Efficacy of Olanzapine in Acute Bipolar Mania:  A Double-blind, Placebo-Controlled Study FREE

ALTHOUGH ADVANCES have been made in the treatment of bipolar disorder, existing therapies are not always effective or are accompanied by adverse effects that lead to noncompliance. The efficacy of lithium and valproate has been established by well-designed clinical trials^1- 3; however, side effects and treatment failures are present with both drugs.^1,4 Typical antipsychotics are also used for the treatment of acute mania, although their side effect profiles are far from ideal.⁵

Olanzapine has also been used for the treatment of bipolar disorder. A 21-day, double-blind, placebo-controlled study found olanzapine to be an effective and safe treatment in acute mania.^6- 7 Limitations of that trial included separation of olanzapine from placebo at week 3 of treatment, rather than earlier, as occurred in other similarly designed modern trials of valproate and lithium in acute mania.^2- 3 Possible reasons for the lack of a more robust separation between drug and placebo were hypothesized to include the following: (1) too slow an increase in olanzapine dosing (ie, acute mania may require more aggressive olanzapine dosing for optimal response); (2) too liberal use of adjunctive lorazepam; (3) inclusion of first-episode patients (who showed a disproportionately high rate of response to placebo); and (4) too short a treatment period. We therefore conducted a second double-blind, placebo-controlled study to further evaluate the efficacy and safety of olanzapine in the treatment of acute bipolar mania, with special attention to the potential methodological limitations of the first trial. Specifically, we conducted a 28-day study of 115 multiple-episode patients from December 1, 1997, through February 28, 1999, that used a more aggressive olanzapine-dosing schedule but permitted less concomitant lorazepam use.

Patients, aged 18 through 70 years, with a DSM-IV⁸ diagnosis of bipolar disorder, manic or mixed, with or without psychotic features,were eligible to be enrolled in this study. Investigators recruited patients from private practices (13 sites), inpatient and outpatient services of university-affiliated centers (10 sites), and a Veterans Affairs facility. In addition, some sites recruited patients through colleague referral, and 6 sites advertised the study in local newspapers. Diagnosis was based on clinical assessment and confirmed by results of the Structured Clinical Interview for the DSM-IV, Patient Version (SCID-P), administered by trained clinicians (including principal and subinvestigators [all physicians] and study personnel with appropriate clinical degrees [PhD in psychology or MSW] and experience). After having the protocol explained to them, patients provided written informed consent to participate in the study. A minimum total score of at least 20 on the Young–Mania Rating Scale (Y-MRS)⁹ was required at the screening visit and on the day of randomization (baseline). At baseline, patients displayed a clinically severe symptom profile, with a mean Y-MRS score of 29.10 (range, 14-49; 1 patient was enrolled with a baseline Y-MRS total score of 14). Patients were excluded with any of the following criteria: serious, unstable medical illness; DSM-IV substance dependence (except nicotine or caffeine) within the past 3 months; and serious suicidal risk.

We conducted a 4-week, randomized, double-blind, parallel study. All psychotropic medication therapy (except benzodiazepines) was tapered during the screening period and discontinued at least 1 day before randomization. Patients were randomized to olanzapine or placebo, in a 1:1 ratio. Computer-generated codes were used to create randomized blocks of clinical trial material kits before study start-up. Each block contained 2 olanzapine and 2 placebo kits. Each kit contained all clinical trial material used by a patient throughout the 4-week study. Personnel at the site assigned a patient the next available kit. Patients were required to be hospitalized for a minimum of 1 week after randomization and were allowed to leave the hospital after that time only if their Clinical Global Impressions–Bipolar Version of Severity of Illness (CGI-BP)¹⁰ mania score was no greater than 3 (mild) and they had at least a 50% reduction in their Y-MRS score. Psychotherapy was permitted, but not controlled for, during the study.

