WHEN I PRESENT my own approach to depression, intentional underprescribing, it may seem that I am ducking an important part of the antidepressant controversy, the argument over whether medication is handed out promiscuously. I share the concern. As antidepressant use has accelerated—as drug residue becomes detectable in our water supply—it has been hard not to imagine that some of it is excessive.
Overprescribing can prove hard to document. I know as much from my work with Gerry Klerman. In the early 1980s, he tasked me with composing a reply for the signature of the surgeon general, Julius Richmond, on whether “mother’s little helpers” such as Valium were being overprescribed for women. Experts on my working group found that women got more mental health treatment in general, starting with psychotherapy. Patients prescribed for arrived with high levels of impairment. Anxiolytic-related suicides had dropped. So had recent prescribing. And so on. Despite widespread unease at the level of drug use—it looked excessive—without research “on the ground,” surveys that asked doctors why they had acted in each case, fault was hard to pinpoint.
The objective research on antidepressant prescribing has been similarly short on detail. Surveys find that many patients on medication have no relevant diagnosis in their charts, but do busy GPs record the justification for each treatment they offer?
My own experience gave no special cause for worry. In my office, I had not seen anyone arrive medicated for indications I considered trivial or cosmetic, not a single case in decades of practice. Correspondence and phone calls I had fielded, a steady stream, for years, told the same tale, substantial need. Meanwhile, I continued to hear from and be referred patients treated with psychotherapy only, and no medication, despite long intervals of debilitating depression.
That is not to say that when medicine was prescribed, I agreed with every choice made. Some of the new patients who arrived on antidepressants I might have treated longer in psychotherapy alone. Certain regimens alarmed me: too many drugs, or unlikely combinations. Still, every patient had a condition that seemed to call for treatment. Sometimes when I tried tapering an implausible fourth medication, depression recurred. I came to respect the out-of-town “cowboy,” the aggressive prescriber who had discovered what worked.
My practice was privileged. Most patients had seen topflight doctors in the past. Still, I imagined that if there were massive unjustifiable prescribing, I would have encountered at least one example.
Lately, the quality of research on prescribing has improved. Probably the best data come from a household survey of people willing to discuss their medical histories. The questionnaire used ignores some mental illnesses and underdiagnoses others. In interviews, respondents may not disclose symptoms that they complained of to their doctors in times of distress. Still, the results present a revealing picture of the match between treatments and ailments.
One widely publicized analysis, from 2014, looked at residents of Baltimore who had been followed with periodic interviews for twenty years. Thirteen percent were using antidepressants. Only 31 percent of those on medicine had ever met diagnostic criteria for major depression. Another 31 percent had been diagnosed with one of four serious anxiety disorders known to respond to antidepressants. But (according to diagnoses arrived at through answers to survey questions) 38 percent had never had any condition where, by the researchers’ standards, high-quality evidence justified prescribing.
Having “ever met” sounds like a liberal criterion. It covers doctors treating current episodes and preventing future ones. Even among patients who fit that description, some may be using medicine unnecessarily. In that light, the figure for the remaining patients, 38 percent, is alarming. Is it almost the rule that people taking antidepressants are on them for no good reason?
The press treated the data that way. The usual news lede declared that two-thirds of patients on antidepressants were not depressed. The website for HLN, the Dr. Drew cable network, ran the headline “Antidepressants Use Being Abused, Study Suggests.” But to my reading, the Baltimore analysis is less about overprescribing than about evidence-based medicine, narrowly taken, and the concern that doctors fail to practice to it.
We know enough about antidepressants to ask whether five diagnoses cover the territory. The researchers believed that research on dysthymia was insufficient and so excluded it as a proper cause for treatment. Likewise, they questioned the efficacy of antidepressants as sleep aids. (Trazodone, a precursor of Serzone, is often used, in low doses, as a sedative. Because it is thought not to be addictive, trazodone is especially popular in the drug-treatment community.) The figures ignore antidepressant use for smoking cessation, premenstrual syndrome, post-traumatic stress disorder, and many other indications. Minor depression was specifically excluded. (We may recall the Dartmouth study, which contained hints that in minor depression antidepressants can be helpful for patients who suffer few but disabling symptoms.) Seven percent of those interviewed were stroke survivors.
The Baltimore analysis was not about whether doctors had prescribed thoughtfully, it was about whether the prescribing went beyond what strict standards of evidence could justify—or rather, it assumed that thoughtfulness meant accepting rigorous constraints on evidence.
By way of contrast, it is instructive to see how other researchers, from Harvard and elsewhere, approached the national version of this same data. The group was examining a different question, whether Americans use mental health services unnecessarily. The investigators created an ever-broadening series of categories to detect reasons for seeking care.
