For many therapists, an air of mystery surrounds the role of psychopharmacology in mental health treatment. It seems like a parallel universe, standing apart from the usual concerns of psychotherapeutic encounters. In the interview below, psychiatrist Steven Dubovsky—author of Clinical Guide to Psychotropic Medications and chair at the University of Buffalo School of Medicine and Biomedical Sciences—offers a step-by-step tour of the complexities of the psychiatric consultation while exploring the overlap between psychopharmacology and psychotherapy.

What’s the process you go through when a therapist sends you a patient for a consultation?

Steven Dubovsky: For any condition, I go through an algorithm of the most likely causes of treatment being stuck. About 40 percent of the time, it’s because the patient’s diagnosis is incorrect, but sometimes people actually try to hang on to their diagnosis, especially if it sounds exotic. Some people say things like “I’m a rapid-cycling bipolar with mixed and psychotic features” with glowing pride, as if they’d happily write it on a name tag. Or, as if the consultation were some kind of show and tell, they might say, “I’ve got multiple personality disorder. Do you want to see some of my personalities?”

Since I always tell people that they can bring along anyone they want for the initial consultation, when I decide that the diagnosis isn’t correct, which I often do, I’ll ask the other person what he or she thinks. This is useful because a lot of times when a patient says one thing, his spouse will be sitting there shaking her head. For example, I might ask the patient, “Are you often irritable at home?” And he’ll say, “No, that’s the one thing I’m not. I’m not irritable.” But the spouse will say, “Not irritable? Didn’t you throw a vase at me last night?” Of course, the patient may have genuinely forgotten about throwing the vase. Part of that is because recall is state dependent. At that moment in my office, the patient is in a different emotional state from when he threw the vase.

What role does diagnosis play in your assessment procedure?

Dubovsky: Well, the correct diagnosis should inform, at least in general, what the treatment might be. Of course, this is a tricky issue because DSM diagnoses, by and large, aren’t that reliable and have questionable validity in predicting treatment response. But where diagnosis really changes your treatment is with people with complaints of depression or attention deficit disorder, or a personality disorder with a lot of affective lability. For example, when someone has an underlying bipolar mood disorder, a prescription of antidepressants often makes the mood disorder worse. In that case, you’re going to take a step back and change the medications completely.

As someone who’s regularly in the role of being a troubleshooter when drug treatment goes off course, what are the two or three factors you usually find are most responsible?

Dubovsky: Common causes of refractory conditions in the cases I see are undiagnosed bipolar mood disorders, medical conditions that are aggravating or even causing the supposed psychiatric illness, and failure to recognize the effect of substance abuse on the patient’s symptoms. But the most common are probably noncompliance, failure to recognize underlying family problems, and clinicians getting too focused on symptoms to get the big picture of the patient’s life.

Again and again in the cases referred to me, I discover that even when the treatment approach being taken isn’t working, the therapist has continued to do more of it. So if you’re talking about the patient’s past and the patient is getting worse, some therapists respond by saying, “We just haven’t resolved this yet. We’ve got to work harder on it.” Or even as the patient has become increasingly overstimulated and disorganized, the therapist might say, “This is proof that we’ve touched a nerve and are finally getting somewhere.” Or the patient says, “I can’t stand this. I feel worse every time I come here.” And the therapist replies, “You have to feel worse in order to feel better.” I remember one unhappy patient who told his therapist, “Is there a law against feeling better? What’s the matter with that?”

In the best cases, how should psychopharmacology and psychotherapy work together?

Dubovsky: In the kind of complicated cases I see, people are rarely cured just by a medicine—or just by therapy, for that matter. When treatment works, the two complement each other. Medications are especially useful with conditions in which avoidance plays a major role, such as obsessive compulsive disorder, various anxiety disorders, and some forms of post-traumatic stress disorder. They can reduce levels of arousal that otherwise make it impossible for patients to integrate any new learning. By reducing those high levels of arousal, they make exposure, relaxation, and cognitive restructuring possible. In contrast, psychotherapy sometimes makes it possible for the medications to work.