The starting dose of olanzapine was 15 mg/d. After the first day of therapy, the daily dose could be adjusted upward or downward, as clinically indicated, by 5-mg increments or decrements within the allowed dose range of 5 to 20 mg/d. Modal dose was defined as the dose that the patient was prescribed for the most number of days. The mean (± SD) modal and median modal doses of olanzapine were 16.4 ± 4.2 mg/d and 20 mg/d, respectively.

Concomitant use of lorazepam was allowed during double-blind therapy up to 2 mg/d for the first 4 days of treatment and thereafter by up to 1 mg/d for the next 6 days. Lorazepam was not permitted beyond the initial 10 days after randomization. Benztropine mesylate was permitted to treat extrapyramidal symptoms (EPSs) up to a maximum of 2 mg/d throughout the course of the study. However, the use of benztropine as prophylaxis was not allowed.

Severity of illness and psychopathologic features were measured by the following rating scales: Y-MRS, Hamilton Psychiatric Rating Scale for Depression–21 Item (HAMD-21),¹¹ CGI-BP, and the Positive and Negative Syndrome Scale (PANSS).¹² Safety was monitored by assessing adverse events, including EPSs (parkinsonism as measured by the Simpson-Angus scale¹³ and akathisia as measured by the Barnes Akathisia scale¹⁴), laboratory values, electrocardiograms (ECGs), vital signs, and weight change. All adverse events reported by patients during the study were recorded and coded using the Clinical Symbol and Thesaurus for Adverse Event Terminology (COSTART) dictionary.

The primary efficacy variable, as defined by the protocol, was the reduction from baseline of the Y-MRS total score after 4 weeks of therapy. Response and euthymia were defined, a priori, as at least a 50% improvement from baseline to end point and as a score of no greater than 12 at end point in the Y-MRS total score, respectively. Interrater reliability assessments with the Y-MRS were conducted before study initiation by measuring the correlation of each rater with the groupwise median score of each item. Raters who did not achieve a correlation of at least 0.80 were not allowed to rate patients in this study.

To further investigate the effect of olanzapine on depressive symptoms, additional analyses were performed. The mean change from baseline to end point on the HAMD-21 score was calculated for all randomized patients and in a subset of patients who presented with moderate to severe depressive symptoms (HAMD-21 score, ≥20 at baseline). In addition, the proportion of patients experiencing a clinically detectable worsening in depressive symptoms at any time during acute therapy was assessed. A worsening of at least 3 points on the HAMD-21 score was used as a definition of clinically detectable worsening of depressive symptoms.

Patient data were analyzed on an intent-to-treat basis. For analysis of last observation carried forward (LOCF) mean change from baseline to end point, patients with a baseline and at least 1 postbaseline measurement were included in the analysis. Four placebo-treated patients and 1 olanzapine-treated patient did not have a postbaseline measure and were excluded from all efficacy analyses. Total scores from rating scales were derived from the individual items; if any single item was missing, the total score was treated as missing.

Continuous efficacy and safety parameters were evaluated using analysis of variance. The models generally included terms for the fixed effects of treatment, investigator, and treatment × investigator interaction. Investigators with fewer than 2 patients per treatment group were pooled as specified in the protocol. Analyses of subgroups included a term for treatment only, owing to sparse data. The LOCF change in the Y-MRS total score was also compared between treatment groups using the baseline Y-MRS score as a covariate to examine change in relation to initial severity; investigator was not included in this model. An examination of the effect of treatment over time was conducted on the Y-MRS total score using a likelihood-based repeated-measures analysis. The Y-MRS total score at each postbaseline visit was used as the response variable, and the baseline Y-MRS total score was used as a covariate. This analysis evaluated treatment and investigator effects along with the treatment × investigator and treatment × visit interactions using an unstructured covariance matrix for the within-patient error as specified in the protocol. In addition, an examination of the therapy difference stratified by treatment time for the Y-MRS total score was performed using a pattern-mixture analysis.¹⁵ A mixed-effects model was used, including the main effects for therapy, visit, treatment time, investigator, and the interaction effects for therapy × investigator, therapy × treatment time, therapy × visit, investigator × visit, and therapy × treatment time × visit. Visit and dropout time were random effects; therapy and investigator were fixed effects in the model. The Kruskal-Wallis test was used to compare treatments for each of the individual items of the Y-MRS. The Fisher exact test was used to analyze treatment effects for categorical efficacy and safety parameters. All cited P values are 2-tailed, with a significance level of .05 as specified in the protocol. Unless otherwise indicated, data are given as mean ± SD.

