The same antipsychotic medication that is effective in the acute psychotic episode of the illness can often be continued in the maintenance phase, where the primary goal is no longer acute treatment but relapse prevention. In an era where recovery becomes the focus of treatment, we expect medications to do more than just prevent relapse (1). Broader recovery-oriented goals include—as much as possible—eradicating or at least attenuating the broad range of symptoms of schizophrenia that can wreak havoc with our patients' lives (2). The burden of pharmacological treatment can also be unrelenting, and we should always endeavor to reduce side effects that cause distress, even if they are not medically serious, and minimize the medical (3) and neurological (4) risks that accompany long-term antipsychotic treatment. These dilemmas have been around for a while (5) and are still with us today (6).

Treatment is straightforward when all of these goals are addressed by the same drug. Unfortunately, we usually need to consider trade-offs when choosing among medication options. What happens if the goal of continued stability collides with the goal of reducing the burden of treatment or persistent symptoms? Is it worth risking current stability to find a treatment that might be more effective or may improve life expectancy many years from now? Finally, if a change in medication is considered, what are the potential benefits and risks of switching from one antipsychotic to another?

In 1974, George Gardos published the first modern report on the risks of switching antipsychotic medications (7). This study examined patients who were randomly assigned to either stay on their current antipsychotic or switch to another antipsychotic. On follow-up after 2 years, researchers observed that changing antipsychotics was associated with an increased risk of destabilization; 44% of the patients who switched medication showed "clinical deterioration," compared with only 12% of the patients who stayed on their current medication. Over 30 years later, in a reanalysis of the Clinical Antipsychotic Trials in Intervention Effectiveness (CATIE) phase I results, Essock and colleagues (8) found that participants entering the CATIE study who had to switch medications were more likely to discontinue that medication than those who were assigned to stay on the medication they entered with. No surprises here. We have to conclude that switching medication for patients who are stable on their current antipsychotic carries some risk of destabilizing treatment for the patient in the near future. In other words, the decision to switch antipsychotic medications cannot be taken lightly, and there should be a sound clinical reason to recommend changing antipsychotics in patients who are stable on their current medication regimen.

Switching for antipsychotic-induced weight gain or dyslipidemia can be one of those sound clinical reasons. It is very painful to see patients make remarkable progress in their recovery only to die at an early age. While the excess morbidity in schizophrenia predated the arrival of newer drugs, any iatrogenic risk will exacerbate the problem. Currently available antipsychotics have a wide range of weight and dyslipidemia liabilities (9). When it comes to pharmacologic interventions for patients with antipsychotic-induced metabolic problems, there has been less confidence that these tolerability differences can be used as an effective intervention strategy (9). In this issue, Stroup et al. (10) report on the safety and effectiveness of switching from any of several antipsychotics associated with weight gain or dyslipidemia to one with a lower liability (aripiprazole). Schizophrenia outpatients who were stable while taking olanzapine, quetiapine, or risperidone who also met criteria for metabolic or weight problems were randomly assigned either to switch to aripiprazole or to stay on their current antipsychotic. The primary switch goal was improved metabolic tolerability. The presence of a nonswitch comparison group provided important information on the benefits and relative risks of changing medication. The study showed that while more of the switch patients discontinued their assigned treatment, changing medication is reasonably safe within carefully monitored clinical research settings. There was no appreciable difference in serious adverse events between those who switched and those who stayed on their medication, and rates of efficacy failure were comparable between those two groups. The cautionary note here is that the safety findings are in the context of a controlled clinical trial in which investigators were motivated to make sure the switch was completed and patients were carefully monitored. In real-world practice, where there is less monitoring during the crossover process, we can expect much higher rates of incomplete and failed medication switches (11).

The Stroup et al. (10) article is one of a series of studies (and the second direct test [12]) of the metabolic benefits of a switch to aripiprazole using a double-blind randomized clinical trial. While aripiprazole was the medication chosen here, the overall findings imply that the expected tolerability changes depend on differences in the metabolic profiles of the patient's current antipsychotic and the antipsychotic that is about to be introduced (13, 14). Once it becomes accepted that changing antipsychotic medication can be an effective intervention for antipsychotic-induced weight gain and dyslipidemia, clinicians must also examine the risks of making such a switch.

Results from these two controlled switch studies are very good news indeed. Until now, there had been insufficient evidence to support switching to another antipsychotic as an evidence-based intervention for reducing weight gain or dyslipidemia associated with antipsychotic medication. Changing antipsychotics was not included as a recommended method for reversing weight gain or dyslipidemia was not included in the 2009 Schizophrenia Patient Outcomes Research Team (PORT) guidelines (9). In my opinion, this was a conservative decision given the magnitude of the problem and the strong evidence of benefits after switching. At the time, the authors of the PORT guidelines were concerned that the safety of switching had not been adequately studied, and the document specifically mentioned the need to await the results of the study that Stroup et al. (10) report here. This omission now needs to be revised. The fact that there are two well-designed randomized controlled trials that establish metabolic benefits and the relatively modest risk of clinical worsening strongly suggests that practice guidelines and public policy should recommend that clinicians consider the value of switching antipsychotics in patients with elevated metabolic risk.

At this point, we know much more about differences in tolerability across antipsychotics than we do about differential efficacy across individual patients. Unfortunately, we lack a solid evidence-based foundation to predict which patients may be most likely to benefit from efficacy switches. For persistent symptoms—aside from the special case of clozapine—we lack the knowledge base or endpoints that can guide us in accepting the therapeutic results of a current medication. Until we have more long-term studies on effectiveness in recovery-oriented outcome measures, we need to do the best we can to exploit the differences in side effect profiles as well as differential efficacy across all available antipsychotic medications to find the best medication for each patient.

Dr. Weiden thanks John Newcomer, M.D., and John Sweeney, Ph.D., for the helpful reviews of earlier drafts of this editorial.