Dr. Davis is Gilman Professor and Ms. Chen is a research data analyst in the Department of Psychiatry at the University of Illinois in Chicago.

Acknowledgments: The authors wish to thank Michael E. Bennett for his assistance in the preparation of this manuscript. 



Should all second-generation antipsychotics (SGAs) be treated as a homogenous group? This article focuses on efficacy differences using results from a meta-analysis of 124 randomized, controlled studies comparing an SGA with a first-generation antipsychotic. Results from a meta-analyses on olanzapine and risperidone registrational trials is discussed. While the focus of this article is on efficacy differences among SGAs, a brief discussion of side effects is included as well. Using evidence-based medicine, the assumption that SGAs constitute a homogenous group is called into question. Since SGAs differ in efficacy and all the major side effects, there is really no property in which they are clearly homogeneous and the choice of antipsychotic medication should be tailored toward individual patients.



This article evaluates the centrally important issues regarding the choice of antipsychotic for schizophrenia through the use of evidence-based medicine. With the development of the second-generation antipsychotics (SGAs) clinicians have considerable choice in which antipsychotic to prescribe (Table 1) and revision of old assumptions and practices are necessary. In our view, double-blind, random-assignment, controlled clinical trials provide quantitative evidence largely free of bias, while most existing guidelines, meta-analyses, and narrative reviews are vague and contradictory. Most guidelines discuss SGAs as a homogenous group (Table 2),1-7 and data will be presented to show that this is not the case.


In the 1960s, the manufacturers of each antipsychotic advertised its drug to be particularly useful for certain symptoms of schizophrenia. At that time, we reviewed the evidence and found that typical, or first-generation antipsychotics (FGAs), have identical efficacy and spectrum of action and generally similar side-effect profiles.8 While this similarity has established a tradition of lumping drugs together, we feel that SGAs differ from each other in almost all respects, and therefore need to be individually characterized.



We performed a comprehensive meta-analysis of 124 randomized, controlled studies that compared SGAs with FGAs in schizophrenia or schizoaffective patients. Using a search strategy similar to that employed in Cochrane reviews, we located studies by searching MEDLINE, International Pharmaceutical Abstracts, CINAHL, PsychINFO, and the Cochrane Database of Systematic Reviews. We also searched reference lists in journal articles, and included data from the Food and Drug Administration Web site. Other sources of data include poster presentations, and unpublished data from Cochrane Database reviews or other meta-analyses, conference abstracts, and manuscripts submitted for publications. We queried investigators to locate additional studies and contacted manufacturers to obtain company monographs. In addition, we performed a meta-analysis9 using pooled raw data from four registrational trials of olanzapine.10-13 We also analyzed the raw data from the two registrational studies of risperidone conducted in the United States and Canada.14-16


Evaluation of Schizophrenia Symptoms

The development of the Positive and Negative Symptom Scale for Schizophrenia (PANSS),17 a 30-symptom scale which replaced the 18-symptom Brief Psychiatric Rating Scale (BPRS)18 has facilitated examination of schizophrenia symptom clusters. The PANSS is divided into three subscales based on theoretical grounds: positive symptoms, negative symptoms, and the general psychopathology. Using the PANSS for assessment of symptoms, Crow and colleagues19 showed that schizophrenia patients with negative symptoms have enlarged ventricles—a finding which has withstood the test of time. In the US, Lindenmayer and colleagues20 carried out a number of factor analyses demonstrating that there are really five clusters of symptoms that best characterize the PANSS items. This was an important extension of our knowledge on schizophrenic symptoms and has withstood verification. A review by Marder and colleagues21 found the same five factors in schizophrenia patients in many countries where translations of the PANSS are used.



Janssen Pharmaceuticals developed specific type-2 serotonin (5-HT2) antagonists, based on evidence linking hallucinogens to serotonin receptors and on increasing knowledge of the biology of serotonin. These 5-HT2 antagonists proved to be helpful with negative and depressive symptoms, but ineffective with the positive symptoms of schizophrenia. These findings led Janssen to consider that dopamine (D)-blocking properties were also necessary for antipsychotics and to focus on risperidone, which blocks both D2 receptors and 5-HT2 receptors.

