WHAT DO WE make of what psychiatrists provide? I mean, fiddling with drug doses, listening, interpreting, everything? Increasingly, researchers have tried to measure the benefits of expert care. We’ll look at this effort from two angles, two takes on how psychiatry contributes to depression treatment.
One line of inquiry concerns psychosomatics. Depression is a common accompaniment to other illness, and public health advocates and cost cutters are intent on treating psychiatric disorders in medical settings. The result has been research on what is called collaborative care.
For example, the COPES project (for Coronary Psychosocial Evaluation Studies—run at Columbia University and elsewhere) asked whether improved psychiatric treatment makes a difference for depressed patients with heart disease. The program—it began in 2005—targeted patients who had suffered a recent heart attack or were experiencing unstable angina, pain that signals reduced blood flow to the heart. Those conditions put patients at high risk of further deterioration or death.
Discussing psychosomatics, we have considered research that takes on all comers. The COPES group used a different approach. They enrolled patients and then waited three months. Only those with persistent heart disease and persistent depression entered the experiment. The run-in guaranteed that everyone had real mood disorder, generally mild-to-moderate depression.
The cardiac patients came as they were, a few already in psychotherapy, a few more on antidepressants. The researchers randomized participants into two groups. One continued to receive their usual medical care. The other got enhanced treatment, either psychotherapy (focused on problem solving) or antidepressants administered by experts. The patients chose.
The researchers also followed a group of similar cardiac patients who were not depressed. There were almost 240 participants in all.
Periodically, patients in the enhanced-care group underwent reassessment. If they were not doing well, they could switch to a different intervention or add one—pharmacotherapy or, for those not yet in it, psychotherapy. The question was whether psychiatric care, done to an acceptable standard, matters.
In their usual care (the control condition), cardiac patients got plenty of assessment, attention, and pills. The grab-bag elements were present.
Over six months, collaborative care yielded an effect size of 0.6, medium-plus. With depressed cardiac patients who are already in good hands, more systematic treatment brings improvements at a level rivaling any found in the mental health arena.
The researchers also tracked “major adverse cardiac events,” such as recurrent heart attacks. The attentively treated group had significantly fewer. Ten adverse events occurred in depressed patients receiving “usual care,” five in the never-depressed group, and three in the group given enhanced psychiatric care. Yes, there was the usual confound: two of the five antidepressants offered were ones that may protect the heart. Still, treating eighty people for six months appeared to spare them seven major cardiac events, a level of benefit that, if it held up, would rival results for any known preventive cardiac intervention.
Comparable outcomes have been found from collaborative care for stroke survivors, women attending obstetrics and gynecology clinics, and less select groups: the elderly, adolescents, and primary-care-clinic patients. Some of these trials omit psychotherapy and offer more attentive pharmacology, on its own, as the experimental condition. The largest body of evidence, for diabetic patients, shows that the benefits of enhanced psychiatric attention begin with decreased depression and extend to better control of blood sugar.
These trials are not perfect—they can hardly be double-blind—but most have the advantage of inclusiveness, attracting a range of patients, many new to mental health care. In the treatment efforts, antidepressants play a prominent role. If they didn’t work—and if skilled fiddling with them made no difference—the trials could hardly show efficacy.
* * *
Researchers have used a different approach to evaluate psychiatrists’ care for their more customary patients, hard-to-treat depressives. Two large, lengthy studies, beginning in 2001 and 2002, have looked at prescribing and, to a lesser degree, psychotherapy, offered in ways that resemble what doctors provide in their offices.
The better-known effort is called STAR*D, Sequenced Treatment Alternatives to Relieve Depression. The trial was not controlled, a downside that had its good points. Depressed patients, recruited in primary-care and psychiatry clinics, could sign on with confidence that they’d get a proven treatment. Those with additional psychiatric or medical disorders were welcome. The intent was to attract a range of patients.
Only the sickest came. For nearly 80 percent of the participants (more than twenty-eight hundred were tested), the disorder was chronic. The average length of the current depressive episode was over two years, generally despite attempts at treatment. The average enrollee had lived with depression on and off for over fifteen years and was now in a seventh episode. Most patients were alcoholic or had other forms of mental illness. The average Hamilton score at entry was 22, moderate depression or, in the inflated nomenclature Kirsch criticized, “severe.” STAR*D showed how psychiatry does with its failures.
In the first phase, patients were put on Celexa, managed by their own doctors. About 30 percent of patients achieved remission within weeks—virtually no symptoms. Responses (including remissions) ran at just under 50 percent.
Commentators considered this outcome disappointing, but is it? A decade and a half into a career of depression, during an episode that’s lasted two years despite treatment, medically complex, sometimes alcoholic patients have a fifty-fifty chance of responding to the first drug they’re offered. Does any specialty do better with its difficult cases?
It’s unlikely that classic placebo effects played a role: these patients had failed on all sorts of pills. Effectively, what was being tested was enhanced care—adequate medication doses in a supportive environment.
In later phases of STAR*D, patients were offered different or additional medicines or psychotherapy. If they stuck with the trial, almost 70 percent of these difficult patients remitted—but the dropout rates were astronomical, and often patients who improved then relapsed. The patients who had been sickest at the start were likeliest to remain depressed.
