Selection of an Oral Antipsychotic: Focus on Efficacy, and Speed of Response¹

When the patient is able to accept oral medications, the physician must decide which antipsychotic to use. The discussion which follows will focus on efficacy, and speed of response. Clozapine will not be addressed here since the selection of clozapine is not based on speed considerations but rather on the history of inadequate response to full courses of two or more antipsychotics. Summary recommendations will be presented first, followed by a critical analysis.

Click here for advice on basic dosing approaches for the atypical antipsychotics.

(1)Osser DN, Sigadel R. Short-term inpatient pharmacotherapy of schizophrenia. Harvard Review of Psychiatry 2001;9:89-104. This material is taken from this paper with updates from later in 2001.

Summary of Recommendations on Efficacy and Speed

1. The weight of the evidence and opinion appears to favor the new generation antipsychotics for first-line use among non-treatment resistant patients with schizophrenia in the short-term, acute inpatient setting. All of the quality evidence that exists involves monotherapy with these agents. (HIGH Confidence) Among the four non-clozapine options currently available in the U.S., olanzapine and risperidone have the most support for superior efficacy compared with the typical antipsychotics. (MODERATE confidence). Quetiapine and ziprasidone have not been demonstrated to be superior to typicals in the studies so far. (MODERATE confidence)
2. Some indirect evidence indicates that risperidone works more quickly than olanzapine on both positive and negative symptoms.(Educated Guess) Specifically: some studies showed risperidone works faster than haloperidol, whereas all studies that we have reviewed found that olanzapine works at the same speed as haloperidol, or at a lower speed.
3. If olanzapine is selected, the evidence suggests that the most rapid effect will be achieved with a starting dose of 15 mg per day, especially in male smokers (or 10 mg for first-onset patients¹). (HIGH Confidence) Lower doses may increase the time required for response.
4. Quetiapine also appears to have a relatively slow onset of action. (MODERATE Confidence)
5. The rate of response to ziprasidone in comparison with other atypical antipsychotics is unclear at this time. In one unpublished, double-blind, multicenter trial comparing ziprasidone vs olanzapine, improvement on ziprasidone appeared significantly greater at one week. But this may have been an artifact of the doses used.

¹Sanger TM, Lieberman JA, Tohen M, Grundy S, Beasley C, Jr., Tollefson GD. Olanzapine versus haloperidol treatment in first-episode psychosis. American Journal of Psychiatry 1999; 156:79-87.

Critical Analysis of Selection of an Oral Antipsychotic with Focus on Efficacy and Speed of Response

Atypicals vs. Typicals

Most major reviews of the pertinent evidence have concluded that the new atypical antipsychotic medications are preferred for first-line use in treatment of initial episodes and exacerbations of schizophrenia.(1, 2, 4, 6, 10, 37) However, there are some dissenting opinions.. The Royal College of Psychiatrists in Britain recently performed a meta-analysis of 52 randomized trials involving 12,649 patients. They concluded that most of the apparent advantages in efficacy of the new generation antipsychotics could be attributed to the use of excessive doses of conventional neuroleptics in the trials. They recommended that initial treatment of episodes of schizophrenia should be with the lowest effective dose of a conventional neuroleptic.(38) In their view, an atypical should be considered only if the neuroleptic is poorly tolerated or if the response is unsatisfactory.

One atypical in particular - quetiapine - has also been the subject of controversy. An evidence analysis by the Cochrane Group (updated May 27, 2000) raised questions about the robustness of quetiapine's efficacy in schizophrenia.(39, 40) The Cochrane Group is well known for its high-quality systematic reviews; they consider all published data and also search extensively for unpublished data. This is done to reduce the problem of "publication bias," which occurs when an overview of the literature reaches a falsely positive impression of a treatment because negative data were never published.(41)

The Cochrane group reviewed 42 papers and reports on quetiapine, including 11 randomized, controlled trials (most of very short duration). They found that dropout rates were very high (36-64%) and only marginally lower in the quetiapine groups than in the neuroleptic control groups. The authors concluded that it was very difficult to interpret the results of these studies and that more studies with longer-term outcomes were needed "before quetiapine's use can be recommended."

