr. Wakefield says as much in his book—repeatedly but unsuccessfully tried to define and sustain a distinction between caused and uncaused depressions. Initially, they hoped that caused or neurotic depressions would prove to be psychological and best treated with psychotherapy while the uncaused or genetic episodes would respond to drugs. No such distinction held up. Ten years ago, the great mind in this field, George Winokur, wrote a valedictory paper titled, “All Roads Lead to Depression: Clinically Homogeneous, Etiologically Heterogeneous.” Winokur did not quite throw in the towel, but he had come to believe that the commonalities among depressions, once under way, were more important than the differences.

Repeatedly, depression has revealed itself as depression, with (almost) the only differentiator being severity of symptoms.

The symptom-based unitary diagnosis is in part a response to the failure of attempts to subdivide the depressions. I should add that biological psychiatrists, like psychoanalysts, understand that stressors are not uniform, that the death of an older relative after a debilitating illness might—or might not—differ from the unexpected death of a vital, young family member. Given the poor relationship between category and meaning it is remarkable that surveys, which form the basis for diagnostics and genetics, do as well as they do in advancing our understanding of disease.

All researchers recognize that the current criteria are flawed. They represent a compromise of convenience, not a means of carving nature at the joints. But the direction of the error remains unclear. Patients with only four symptoms (or only 10 days of symptoms, or symptoms of mild severity) may suffer substantially, function poorly, and go on to quite bad outcomes. Meanwhile, groups of depressed patients who say they have suicidal thoughts have only slightly worse outcomes than those without. Research that attempts to bound depression sharply generally ends in finding it seamless.

So: Our diagnostic system is inexact. It is not clear whether it is over- or under-inclusive. Cause has been an inherent element in mood disorder for decades. All the same, past research failures and current theories suggest that setting aside caused depressions will not create a cleaner version of the disorder.

Nor, I think, would Dr. Wakefield’s reformulation impact clinical practice. I have referred to my acknowledgment in Against Depression that I don't rush to treat brief, moderate depression. My colleagues, including general practicioners, have the same leanings. Evidence comes from Dr. Wakefield’s own work, his reanalysis of the NCS data.

There, Dr. Wakefield took the bereavement exclusion and extended it to other losses, removing from the category of depression episodes that were set off by, say, a divorce, and were not characterized by markers of severity like morbid preoccupations. Here, I am turning to the principal piece of evidence that Dr. Wakefield brings to bear in his critique of the current method for defining depression. Since his analysis of the data is so important, I want to spend a few minutes with it. I hope you will excuse me if what follows sounds technical; I think it gets to the crux of the problem before us.

What happens to the patients, with early, milder, caused syndromes, the ones that Dr. Wakefield says are misdiagnosed if we call them depressed? In the NCS, fewer than five per cent of people with uncomplicated triggered depressions had
ever in their lifetime taken a medication to treat depression. These patients were about thirty-four years old and had gone through an average of about three of these episodes. So: sixteen years of adulthood, recurrent symptoms in response to stress, and almost no prescriptions.

A reasonable estimate—I can say later how I arrived at it—is that five per cent of American adults took an antidepressant in 1992. So the
one-year rate for Americans in 1992 may well have been the same as the sixteen-year rate for the interviewees Dr. Wakefield picked out. And it is not all clear that the syndromes he identified were uncomplicated, according the bereavement criterion; Dr. Wakefield’s analysis shows that these “uncaused,” uncomplicated depressions lasted not two but five months on average.

Annual probabilities compound. Think about it: if the independent probability of getting an antidepressant in a given year were only one per cent, the odds of receiving one ever in sixteen years is almost fifteen per cent. Dr. Wakefield seems to have identified a group of Americans almost uniquely shielded from exposure to antidepressants. Doctors just do not prescribe for people who respond to adversity with mild depressive symptoms.

There may be another reason that doctors did not prescribe for certain patients labeled depressed in the NCS: the patients were
not depressed, not even under the DSM system. You will recall that symptoms are not enough to qualify a person for a diagnosis; those symptoms must cause moderate impairment. For reasons we can discuss later, the original NCS analysis did not include the severity criterion. That’s right—a good percentage of the people counted as depressed in the NCS did not qualify for the ordinary diagnosis of depression. And the problem was a failure to ask about the factor that comes closest to Dr. Wakefield’s own criterion for disorder: harmful dysfunction.

