Efficacy of Olanzapine in Combination With Valproate or Lithium in the Treatment of Mania in Patients Partially Nonresponsive to Valproate or Lithium Monotherapy FREE

THE EXPERT Consensus Guidelines Series, published in the year 2000, recommends lithium and valproate as first-line treatments for bipolar mania.¹ However, up to 40% of patients respond poorly to monotherapy with either treatment.² When monotherapy fails, the guidelines recommend combination therapies. A number of authors have recently reviewed the use of such cotherapies for bipolar mania. Freeman and Stoll³ concluded that the combination of lithium and valproate is better tolerated and more efficacious in maintenance therapy than other combination treatments.

Typical neuroleptics have been suggested to be superior in efficacy to lithium monotherapy.⁴ Conversely, the addition of a mood stabilizer to conventional antipsychotic therapy seems superior to antipsychotic agents alone.⁵ In support of this, Müller-Oerlinghausen et al⁶ compared the efficacy of combined therapy with conventional antipsychotics and valproate vs valproate monotherapy in patients with bipolar or schizoaffective disorder and found combination therapy to be superior to monotherapy.

Olanzapine, an atypical antipsychotic, has been shown in 2 placebo-controlled studies to have acute antimanic effects.^7- 8 Moreover, a previous report has suggested that olanzapine is effective when used in combination with other psychotropic agents.⁹ The present study was conducted to investigate the efficacy and safety of combined therapy with olanzapine and either valproate or lithium compared with valproate or lithium monotherapy.

All patients were diagnosed as having bipolar disorder, manic or mixed episode, with or without psychotic features, using the Structured Clinical Interview for the DSM-IV¹⁰(SCID).¹¹ Patients had to have at least 2 previous depressed, manic, or mixed episodes as well as a Young Mania Rating Scale¹² (YMRS) total score of 16 or greater at visit 1 and visit 2 (2-7 days later). Patients were required to have had a documented trial of treatment, with a therapeutic blood level of lithium (0.6-1.2 mmol/L) or valproate (50-125 µg/mL), for at least 2 weeks immediately prior to visit 1. Patients were included only if they showed inadequate response to monotherapy (YMRS total score ≥16). Prior to participation, all patients signed an informed consent document approved by their study site's institutional review board.

Participants in the study initially entered a 2- to 7-day screening and washout period (study period 1) during which all concomitant medications other than lithium or valproate were discontinued. Patients already receiving valproate or lithium continued to do so throughout the study. Patients receiving other forms of treatment started receiving either lithium or valproate at investigator discretion for the 2 weeks immediately prior to visit 1. Plasma levels of the medications were documented to be within the therapeutic ranges. Only patients scoring greater than or equal to 16 on the YMRS were randomized to receive concurrent treatment combined with either olanzapine or placebo(study period 2).

Study period 2 consisted of a 6-week acute, double-blind phase, during which levels of lithium or valproate were maintained within the therapeutic range. Patients were assessed weekly. Patients were randomized 2:1 to receive either olanzapine (flexible dose range of 5, 10, 15, or 20 mg/d) added to valproate or lithium or placebo added to valproate or lithium. Olanzapine therapy was initiated at 10 mg/d. To maintain blinding, treatment took the form of two 5-mg capsules (either olanzapine or placebo), titrated up in increments of 1 capsule or down by any number of decrements at investigator discretion as indicated by each patient's tolerance. Patients unable to tolerate the minimum dose were discontinued. Patients were permitted adjunctive use of benzodiazepine (≤2 mg/d of lorazepam equivalents) for no more than 14 days cumulatively. Anticholinergic therapy (benztropine mesylate, ≤2 mg/d) was permitted throughout the study for treatment of extrapyramidal symptoms but not for prophylaxis. Aside from study drugs, benzodiazepines, and anticholinergics, no other drugs were permitted during the study.