Adding to the concerns raised by the Cochrane group, a recent small study of 23 stable schizophrenia patients switched from conventional neuroleptics or risperidone to open-label quetiapine found that only 5 patients completed 77-96 weeks of treatment.(42) The intention was to continue treatment for three years. Patients kept having breakthrough symptoms despite increasing dosage. When quetiapine was approved for marketing, the manufacturer discontinued the study. The authors speculated in their poster that a possible mechanism to explain these clinical findings could be quetiapine's weak dopamine-D2 receptor binding and tendency to induce a proposed dopamine receptor supersensitivity.(43)

In contrast to this experience, more positive results with quetiapine were presented in another poster session.(44) The researchers described a multicenter, open-label trial involving 751 patients with different psychotic diagnoses. Patients were started on either quetiapine or risperidone and maintained for 16 weeks. It is unclear whether they were acutely ill on entrance to the study or whether their illnesses were stable. Patients did improve on both treatments, and the dropout rate appeared to be only about 15%. The mean dose of quetiapine at the end of the study was 317 mg and for risperidone it was 4.5 mg. Extrapyramidal symptoms were greater with risperidone.

Cochrane Group analyses have been done on 20 published studies involving olanzapine and on 14 studies of risperidone. They concluded that these atypicals were clearly effective antipsychotics, with perhaps marginally greater benefit and tolerability compared with haloperidol.(45, 46) The Cochrane group has also published commentary on the Royal College meta-analysis(38) of the trials of atypical antipsychotics.(47) They noted that the Royal College did not pay sufficient attention to the higher drop-out rates on conventional neuroleptics, and this inflated the impression of efficacy of the neuroleptics. In their view, the recommendations derived from the study "could harm people with schizophrenia."

The Cochrane group has also prepared a report on ziprasidone based on the premarketing studies.(47a) There were three comparisons with placebo and one with haloperidol 15 mg. So, there was very limited data for determining how ziprasidone performs compared with other antipsychotics. The comparison with haloperidol showed no difference in efficacy. Since that review, there is a poster presentation available showing a double blind, multicenter comparison with olanzapine involving 269 patients.(47b) At six weeks, the response rates were equivalent, with a slight numerical (non-significant) difference in favor of ziprasidone. Mean dose of ziprasidone was 130 mg and olanzapine, 11 mg. 52% of the ziprasidone patients completed the study compared with 63% of the olanzapine patients.

When all reviews, evidence, and opinions are combined, it appears reasonable to state that the atypicals, particularly risperidone and olanzapine, seem to offer somewhat improved efficacy and potential for a reduced (or at least, different) side effect burden compared to the typicals. They may be the preferred antipsychotics for general use. (MODERATE Confidence) However, further analysis is necessary to see when there might be a preference among the atypicals. For this purpose, we examined the four available comparisons of risperidone and olanzapine in acute inpatient treatment (one with a small number of patients, and one retrospective).(48-51) In addition there are several studies of atypicals in comparison with conventional neuroleptics that add important information about the rate of response in the first inpatient week. The relevant information is narrated and discussed below.

Speed of Response: Efficacy after 1 Week

Risperidone, Olanzapine, Risperidone vs. Olanzapine, Quetiapine, Ziprasidone

Risperidone

Risperidone at 2, 6, 10, or 16 mg per day was compared with 20 mg of haloperidol (higher than the ideal dose) and placebo in a randomized double-blind study involving 513 persons with chronic schizophrenia.(61) All were inpatients, and most had been hospitalized longer than 6 months and presumably had not responded well to conventional antipsychotics. After 1 week total scores on the Positive and Negative Syndrome Scale (PANSS) decreased by nine points with risperidone (6-16 mg) and four points with haloperidol (p = 0.05). Risperidone at 2 mg decreased the PANSS total by six points, which was not significantly different from haloperidol's four-point effect. The weekly results for each of the five PANSS factor scores (positive symptoms, negative symptoms, disorganization, hostility, excitement) are shown in figures. Although the factor scores at 1 week are not mentioned in the text, the 1-week data points in the graphs indicate trends toward superior response with risperidone (6-16 mg) compared to haloperidol and placebo. These trends appear to increase in the data for week 2. There is no analysis of the possible statistical significance of these visually apparent trends.

Three smaller controlled studies comparing risperidone with a typical neuroleptic in patients with chronic schizophrenia(62-64) show data at the 1-week point. Two of the three confirm a faster response to risperidone at 1 week: one (n = 44) with haloperidol as the control (p = 0.05 in favor of risperidone on the PANSS total),(63) and one (n = 98) with the thioxanthine derivative zuclopenthixol as the control (p = 0.03 in favor of risperidone on a global impression of improvement).(62) The third study(64) (n = 107) compared risperidone to perphenazine and showed no difference between the drugs in PANSS total scores at 1 week.