In our opinion, the best studies of risperidone efficacy in acute patients were a US registrational trial conducted by Marder and Meibach15 and a Canadian study conducted by Chouinard and colleagues.16 Our meta-analysis of the combined raw data from these two studies focused on a more detailed differential efficacy analysis.14 The larger sample size of the combined data sets allowed for a detailed examination of individual symptoms and clusters of symptoms. We factor-analyzed the PANSS with a focus on the dimensions of schizophrenia. In the combined data set, the active comparator was haloperidol—the FGA comparator used in almost all contemporary clinical trials. While the theoretically postulated positive symptoms and negative symptoms scales were approximately correct in that most of these symptoms cluster in the appropriate subscales, not all items clustered in the theoretically defined scale. Therefore, the traditionally defined scales were not fully accurate in terms of their factor structure.

Analysis of the five factors of the PANSS showed that haloperidol produced more improvement than placebo on positive symptoms, and had some modest benefits on the cognitive (thought disorder) factor. However, haloperidol was ineffective on the negative symptoms, impulsivity/hostility, and anxiety/depression factors.

Risperidone 6 mg/day was substantially more effective than placebo on all five factors. Since all FGAs and SGAs are superior to placebo, the real question is how they compare with each other. Risperidone at 6 mg/day (or the pooled 6-, 10-, and 16-mg/day doses) proved to be more efficacious than haloperidol on all five factors (Figure 1). The best overall effect is observed at the 6-mg risperidone dose. It is unclear whether the lower optimum response at the 10 and 16 mg/day doses is due to statistical variation or reflects a real fall-off of efficacy at higher doses.

There is clear evidence that risperidone doses of 4–6 mg/day are clearly optimal doses and Patient Outcomes Research Team (PORT) guidelines recommend a daily dose of 4–10 mg.6 There has been some discussion among psychiatrists that the dose range may be between 2–4 mg/day, but certainly for elderly and first admission patients, one should use target doses of 2–4 mg/day.22 Our analysis of adult schizophrenics shows that the efficacy at the 2-mg dose is 50% less than the efficacy at the 6-mg dose. For the typical adult schizophrenic patient we believe decreasing the dose to 2 mg/day would be too low a dose to be fully effective for most patients.

On a descriptive level, risperidone was better than haloperidol for positive symptoms, although haloperidol was also effective. Risperidone was also quite effective in treating negative symptoms, anxiety/depression, and impulsivity/hostility, while haloperidol was not effective at all. Haloperidol showed some modest benefit in the cognitive (thought disorder) factor, but risperidone was more effective. We also compared risperidone and haloperidol on each of the 30 individual schizophrenia symptoms tested on the PANSS. Risperidone showed greater improvement than haloperidol on the majority of symptoms. Many of the symptoms were not improved by haloperidol.

To further define the spectrum of action of these drugs, we developed a haloperidol-responsive scale and a haloperidol-nonresponsive scale based on items of which haloperidol was shown to be effective or ineffective. The scales are described in a previous publication.14 About two thirds of the better improvement was achieved on the haloperidol-nonresponsive scale. We emphasize that risperidone produces a wider range of effects than haloperidol. Many of these symptoms are important for function in the community. Those who focus on positive symptoms, but recognize that drugs such as risperidone affect negative and affective symptoms, hold a somewhat similar position to us. This is an important point to clarify so that disagreements are not merely semantic. It is possible that two antipsychotics could have the same overall efficacy but one could be better for positive symptoms at the expense of negative symptoms or vice versa. This was not the case for risperidone. Risperidone was better than haloperidol on symptoms that improved with haloperidol. One third of risperidone’s increased efficacy was due to an even greater effect on symptoms already improved by haloperidol.


Speed of Action

The time course of improvement with risperidone and haloperidol is presented in Figure 2. Note that although risperidone at 6–16 mg/day produces a much better symptom reduction than haloperidol, the speed of action is not that different (Figure 3). Risperidone achieved almost its full effect by 2 weeks. SGAs have been described as having a slower onset of action than the FGAs, but inspection of the mean rate of response does not support this.