Recently, I saw a depressed patient who refused to take the Zoloft his primary-care physician had prescribed. At the end of his first session with me, I said, “An antidepressant wouldn’t be recommended in your case. You have a bipolar mood disorder, not a unipolar depression. You need a mood-stabilizing medication.”

“Well, I’m not taking any of those medications,” he said. “They’re only for crazy people.” When I asked him where he got that idea from he answered, “My mother had a bipolar mood disorder, and she was sure crazy. I’m not like her.” After a fair amount of work getting him to accept the idea that a medication might help him regulate his mood better, he said, “Maybe there’s a medicine that might help, but I want something that nobody would really think is a mood-stabilizing medicine. I don’t want people to think I have a bipolar mood disorder.”

This particular patient had experienced the tragic and traumatic loss of a child but was incapable of grieving because his mood was completely out of control. Essentially, every time he thought about this loss, he’d become manic and start yelling at people. When we finally agreed on a medicine that settled down his mood, he was able to work on this unresolved grief. But as he made progress, his wife, who felt deeply responsible for this loss, grew furious because she couldn’t tolerate the guilt she felt. Once this pattern emerged in the family, however, it became possible for both the husband and wife to move forward in their lives together.

What advice might you give nonmedical therapists about how we can be more effective with the kinds of patients you’re talking about?

Dubovsky: Number one, psychiatric medicines don’t magically cure everything, but they aren’t placebos either. There’s scientific evidence that they’re clearly effective for certain types of conditions. In fact, for simple, uncomplicated types of depression and for some simple, uncomplicated cases of anxiety, primary-care physicians can do just fine prescribing them without our involvement. But when cases get more complicated, you need a wider, more discerning perspective.

As a therapist, if your therapeutic approach isn’t producing the desired effect, you’ll want to go through the kind of algorithm I was talking about earlier. Is the diagnosis right? Is this the right treatment? Is anything else going on that might be interfering with the treatment? What else is happening with the patient’s family? Have I involved the family? You can’t just say, “Let’s do more of the same.” You need to have a hypothesis about the expected outcome within a specified period of time. If that outcome doesn’t occur, your hypothesis is disconfirmed and you have to go back and form a new one.

We have far more psychopharmacological choices today than we had 25 or 30 years ago. Do you feel as if the quality of treatment that’s offered to patients has improved?

Dubovsky: I think the amount of treatment has certainly increased. Whether its effectiveness has increased, I don’t know. For one thing, if you look at studies of psychiatric medicines, almost all the studies have been designed and conducted by the pharmaceutical industry.

These studies follow a protocol whose purpose is to get FDA approval for the drug, not to find out what works best for the patient. They usually use a method called sample enrichment. First, you treat patients openly with a medicine and you see if it works. If it does, then you randomize them to that treatment or either comparison treatment or placebo group conditions. But you kick out everybody it didn’t work for. So the best you can say at the end of that study is that if it worked at first, it’s likely to keep working.

The other problem is the outcome measures in most of these studies are symptom-rating scale scores. To certify an antidepressant as effective for depression, it has to make you 50 percent less depressed on a depression-rating scale. Now, tell me how many people are happy to only be half as depressed as they were? And for an antipsychotic drug to be determined to be effective by the FDA, you only have to be 30 percent less psychotic. Well thanks a lot! Plus, you’re looking at segmented populations. Who’s excluded from these studies? Substance abusers, suicidal patients, patients with any type of medical or psychiatric comorbidity, patients with treatment resistance. So the data we have are quite limited.

That’s sounds pretty discouraging, but at the same time you have an optimistic view of our ability to help patients under the right treatment conditions.

Dubovsky: I rarely see patients that I don’t help—less than one percent—and I’m seeing really sick patients. But it takes a lot of work, especially with complex conditions. Typically, you have to combine different kinds of treatments. To my mind, this isn’t discouraging at all. It’s what happens in medical treatment with conditions like HIV or cancer, or even hypertension. Nobody starts out with a single treatment anymore. The research shows that patients respond best when it’s the right combination of treatments. But you have to figure out how to sequence them and how to combine them to cure a particular condition or suppress it maximally, rather than just make it a little better.