Psychiatry was divided about the meaning of these results. While many patients remained symptomatic, the pace of STAR*D was deliberate, the treatments were not aggressive, and even so, with the first or second medication used, most participants emerged from depression. Participants who entered with months or years of mood disorder mostly responded within weeks, a sign, in my view, that psychiatric care can make an important difference.
Although STAR*D was more flexible than typical drug trials, it left room for a more interesting inquiry. What if psychiatrists are left to their own devices? Can they do better?
Here’s where fascinating, underreported—and misreported—research from the same era comes in. The Penn-Vanderbilt team, joined by doctors at Rush, in Chicago, undertook a study of vigorous prescribing. Experienced pharmacologists were invited to push depressed patients hard, using any medication or combination of medications, and get them into remission. Jan Fawcett, who had helped design the NIMH collaborative trial and is now in his eighties, oversaw the prescribing.
Besides antidepressants, half of the patients (there were 452 in all) received intensive cognitive psychotherapy of a sort that, in prior research, had proved highly effective with depression. In the new study, psychotherapy would continue, sometimes twice weekly, for eighteen months or until patients’ symptoms disappeared.
Once patients had enjoyed four weeks in remission (or substantial responses just shy of it), doctors worked to get them to recovery—six more good months. Because it focused on remission and recovery, I think of this trial as the R&R study.
The treatment was insistent. If a patient was already on an antidepressant, the pharmacologists raised it quickly to its upper limit. They waited only briefly for signs of progress. If none appeared, the doctors added a second medication or tapered the drug and chose a different one.
Not all the details have appeared in print, but between one article and a series of conference presentations, an overview has emerged. As in the STAR*D study, the patients were challenging, with severe or chronic depression, sometimes in conjunction with other conditions. As in STAR*D, the average starting Hamilton score was 22.
In R&R, the remedies used were diverse. Often, patients ended up on two or three drugs, including augmenters, like thyroid and lithium. Some patients needed antipsychotic medications. A few did well with older antidepressants, like imipramine. A handful responded to amphetamine or medicines like Valium, often in combination with conventional antidepressants. But many patients remitted on a single antidepressant. Selection and (full) dosage appeared to be important.
By the eighteen-month mark, about 80 percent of those who started the trial had made it to some level of remission. Many had gotten there much earlier. Almost everyone who remitted also stayed well for six months or more.
Psychotherapy played a limited role. In the R&R study, when 77 percent of patients remitted on medication, combined treatment pulled that figure up only to 80 percent. On its own, in prior research, perhaps with less ill patients, that same cognitive therapy had achieved a 40 percent remission rate in sixteen weeks. The R&R study stands as further evidence that the effects of psychotherapy and antidepressant medication are not additive. For symptom relief, most of what medication and talk accomplish together can be done by medication alone.
The press gave a blindered view of the R&R results. Most reports focused on a subgroup of participants who benefited from psychotherapy. It inoculated certain patients, those who arrived with “severe and nonchronic depression”—isolated intense episodes—against relapse.
But when you divide data all sorts of ways, all sorts of patterns emerge. Patients with more common levels of depression—Hamilton scores below 22, typical for outpatients—seemed to do better on medication alone (82 percent remitted) than on combination treatment (72 percent).
Looking at all patients with chronic depression, more (70 percent) recovered on medicine alone than on medicine-plus-psychotherapy (63 percent). That difference was not statistically significant, but even if therapy did nothing to impair the effects of medication, the result hardly made the case for combined treatment. No one was keen to broadcast this finding, that for chronically depressed patients offered vigorous prescribing, psychotherapy adds little.
Now, I don’t believe that result, any more than I believe that psychotherapy has a special role for severe and nonchronic depression. It may be that, on its own, skilled pharmacologic care is hard to beat. Perhaps the prescribing doctors were so evidently expert that patients invested little in the therapy. What happens in the office of a psychiatrist who does both jobs, talk and prescribing, may be a different matter.
Still, the R&R study presents a challenge to evidence-based medicine and the press: Are we evenhanded in our consideration of test results? Where must we let numbers tell their story, and where may we reserve room for judgment? In the treatment of chronic depression, to favor combined treatment, psychotherapy augmented by medication, is to favor common sense over formal evidence.
The psychotherapy findings aside, the R&R results are reasonably encouraging. With the passage of time or intensive medication management, hard-to-treat patients emerge from depression, a result that holds throughout the severity spectrum. The recoveries have substantial stability. (In a presentation, the researchers said that even patients who “relapsed” tended not to return to their prior level of depression.) Some of this depression would have remitted with no treatment; most episodes end, give or take residual symptoms. But the unimpressive effect sizes we’ve seen in brief drug-company trials translate into substantial effectiveness in practice.
Treated by the R&R pharmacologists, the vast majority of patients made it to remission, and with low dropout rates. Almost all completers got better and stayed better. If the study’s participants were comparable to those in STAR*D, then the exercise of judgment had made a difference. We may wonder, What is it that a skilled psychopharmacologist might know?
