THE MAINTENANCE LITERATURE is clear—antidepressants ward off depression—yet by itself that observation does not tell doctors what to do. Antidepressants may be stabilizing, and still we may not want patients on them indefinitely.
Wherever possible, I taper patients off. If they are not seeing me regularly, I will catechize them on self-monitoring for deterioration. If they slip, I will restart medication promptly, often with a refresher course of psychotherapy.
Why? Especially with recurrently depressed patients, why discontinue medication when maintenance is protective and mood disorder is harmful? I could answer in terms of antidepressants’ side effects, which are many. Antidepressants may increase the rate of cataract formation in the eyes. Certain of the drugs have been associated with a change in electrical conduction in the heart. The drugs may affect calcium absorption into bone. Older patients on medication are at higher risk for automobile accidents and perhaps falls. (This last issue exercises Peter Gøtzsche the most; he believes that the rate of falls in medicated elderly patients has been underestimated.)
But in truth, I was as conservative a prescriber before these specific risks were identified, and at a time when estimates of the protective benefits of antidepressants were, if anything, higher.
No, I have always liked to keep drug exposure to a minimum because when medication is used chronically, bad things tend to happen. One of my child-psychiatry teachers, the late Donald Cohen, used to say, “I rarely have to sit down with a mother and tell her, ‘You know that drug I’ve been prescribing for your son? We’ve just found that it makes children smarter.’” Actually, there have been reports along these general lines with antidepressants. I have mentioned evidence that the medications may protect memory in chronically depressed women. But the more usual pattern is for the late-appearing news to be bad. In Listening to Prozac, composed before long-term use of antidepressants was common, I wrote, “We know that some drugs, especially ones that are taken chronically, will have unknown or even late-appearing (tardive) side effects.”
When I say that, in clinical decision making, experience balances research findings, part of what I mean is that doctors learn from practice altogether, from work with patients with many diagnoses. The influences may be mixed. I recall studies of antipsychotic medications for schizophrenic patients that found “drug holidays” to be counterproductive. If you took patients off medicine, they risked sharp relapses, requiring vigorous prescribing, so that the overall exposure—milligrams of drug taken—increased. I saw precisely that problem in my training years. In contrast, some research suggests that when people with a tendency toward manic depression are given antidepressants without accompanying mood-stabilizing drugs, even if the acute problem resolves, the rate of future episodes may increase. We’re applying lots of not-perfectly-relevant knowledge to the case at hand.
In this context, I want to mention what I call the “nightmare scenario,” the notion that antidepressants are counterproductive. Skeptics have expanded on Ross Baldessarini’s concern over rapid withdrawal and asked whether the apparent benefits of maintenance speak to addiction. Might medicine make patients dependent on the cure? The extreme version of this concern is that in the very long term, treatment makes patients vulnerable. While they are on medication, they stay well. When they come off, they will be engulfed by depression.
If I find this claim implausible, it’s because I take people off antidepressants all the time. A handful—they took antidepressants early in their treatment and later relied on psychotherapy alone—visit regularly, a few times a year. They do fine—better than they had done in the early going, before ever taking medicine. This experience has led me to read the “nightmare” literature critically.
It tends to arise from a particular sort of analysis. A researcher will contrast patients who have done well in the placebo arm of a drug trial to patients who have done well in the medication arm of a drug trial. How do they fare once the experiment has ended?
In a follow-up phase, the patients on placebo will remain on placebo, while the patients on medication will be switched to placebo. There will be more relapses in the group that discontinued the antidepressant. The skeptics conclude that medication merely postponed trouble and perhaps did harm. Recovering on placebo leads to a steady course; recovering on medication predicts instability once the antidepressant is withdrawn.
Having seen the differential sieve in action, we will object that this result—more relapses in follow-up from the antidepressant arm—will occur if medication is effective and placebo pills are not.
Through its nonaction, placebo identifies the least shaky participants in any trial: they recover even on dummy pills. The short-term trial serves as a sieve, separating out ill people and passing through (for further study) healthy ones. Patients who recover on antidepressants are a more vulnerable group. Some will have had serious illness and responded to the medicine. The treatment arm of the trial is a broad-gauge sieve that lets through sick people.
Patients who do well on placebo will continue to do well. That’s what healthy people do. In contrast, off medication, people who responded to it may falter. That’s what sick people do. They may nonetheless fare better than they would have without treatment. Because they are effective, the drugs carry ill patients, ones who don’t respond to the placebo grab bag, through the original phase of treatment.
A trial in which some patients enter via response to placebo and some enter via response to medication is not randomized. It suffers from susceptibility bias.
I would be tempted to say that, on the continuum from randomized trials to storytelling, the nightmare reports stand at the anecdote end, except that anecdotes have the virtue of transparency. They are what they seem. But if we were to grant some such status, we might say that the nightmare calculations amount to a cautionary tale with this moral: There’s simply too much that we don’t know about the powerful medications.
I share that viewpoint, which is why, ample evidence of efficacy notwithstanding, I often find myself reluctant to prescribe. In particular, I am ambivalent about prescribing for young people, a category I extend through college age into the midtwenties. Some of the hesitancy is research based. A study has found that for adolescent monkeys, yearlong Prozac treatment has a lasting impact on serotonin transmission in the brain. That nugget of evidence—of unknown significance—comes amid a host of vague concerns. It is worrisome to intervene directly in the developing nervous system. Beyond the wish to ease suffering here and now, what makes me prescribe—and, however hesitantly, I do it all the time—are counterbalancing concerns about the enduring effects of depression. It’s important to interrupt progressive diseases. Depression now predicts recurrent depression down the road.
The situation is the reverse of the one we imagine if we listen to those who question the evidence base for antidepressant use. Well-documented hazards lead us to prescribe; unsubstantiated fears stay our hand—a balance that, to my mind, constitutes good judgment.