A total of 115 patients were enrolled in the study. Mean age was 39 years; 80.0% were white, and 50.0% were men. Based on DSM-IV criteria using the SCID-P, 42.6% of the patients were in a mixed episode and 55.7% were experiencing psychotic features. Of those 64 patients with psychotic features, 47 (73.4%) were experiencing mood-congruent psychotic features. There were no statistically significant differences in any demographic or illness characteristics between treatment groups. Historical illness characteristics and previous medication use and response are presented in Table 1. A statistically significant greater number of patients randomized to the placebo-treated group had a history of previous response to valproate than in the olanzapine-treated group (P = .02, Fisher exact test). Frequency of recorded medication use at the beginning of the screening period included benzodiazepines and/or hypnotics (68.7%), anticonvulsants (23.5%), typical antipsychotics (16.5%), anticholinergics (14.8%), lithium (9.6%), atypical antipsychotics (7.8%), and antidepressants (4.3%). Study completion and discontinuation summary details are presented in Table 2. Frequency of study completion was significantly greater (P = .04; Fisher exact test) in the olanzapine group (61.8%) compared with the placebo group (41.7%). There were no significant differences between groups regarding reasons for discontinuation.

The primary efficacy measure was the change in Y-MRS score from baseline to end point (LOCF), after up to 4 weeks of acute double-blind treatment. The olanzapine group experienced a 6.65-point greater mean improvement in Y-MRS total score compared with the placebo group (F_1,86 = 12.47; P<.001). The impact of initial severity on LOCF change in Y-MRS score was not significantly different between the treatment groups (F_1,106 = 2.19; P = .14; Figure 1). In addition, olanzapine-treated patients demonstrated a statistically significant greater mean improvement on the CGI-BP severity of mania, CGI-BP severity of overall bipolar illness, and PANSS total and positive scores compared with placebo-treated patients (Table 3). Efficacy subgroup analyses were also performed based on the presence or absence of psychotic features and between patients in a manic or a mixed episode. Olanzapine-treated patients exhibited no statistically significant difference in the mean change in Y-MRS scores for any of these subtypes. For olanzapine-treated patients, the antimanic effect in patients with and without psychotic features was similar.

The olanzapine group consistently showed greater LOCF mean improvement on Y-MRS total score; HAMD-21 total score; CGI-BP mania, depression, and overall bipolar illness scores; and PANSS total, positive, and negative scores compared with the placebo group at each week. Olanzapine-treated patients demonstrated a statistically significant greater improvement in the mean change from baseline in the Y-MRS total score at the first postbaseline observation at week 1 (F_1,86 = 4.78; P = .03) (Figure 2). This statistically significant separation from the placebo group was maintained during the 4-week study. In addition, treatment differences were statistically significant at each week for CGI-BP severity of mania and overall bipolar illness scores and PANSS total and positive scores.

An examination of treatment effect over time using a repeated-measures analysis was conducted on the Y-MRS total score as specified in the protocol. Olanzapine demonstrated a statistically significant greater treatment effect compared with placebo (F_1,207 = 10.47; P = .002). The superior treatment effect of olanzapine was evident at week 1, and the superiority was maintained over time (Table 4).

A post hoc examination of the effect of dropout time on treatment result was performed. Patients who dropped out at week 1 had a similar response regardless of therapy. Placebo-treated patients who dropped out at weeks 2 or 3 had minimal response. On the other hand, olanzapine-treated patients who dropped out at week 2 or 3 did have some improvement (Table 5). To estimate the therapy difference stratified by treatment time, a pattern-mixture analysis¹⁵ was performed. The results of this analysis were similar to the results of the LOCF and repeated-measures visitwise analyses. The main difference was that in this analysis, there was no statistical separation at week 1 (Table 6).