Patient assessments were conducted by mental health care professionals, including psychiatrists, psychologists, nurses, and other mental health caregivers with a clinical degree or certification. Raters were trained in the use of the SCID and symptom-rating scales before study initiation. To ensure high interrater reliability, investigators were required to achieve a reliability coefficient of 0.75 or greater. The primary measure of efficacy to assess severity of manic symptoms was the mean change from baseline to end point in the YMRS total score. Secondary measures included the 21-item Hamilton Depression Rating Scale¹³ (HAMD-21); the Positive and Negative Syndrome Scale¹⁴; and the Clinical Global Impressions Severity of Bipolar Disorder scale¹⁴(CGI-BP) total scores, and mania and depression subscale scores. Clinical responses on the YMRS and HAMD-21 were defined a priori as an improvement of 50% or greater. Clinical remission (euthymia) was defined a priori as achievement of a YMRS total score of less than or equal to 12. A subsample of patients with moderate to severe depressive symptoms was defined by a current mixed episode and a HAMD-21 total score of 20 or greater at baseline. Secondary assessments, also defined a priori, included analyses of treatment differences following stratification by the current course of illness, the presence or absence of psychotic features, and the use of lithium or valproate.

Scales for the assessment of neurologic adverse events included the Simpson-Angus Scale,¹⁵ the Barnes Akathisia Scale,¹⁶ and the Abnormal Involuntary Movement Scale.¹⁴ Assessment of vital signs, weight, and clinical laboratory analytes (including prolactin, nonfasting glucose, and electrolyte levels and hematologic analysis) was performed at each visit. Serum concentrations of mood stabilizers were collected at every visit.

Data were analyzed on an intent-to-treat basis,¹⁷ included all patients who met the entry criteria (including inadequate responsiveness to the minimum 2-week prior treatment with lithium or valproate), and provided both a baseline and at least 1 postbaseline data measurement. Total scores from rating scales were derived from the individual items; if any item was missing, the total score was treated as missing. All tests were 2-sided, with an α level of .05. Analysis of variance (ANOVA) models were used to evaluate continuous data, including terms for treatment, investigator, and treatment-investigator interaction. The linear model for this analysis included terms for baseline, treatment, investigator, treatment-investigator interaction, visit, and treatment-visit interaction. The Fisher exact test was used for categorical analyses, including laboratory values, vital signs, and treatment-emergent adverse events. Data are given as mean (SD) unless otherwise indicated.

A total of 501 patients entered the screening phase and 344 patients were randomized and enrolled (33 US centers, 5 Canadian), with a mean enrollment of 9 patients per site. Patients were recruited from both academic and nonacademic sites from existing clinical patient populations seeking treatment at those sites. Of the 344 randomized patients, 322 came from outpatient centers. The other 20 (cotherapy, n = 16; monotherapy, n = 4) came from inpatient settings. Patients were initially screened on the basis of face-to-face interviews, medical record reviews, and information obtained from family members and referring clinicians. Reasons for lack of enrollment included entry criteria not met(86 patients, including 24 failing to meet the YMRS total score criterion of ≥16); patient decision or loss to follow-up during the screening phase(58); investigator decision (8); protocol violation (4); and a single death that occurred before completion of screening or exposure to the study drug. Ultimately, 229 patients were randomized to receive olanzapine cotherapy and115 to receive monotherapy (Table 1). One patient in the monotherapy group received both valproate and lithium and accordingly was excluded from the subgroup analyses. The median duration of mood-stabilizer therapy prior to randomization was 67 days; 203 patients had a duration of therapy longer than 6 weeks. One patient in the monotherapy group and 9 in the cotherapy group had no postbaseline measures and were excluded from all efficacy analyses.

The percentage of patients completing the study was roughly equal in the 2 treatment groups (cotherapy, 69.9%; monotherapy, 71.3%). Significantly more patients in the monotherapy group discontinued treatment due to lack of efficacy (12.2% vs 3.1%; P = .002), whereas significantly more patients in the cotherapy group withdrew due to adverse events (10.9% vs 1.7%; P = .002) (Table 2).

Patient demographics and illness characteristics were not significantly different between the cotherapy and monotherapy treatment groups overall (Table 1). In the overall study group (n= 344), the mean age was 40.6 (11.1) years. One hundred sixty-five patients(48.0%) had mixed episodes at enrollment; the remainder had pure manic episodes. Overall baseline mean YMRS total scores for the olanzapine cotherapy (n =220) and monotherapy (n = 114) groups were 22.31 (5.39) and 22.67 (5.15), respectively, and mean HAMD-21 scores were 14.52 (8.46) and 13.54 (7.63), respectively (Table 3).

