Effectiveness of Switching Antipsychotic Medications

In clinical practice, people with schizophrenia sometimes change medication in hope that the next medication will work better than the current one. The analyses of the CATIE phase 1 findings presented here suggest that the likelihood of finding a medication that fits better than the current medication (as measured by time to all-cause discontinuation) is a function both of the medication being switched from and the medication being switched to, and that such switches have only modest success rates among the medications used in CATIE phase 1. The subgroup analysis of CATIE patients entering the study on a regimen of olanzapine monotherapy indicated that, on average, it may be better to stay with olanzapine than to switch to one of the other antipsychotics used in phase 1. Similarly, analyses of patients receiving double-blind olanzapine and risperidone treatment indicated that stayers, on average, fared somewhat better than patients newly switched to these two medications (about half of stayers versus over 70% of switchers discontinued their assigned medication).

When individuals who did not switch medications at study entry were excluded from the analyses, the core findings of CATIE were supported ( Figure 3 ), although statistical significance was attenuated both because of loss of power with the reduced sample size and because removing stayers removed a number of individuals receiving olanzapine who tolerated that medication relatively well. As shown in Figure 3, the separation between the olanzapine curve and the other curves is still apparent when stayers are excluded, but the separation is narrower than when all participants are included. Under the CATIE study conditions where side effects were monitored, individuals newly treated with olanzapine (olanzapine switchers) were more likely to discontinue their assigned medication due to intolerability than were individuals assigned to olanzapine who were already receiving it at study entry (olanzapine stayers). Olanzapine switchers also were more likely to discontinue their assigned medication due to intolerability than were individuals receiving risperidone at study entry who were assigned to risperidone (risperidone stayers) and individuals who were newly treated with risperidone (risperidone switchers) (data in online supplement). These limitations in olanzapine’s tolerability acted to offset some of its superior efficacy as originally reported (3) .

The finding that risperidone or olanzapine stayers fared better than switchers raises the more general question of the extent to which switching from any nonclozapine oral antipsychotic monotherapy to another is a hopeful treatment strategy. These findings suggest that, unless the clinical situation requires a medication change, prescribers should take steps to optimize current medication regimens (e.g., via dosage changes, behavioral or psychosocial interventions, adjunctive medications) before switching medications. Certainly, individuals’ desires to try a different medication should be respected. What we are stressing is the clinical appropriateness of working diligently with patients to maximize the effectiveness of the current antipsychotic before taking on the risks associated with a medication change in hopes that the next one will be more satisfactory (i.e., taken for a longer period of time). Short of a compelling clinical indication to switch medication, different individuals (whether patient or prescriber) have different thresholds for being willing to continue a medication given limitations in effectiveness or side effects. For example, one individual might be willing to accept an increase in a particular side effect in exchange for symptom improvement, while another individual may choose to change to a medication with less symptom control in order to ameliorate particular side effects. On the one hand, changing medications offers the possibility of further symptom improvement or reduction of side effects experienced with the current antipsychotic medication. On the other hand, changing medications risks destabilizing a formerly stable person (8) as well as increases the burden on already stretched clinical resources from both increased monitoring following a medication change and increased clinical attention if symptom exacerbations occur. The field has very few randomized trial studies examining the relative effectiveness of staying with one’s current antipsychotic treatment versus switching to another (8, 9), hence this opportunity to examine the phase 1 CATIE data from this perspective is particularly timely.

In cases where medication changes are clinically warranted or where individuals prefer to try something new, the analyses of CATIE phase 1 presented here along with the results of CATIE phase 2 may help inform medication choice. Specifically, in phase 2 of CATIE, some of the people who discontinued their phase 1 medication stayed with their new phase 2 medication for the remainder of the study, indicating a successful antipsychotic medication switch. When clozapine was a treatment arm, such successful switches were significantly more likely, when considering discontinuation for any reason, among individuals who switched to clozapine relative to quetiapine and risperidone but not to olanzapine (10) . In addition, individuals receiving clozapine were significantly less likely to discontinue treatment for reasons of inefficacy than were individuals receiving any of the other atypicals (10). When clozapine was not a treatment arm and when considering those who discontinued the previous medication for reasons of inefficacy, olanzapine was superior to quetiapine and ziprasidone, and risperidone was superior to quetiapine (11) .