In the analysis of the individual items of the Y-MRS, olanzapine-treated patients exhibited a statistically significant greater mean improvement than placebo-treated patients on the following items: elevated mood (χ²₁ = 9.11; P = .003), sleep (χ²₁ = 12.33; P<.001), language-thought disorder (χ²₁ = 4.66; P = .03), content (χ²₁ = 8.48; P = .004), and disruptive-aggressive behavior (χ²₁ = 6.64; P = .01).

Responders were classified as patients with an improvement of 50% or more in Y-MRS total score from baseline to end point (LOCF). The olanzapine group demonstrated a significantly greater response rate compared with the placebo group (64.8% vs 42.9%, respectively; Fisher exact test, P = .02). Patients achieving a Y-MRS total score of at least 12 at the final visit of the acute phase were considered to be euthymic. A statistically significant greater number of olanzapine than placebo-treated patients met the euthymia criterion for mania (61.1% vs 35.7%, respectively; Fisher exact test, P = .01).

The analysis of change in HAMD-21 score from baseline to end point for all randomized patients showed a similar improvement in olanzapine- and placebo-treated patients (−7.83 ± 7.79 vs −4.45 ± 6.95, respectively; F_1,86 = 2.91; P = .09). In patients who presented with moderate to severe depressive symptoms (HAMD-21 score, ≥20 at baseline), a statistically significant greater improvement in olanzapine compared with placebo-treated patients was observed on the change in HAMD-21 score from baseline to end point (−12.29 ± 8.79 vs −6.81 ± 8.43, respectively; F_1,40 = 4.24; P = .05) (Figure 3). Using a 6-item subscale score of the HAMD-21 to reflect a core mood factor^16- 17 (items 1, 2, and 7-10,), there was no significant difference in change from baseline to end point when comparing all olanzapine- and placebo-treated patients (−3.06 ± 4.24 vs −2.04 ± 3.69, respectively; F_1,86 = 0.30; P = .59), or in the subset of patients who presented with moderate to severe depressive symptoms at baseline (−5.52 ± 4.72 vs −3.19 ± 4.34, respectively; F_1,40 = 2.78; P = .10).

The effect of olanzapine on induction of depressive symptoms was also investigated. A worsening in the HAMD-21 score of at least 3 points was used as a definition of a clinically detectable worsening. The percentage of olanzapine-treated patients who experienced a clinically detectable worsening in depressive symptoms at any time during double-blind therapy was similar to that seen in placebo-treated patients (11.1% vs 17.9%, respectively; P = .42, Fisher exact test).

The categorical rates of patients who received at least 1 dose of benzodiazepine were 36 (65.5%) of 55 patients and 44 (73.3%) of 60 patients in the olanzapine and placebo groups, respectively. The between-treatment group difference in categorical use was not statistically significant P = .42, Fisher exact test). Of those patients treated with a benzodiazepine, placebo-treated patients had a higher mean daily dose (0.74 mg/d) compared with olanzapine-treated patients (0.55 mg/d) (F_1,55 = 1.06; P = .31).

Adverse events that originally occurred or worsened in severity during double-blind therapy were considered treatment emergent. One patient in the placebo group (agitation) and 2 patients in the olanzapine group (unintended pregnancy and rash) discontinued treatment because of an adverse event. The only treatment-emergent event with a statistically significant more frequent occurrence in the olanzapine group compared with the placebo group was somnolence (P<.001, Fisher exact test) (Table 7). The only treatment-emergent event with a statistically significant more frequent occurrence in the placebo group was agitation (P = .03, Fisher exact test).