When you follow that approach, however, it’s hard to get definitive data. There are just too many treatment variables. I think this has inhibited our scientific progress. The other thing that’s inhibited our progress is that treatments in psychiatry are probably more effective for certain dimensions or features of illness than for the illness itself. For example, some SSRIs that work for depression may be a little bit better for depression with rumination or depression with a lot of anxiety or for slowing down arousal, whereas other antidepressants, like Wellbutrin, may be a little better for patients who are already more slowed down. The question is how do you combine these treatments? And that’s where our data are limited.

Looking ahead, what are the developments you see right now that are going to lead to a higher psychopharmacological standard of care?

Dubovsky: One ray of light is the movement to defining endophenotypes in psychiatric disorders. An endophenotype is an observable piece of data—it may be physiological or psychological, and it’s generally present in more than one diagnosis. One example of a well-defined endophenotype is defective sensory gating. What this refers to is the inability to shut off irrelevant information. So let’s say I put some headphones over your ears and do an EEG, and I put a click through the earphones. The first time that click goes through your temporal lobe it gives a little spike like, “What was that?” The second time I put the click through, you don’t get the same spike because you decided it was irrelevant. If you’re schizophrenic, you put that click through, get a deflection, put another click through, and get a bigger deflection—and it never shuts off. This is because people with schizophrenia can’t gate or filter out irrelevant information. You can say that’s a marker for schizophrenia, except it also appears in bipolar disorder. It also appears in relatives of people with schizophrenia, and it occurs in a number of other conditions.

It turns out there are medications that will improve sensory gating regardless of the diagnosis it’s associated with. In fact, one substance that clearly improves sensory gating is inhaled nicotine. There are other medications as well that work on this and on a variety of conditions for the gating problem.

Another example of an endophenotype is anhedonia, the inability to experience pleasure. This transcends depression. It’s a feature of depression, but it’s seen in other conditions as well. This appears also to be a reproducible trait. So the more we’re defining these reproducible traits, we can aim our treatments at them, rather than a diagnosis, because the diagnoses are quite unreliable. In fact, the next big step for psychopharmacology may be aiming your treatment not at the diagnosis, but at more precise aspects of people’s functioning. So let’s say we’ve helped you filter out information better. What’s the rest of your problem? Well, maybe every time you get rattled, you start hearing voices. Okay, now we need something aimed at hearing voices, which would be another endophenotype. So as treatments can be studied as they act on specific dimensions of illness, we’re going to get much more predictable results.

Another point has to do with some of the genetic studies. Let’s say you want to look at the genetic contributions to bipolar illness. Well, there are probably 50 types of bipolar illness. Some have an early onset, some have a late onset, some have a lot of family loading, some have no family loading. Those are different conditions with different treatment responses, so we’re starting to look at different markers within that broad diagnosis.

Here’s one example. I’ve done quite a few studies over the years, replicated in multiple laboratories, that show that bipolar disorder is associated with an elevated level of free calcium ion concentration inside peripheral cells—no matter what peripheral cell you study. We’ve studied platelets mostly, but other cells as well. Not everybody has this marker with a bipolar mood disorder, but a subset do. And that elevated calcium signal seems to predict a better response to Lithium than do other types of bipolar illness. That’s an example of a type of marker that may end up producing a more specific guide to treatment and a more customized approach than what we’re capable of right now.

Do you have any final words of wisdom for us?

Dubovsky: I think the main take-home message to anybody working in psychiatry is don’t give up. We have a vast array of treatments, and I’ve used most of them. Most of them work for somebody; it’s just not clear yet who responds best to what and what combinations of treatments are best for whom. So that’s the next step in our research. Until that information becomes available to us in controlled studies, we have to be clinical scientists. We have to do hypothesis testing and keep working until we get it right. If we’re persistent, I think we can cure or substantially help a lot of people, so long as our persistence doesn’t involve doing the same thing that doesn’t work over and over again.

Steven Dubovsky

Steven Dubovsky, MD, is Professor and Chair of the Department of Psychiatry at the University at Buffalo, SUNY and Adjoint Professor of Psychiatry and Medicine at the University of Colorado.