Simone is a 45-year-old African American woman whose schizophrenia was diagnosed when she was 22 years old. She has had about 5 brief hospitalizations for exacerbations of her symptoms, usually precipitated by nonadherence when her medications were changed to something she did not agree with. She lives in a basement apartment owned by her family and attends a day treatment program 3 times a week. Simone smokes about 10 cigarettes a day when she is able to afford them. She does not use street drugs or alcohol(Drug information on alcohol). Simone has a fraternal twin, Sabella, who also has schizophrenia.
Simone responds reasonably well to quetiapine(Drug information on quetiapine) or risperidone(Drug information on risperidone), but not to haloperidol(Drug information on haloperidol), and she objects to extrapyramidal side effects, particularly akathisia. Simone’s mother and all of her maternal aunts have type 2 diabetes mellitus. Her body mass index is 29 kg/m2, and she eats high-fat foods and does not exercise. She is willing to take either quetiapine or risperidone and her clinician prescribes generic risperidone.
Simone presents with a number of conflicting issues, and the best medication option for Simone at this juncture is difficult to immediately ascertain. Fortunately, there are clinical trials that can help inform the clinician in making thoughtful individualized treatment decisions about antipsychotic medication switches. Simone is representative of ambulatory patients with schizophrenia who have been ill for a decade or more. Thus, she resembles the patients enrolled in the CATIE schizophrenia studies.3-6
Approximately 1500 patients participated in at least one part of CATIE, and the study remains the largest randomized pragmatic clinical trial to date that compares the overall relative effectiveness of several of the atypical antipsychotics with each other as well as with a representative typical antipsychotic (perphenazine). Time to all-cause discontinuation was the primary outcome measure and represents the “real-world” usefulness of a medication. The assumption is that a patient will continue to use a medication (and a clinician will continue to prescribe it) if the medication is considered by all parties to be effective, safe, and tolerable.
One of the basic tools of evidence-based medicine used to appraise clinical trial data is number needed to treat.7 Number needed to treat answers the question: for how many patients do I need to prescribe medication X instead of medication Y before I would expect to encounter one additional success? If the number needed to treat for success of medication X versus medication Y is 5, then it would take 5 patients to be treated with X instead of Y before one additional patient has a successful outcome. The higher the number needed to treat, the less of a difference there is between medications X and Y. The lower the number needed to treat, the more compelling the difference is. (See Sidebar for how to calculate number needed to treat.)
CATIE outcomes can be reinterpreted using number needed to treat.8 Depending on the phase of CATIE, different antipsychotics had different rankings for overall effectiveness.9 The usefulness of CATIE in informing medical decision making is that the study consisted of several phases during which patients could discontinue their initial randomized medication and be re-randomized to another. Thus CATIE actually mimics clinical practice: if one antipsychotic does not work for any reason, another antipsychotic is tried. (See Table 1 for a summary of CATIE methodology and Table 2 for a summary of patient characteristics.)
Phase 1 demonstrated that all-cause discontinuation was lowest with olanzapine(Drug information on olanzapine); risperidone and perphenazine(Drug information on perphenazine) also had advantages in terms of number needed to treat. Switching away from risperidone or olanzapine may also have an impact on all-cause discontinuation. This was observed in a post hoc analysis of patients randomly assigned to olanzapine or risperidone who also continued to take their baseline medication. These patients had significantly longer times until discontinuation than those who switched antipsychotics.10 Thus, switching antipsychotics is not necessarily a benign process.
An important caveat is that we do not know what exactly motivated the patients to participate in CATIE, nor do we know with any precision how effective or tolerable their medication regimens were before randomization.
What are the relative rankings of the different antipsychotics in phase 1? We can calculate the number needed to treat for pair-wise comparisons on all-cause discontinuation. The 95% confidence interval (CI) is provided for all comparisons that were statistically significant at the P < .05 level (Table 3).
The lower the number needed to treat value, the bigger the advantage is for the antipsychotic relative to its comparator. For example, for every 6 patients randomized to olanzapine instead of quetiapine in phase 1, one additional patient will have completed the entire 18 months on the initially randomized medication (ie, successful in avoiding all-cause discontinuation). Likewise, for every 13 patients randomized to risperidone instead of quetiapine in phase 1, one additional patient will have avoided all-cause discontinuation. For this latter comparison, the 95% CI is considered “wide,” with 95% certainty, the actual difference between risperidone and quetiapine on all-cause discontinuation is somewhere between a low of every 7 patients to a high of every 54 patients.
CATIE also answers the question about what agent to switch to after a patient has had an unsuccessful trial with perphenazine. Phase 1B was an interim phase for those patients who had been randomized to perphenazine in phase 1 but who discontinued for any reason. Inadequate therapeutic effect was the reason why 48% of the participants in phase 1B discontinued perphenazine, and 32% discontinued it because of unacceptable adverse effects. Thus, the enrollees in phase 1B represent patients who did not do well enough because of either efficacy or tolerability issues on an older agent with relatively “tight” dopamine(Drug information on dopamine) D2 receptor binding (perphenazine). In this phase, quetiapine (and olanzapine) performed relatively well, whereas risperidone, a “tighter” D2 blocking agent than olanzapine or quetiapine, did not (Table 4). This difference between risperidone and quetiapine is the opposite of what was observed in phase 1 and, as will be discussed later, in phase 2T. Phase 1B demonstrates that number needed to treat values for all-cause discontinuation are highly dependent on which phase of CATIE is being scrutinized.