Emergence of EPSs was low, and anticholinergic use was negligible for both groups (mean, 0.33 ± 0.26 vs 0.22 ± 0.17 mg/d for olanzapine vs placebo-treated patients, respectively; F_1,3 = 2.28; P = .23). Olanzapine-treated patients experienced an improvement in parkinsonism (Simpson-Angus) and akathisia (Barnes) from baseline to end point, whereas placebo-treated patients experienced a worsening in parkinsonism and an improvement in akathisia. The difference in mean change at end point between olanzapine- and placebo-treated patients was not statistically significant for parkinsonism (−0.27 ± 1.16 vs 0.13 ± 1.61, respectively; F_1,86 = 1.81; P = .18) or akathisia (−0.40 ± 0.83 vs −0.16 ± 0.76, respectively; F_1,87 = 2.07; P = .16).

Significant differences between groups were observed for mean change in supine systolic blood pressure from baseline to end point (5.04 ± 15.98 vs −3.86 ± 17.92 mm Hg for olanzapine vs placebo-treated patients, respectively; F_1,86 = 9.84; P = .04). Patients in the olanzapine group had a statistically significant larger mean weight gain compared with the placebo group (2.11 ± 2.83 vs 0.45 ± 2.31 kg, respectively; F_1,88 = 4.22; P = .002).

The only statistically significant difference in laboratory values between treatment groups occurred in alanine aminotransferase (ALT) and aspartate aminotransferase (AST). Patients in the olanzapine group had a statistically significant higher incidence of increased ALT and AST levels than patients in the placebo group (ALT, 21.6% vs 3.9%; P = .02, Fisher exact test; AST, 17.3% vs 0%; P = .003, Fisher exact test). High and low limits that are of potential clinical concern were determined a priori for each laboratory test. Patients who met these criteria at end point or for 2 consecutive observations were defined to have a potentially clinically significant abnormality. One olanzapine-treated patient met this definition for ALT (≥165 U/L) and AST levels (≥150 U/L). No patients displayed clinical symptoms of hepatic dysfunction at any time during the study, and none were discontinued because of abnormal results of liver function tests.

There were no statistically significant differences in ECGs between treatment groups, although fewer olanzapine than placebo-treated patients experienced a treatment-emergent ECG abnormality (13.3% vs 33.3%, respectively; P = .11, Fisher exact test). No statisticallysignificant differences were noted between olanzapine- and placebo-treated patients with respect to treatment-emergent corrected QT interval prolongation (≥430 ms in men or ≥450 ms in women) (4.3% vs 5.1%, respectively; P>.99, Fisher exact test).

This 4-week, double-blind, placebo-controlled trial suggests that olanzapine is an effective and safe treatment in acute bipolar mania. Olanzapine demonstrated an improvement of 14.8 points from baseline to end point on the Y-MRS and achieved statistical significance compared with placebo, which demonstrated an improvement of 8.1 points. This differential change of 6.6 points, in relation to the variability present in the data, corresponds to an effect size of 0.5. As described by Bowden and colleagues,¹⁸ this effect size matches the conventional definition of a medium effect size of 0.5, which can "usually be discerned by a trained clinician." The antimanic effect of olanzapine was evident at the first postbaseline observation after 1 week of acute therapy and maintained throughout the 4-week trial. In olanzapine-treated patients, 65% responded (Y-MRS improvement, ≥50%) and 61% (Y-MRS, ≤12) achieved euthymia at their final visit. Most important, olanzapine achieved statistical significance on the elevated mood item on the Y-MRS, suggesting efficacy in core manic symptoms. In addition, olanzapine exhibited antimanic efficacy in the main subtypes of acute bipolar mania. The Y-MRS mean score change observed in the olanzapine group was similar for patients with and without psychotic features, and for patients in manic and mixed episodes.

Some differences emerge with the results of this study compared with the only other placebo-controlled study of olanzapine in acute mania.⁷ Differences in the study design include a starting dose of 15 instead of 10 mg, duration of 4 weeks rather than 3 weeks, and approximately half of the allowed dose of lorazepam. In terms of efficacy, our present study achieved a similar mean difference as the previous study between the olanzapine and placebo groups in the Y-MRS study (−6.7 points [P<.001] vs −5.4 points [P = .02], respectively). A larger percentage of olanzapine-treated patients in our study were classified as responders (35 [64.8%] of 54 patients vs 34 [48.6%] of 70 patients). In addition, in our study, a statistically significant difference was first observed at week 1 as opposed to week 3 for the previous study. The more robust results in our study could be explained by the higher starting dose (15 vs 10 mg) or by patient variability.