The most recently published study was a post-hoc subanalysis of data from a European double-blind, randomized trial comparing risperidone and haloperidol in chronic schizophrenia.(64a) In this study, the dose of risperidone was 4 mg and haloperidol dose was 10 mg, which are closer to what are considered the optimal doses of each. The haloperidol is still perhaps a bit high, but lower than the 20 mg used in the other comparison above. The results once again showed that PANSS scores improved significantly more on risperidone in the first week(-9 vs -6, p <.05).

Olanzapine

Olanzapine at starting daily doses of 5, 10, and 15 mg was compared with 15 mg of haloperidol per day and placebo in a randomized double-blind study of 335 inpatients with an acute exacerbation of schizophrenia.(56) The authors showed weekly results on the BPRS and its positive symptom cluster in figures but did not discuss these data in the text or indicate whether any of the apparent differences were significant. Examination of the figures indicates that after 1 week of treatment, BPRS total scores were reduced the most (12 points) by haloperidol. The highest dose of olanzapine, 15 mg, had about the same effect as haloperidol, with BPRS total scores dropping by 11 points. Lower doses of olanzapine (5 or 10 mg) appeared to be less efficacious than haloperidol 15 mg or high-dose olanzapine at 1 week, with BPRS total score reductions of six points for the 5-mg dose and eight points for the 10-mg dose. Placebo lowered the BPRS score by only three points. In the figure showing BPRS positive symptoms, these same trends were present to a slightly greater degree than with the BPRS total score. Although less information is given about the population treated than in the risperidone study discussed above,(61) the patients appear to be fairly similar: mean age 36 years (vs. 37 in the risperidone study), 91% diagnosed with the chronic subtype of schizophrenia (vs. 100% in the risperidone study), suffering from clinically severe illness, and predominantly male.

This same olanzapine study(56) also showed data points at week 0.5 (i.e., day 3 or 4) in the figures. Inspection of these figures gives some indication of what to expect from olanzapine and haloperidol in the critical early days of acute inpatient treatment. Perhaps unexpectedly, haloperidol at 15 mg seems to be the clear winner over olanzapine at any of the three doses. Haloperidol lowered the BPRS total by 7.5 points; the improvements from olanzapine 5, 10, and 15 mg were three, five, and four points, respectively. The drop with placebo was four points. Again, the positive symptom cluster showed a slightly wider margin in favor of haloperidol. These numbers were not analyzed for their possible statistical significance. Nevertheless, this soft evidence does raise questions about the rapidity of olanzapine response over the first few days of treatment. Indeed, not until week 4 were results on the BPRS total and positive symptom scores with the lower doses of olanzapine equivalent to those with haloperidol. By 6 weeks, the medium and high doses of olanzapine (then at means of 11.6 mg and 16.3 mg per day) finally surpassed the haloperidol (then at 16.4 mg per day), particularly on negative symptoms. Negative symptoms worsened markedly with haloperidol between weeks 3 and 6 but continued to improve with all doses of olanzapine. This suggests that haloperidol-induced secondary negative symptoms were accounting for a large portion of the end-point efficacy differences favoring olanzapine.

In a larger controlled study of olanzapine,(57) 1996 outpatients and inpatients at 174 sites were given flexible doses of olanzapine or haloperidol for 6 weeks. The mean daily dose by end-point was 13.2 mg for olanzapine and 11.8 mg for haloperidol, but during the first week the daily dose for each drug was 5 mg. BPRS results at 1 week are shown in a figure: there was a mean drop of four points in each group. By the end of 2 weeks, at which time both drugs could have been increased to as much as 10 mg per day, the BPRS drop was just over six points in each group. Again in this study, low-dose olanzapine produced only modest improvement on the BPRS during the first week, which was no greater than that seen in the haloperidol control group. The population was similar to those in the other studies: mean age of 38, diagnosis of the chronic subtype of schizophrenia in 87%, and history of unsatisfactory response to the last neuroleptic in 77%. Thus, olanzapine was no better than haloperidol over the first week in a population that one would expect to be fairly resistant to haloperidol.

Risperidone vs. Olanzapine

The suggestion that response is more rapid to risperidone than to olanzapine was not confirmed in the single published head-to-head comparison of these two atypicals.(48) In a randomized double-blind trial involving 339 patients with schizophrenia, schizoaffective disorder, or schizophreniform psychosis (sponsored by Lilly), no significant difference was seen between the two medications in the PANSS-total at any point from week 1 through week 6. Both produced an eight-point improvement after 1 week. Dosage of risperidone was 1 mg twice daily on day 1, 2 mg twice daily on day 2, and 3 mg twice daily on days 3-7. Olanzapine dosage was 15 mg daily during the first week. It is possible that this study showed no differences in early response because the patient population was different than those in the above-described studies of olanzapine and risperidone. For example, the patients seemed less chronically ill: the mean age was 36 and the age of onset was 23, but only 42% were inpatients, and only 40% were diagnosed as having the chronic subtype of schizophrenia. In addition, the use of a high initial dose of olanzapine (15 mg) would have maximized results from this antipsychotic and minimized differences from risperidone.