A large randomized, double-blind prospective study of maintenance treatment demonstrated that risperidone produced a greater decrease in relapse than haloperidol as well as a significantly greater improvement on all five factors of schizophrenia.23  In order to evaluate this finding and the findings from the pooled US and Canadian studies, a meta-analysis of all the randomized studies comparing risperidone to FGAs was performed. Risperidone was shown to be consistently more effective than FGAs on both total symptom reduction and on the traditional positive and negative scales. This finding indicates that the effect observed in both the North American clinical trials and the maintenance trial are consistently observed in most trials.



We did a meta-analysis of the registrational trials of olanzapine.9 Eli Lilly and Company has conducted four large trials using the following doses: (1) olanzapine at 1 and 10 mg/day versus placebo11; (2) olanzapine at 5, 10, and 15 mg/day versus haloperidol and placebo10; (3) olanzapine at 1, 5, 10, and 15 mg/day versus haloperidol12; and (4) a large flexible-dose international study of haloperidol versus olanzapine.10-13 In the 5-, 10-, and 15-mg/day dose studies, the clinicians could adjust the olanzapine dose by ±2.5 mg so that the average doses administered were approximately 7, 12, and 16 mg. Olanzapine was substantially more effective than haloperidol and significantly effective on most of the 30 schizophrenia symptoms.

Once again, the real question is how olanzapine compares with FGAs. The same approach used with risperidone was utilized. Factor analysis verified that the PANSS is best described by the same five factors. As with risperidone, olanzapine produces more improvement than haloperidol on all five factors and on most symptoms (Figure 4). We also identified haloperidol-responsive and haloperidol-nonresponsive scales, and showed that olanzapine was more efficacious than haloperidol on both the responsive and nonresponsive scales. While two thirds of olanzapine superiority was on the haloperidol nonresponsive scale, it was also clearly superior on the haloperidol-responsive scale. Thus, it can be said that olanzapine is a better haloperidol than haloperidol and clearly benefits many symptoms not improved at all by haloperidol.

The clinical profile of olanzapine and risperidone are similar. Both drugs produce significantly more improvement than FGAs on all five factors. However, there are two minor differences. First, while olanzapine produces significantly more improvement than haloperidol on the impulsivity/hostility factor, the magnitude of the difference is greater with risperidone 6 mg/day (Figures 1 and 4). Second, while risperidone produces more improvement than haloperidol on anxiety/depression, olanzapine produces a statistically greater superiority over haloperidol than risperidone. Whether the slight difference in the symptoms-improvement profile is real and can withstand the test of time remains to be determined.

It is our opinion that indirect comparisons of two drugs (in this case, olanzapine and risperidone) to a common third drug are not quantitative. Comparison of efficacy is a measure of relative difference, not an absolute difference, and the comparator is a flexible yardstick. Head-to-head trials are required for comparative efficacy. A meta-analysis of all the double-blind random-assignment studies comparing risperidone and olanzapine was performed, and a significant difference was not produced.


Speed of Action

Lilly has conducted several studies10-13 of intramuscular olanzapine compared to haloperidol or benzodiazepine in acutely ill patients. Olanzapine produces a rapid response in the first day or so. This is important because some have said that the onset of the beneficial effect of SGAs is gradual. We have plotted the time to response of olanzapine and haloperidol (Figure 5). There was no significant difference in response in the first 2 weeks, but olanzapine gradually proved to be superior to haloperidol (Figure 6). Since the nonsignificant trends favored olanzapine, it would be difficult to argue that olanzapine had a slower response rate. The acute intramuscular studies are critical in this respect. Olanzapine exhibited as rapid a response as haloperidol. Since there is evidence that benzodiazepines have considerable efficacy on schizophrenia symptoms in the first few days, we feel that benzodiazepine augmentation is often useful in the first few days, particularly for agitated patients.



While use of olanzapine during maintenance treatment has not been studied in a comparable comparative study, the follow-up of the acute study showed that olanzapine continues to be superior to haloperidol in this phase. As more haloperidol patients drop out, this study would tend to minimize the olanzapine superiority, yet olanzapine continued to be superior.