In our study, 43% of placebo-treated patients responded (improved ≥50% in Y-MRS score), which was higher compared with our previous study and any other modern, similarly designed trials in mania, which found placebo response rates of 24%,¹⁹ 11%,³ and 25%.² Nonetheless, despite the high placebo response, the difference between placebo- and olanzapine-treated patients was statistically significant. The higher proportion of patients with a history of previous response to valproate in the placebo group may have contributed to the high placebo response, as has been suggested by Keck and colleagues,²⁰ being higher in non–treatment-resistant patients. The longer trial duration may have allowed more placebo-treated patients to achieve responder status. Analysis of the data from baseline to week 3 revealed a much smaller placebo effect than the data from baseline to week 4. In contrast, a much smaller difference between the week-3 and week-4 results were observed among olanzapine-treated patients. Beyond the increased duration of study in this clinical trial, it is difficult to speculate why the placebo response was so high. However, it is important to note that placebo response is not the same as spontaneous remission. In patients with major depression, placebo response has been reported to be as high as 70%²¹; however, improvement of patients assigned to a waiting list is minimal.^22- 23 To our knowledge, similar studies have not been conducted in patients with bipolar disorder. Brown²⁴ suggested that the benefits of placebo in psychiatric disorders to some extent have similarities with the benefits of psychotherapy, ie, they both provide "expectation of improvement, support, and mobilization of hope." In addition, the large percentage of patients with rapid-cycling illness (45 [39.1%] of 115 patients) combined with the longer trial duration could have accounted for the placebo response at the time of the rating. Keck and colleagues²⁰ recently concluded that a number of variables can contribute to placebo response in acute mania, including the presence of patients with a rapid-cycling course and spontaneous remissions, the latter more likely to happen in trials of longer duration. They also concluded that a higher placebo response is likely to be observed in large multicenter trials like the present one, because of the potentially larger variance in interrater reliability across the many study sites.

Depressogenic effects of antipsychotics in bipolar disorder are of great concern.^25- 26 Lack of depressogenic effect was suggested by the percentage of patients who experienced a worsening in depressive symptoms (worsening in HAMD-21 score of ≥3 at any time), ie, 6 (11.1%) of 54 olanzapine- and 10 (17.9%) of 56 placebo-treated patients (P = .42, Fisher exact test). In addition, the analysis of the change in HAMD-21 score from baseline to end point for all randomized patients revealed no statistically significant difference between olanzapine- and placebo-treated patients (−7.8 vs −4.5, respectively; F_1,86 = 2.91; P = .09). Furthermore, in patients with moderate to severe depressive symptoms (HAMD-21 score, ≥20 at baseline), a statistically significant improvement in olanzapine compared with placebo-treated patients was observed on the change in HAMD-21 score from baseline to end point (F_1,40 = 4.24; P = .05). These results are in contrast to those of the first trial, where depressive symptoms improved by the same amount in both groups (−3.0 vs −2.9 for the olanzapine vs placebo-treated patients, respectively; F_1,103 = 0.03; P = .87).⁷ The apparent efficacy of olanzapine in moderate to severe depressive symptoms in mania in our present study, but not in the first study, may be partly explained by a higher rate of mixed episodes and higher mean HAMD-21 score at baseline (17.3 ± 9.2 in our present study vs 12.6 ± 7.2 in the first study). However, the first and present study results suggest that olanzapine does not induce depression during acute therapy, and the results from our present study suggest that for patients with more severe depression, olanzapine may improve symptoms of depression in patients with acute manic or mixed illness.