In a given year, over a quarter of those who sought help had never qualified for a diagnosis of mental illness. But an additional 10 percent had revealed “some other indicator of possible need,” such as a history of hospitalization for mental illness or a report of a recent substantial stressor like rape. (It is a commentary on the limitations of the interviews that a number of people who had spent time in a mental hospital had, on the basis of symptoms they disclosed, been categorized as never having suffered a mental illness.) Many of those in treatment fell one symptom shy of a full diagnosis. And so on. In the end, the analysis was left with 8 percent of respondents who had received services for no apparent reason—and many of them had been seen in “alternative” settings, for interventions like prayer. It may be that almost everyone who consults a doctor for mental health care arrives with substantial need.
I am not suggesting that needing mental health services is the same as needing antidepressants. I mean only that survey data can be utilized in different ways. You can propose strict standards and report on instances in which caregivers fail to meet them. Or you can consider indicators of distress and disability to estimate how much treatment may be, in a broad sense, warranted. The first method asks whether doctors follow certain rules. The second is more generous: Do the data hint that doctors and patients may be acting sensibly in the face of disorders for which treatment is always, finally, empirical?
Let’s imagine a Baltimore householder who will (shortly, when interviewers arrive) land in the 38 percent. This patient shows up at a mental health clinic. He has, perhaps, a history of recent stroke or of dysthymia. Or he has achieved sobriety and now finds his stability threatened by insomnia—sleeplessness that has not responded to what the doctor recommended at a prior visit, attention to sleep hygiene. He has, our hypothetical patient, disabling minor depression—persistent, despite weeks of pastoral counseling. He arrives seeking relief from a medical problem, a typical psychiatric problem.
What should his doctor do?
Evidence-based medicine offers scant guidance. Regarding dysthymia and much else, the Baltimore researchers judge the pharmacology trials inadequate and would be bound to consider psychotherapy trials less convincing yet. What is at issue is not the psychiatric equivalent of routinely dispensing antibiotics for the common cold. Instead, when doctors prescribe antidepressants, often we are seeing the resolution of clinical dilemmas in the face of limited research.
In almost all cases, the need will be substantial—and scattered, imperfect evidence will suggest that antidepressants can help. A fair proportion of that ominous 38 percent may represent thoughtful patient care.
From a clinician’s standpoint, the 38 percent figure stands as a judgment on the research base, which should be broadened and, equally, against the notion that strict adherence to evidence-based medicine is possible—a judgment against the requirement that, failing validation by randomized trials, doctors must not act. Practice on that principle is impossible. Most of what patients arrive with stands beyond the reach of well-researched treatment, and for some conditions doctors nonetheless have a fair notion of what might be worth trying.
The evidence on trazodone as a sedative is mixed, and yet, prescribing, I have had the occasional gratifying success. A patient with chronic insomnia begins to sleep through the night, and as a bonus, she finds relief from the chronic low-level anxiety that had been with her on waking. I have had patients whom I weaned off complex medication combinations and who continued to do well with, as their sole treatment, a low dose of trazodone at bedtime.
Nothing about the try-this-then-that approach to, say, disabling minor depression is unusual. Consider the treatment of headache. With a migraine sufferer, a neurologist will run a series of practical trials, each using a medicine of limited efficacy, until one does the trick, interrupting attacks, decreasing their frequency and intensity, or preventing them outright. Knowing more would be better—what’s likely to work for whom and why. But there’s no shame in employing expedients that don’t, on a population basis, yield highly favorable numbers needed to treat. That an intervention works decisively for some people is evidence, too.
Our wishes stand as a fair test of our commitment to narrow-gauge evidence. Thinking of members of that ill-defined group, the 38 percent, do we wish that many of them had received psychotherapy?
If so, should we disown the wish? Not every figure that finds its way into print (that low effect size for psychotherapy!) should shape doctors’ practices.
Whatever our wishes, the reality is hardly generous. In a given year, almost a third of American adults with major depression receive no medical attention for it.
No one knows how often antidepressants are misused. If we were to study the “health-care delivery” question better, through having specialists rediagnose patients treated in the regular course of things, we would still need standards. Regarding appropriateness, there is substantial disagreement. We may differ even (thinking of Doris Mayer) about what counts as a favorable outcome. Is lessening symptoms always for the best?
I don’t doubt, finally, that antidepressants are overprescribed, prescribed inaccurately and unthinkingly, prescribed with poor follow-up, prescribed for too long, and the rest. There are medical horrors I never see—results of treatment in Medicaid mills, for instance, horrors that extend beyond mental health care. Nursing home physicians seem to overuse psychotherapeutic drugs. I worry a great deal about an area of practice I have little access to, the treatment of children. My impression is that antidepressants work unreliably in children and carry serious risks—known ones and, more frighteningly, unknown ones hinted at by our knowledge about the vulnerability of the developing brain.
But thinking about the different takes on the door-to-door survey data: doctors are prescribing for people in distress and generally with plausible reasons for doing so. I take minor comfort from what I have seen personally, judicious use of medication in adults. I confess to faith in my colleagues; with infrequent exceptions, I respect their work. And always, I worry about stigma. How much of our concern do we expend on that unfortunate remnant, the third of sufferers who receive no attention at all for this highly treatable affliction, depression?