Dose Response

The best data on olanzapine dose-response came from the US trial10 because there was a placebo group for perspective. The rate of response is linear over time. The dose-response curve is a sigmoid curve where the curve levels off after the drug produces all possible effect. Olanzapine has not reached this leveling off yet. Since the goal of treatment is to achieve maximum efficacy, assuming the patients can tolerate the drug, the best dose is the point at which the response levels off. Interestingly, many practitioners are increasing the dose of olanzapine up to 20–30 mg/day, but there is little data from controlled studies on possible efficacy advantages and side-effect disadvantages on this dosing strategy.


Distribution of Outcome Variables

The degree of symptom reduction as measured by the PANSS or BPRS was normally distributed in the response to all the different drug groups in the pivotal studies of both olanzapine and risperidone (Figure 7). There was no evidence of a bimodal curve that would allow patients to be classified as responders or nonresponders based on how much they responded. Although the 20% criteria (of PANSS or BPRS reduction) is frequently used as a definition of responder, other cut-off points have also been chosen to favor one drug over another. An important problem with the arbitrary definition of responders is that a given manufacturer can choose the optimal cut-off point for their drug, which creates a systematic bias. If all studies consistently use the 20% criteria, the bias would be avoided. A better method of measuring improvement is the covariate-adjusted change from baseline (with baseline as covariate). When the underlying distribution is continuous, information is lost by setting an arbitrary dichotomy. Fifty percent of information is lost when both continuous variables of a correlation are dichotomized.



Although sertindole has been withdrawn from the market, it is of academic interest since its overall efficacy is identical to haloperidol. Sertindole exhibited a trend for better efficacy for negative symptoms and less optimal efficacy for positive symptoms. This illustrates the trade-off of one type of symptom for another. While a drug may be superior in one respect, it may be less effective in another, although overall efficacy may be identical to that of the FGAs.



Quetiapine has been investigated in four large randomized trials.24-27 While two of the studies found that the drug’s efficacy was identical to the FGAs,24,27 the other two found the drug to be slightly but not significantly different than the FGAs—one showed it to be more effective,26 while the other showed the drug to be less effective.25 Full data on quetiapine’s profile on positive or negative symptoms have not been systematically presented.



Ziprasidone is on the market in the US, but most of its efficacy studies have not been published. The FDA Web site probably has the best data on ziprasidone.28 Initially, ziprasidone was used at an overly low dose, so most of the registrational studies defined the dose-response curve against placebo. No data exist to indicate that ziprasidone has efficacy above placebo below a dose of 120 mg. The two 80-mg studies have shown somewhat disparate findings, indicating only that efficacy cannot be consistently displayed at the 80-mg dose in acute patients.28 Only at 120, 160, or 200 mg, is there clearly marked efficacy (highly significant) compared to placebo.28 One of these studies using an active comparator (haloperidol), demonstrated that ziprasidone was somewhat less efficacious than haloperidol.28 It is unclear whether this was statistically significant. Another study29 showed that the ziprasidone and olanzapine did not differ greatly in efficacy. Therefore, we view ziprasidone to be at least as efficacious as FGAs. A large maintenance study demonstrated dose-related efficacy against placebo.28



All the planned registrational studies for aripiprazole are mostly completed and have been reported at meetings, but none have been published yet. Our impression is that aripiprazole has about the same efficacy as haloperidol. A study comparing the drug with risperidone showed aripiprazole to be slightly less efficacious than risperidone, but the difference was not significant.30 This would suggest that aripiprazole efficacy may fall between that of FGAs and risperidone or olanzapine. The dose-response curve indicated that the dose producing full efficacy plateaus at about 10 mg/day (possibly at a lower dose in one study).



Clozapine has been studied in double-blind studies with FGAs. Meta-analyses conducted by Wahlbeck and Colleagues,31 Leucht,32 Geddes and colleagues,33 and the authors of this article (unpublished data, 2002) demonstrated that clozapine is substantially more efficacious than FGAs. Weekly white blood cell monitoring has reduced the incidence of death from agranulocytosis to almost zero. Since clozapine has more side effects than the FGAs, it is clearly a second-line drug. Because it is the most efficacious drug, it should be considered in all patients with incomplete response.


