Ordinarily Well: The Case for Antidepressants



ONE OF THOSE bright spots is work with mild major depression, which tends to do well with what psychiatrists offer, medication and psychotherapy, a wonderful truth, since “mild” is serious enough. The label major depression applies only to highly symptomatic episodes—intervals involving more intense suffering, more widespread impairment, than dysthymic patients experience on most days. Even in less severe cases, where some freedom of action remains, depressed patients will be hobbled by apathy, low energy, and dark views of the self and the world.

I think of a stuck writer, a long-form journalist, Maggie. It’s not just that she can’t make progress on the next chapter. She no longer believes that finishing will make a difference. She can’t sit at the computer, is phobic about the office altogether. Research is no easier. She won’t pick up the phone, much less travel. The miasma has spread to engulf her day job—teaching at a community college. Nothing matters. Besides, she’s worthless.

How depressed is Maggie? She eats and sleeps. She drags herself to campus when she needs to and even attends committee meetings. By the numbers, she barely makes the diagnosis. But the fixed perspective leads her in circles, bumping her up against the same imagined obstacles. Can we be said to be doing psychotherapy?

Within days of taking an antidepressant, Maggie finds that she can approach her desk. She begins answering e-mails, then revising lectures. She turns to her manuscript in progress. Once more, Maggie thinks aloud in front of me, weighing competing means for shaping material. Soon we are on to more abstract issues: commitment, values, and goals. All that had been missing was the whole person. Who was this woman who foresaw only failure?

Before antidepressants—before I gained the courage to prescribe them—mild major depression might go on and on, leaving patients demoralized. I might ask, Didn’t you win awards for teaching? In my hands, logical guidance rarely broke the logjam. Medication did.

No one doubts these observations. Antidepressants do well with the lesser forms of major depression. The problem is, placebos do, too, in some research. For all that critics cavil, there’s little doubt that doctors should and will use medication to shorten profound depressive episodes. It’s about less disabling presentations that serious people disagree.

The notion that antidepressants work best for grave forms of mood disorder is called the severity hypothesis. It dates to the earliest days of the psychopharmacologic era, when lesser depression was psychotherapy’s domain.

But attempts to subdivide depression—neurotic and vital, psychogenic and biogenic, exogenous and endogenous—failed. Even if it begins in what looks like neurosis, as depression worsens or persists, it accretes symptoms that doctors think of as physical. It is hard to specify a subtype—a cluster of traits—that will not respond to medication.

Roland Kuhn had presumed that imipramine would prove effective for vital (biological) and not reactive (more psychological) depression. But when he prescribed for patients with “caused” depressions—with polio or after a criminal abortion—they responded. In Gerry Klerman and Jonathan Cole’s collection of studies, medication did best for outpatients.

The severity hypothesis never died, but in practical terms the newer antidepressants cast it in shadow. Well tolerated, Prozac and Zoloft and the rest were prescribed for every level of illness. Because they mute the debilitating self-consciousness and social anxiety that accompany depression, the medicines were well suited to the lesser variants. Between the existing research and patients’ experience, the question of usefulness in mild major depression answered itself.

If some experts retained misgivings, Irving Kirsch was not one of them, not at first. Responding to his “Emperor” paper, the one based on FDA data, readers complained that Kirsch had failed to separate out severe depression. Writing in 2002 and again in 2005, Kirsch replied that there was little reason to believe that efficacy follows severity. He had performed a quick analysis of the FDA studies and “found no relation between severity and antidepressant effect.” In a 2003 overview, outpatient trials had shown larger drug benefits than inpatient trials. Kirsch referenced research by NICE, the British clinical-standards institute, in which patients with severe depression proved more responsive than patients with higher (very severe) or lower (moderate) grades of illness.

Depending on the selection of studies, you could find any pattern you pleased: uniform efficacy, more efficacy in severe depression, more in mild depression, or more in the middle.

As for the categories, Kirsch noted inflation in labeling. What was now called severe depression had once been moderate. Antidepressants’ best use might be in the treatment of standard, midrange depression—if there was a pattern at all. Of course, Kirsch also argued that, in general, antidepressants’ inherent efficacy was minimal.

Kirsch changed his opinion of the severity hypothesis. In 2008, after reanalyzing the FDA files, Kirsch wrote that antidepressants are useful only in very severe depression. Below Hamilton scores of 28, antidepressants no longer met the NICE standard: three points of extra benefit, drug over placebo.

Researchers using the same data but different methods pushed the cut point down to 26 or 24. The vast majority of trials in the FDA collection involved groups with average Hamilton scores of 24 and above, so where the drugs had been tested, they worked. Most patients with Hamilton scores below 24 had taken Serzone.

Severity turns out to be a tough issue to investigate.

Fewer Hamilton factors apply in milder depression. You can’t reverse suicidality in someone who’s not suicidal. When a medicine helps less ill patients on the Is life worth living? front, the scale will not register the change.

When a patient with very severe depression recovers fully, he can lose 25 or 30 Hamilton points. In a comparable trial of mild major depression, a patient who wipes the slate clean can contribute only, say, 12 points. Each patient who remits contributes little to the average number of points lost per patient.

The term of art for this problem is floor effect. When you start near the floor, you can’t make great leaps down. For medications, like antidepressants, that bring strong benefits to certain individuals (rather than minor benefits to many), floor effects cause underestimates of the efficacy in the treatment of less ill patients.

One way to avoid this distortion is to look at categories. For instance, in each group how many people lose half their symptoms? That’s what regulators from Sweden, England, France, and Holland did. Stung by critiques of trials submitted to the FDA, the Europeans reanalyzed all the data in their files, which were more complete and detailed than the FDA’s.

Response rates were uniform. “There was no evidence of a diminishing magnitude of effect with lower severity at baseline,” the regulators noted. “The responder criterion is a relative measure, and thus floor effects, which might contribute to the diminishing effect size observed with lower average [Hamilton] score at baseline, are avoided.”

John Krystal’s analysis of trajectories came from studies of patients with average Hamiltons of 20, typical in outpatient clinics. Krystal demonstrated important benefits for patients who, at the start of treatment, were nothing like severely depressed.

Using the same studies of Prozac, Paxil, Lexapro, and Cymbalta, Krystal asked about melancholic depression, in which patients experience impairment in energy, sleep, and appetite. Unexpectedly, nonmelancholic patients (with depressions that look less physiological) proved likelier, on medication, to enter a favorable trajectory. That’s what clinical experience—what my experience—shows: antidepressants start neurotically depressed patients in the right direction.

To my mind, these explorations—of response, trajectory, and depression subtype—provide adequate answers to Kirsch’s challenge. The apparent pattern, less efficacy for milder depression, reflects floor effects. Once you take into account how antidepressants work, holding some patients back and helping others a lot, then in data much like the FDA’s, medications prove effective across the board.

The severity hypothesis looks like the antidepressant controversy overall—less worrisome once you consider people, rather than averages. But there remained a big problem with this debate, centered on drug-agency data. Hoping to keep placebo responses manageable, Pharma had run tests on very sick patients. With rare exceptions, trials excluded people with mild major depression. Conclusions about them were based on extrapolation, the extension of lines on graphs, guesswork.

Once again, Kirsch had set an agenda—putting the severity hypothesis back in play. His challenge made researchers hungry to look at files of individual patients, especially patients with lesser forms of major depression—and in more thoughtful trials, not ones run by drug companies to satisfy regulators.

If you find an error or have any questions, please email us at admin@doctorlib.info. Thank you!