Mean modal dose of olanzapine in the cotherapy group (n = 224) was 10.4(4.9) mg/d). Mean plasma levels of lithium among the cotherapy (n = 74) and monotherapy (n = 41) patients were 0.76 (0.16) and 0.82 (0.19) (F_1,86 = 4.26; P = .04) mEq/L, respectively, while mean plasma levels of valproate for cotherapy (n = 145) and monotherapy (n= 73) were 63.6 (18.4) µg/mL and 74.7 (18.6) µg/mL, respectively(F_1,188 = 18.38; P<.001). Benzodiazepine use was not statistically different between patients in the cotherapy (66/229[28.8%]) and monotherapy (39/115 [33.9%]) groups (P= .38).

Both groups of patients improved during the course of treatment as indicated by the primary measure of efficacy, the YMRS total score (Table 3). However, the olanzapine cotherapy group (n = 220) showed a mean decrease in YMRS total score of 13.1 (8.53) points, corresponding to a 58.8% improvement from baseline compared with a decrease of 9.10 (9.36) points F_1,276 = 9.08; P = .003) for the monotherapy group (n = 114), which corresponded to an improvement of 40.1%.

Itemwise analysis of the YMRS revealed that, compared with monotherapy, olanzapine cotherapy brought about significantly greater improvement at end point on the items of Irritability (cotherapy, −1.82 [2.09], n = 220; monotherapy, −1.02 [2.37], n = 114; F_1,276 = 5.69; P = .02); Speech (cotherapy, −2.45 [2.03], n = 220; monotherapy, −1.63 [2.53], n = 114; F_1,276 = 5.24; P = .02); Language/Thought Disorder (cotherapy, −0.94[0.91], n = 220, monotherapy, −0.72 [1.00], n = 114; F_1,276= 5.34; P = .02); and Disruptive/Aggressive Behavior(cotherapy, −1.18 [1.64], n = 220; monotherapy, −0.46 [1.77], n = 114; F_1,276 = 10.16; P = .002).

Clinical response was defined a priori in the protocol as improvement of 50% or greater from baseline to end point in the YMRS total score. On this basis, 149 (67.7%) of the 220 patients in the olanzapine cotherapy group responded to treatment compared with 51 (44.7%) of the 114 patients in the monotherapy group (P<.001). In addition, time to response was significantly shorter for cotherapy (P = .002, log rank test), with a median response time of 18 days for cotherapy vs 28 days for monotherapy.

Clinical remission was defined a priori in the protocol as achievement of a YMRS total score of less than or equal to 12. On this basis, 173 (78.6%) of the 220 patients in the olanzapine cotherapy group demonstrated evidence of remission. In the monotherapy group, 75 (65.8%) of the 114 evaluated patients demonstrated evidence of remission. This difference in remission rates was also significant (P = .01). Time to remission was significantly shorter in the cotherapy group (log rank test, P = .002), with a median remission time of 14 days for cotherapy vs22 days for monotherapy.

Compared with the patients in the monotherapy group, patients in the olanzapine cotherapy group showed significantly greater improvement on the HAMD-21 at each time point throughout the study. By week 6, the cotherapy group (n = 220) experienced a mean last observation carried forward decrease in HAMD-21 scores of 4.98 (7.61) points, significantly greater (F_1,276 = 18.05; P<.001) than the decrease of 0.89(6.90) points in the monotherapy group (n = 114). An exploratory itemwise analysis showed significantly greater improvement in the dimensions of depressed mood, feelings of guilt, suicidality, early insomnia, anxiety-psychic, and paranoid symptoms.

Analysis of end point HAMD-21 scores conducted in the subset of patients experiencing a mixed episode with moderate to severe depressive symptoms at baseline (HAMD-21 total score ≥20 at baseline) showed a decrease of 10.31(8.19) points for olanzapine cotherapy (n = 51) compared with 1.57 (7.73) points (F_1,70 = 17.50; P<.001) for monotherapy (n = 21). Within this subset, 43.1% of patients in the cotherapy group showed ≥50% improvement of depressive symptoms compared with 9.5% in the monotherapy group (P = .006).

Other secondary measures of efficacy included the Positive and Negative Syndrome Scale (total; Positive, Negative, and Cognitive clusters; and Hostility subscores) and the CGI-BP (overall, Severity of Mania, and Severity of Depression). Olanzapine cotherapy brought about significantly greater improvement than monotherapy on patients' last observation carried forward, Positive and Negative Syndrome Scale total, and Hostility item scores, as well as on the CGI-BP overall and Severity of Depression scores (Table 3).