There were no statistically significant differences between treatment groups in these measures of EPSs (akathisia and parkinsonism as measured by the Barnes Akathisia scale and Simpson-Angus scale, respectively). In addition, very few EPSs emerged, and anticholinergic medication use was negligible for both treatment groups. The mild worsening of parkinsonian symptoms experienced by patients randomized to placebo was not statistically significant. However, it may have resulted from withdrawal effects of previous antipsychotic and anticholinergic treatments. Olanzapine-treated patients may not have exhibited this mild worsening of parkinsonian symptoms because of the affinity of olanzapine to different neurochemical receptors.

In regard to other safety measures, 2 patients receiving olanzapine discontinued its use because of an adverse event (unintended pregnancy and rash). Somnolence was the only treatment-emergent adverse event that occurred in a statistically significant higher percentage of olanzapine-treated patients compared with placebo-treated patients (38.2% vs 8.3%, respectively; P<.001, Fisher exact test). In addition, there were no clinically significant changes in vital signs, laboratory values, or ECGs. Although no patients in this study discontinued treatment with olanzapine secondary to weight gain, a mean weight gain of 2.1 kg was observed in a 4-week period. Although this amount of weight gain should not be a safety concern, it may lead to treatment noncompliance in some cases.

A possible limitation of this study is the lack of an active comparator, which has been advocated as a tool in aiding the validity of placebo-controlled studies.² Further comparator studies need to be completed.

In our clinical trial, olanzapine demonstrated statistically significant greater efficacy in the treatment of acute bipolar mania compared with placebo. Antimanic effect was observed in patients with or without psychotic features and in patients in manic or mixed episodes. Further studies in maintenance treatment and bipolar depression need to be conducted to determine the mood stabilizing properties of olanzapine. In addition, studies comparing olanzapine with current mood stabilizers, such as lithium and valproate, will help to further define olanzapine's role in bipolar mania.

Accepted for publication April 11, 2000.

This work was sponsored by Lilly Research Laboratories, Indianapolis, Ind.

We acknowledge that the following institutions and individuals, in addition to the authors, participated in the Olanzapine HGGW Study Group: Northwest Behavioral Medicine, Marietta, Ga: Michael C. Banov, MD; Psychiatric Institute of Florida PA, Orlando: Scott West, MD; Mesa Vista Hospital, San Diego, Calif: Michael Plopper, MD; Raleigh, NC: Richard Weisler, MD; Fabre Research Incorporated, Houston, Tex: Louis Fabre, MD; Biological Psychiatry Program, University of Cincinnati, College of Medicine Cincinnati, Ohio: Susan L. McElroy, MD; Department of Psychiatry, University of Cincinnati, College of Medicine: Paul E. Keck, Jr, MD; Milwaukee Psychiatric Hospital, Milwaukee, Wis: Richard Wang, MD, PhD; Stanley Center for Treatment of Bipolar Disorders, Western Psychiatric Institute and Clinic, University of Pittsburgh Medical Center, Pittsburgh, Pa: K. N. Roy Chengappa, MD; The Psychiatric Institute, Department of Psychiatry, University of Illinois, Chicago: Philip G. Janicak, MD; Northbrook Research Center, Brown Deer, Wis: Jeffrey Simon, MD; Albert Einstein Medical Center, Philadelphia, Pa: Cherian Verghese, MD; Princeton Biomedical Research, Princeton, NJ: Jeffrey Apter, MD; Hawaii Clinical Research Center, Honolulu, Hawaii: Dennis Mee-Lee, MD; Veterans Administration at Brentwood, Los Angeles, Calif: Robert H. Gerner, MD; John Umstead Hospital, Butner, NC: Joseph McEvoy, MD; Rhode Island Hospital, Providence: Gabor Keitner, MD; Elgin Mental Health Center, Elgin, Ill: Mohammed Alam, MD; Washington University, St Louis, Mo: Raj Nakra, MD; and Northwest Clinical Research Center, Bellevue, Wash: Arifulla Khan, MD.

We deeply appreciate the patience and dedication of all clinical research and hospital staff involved in this clinical study.

Reprints: Mauricio Tohen, MD, DrPH, Lilly Research Laboratories, Lilly Corporate Center, Indianapolis, IN 46285 (e-mail: m.tohen@lilly.com).