How many subjects were mistakenly counted as depressed? Darrel Regier reran the data using proxies for the severity criterion and found an over-diagnosis of 20 per cent, roughly the same percentage Dr. Wakefield reports. (Dr. Wakefield's monograph brackets the over-diagnosis between fifteen and 25 per cent.) The groups newly excluded in the two reanalyses may contain the very same people. In Dr. Wakefield’s version , only twelve per cent of subjects in the “uncomplicated loss-triggered” category said that their episodes interfered with life a lot. So Dr. Regier, using the
DSM, and Dr. Wakefield, using the loss exclusion, make the same cuts, for different reasons. Again: of the subjects Dr. Wakefield focuses on, the ones he calls merely sad, 88 per cent would have been diagnosed by DSM, properly applied, as not depressed. It’s not clear that Dr. Wakefield has found any over-diagnosis, relative to the current diagnostic algorithm. Nothing in the NCS data, nothing in Dr. Wakefield’s application to that data of his narrower definition of depression, suggests that we are losing sadness.

One further point about the NCS data. Almost all the subjects interviewed mentioned some cause in association with their syndrome. If you look at those who were diagnosed depressed, fewer than ten per cent either named no cause or failed to answer the survey question. Depression most often is associated with some change in a person’s life. (Parenthetically, we don’t know that people are always right about the direction of causation; in another of my books, Should You Leave?, I suggest that depression causes divorce as often as the reverse—though people tend to recall events as causing mood change.) Casting suspicion on caused depressions risks removing from the disease category the vast majority of those patients who go on to heart and blood vessel disease, recurrent mood disorder, and the other serious consequences. Dr. Wakefield’s proposal may look incremental, but it is radical. For the most part, causation—certainly, the immediate trigger—is characteristic of depression and always has been.

So much for practice. Let’s spend a minute on
theory by looking further at the bereavement exception. It has two justifications and an explanation.

The explanation is political. It’s simply hard to tell the public why grief should sometimes be considered depression. But many serious researchers make that grouping. In behavioral genetics, where it is critical to distinguish subtle signals from loud noise, researchers tend not to exclude bereavement. If stress triggers depression, the result is counted as depression. The character of the stress is irrelevant. There is some external justification for this choice. Grossly, bereavement looks like depression in some of its consequences. A recent overview in the
Lancet found that loss of a loved one causes startling increases cardiac death and suicide. Like the symptoms of depression, those of bereavement respond to psychotherapy and antidepressants. Some bereavement is depression.

The opposing case, that bereavement is exceptional, is twofold. The death of a spouse or child is a stronger, more disruptive stressor than others we recognize. And grief differs from depression: grief is at once more various and more particular. Bereavement can look like an anxiety disorder (or even psychosis) rather than depression, or the syndromes can be mixed. And the depressive symptoms of bereavement have their own cast. The grieving tend not to have psychomotor retardation—that’s why that consideration enters into the standard manual. They tend to ruminate over sins of omission, not commission. They cry more than the depressed. As I have said, depression is not really like sadness. It’s more an oppressive flattening of feeling.

But when bereavement looks just like depression, it contains much of the risk of depression.

The choice might be impractical clinically, in terms of the distinctions that would need to be specified, but in scientific terms, it might make more sense to lift the bereavement exclusion than to extend it to other stressors. Parenthetically, in Dr. Wakefield’s data, when you compare uncomplicated depression to uncomplicated bereavement, many more bereaved people receive medication. Partly, that’s because bereaved people can be depressed; partly, it’s because medication use is a poor marker for depression.

I hope I have been responsive to the notion of debate by turning immediately to these issues of interest to Dr. Wakefield. They are less important to me. If we all agreed that depression is a grievous disease, but that if relatively mild it should not be diagnosed until day 56, so long as we did that well—so long as we followed the early, milder cases attentively, monitoring the patients for any deterioration—that change might not disturb me. More to the point, it might strike a prudent public health policy maker as a reasonable alternative to applying a more expansive standard haphazardly. But here we are discussing pragmatism, not science or theory. There is no scientific or theoretical reason for demanding a more stringent set of criteria for caused than for uncaused depressive episodes.