Subgroup analyses, defined a priori, were conducted on baseline to end point YMRS total scores. No significant interactions were seen between previous exposure to psychotropics (antidepressants, antipsychotics) and therapy (cotherapy, monotherapy). However, among all patients without psychotic features, olanzapine cotherapy was significantly more efficacious than monotherapy (cotherapy: −13.25[7.76], n = 150; monotherapy: −8.32 [8.68], n = 76; F_1,196= 16.97; P<.001). Among patients without psychotic features, olanzapine cotherapy was more effective than monotherapy regardless of whether patients received lithium or valproate. However, among patients with psychotic features, responses to treatment were not different between the cotherapy and monotherapy groups regardless of whether patients received lithium or valproate—this despite the lack of association between the presence of psychotic features and the differential effect of therapy (ANOVA test of interaction: F_1,274 = 0.60; P= .44).

Among patients with a current mixed episode, olanzapine cotherapy was superior to monotherapy (cotherapy: −12.92 [8.37], n = 121; monotherapy: −7.46[10.15], n = 54; F_1,146 = 17.31; P<.001). However, among patients presenting with pure mania, the treatment difference did not achieve statistical significance (cotherapy: −13.34 [8.77], n = 99; monotherapy: −10.57 [8.40], n = 60; F_1,129 = 2.95; P = .09). The superiority of olanzapine cotherapy over monotherapy seen in patients with mixed episodes was found only in patients receiving valproate (cotherapy: −13.18, [8.49], n = 84; monotherapy: −7.48[10.74], n = 42; F_1,124 = 10.53; P = .002), whereas the treatment difference seen with lithium did not achieve statistical significance (cotherapy: −12.32 [8.15], n = 37; monotherapy: −7.42[8.14], n = 12; F_1,47 = 3.28; P = .08), again despite the lack of association between course of illness and the differential effect of therapy (ANOVA test of interaction, F_1,274 = 0.14; P = .71).

Finally, among patients receiving valproate, olanzapine cotherapy brought about significantly greater improvement in YMRS total scores compared with patients receiving valproate monotherapy (cotherapy: −12.85 [8.64], n = 146; monotherapy: −8.39 [9.76], n = 72; F_1,188 = 13.44; P<.001). Among patients receiving lithium, the greater improvement seen with olanzapine cotherapy relative to monotherapy did not achieve statistical significance (cotherapy: −13.62 [8.36], n = 74; monotherapy: −10.39 [8.69], n = 41; F_1,86 = 3.74; P = .06). The type of mood stabilizer was not associated significantly with a differential effect of cotherapy compared with monotherapy (ANOVA test of interaction, F_1,273 = 0.74; P = .39).

No statistically significant changes from baseline were seen in extrapyramidal symptoms on the Simpson–Angus Scale, Abnormal Involuntary Movement Scale, and Barnes Akathisia Scale. Rates of adverse events (Table 4) more frequently reported in the cotherapy group included somnolence, dry mouth, weight gain, increased appetite, tremor, and speech disorder (cotherapy: slurred speech n = 14, speaking difficulties n = 1; monotherapy: stuttering n = 1). In the cotherapy group, 25 patients discontinued treatment due to adverse events (Table 1). Six (2.6%) discontinued due to somnolence, 3 (1.3%) due to weight gain, and3 (1.3%) due to peripheral edema. The remaining 13 discontinuing patients in the olanzapine cotherapy group withdrew due to 13 different adverse events(1 patient per event class). The 2 patients (1.7%) in the monotherapy group who discontinued both withdrew due to depression (Table 1). Post hoc subgroup analysis showed that, among patients receiving valproate, a significantly higher incidence of the adverse event"dizziness" was seen in patients receiving cotherapy (16.3% vs 4.1%; P = .009).

No statistically or clinically significant differences emerged between the monotherapy and olanzapine cotherapy groups for changes in vital signs. However, the cotherapy group experienced a 3.6% increase in body weight, significantly higher than that seen in the monotherapy group (cotherapy: 3.08 [3.04] kg, n = 219; monotherapy: 0.23 [2.48] kg, n = 113; F_1,302 = 73.88; P<.001). With the exception of a greater incidence of treatment-emergent elevated prolactin levels (upper limit: 0.81 nmol/L for men, 1.05 nmol/L for women) at end point in the cotherapy group (19.1% vs4.3%; P = .001), there were no other statistically and clinically significant differences in treatment-emergent laboratory test result abnormalities at end point, including nonfasting glucose levels, between the olanzapine cotherapy group and the monotherapy group.

