Switch or Stay?

Psychiatrists caring for individuals with schizophrenia are frequently faced with a difficult choice: should a patient stay on an antipsychotic regimen that has not relieved all the symptoms of schizophrenia, or should the patient be switched to another medication? There can be considerable urgency to switch. Antipsychotics are usually only partially effective, leaving patients with a substantial symptom burden. In addition, antipsychotics can cause side effects that are discomforting and unhealthy. The temptation to look for better treatments is continuous and reasonable. But we are reminded by Dr. Susan Essock and colleagues in a paper in this issue of the Journal, written from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) data, that there is another factor to evaluate in making this decision, perhaps more subtle than the potential advantages of new pharmacology, and that is the therapeutic cost of changing medication. Essock et al. have analyzed the phase 1 CATIE study for those individuals who “switched” from or “stayed” on the medication regimen they were receiving at study entry and have shown that those who switched, no matter from what drug or to what medication, always fared worse. Switching is a not a risk-free decision for the patient.

Essock et al. used the CATIE data set in a deceptively simple way to measure the vulnerability of switching to a new medication compared with remaining on the same antipsychotic regimen. As could be anticipated, many CATIE participants were already receiving second-generation antipsychotic drugs before being randomly assigned to their phase 1 CATIE medications. Thus, some of the CATIE volunteers “stayed” with the medications that had stabilized them and others “switched” to a new CATIE medication. Seventy-seven out of 314 olanzapine volunteers “stayed” with olanzapine, and 62 out of 321 risperidone volunteers “stayed” with risperidone. In general, the “stayers” did better than the “switchers” for each medication, but especially for olanzapine. Thus, if patients are taking olanzapine or risperidone and switch to another medication, they will probably have a poorer outcome than if they did not switch. Moreover, the switching vulnerability is different for each of the medications that they examined and was worst for olanzapine.

The Essock et al. findings do not address the issue of switching medications early in the course of the illness. The CATIE patients had been treated with antipsychotics for an average of more than 14 years. It is likely that these patients had experienced trials with a number of antipsychotics. In part they may have ended up on their current medication regimen because it produced better efficacy, did not have severe side effects, and they did not dislike it. It is not surprising that they ended up receiving a drug that was better for them than a random selection.

The “switch or stay” question is harder to answer in psychiatry than in most other medical specialties. Internists can monitor objective laboratory or clinical measures reflecting disease (e.g., hematocrit or blood pressure). Schizophrenia almost always responds incompletely to treatment; a patient’s optimal response is always a matter of judgment and usually difficult to assess, quantitate, and record in the chart. Furthermore, changes in symptoms may evolve over several weeks. It is important that clinicians monitor and record efficacy and side effect changes, thereby providing a picture over time of how well a patient is doing with how much short- and long-term side effect burden. Careful monitoring may preserve the clinically useful information of why a patient may do better with a given drug.

The risk of switching should not be ignored in the face of advertisements and inducements for new medications for psychosis. This noted, clinicians must always be alert for anything better in schizophrenia therapeutics. For example, findings from the second phase of CATIE (reported in the May 2006 issue of the Journal [pp. 600–610]), which were not analyzed by Essock et al., suggest that switching to clozapine may be worth the risk. If patients have not had a trial with olanzapine, this might also be considered. However, the health risks associated with both of these antipsychotics are not trivial. Careful monitoring of the metabolic effects of both of these agents should be an important component of the treatment plan. Whatever is done, it is important to remember that to calculate correctly the potential benefit of a new opportunity, a clinician should consider both 1) the therapeutic opportunity of a new drug, and 2) the vulnerability to stability of leaving the old drug. (Of course, some individuals appear to acquire new treatments only without having old ones replaced, solving the “switching” vulnerability but only at the expense of generating complexly over-treated persons, no better results, and more side effects.)

There were problematic aspects of this study and missed opportunities with this large data set. One could imagine that switching within the CATIE study with its perceived nonequivalent dosing would have generated some switches where the equivalent medication dose was lower in the “switched to” than the “switched from” condition. This could explain why “switching from” olanzapine to risperidone was the worst condition. Also, one of the primary findings of CATIE was that medications that may have marginal advantages on measures of effectiveness may be associated with adverse side effects that are unacceptable to patients and clinicians. It would have been helpful if the authors had also analyzed the effects of staying versus switching on side effect measures such as weight, lipids, and glucose. If staying with olanzapine treatment is associated with fewer discontinuations for psychosis, but changing to another antipsychotic is associated with a lower risk of serious side effects such as heart disease, then what should the clinician do? Patients and clinicians face this very difficult question nearly every day.

Dr. Essock was not originally a member of the CATIE collaboration, but she felt that the effects of switching versus not switching at initial drug assignment should be assessed. Jeffrey Lieberman, M.D., CATIE’s director, made the data available to her, and the resulting new collaboration adds unexpectedly valuable information. This open door to the data, part of the vision of NIMH in supporting this program, is a striking example of collegiality in the service of providing the best guidance for patients and their doctors.