By contrast, another comparison of risperidone and olanzapine,(51) published only in abstract form, does suggest differences. This study was a retrospective analysis of the treatment of 601 patients with schizophrenia or schizoaffective disorder at 11 centers in Europe. Mean time to improvement was 14 days with risperidone and 23 days with olanzapine (n = 366; p = 0.0008). The mean final daily doses of risperidone were 4.9 mg and 14.9 mg, respectively; no information was provided on the doses during the first week. Efficacy rates at the completion of the trials were equivalent (77% and 78%, respectively). Poor outcome was strongly correlated with concomitant use of other neuroleptics. This study, which was not funded by a pharmaceutical firm, adds support to the proposition that risperidone works more quickly than olanzapine in some populations. It also suggests that although physicians often add a second neuroleptic when response is slow, this does not convert the patients to rapid responders.

Quetiapine

Three published studies of quetiapine(65-67) have provided outcome information at 1 week, but no comparisons with risperidone or olanzapine. The results indicate a weak early effect. Quetiapine performed best in a 6-week randomized double-blind trial in which it was compared with the neuroleptic chlorpromazine in 201 patients with acute exacerbation of chronic schizophrenia. The mean daily dose at the end of the study was 407 mg of quetiapine or 384 mg of chlorpromazine. Both drugs were started at 25 mg three times daily. Many outcome measures were employed during the study, but for week 1 only the Clinical Global Impression of Improvement was used. At the end of the first week, 60% of quetiapine patients had at least minimal improvement, compared with 61% on chlorpromazine; 18% of both groups showed "much improvement." This study suggests that quetiapine produces a pattern of early response more like that of olanzapine than that of risperidone-that is, quetiapine appears no better than a standard neuroleptic in a population that one would expect to be poorly responsive to the standard neuroleptic.

In a study comparing quetiapine and haloperidol in 448 hospitalized patients, with mostly chronic schizophrenia,(66) the PANSS total dropped by 5.3 points with quetiapine and 7.4 points with haloperidol after 1 week, a nonsignificant trend in favor of haloperidol. Mean daily doses at end-point were 455 mg for quetiapine and 8 mg for haloperidol. The third report with 1-week data was a placebo-controlled trial of quetiapine in 109 patients with schizophrenia.67 Patients were started at 25 mg three times daily and titrated upward. After 1 week on quetiapine, there was no change in the Clinical Global Impression. The BPRS total improved by just over three points, which may be compared with the four-point improvement seen with olanzapine and haloperidol in the large study reviewed above.(57)

Ziprasidone

There is a poster presentation available showing a double blind, multicenter comparison of ziprasidone with olanzapine in 269 acute patients with schizophrenia or schizoaffective disorder.(47b) At six weeks, the response rates were equivalent, with a slight numerical (non-significant) difference in favor of ziprasidone. In the first week, patients were given 5 mg of olanzapine for two days, and then 10 mg for the remainder of the first week. This is a dosing strategy, as discussed above under olanzapine, that produces a slow response, not as fast as haloperidol 10 mg. Ziprasidone was begun at 40 mg bid for the first two days, and 80 bid for the rest of the week. This is an optimal dosage strategy for ziprasidone. By the end of one week, BPRS scores decreased 7 points on ziprasidone, and 4 points on olanzapine. CGI and PANSS showed similar differences. Thus, ziprasidone seemed to work faster, but the doses used did not give olanzapine its best chance to do well. A comparison of ziprasidone and risperidone would be of interest.

Concomitant Second Antipsychotic

A survey in several outpatient clinics(15) showed that when risperidone is prescribed, a concomitant second antipsychotic is also given in 4-7% of patients. The rate with olanzapine was 6-25%; with quetiapine, 6-22%. The statistical significance of such differences was not reported, but these data are consistent with the idea that olanzapine and quetiapine sometimes work more slowly, and that clinicians are adding other medications because waiting for improvement is difficult. No information about combination treatments with ziprasidone is available.

RETURN TO SUMMARY OF RECOMMENDATIONS