Dose Effect of First-Generation Comparison Drug

The results of the meta-analysis by Geddes and colleagues33 showed that the SGAs olanzapine, amisulpride, risperidone, and clozapine were more effective than FGAs. However, Geddes and colleagues concluded that the observed efficacy advantage of the SGAs was only observed in studies where excessively high haloperidol doses were used as comparison, thus there is no real evidence that show SGAs to be more efficacious than FGAs.

Our meta-analysis of 124 randomized double-blind studies failed to replicate the finding of Geddes and colleagues.33 We observed that many clozapine trials used higher comparison doses and clozapine was substantially more effective than FGA. In contrast, sertindole and quetiapine, which were approximately equally effective to their FGA comparison drug, used lower doses of the comparison drug. We suspect that the observed association of high dose comparator with a greater SGA-FGA effect found by Geddes and colleagues33 was due to clozapine efficacy and the reverse due to the lesser efficacy of quetiapine and sertindole. When we examined each individual SGA versus FGA, we found no tendency for higher doses of the comparison drug to be less efficacious. We feel that the hypothesis presented by Geddes and colleagues is not supported by empirical evidence.


Side Effects

There are qualitative differences in side effects among SGAs. Clinicians need to balance the frequency, reversibility, and seriousness of side effects for each patient. Common side effects can be assessed in controlled clinical trials, but rare side effects are difficult to quantify. The FGAs produce a high incidence of acute extrapyramidal symptoms (EPS), which can be troubling, and dystonia, which can be alarming and painful. Tardive dyskinesias (TD) develops in about 3% to 4% of patients per year, each in a cumulative fashion. Severe cases can be socially disfiguring and irreversible.

The introduction of the newer SGAs (eg, olanzapine, risperidone) marks a major advance in the safety of antipsychotics. The major side effects of these drugs are extrapyramidal symptoms and tardive dyskinesia. With drug-induced movement disorders reduced to virtually zero, treatment is greatly simplified. Yet, as with any group of new drugs, other problems emerge. In our discussion of side effects we will emphasize our meta-analyses as they contribute to the understanding of the risks and benefits of these new drugs.

While the SGAs clearly cause less EPS than the FGAs by some margin, they differ among themselves in the occurrence of all the major side effects including parkinsonian symptoms, TD, prolactin elevation, weight gain, diabetes, sedation, and QT prolongation.34 Low-dose clozapine can help dopa-induced hallucinations in parkinsonian patients without aggravating their parkinsonism. Therefore, an argument can be made that clozapine does not cause EPS. Furthermore, since there is an extensive knowledge base on clozapine worldwide, any EPS associated with its use would be well documented by now. We feel the time has come to conclude that clozapine does not cause EPS.

In blinded clinical trials, it has been shown that a number of patients randomized to placebo have EPS ratings greater than zero.9,14 It is unclear whether this observation is some sort of persistent parkinsonian symptom that continues for some months despite placebo treatment. We feel a good possibility may be rater error. Close examination of normal individuals reveals the presence of some mild degree of movement disorder. Even world-class marksmen have some degree of essential tremor. The amount of movement disorder observed in normal subjects increase with age. Therefore, a rating on an EPS scale above zero does not mean that the drug causes EPS. It reflects a background frequency of EPS seen in normal people. There are even some subtle signs of TD-like symptoms in normal elderly patients. Thus, it is difficult to conclude that a new drug does not cause EPS. Scores on standard scales such as the Barnes Akathisia Scale and the Simpson-Angus Scale are not zero, but change from baseline show a decrease of the scores.

Review of our analysis of the combined olanzapine data set showed no significant difference in side effect from placebo. Data from two registrational studies10,11 where olanzapine was compared against placebo were pooled. Haloperidol showed a definite increase in EPS ratings, which was massively statistically significant. The most interesting comparison was with olanzapine and placebo. Ratings of both Simpson-Angus EPS and the Barnes Akathisia scales dropped below baseline with olanzapine and placebo. There was no statistically reliable difference between olanzapine and placebo on either scale. We could demonstrate no dose-related EPS on the Simpson-Angus Scale or akathisia on the Barnes Akathisia Scale. Even when we combined data from two large studies we were unable to demonstrate statistically that olanzapine caused any EPS or akathisia. However, this does not mean that olanzapine does not cause the side effect, it only means that it could not be measured even in the pooled large sample studies.