I want to end by expressing concern over less technical issues—Dr. Wakefield’s repeated use in his book of the terms loss and sadness. For instance, he begins by saying, “By depressive disorders, we mean sadness that is caused by a harmful dysfunction of loss response mechanisms.”

Bereavement entails loss. But loss is a narrow word for stress or adversity. Depression has any number of social and psychological triggers—consider spousal abuse. In what is probably the most respected research in the behavioral genetics of depression, the following stressors were found to be influential: being robbed or assaulted, facing housing or financial problems, losing a job or encountering serious work problems, encountering major marital problems or undergoing divorce or separation, losing a close confidant, and encountering illness or death or a grave crisis in the life of a child, parent, or sibling, or entering into substantial interpersonal conflict with one of those relatives. Some of these misfortunes are losses, some are not. The following sequence is common in depression: pressures at the workplace trigger depression, which leads to job loss. Isn’t stress the operative factor here?

Caused depressions need not resemble grief. Dr. Wakefield uses a bereavement metaphor to gain sympathy for his view that caused episodes are not depression, but that metaphor ignores the best-developed model of the disease. Depression looks much more like a global decompensation of mind and brain in the face of recurrent adversity, whether or not the adversity resembles loss.

And depression and sadness are only vaguely linked. Consider the form of depression called pseudodementia, where patients appear to have the cognitive impairment of late-stage Parkinsonism but respond to antidepressants; these patients turn out to have personal and family histories of depression. Depression is not about sadness, it is about a disruption of a number of functions of brain and mind, resulting in apathy, hopelessness, and mental anguish.

Dr. Wakefield makes this distinction in his monograph (“We use the term ‘sadness’ as a generic label for normal and abnormal depressive responses to various losses”), but it does not inform the line of thought in his book. Caused depressions need not arise from loss; and sadness need not be the signal feature of depression. Why frame a discussion about depression in terms of loss and sadness? Dr. Wakefield is making a cultural point at the expense of medical accuracy; surely we would worry less about
The Loss of Apathy.

Here is where I will explain my talk’s title.

It seems to me that one of our cultural tropes, a way of expressing humanism and sophistication, is to say that we Americans scant or squelch true emotion, we look for the quick fix, we have, as a hypercapitalist culture, a systematic investment in inauthentic happiness.

I don’t entirely disagree with this position. My favorite among my own books is my novel, Spectacular Happiness. It takes its title from The Society of the Spectacle, a radical critique that applies Marxist and anarchist theory to cultures in which celebrity has supplemented wealth as a social good.

But I take exception to critiques that embody this tendency—blind meliorism—in our treatment of depression. The target is too easy, the metaphors are too automatic. And too often the evidence is lacking.

In the many books that make the same point—that our culture fears pain and overvalues happiness —the authors demand a much higher level of certainty for medical than for social scientific truth. They seem simply to
know that doctors prescribe thoughtlessly and promiscuously. They seem simply to know that Americans are heedlessly self-indulgent and that antidepressant use is a cause and effect of that defect. The position of these critics of medical psychiatry—and here I am pointing not at Dr. Wakefield but at many others—is frankly smug and superior. Patients' lives are at stake. There is room for both perspectives: If the culture is sometimes superficial, still, one of its strengths is the increasing attention to and destigmatization of depression as an ordinary disease.

Yes, we should be more contemplative. Yes, our fate is characterized by absurdity. Yes, the melancholic perspective has its appeal. But we need to separate that series of contentions from this other debate over the proper diagnosis and treatment of medical conditions. In that domain, we might acknowledge that we are making reasonable, though always frustratingly slow, progress against the particular disease depression, not least through our willingness to see depression even in cases where the provocation for particular episodes is apparent.

[A note on the text: When discussing my own writing, generally I speak ex tempore or from brief notes. Because at this meeting I was critiquing a colleague’s work, and because I wanted to cover a broad territory compactly, I wrote out the whole of my remarks. When delivering them, I revised on the fly, suppressing certain sentences and adding others. From memory, I have inserted some of the ad libbed thoughts here. During the question period, it became clear that a couple of points remained unclear to the audience. I have taken phrases from one or two of my answers and stitched them in as well. But for the most part, the talk as represented here can be thought of as the “advance text” of a delivered speech.]