This double-blind, placebo-controlled study suggests that, in patients with inadequate responses to at least 2 weeks of lithium or valproate monotherapy, the addition of olanzapine may confer additional significant efficacy. No significant changes were seen in extrapyramidal symptoms but a number of adverse events were reported more frequently by patients receiving cotherapy.

The findings in this patient population, characterized as partially nonresponsive to monotherapy, are comparable with those of Müller-Oerlinghausen et al,⁶ who compared the efficacy of combined therapy with conventional antipsychotics and valproate vs valproate monotherapy in patients with bipolar or schizoaffective disorder. Sachs¹⁸ recently reported preliminary results comparing the addition of risperidone, haloperidol, or placebo to lithium or valproate, and found that the addition of risperidone showed a significantly improved response compared with monotherapy.

Considering the exclusion of patients who showed a response after 2 weeks of monotherapy, our patient population should be classified as lithium- or valproate-nonresponders or partial responders. A significantly larger percentage of patients who received cotherapy with olanzapine achieved euthymia than did patients receiving valproate or lithium monotherapy. A potential limitation of this finding is the reliance on a single definition of euthymia, which in this study was based on a YMRS total score less than or equal to 12 as suggested in the original publication of the scale.¹² It is possible that another definition may have led to substantially different remission rates.

One of the most interesting findings was the improvement in depressive symptoms in this population of patients with bipolar manic and mixed episodes. This was particularly striking among patients with moderate-to-severe symptoms of depression, who are particularly resistant to antimanic treatment.¹⁹ Among these patients, a nearly 7-fold greater improvement on the HAMD-21 scale was seen in the cotherapy group. Furthermore, nearly5 times as many patients in the cotherapy group showed at least a 50% improvement of depressive symptoms. These findings suggest that cotherapy may provide substantial efficacy against depressive symptoms in this patient population.

Subgroup analysis also produced noteworthy findings. In patients without psychotic features, cotherapy was significantly superior to monotherapy. However, in those with psychotic features, the difference did not reach statistical significance, doubtlessly due to the greater statistical power in the group of patients without psychotic features.

Patients with mixed episodes who received cotherapy had a significantly larger response rate than did those who received monotherapy. By contrast, patients who had purely manic episodes showed no significant difference between treatments. The response to cotherapy was similar in magnitude in both patient types. However, patients with mixed episodes showed weaker responses to monotherapy than did patients who were purely manic. These results again suggest that patients who had bipolar disorder whose illness characteristics include the presence of depressive symptoms may particularly benefit from the addition of olanzapine.

Overall YMRS responses to cotherapy with valproate were similar in magnitude to responses to cotherapy with lithium. On a numeric basis, response to lithium monotherapy was larger than the response to valproate monotherapy. This, plus the larger number of patients receiving valproate, might explain the significant difference between responses to valproate cotherapy vs valproate monotherapy and the nonsignificant difference between responses to lithium cotherapy vs lithium monotherapy.

Adverse events occurring at a significantly higher rate in the cotherapy group and severe enough to result in discontinuation included somnolence and weight gain. Importantly, however, the olanzapine cotherapy group did not experience a significant increase in nonfasting plasma glucose levels or any treatment-emergent hyperglycemia. Weight gain during treatment with olanzapine has been described previously.^20- 21 However, lithium and valproate are both known to be associated with weight gain.^22- 23 Weight gain in the cotherapy group was similar to that reported for olanzapine monotherapy, suggesting that there is no clear synergism between olanzapine and either lithium or valproate in causing weight gain.

Our study has several limitations. First, assignment to valproate or lithium was not randomized but reflected the treatment preferences of clinicians and investigators. The larger recruitment of patients receiving valproate monotherapy reflects the current practice in the United States of a more extensive use of valproate than lithium by patients who have bipolar disorder with manic or mixed episodes. Because the study was not powered to show significant differences in the primary outcome variables stratified by mood stabilizer, significant differences were found only when comparisons were made between monotherapy and cotherapy; few differences were found when patients were stratified by mood stabilizer. Second, our sample was restricted to partial responders to lithium or valproate. Patients received monotherapy for only 2 weeks, with mean blood levels of 0.76 mmol/L and 63.6 µg/mL for lithium and valproate, respectively. In clinical practice, practitioners may be inclined to maximize the dose of monotherapy treatment before initiating cotherapy treatment. Finally, the lack of an olanzapine-monotherapy arm prevented us from considering a possible synergistic effect between olanzapine and the mood stabilizer. Moreover, the lack of a comparator arm, such as a mood stabilizer plus another agent, also limits our conclusions about olanzapine's unique effects when added to mood stabilizers. It would be of considerable benefit to investigate the efficacy of olanzapine cotherapy vs a combination therapy consisting of lithium plus valproate, thereby more accurately reflecting current clinical practice.