It is also possible that a clinician could see clearly in a case with definite symptoms that which is hidden behind the noise of rating scale evaluations. There is some evidence that olanzapine may not be useful in treating dopa-induced hallucinations in patients with Parkinson’s disease, but large scale studies are absent. We feel there is a qualitative difference between olanzapine and risperidone on the ability to produce EPS. With risperidone there is a very low level, which may not be clearly statistically distinguishable at doses of 4–6 mg but is distinguishable at higher doses. In addition, dystonias, characteristic of EPS, clearly occur with risperidone. It is difficult to evaluate TD because it is necessary to have large sample sizes over a number of years. Individual cases are not very helpful because most patients have been treated with many drugs and unusual circumstances can complicate an individual case. Given that risperidone has been used for close to a decade and olanzapine for several years, the absence of TD reported in large-scale studies or a large number of case reports suggest some degree of safety.

Risperidone causes EPS but its incidence is markedly lower than that of FGAs, such that at low doses it is not statistically distinguishable from placebo.8,9 Compared with FGAs, risperidone showed slightly greater elevation in prolactin levels, whereas most of the other SGAs have a favorable prolactin profile. While olanzapine and risperidone clearly cause weight gain,34 some of the newer SGAs may not cause weight gain.

There is a substantial interest in new onset diabetes associated with olanzapine and clozapine use, which is under active investigation at this time. Wirshing and colleagues35 were the first to describe an increase in incidence of new onset diabetes with SGAs. Another study by Wirshing and colleagues36 found that triglyceride levels were statistically significantly increased in olanzapine and clozapine patients. Consequently, there is no doubt from case reports and administrative databases that cases of new-onset diabetes are occurring in SGAs such as clozapine and olanzapine. A popular journal article will generate increased number of reports to public databases, such as the FDA. One cannot exactly quantitate the incidence of side effect from such data. How much of the incidence of new-onset diabetes is attributable to weight gain awaits further investigation.

Gianfrancesco and colleagues37 analyzed a claims-based database and found a statistically significant increase in reimbursement claims for new-onset diabetes following the use of clozapine (odds ratio [OR]=7.4), olanzapine (OR=3.1; relationship was found to be dose-related), low-potency FGAs (OR=3.5), and high-potency FGAs (OR=2.1). The authors failed to find a significant increase in diabetes with risperidone. Patients on antipsychotics should be weighed at each visit and if weight gain is observed, it should be managed by diet and exercise, and/or the patient should be switched to a different drug with a lower propensity for weight gain. Because psychiatrists now have to deal with weight gain as a possible side effect of antipsychotic use, they need to join with internists in helping patients maintain normal weight in the prevention of cardiovascular disease and stroke.

Ziprasidone causes a modest increase in QT interval, which raises the possibility of the propensity for sudden death, particularly in patients with long QT syndrome or in patients taking drugs which also prolong the QT interval.38 Ziprasidone has been on the market for a while, and the absence of such reports is reassuring. It is important to recognize that with rare side effects it is difficult to be sure that a drug does or does not cause a side effect, particularly in the absence of any systematic surveillance.

Since the SGAs are not a homogeneous group, there may be tradeoffs. The clinician may trade a minor difference in efficacy for an advantage in side effects, or choose a slightly higher incidence of one side effect in order to avoid another. Because EPS and TD are serious side effects associated with FGAs, it can be said that the SGAs as a group have a superior safety profile. However, the classes of drugs differ from each other qualitatively, thus the clinician may wish to balance differences in efficacy and in side-effect protocols when deciding on a drug for a given patient.