This study has several limitations worth noting. First, the post hoc analyses were exploratory and, as such, carry concerns about spurious findings from multiple comparisons. Thus, the results should be interpreted as preliminary, and future randomized clinical trials should address the stay versus switch question directly. Second, sample sizes were only sufficient to evaluate with confidence the relative effectiveness of staying versus switching for two antipsychotic medications, olanzapine and risperidone. We recognize that other antipsychotic medications may behave differently under these stay or switch conditions. Indeed, randomized trial evidence indicates that individuals with treatment-resistant schizophrenia are better off switching to clozapine than staying with conventional antipsychotics, at least with respect to lowered likelihood of rehospitalizations (9) . Hence, future studies should examine this question for other agents and for larger numbers of stayers. Because so many individuals with chronic disorders gain some benefit from their current medication (so-called “partial responders”), the “stay versus switch” question has great day-to-day clinical significance and is not addressed by designs where individuals switching to one medication are compared with individuals switching to another medication. Third, dosages of the medications in the double-blind condition may have advantaged some medications (e.g., olanzapine) and not others (e.g., quetiapine and ziprasidone). Finally, we note that these results apply to individuals who were enrolled in the CATIE phase 1 trial, and these individuals varied widely in clinical severity. Because random assignment to the medication taken at baseline was possible for participants taking one of the study drugs, staying on this medication had to be considered a viable clinical option. Those who were completely satisfied with their medication would not have been expected to enter the study; thus, CATIE participants who were taking one of the study drugs at baseline were incompletely satisfied with this drug but did not absolutely require a change.

The results of these analyses are consistent with a critical finding in all previous reports from CATIE: the available antipsychotic medications have considerable limitations. We would expect that enhanced availability of evidence-based psychosocial treatments would improve the overall effectiveness of treatment and promote recovery.

Perhaps the most important finding from the present report is that, even in randomized clinical trials, there can be biases in estimates of treatment response because of differences in the study participants’ exposures to treatment prior to the trial, and that random assignment to treatment condition alone is not a sufficient safeguard to eliminate this bias. For example, as demonstrated in these analyses, comparisons of medication effectiveness need to take into account whether medications being compared were each newly initiated.

Received Aug. 14, 2006; revision received Oct. 17, 2006; accepted Oct. 24, 2006. From the Department of Psychiatry, Division of Health Services Research, Mount Sinai School of Medicine; the Mental Illness Research, Education, and Clinical Center, James J. Peters VA Medical Center, Bronx, N.Y.; Quintiles, Inc., Research Triangle Park, N.C.; University of North Carolina at Chapel Hill; the Departments of Psychiatry and Epidemiology and Public Health, Yale Medical School, New Haven, Conn.; VA New England Mental Illness Research and Education Center, West Haven, Conn.; the Department of Psychiatry, Columbia University College of Physicians and Surgeons and the New York State Psychiatric Institute, New York; and the CATIE Study Investigators Group (investigators from 56 U.S. sites presented in the data supplement that accompanies the online version of this article). Address correspondence and reprint requests to Dr. Essock, Division of Health Services Research, Department of Psychiatry, Box 1230, Mount Sinai School of Medicine, One Gustave L. Levy Place, New York, NY 10029-6574; [email protected] (e-mail).Dr. Davis is an employee of Quintiles and reports having received consulting fees from Eli Lilly and Pfizer. Dr. Stroup reports having received funding for consulting or speaking from AstraZeneca, Eli Lilly, Janssen Pharmaceutica, and Pfizer. Dr. Rosenheck reports having received research funding from AstraZeneca Pharmaceuticals, Bristol-Myers Squibb, Eli Lilly, and Wyeth and consulting fees from Bristol-Myers Squibb, Eli Lilly, GlaxoSmithKline, and Janssen Pharmaceutica. Dr. Lieberman reports having received research funding from AstraZeneca Pharmaceuticals, Bristol-Myers Squibb, GlaxoSmithKline, Janssen Pharmaceutica, and Pfizer and consulting and educational fees from AstraZeneca Pharmaceuticals, Bristol-Myers Squibb, Eli Lilly, Forest Pharmaceuticals, GlaxoSmithKline, Janssen Pharmaceutica, Novartis, Pfizer, Solvay, and Wyeth. Drs. Essock and Covell report no competing interests.Supported by NIMH contract N01 MH90001 and NIMH grants R01 MH59312 and R03 MH 071663. AstraZeneca Pharmaceuticals, Bristol-Myers Squibb, Eli Lilly and Company, Forest Pharmaceuticals, Janssen Pharmaceutica, Novartis, Otsuka Pharmaceutical, Pfizer Inc, Schering-Plough, and Zenith Goldline Pharmaceuticals provided medications for the studies.The authors thank Carlos Jackson, Ph.D., Sue Marcus, Ph.D., and Nina Schooler, Ph.D. for their help in discussing the issues presented in this article.