* * *
There’s a gap between what academics believe and what doctors act on. Conventional wisdom has it that symptom patterns give no clue as to which medicine will help whom. All the same, clinicians try to match treatments to patients.
Here I want to edge myself out on a thin limb. I see indications that the conventional wisdom is mistaken.
The commonest notion about drug selection runs along these lines: Medicines like Celexa that work via serotonin will help patients whose depression is colored by anxiety, where that term embraces timidity, obsessiveness, social unease, and self-doubt. That was the group I offered Prozac when it became available. In contrast, medicines that have mixed effects or work via norepinephrine—desipramine, Wellbutrin, perhaps Effexor, and especially reboxetine—will most help patients with lethargic depressions, where low energy is a distinguishing feature.
That’s not to say that certain medicines treat only certain symptoms. Depression is a syndrome; as it remits, symptoms tend to lift in concert. When anxious patients get better on Prozac, generally their energy will return. When reboxetine works, it relieves anxiety as well as lethargy. But perhaps symptom patterns can predict who will respond to what.
In 2005, a team from the University of Catania, in Sicily, published a study of targeted prescribing. They had been working with depressed stroke survivors. The prevailing view was that in poststroke depression, brain pathways that use serotonin are disrupted, so that medicines like Prozac work best. In a bold move, the doctors in Catania had evaluated reboxetine, whose direct effect is on norepinephrine.
The twist was that they limited the trial to stroke patients with depressions characterized by passivity, low energy, flattened facial expression, drowsiness, and slowed speech and movement. If norepinephrine-based antidepressants treat lethargic depression, perhaps reboxetine would help these stroke victims after all.
The Italian trial enrolled only thirty-one patients, but the results were spectacular.
There were no dropouts. Psychosomatic research has that advantage. Stroke survivors attend checkups. Here, patients given placebo did not change at all. At zero, four, eight, and sixteen weeks, their average Hamilton scores ran 24, 24, 24, and 23—substantial depression. The figures for patients on reboxetine dropped steadily: 24, 15, 12, and finally 9—mild or subthreshold mood disorder. The identifiers (passivity and so on) had picked out reboxetine-responsive patients from a group thought to need drugs from a different class.
In another study, the same research team divided a larger sample of stroke survivors into two groups, those with anxious depression and those with lethargic depression. Half in each category got Celexa and half, reboxetine. The wrong drug (Celexa for patients with bodily slowing, reboxetine for anxious patients) did nothing. The right drug lowered Hamilton scores decisively.
Perhaps it is because they were conducted on neurological patients that these experiments have gotten no play. But the Sicilian research suggests that the clinical lore is right: the symptom picture offers hints about who will respond to what.
Later, Richard Metzner, a clinician at the Semel Institute for Neuroscience at UCLA, attempted a loose replication of the STAR*D trial, only matching patients, 117 in all, to antidepressants through a system similar to the one used in Sicily. He achieved a 59 percent remission rate at six weeks, about twice the benefit obtained with Celexa alone in the first phase of STAR*D.
The standard view remains that there is no reliable way to select antidepressants. But if, as these small and quirky studies suggest, prescribing is an art, then conventional drug trials, where random assignment determines the treatment, underestimate antidepressants’ potential.
Whatever the truth about clinical expertise, psychiatrists will almost certainly improve at matching drugs to patients. Researchers have begun identifying genetic markers that correlate with response to one or another sort of antidepressant. In most studies, the predictive power is weak. You would need to genotype dozens of patients to produce one remission more than you would get through randomly choosing a medication.
But a recent trial found a genetic variant with a “number needed to screen” of 10: you would need to gene-test only ten patients to obtain one remission more than you would get by randomly assigning patients to take an SSRI (Zoloft or Lexapro) or, in contrast, an antidepressant that directly affects the brain’s handling of serotonin and norepinephrine alike (a form of Effexor). This level of advantage might not be quite enough to be employed clinically, but even a little further progress might make insurers willing to pay for testing to avoid weeks of futile treatment efforts.
If the testing pans out, then even using today’s drugs, efficacy results will rise and emperor’s-new-clothes claims will fade. It is implausible that genetic markers are specifying responses to one placebo as opposed to another—which is also to say that even now it is absurd to deny that Zoloft, Lexapro, and Effexor are real antidepressants.
In the genetic studies, patients follow the pattern that David Healy, in Wales, had noticed. They do better globally, reporting fewer side effects, on their preferred antidepressant (the one that genotyping predicts will work for them) and experience adverse responses—low efficacy and numerous side effects—on nonpreferred ones. That’s why average response rates in drug trials are misleading: the figures, weighed down by results from patients who worsen, mask marked improvement in those whom treatment helps.
We now have a better-rounded picture of antidepressants’ potential. We’ve known that patients with uncomplicated mood disorders, the type treated by internists and family doctors, have high response rates on the first drug they’re prescribed. With tough cases, it can take longer to find the right regimen, but with expert prescribing, over three-quarters of those who start treatment enjoy substantial improvement. This ability to make a difference—with difficulty, in time—with hard-to-treat patients is another reason that ours has been the antidepressant era.