While side-effect profile of a drug is probably the most important criterion when choosing an antipsychotic, there are important differences with regard to efficacy of these drugs as well. Efficacy differences of SGAs, in particular, have not been fully appreciated. Differences in both side effects and efficacy necessitate that SGAs not be treated as a homogeneous group. The clinician should evaluate SGAs individually for each patient.  PP



1.    Bhanji NH, Tempier R. Managing schizophrenia during the stable phase: is there consensus among practice guidelines? Can J Psychiatry. 2002;47:76-80.
2.    Practice guideline for the treatment of patients with schizophrenia. American Psychiatric Association. Am J Psychiatry. 1997;154(suppl 4):1-63.
3.    Canadian clinical practice guidelines for the treatment of schizophrenia. The Canadian Psychiatric Association. Can J Psychiatry. 1998;43(suppl 2):25-40.
4.    McEvoy JP, Scheifler PI, Frances A. The Expert Consensus Guideline Series: treatment of schizophrenia 1999. J Clin Psychiatry. 1999;60(suppl 11):4-80.
5.    Agence National Pour le Development de L Evaluation Medicale. Strategies Therapeutiques a Long Terme Dans les Psychoses Schizophreniques. Text du Consensus. Paris: Agence Nationale pour le Development de L Evaluation Medicale; 1994.
6.    Lehman AF, Steinwachs DM. Translating research into practice: the Schizophrenia Patient Outcomes Research Team (PORT) treatment recommendations. Schizophrenia Bulletin. 1998;24:1-10.
7.    Miller AL, Chiles JA, Chiles JK, Crimson ML, Rush JA, Shon SP. The Texas Medication Algorithms Project (TMAP) schizophrenia algorithms. J Clin Psychiatry. 1999;60:649-657.
8.    Klein DF, Davis JM. Diagnosis and Drug Treatment of Psychiatric Disorders. Baltimore, Md: Williams & Wilkins; 1969.
9.    Davis JM, Chen N. The effects of olanzapine on the 5 dimensions of schizophrenia derived by factor analysis: combined results of the North American and international trials. J Clin Psychiatry. 2001;62:757-771.
10.    Beasley CM Jr, Tollefson G, Tran P, Satterlee W, Sanger T, Hamilton S. Olanzapine versus placebo and haloperidol: acute phase results of the North American double-blind olanzapine trial. Neuropsychopharmacology. 1996;14:111-123.
11.    Beasley CM Jr, Sanger T, Satterlee W, Tollefson G, Tran P, Hamilton S. Olanzapine versus placebo: results of a double-blind, fixed-dose olanzapine trial. Psychopharmacology. 1996;124:159-167.
12.    Beasley CM Jr, Hamilton SH, Crawford AM, et al. Olanzapine versus haloperidol: acute phase results of the international double-blind olanzapine trial. Eur Neuropsychopharmacol. 1997;7:125-137.
13.    Tollefson GD, Beasley CM Jr, Tran PV, et al. Olanzapine versus haloperidol in the treatment of schizophrenia and schizoaffective and schizophreniform disorders: results of an international collaborative trial. Am J Psychiatry. 1997;154:457-465.
14.    Davis JM, Chen N. Clinical profile of an atypical antipsychotic: risperidone. Schizophr Bull. 2002;28:43-61.
15.    Marder SR, Meibach RC. Risperidone in the treatment of schizophrenia. Am J Psychiatry. 1994;151:825-835.
16.    Chouinard G, Jones B, Remington G, et al. A Canadian multicenter placebo-controlled study of fixed doses of risperidone and haloperidol in the treatment of chronic schizophrenic patients. J Clin Psychopharmacol. 1993;13:25-40.
17.    Kay SR, Fiszbein A, Opler LA. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr Bull. 1987;13:261-276.
18.    Overall J, Gorham D. The Brief Psychiatric Rating Scale. Psychological Reports. 1962;10:799-813.
19.    Crow TJ, MacMillan JF, Johnson AL, Johnstone EC. A randomised controlled trial of prophylactic neuroleptic treatment. Br J Psychiatry. 1986;148:120-127.
20.    Lindenmayer JP, Bernstein-Hyman R, Grochowski S. A new five factor model of schizophrenia. Psychiatr Q. 1994;65:299-322.