In summary, our findings suggest that, in patients with bipolar manic or mixed episodes who demonstrate inadequate responsiveness to at least 2 weeks of mood-stabilizer monotherapy, the combination of lithium or valproate plus olanzapine may provide additional efficacy compared with either agent alone. Patients treated with combination therapy experienced more adverse events but none seemed to be life-threatening. The response in patients without psychotic features and the improvement of depressive symptoms suggests that the combination of olanzapine and lithium or valproate may have mood-stabilizing properties in the acute treatment of bipolar manic or mixed episodes.

Accepted for publication June 26, 2001.

This study was sponsored by Eli Lilly & Co, Indianapolis, Ind.

This study was presented at the annual meeting of the American Psychiatric Association, Chicago, Ill, May 15, 2000.

Study Sites Principal Investigators

North Bay Psychiatric Hospital, North Bay, Ontario: S. Adams, MD; The International Mood Center, La Jolla, Calif: H. S. Akiskal, MD; University of Pennsylvania, Philadelphia, Pa: J. D. Amsterdam, MD; Princeton Biomedical Research, Princeton, NJ: J. T. Apter, MD; University of Texas Health Science Center, San Antonio: C. L. Bowden, Jr, MD; Case Western Reserve University School of Medicine, Cleveland, Ohio: J. R. Calabrese, MD; McLean Hospital/Harvard Medical School, Belmont, Mass: F. Centorrino, MD; Western Psychiatric Institute and Clinic/University of Pittsburgh Medical Center, Pittsburgh, Pa: K. N. R. Chengappa, MD; Reno, Nev: M. De Priest, MD; Falls Church, Va: D. G. Daniel, MD; Clinical Investigation Specialists Inc, Little Rock, Ark: B. Diner, MD; Pharmacology Research Corporation, Salt Lake City, Utah: J. M. Ferguson, MD; Polyclinique Saint Laurent, Quebec City, PQ: M. J. Filteau; Riverview Consultants, Calgary, Alberta: C. Gorman, MD; Baltimore Veterans Affairs Medical Center, Baltimore, Md: P. Hauser, MD; St Luke's Roosevelt Outpatient Clinic, New York, NY: S. E. Hyler, MD; Montefiore Medical Center/Albert Einstein College of Medicine, New York: N. Iqbal, MD; Center for Addiction and Mental Health, Toronto, Ontario: S. Kennedy, MD; Stanford University School of Medicine, Stanford, Calif: T. A. Ketter, MD; Clinic Bois de Boulogne, Montreal, Quebec: J. Leblanc, MD; Birmingham Psychiatric Pharmaceutical Studies, Birmingham, Ala: H. E. Logue, MD; University of Cincinnati Medical Center, Cincinnati, Ohio: S. L. McElroy, MD; University of New Mexico Health Sciences Center, Albuquerque: H. G. Nurnberg; Veterans Affairs Medical Center, Dallas, Tex: F. Petty, MD, PhD; Mesa Vista Hospital, San Diego, Calif: M. Plopper, MD; Bio Behavioral Research Center, Decatur, Ga: R. A. Riesenberg, MD; Massachusetts General Hospital, Boston, Mass: G. S. Sachs, MD; University of Texas Southwestern Medical Center, Dallas: T. Suppes, MD, PhD; Crozer Chester Medical Center, Philadelphia: S. D. Targum, MD; University of Colorado Health Sciences Center, Denver, Colo: M. Thomas, MD; Medstream Inc, Milwaukee, Wis: R. I. H. Wang, MD, PhD; Raleigh, NC: R. H. Weisler, MD; Psychiatric Institute of Florida, Orlando: S. A. West, MD.

Corresponding author and reprints: Mauricio Tohen, MD, DrPH, Lilly Research Laboratories, Eli Lilly & Co, Indianapolis, IN 46285.