21.    Marder SR, Davis JM, Chouinard G. The effects of risperidone on the five dimensions of schizophrenia derived by factor analysis: combined results of the North American Trials. J Clin Psychiatry. 1997;58:538-546.
22.    Emsley RA. Risperidone in the treatment of first-episode psychotic patients: a double-blind multicenter study. Risperidone Working Group. Schizophr Bull. 1999;25:721-729.
23.    Csernansky JG, Mahmoud R, Brenner R. A comparison of risperidone and haloperidol for the prevention of relapse in patients with schizophrenia. N Eng J Med. 2002;346:16-22.
24.    Arvanitis LA, Miller BG. Multiple fixed doses of “Seroquel” (quetiapine) in patients with acute exacerbation of schizophrenia: a comparison with haloperidol and placebo. The Seroquel Trial 13 Study Group. Biol Psychiatry. 1997;42:233-246.
25.    Copolov DL, Link CGG, Kowalcyk B. A multicentre, double-blind, randomized comparison of quetiapine (ICI 204,636, ‘seroquel’) and haloperidol in schizophrenia. Psychol Med. 2000;30:95-105.
26.    Emsley RA, Raniwalla J, Bailey, Jones AM. A comparison of the effects of quetiapine (‘seroquel’) and haloperidol in schizophrenic patients with a history of and a demonstrated, partial response to conventional antipsychotic treatment. PRIZE Study Group. Int Clin Psychopharmacol. 2000;15:121-131.
27.    Peuskens J, Link CG. A comparison of quetiapine and chlorpromazine in the treatment of schizophrenia. Acta Psychiatr Scand. 1997;96:265-273.
28.    NDA 20-919: Zeldox (ziprasidone mesylate IM, Pfizer.). Pfizer Global Research & Development. Available at: http://www.fda.gov/ohrms/dockets/ ac/01/briefing/3685b2.htm. Accessed May 2002.
29.    Simpson G, Romano SJ, Horne RL, Weiden P, Pigott T, Bari M. Ziprasidone vs olanzapine in schizophrenia: results of a double-blind trial. Paper presented at: Annual Meeting of the  American Psychiatric Association; May 15-21, 2001; New Orleans, LA.
30.    Carson WH, Saha A, Ali M, Dunbar GC, Ingenito G. Aripiprazole and risperidone vs. placebo in schizophrenia and schizoaffective disorder. Paper presented at: Annual Meetin g of theAmerican Psychiatric Association; May 5-10, 2001; New Orleans, LA.
31.    Wahlbeck K, Cheine M, Essali A, Adams C. Evidence of clozapine’s effectiveness in schizophrenia: a systematic review and meta-analysis of randomized trials. Am J Psychiatry. 1999;156:990-999.
32.    Leucht S, Pitschel-Walz G, Abraham D, Kissling W. Efficacy and extrapyramidal side-effects of the new antipsychotics olanzapine, quetiapine, risperidone, and sertindole compared to conventional antipsychotics and placebo. A meta-analysis of randomized controlled trials. Schizophr Res. 1999;35:51-68.
33.    Geddes J, Freemantle N, Harrison P, Bebbington P. Atypical antpsychotics in the treatment of schizophrenia systematic overview and meta-regression analysis. BMJ. 2000;321:1371-1376.
34.    Marder SR, Essock SM, Miller AL, et al. The Mount Sinai conference on the pharmacotherapy of schizophrenia. Schizophr Bull. 2002;28:5-16.
35.    Wirshing DA, Spellberg BJ, Erhart SM, Marder SR, Wirshing WC. Novel antipsychotics and new onset diabetes. Biol Psychiatry. 1998;44:778-783.
36.    Wirshing DA, Boyd JA, Meng LR, Ballon BA, Marder SR, Wirshing WC. The effects of novel antipsychotics on glucose and lipid levels. J Clin Psychiatry. 2002;63:856-865.
37.    Gianfancesco FD, Grogg AL, Mahmoud RA, Wang R-H, Nasrallah HA. Differential effects of risperidone, olanzapine, clozapine, and conventional antipsychotics on type 2 diabetes: findings from a large health plan database. J Clin Psychiatry. 2002;63:920-930.
38.    Glassman AH, Bigger JT Jr. Antipsychotic drugs: prolonged QTc interval, torsade de pointes, and sudden death. Am J Psychiatry. 2001;158:1774-1782.