Effectiveness and medical costs of divalproex versus lithium in the treatment of bipolar disorder: Results of a naturalistic clinical trial

Journal of Affective Disorders 86 (2005) 183 – 193 www.elsevier.com/locate/jad

Research report

Effectiveness and medical costs of divalproex versus lithium in the treatment of bipolar disorder: Results of a naturalistic clinical trialB Dennis A. Revickia,T, Robert M.A. Hirschfeldb, Eileen P. Ahearnc, Richard H. Weislerd, Cynthia Palmere, Paul E. Keck Jr.f a

Center for Health Outcomes Research, MEDTAP International, 7101 Wisconsin Avenue, Suite 600, Bethesda, MD 20814, United States b Department of Psychiatry, University of Texas Medical Branch, Galveston, TX, United States c Mental Health Services, William S. Middleton Veterans Hospital, Madison, WI, United States d Raleigh, North Carolina and Duke University Medical Center, Durham, NC, United States e Center for Organization and Delivery Studies, Agency for Health Research and Quality, Rockville, MD, United States f Psychopharmacology Research Program, Department of Psychiatry, University of Cincinnati College of Medicine, Cincinnati, OH, United States Received 24 September 2003; accepted 12 January 2005

Abstract Objective: The clinical, quality of life (QOL), and medical cost outcomes of treatment with divalproex were compared with lithium in patients with bipolar I disorder over 1 year. Methods: In a pragmatic, randomized clinical trial, 201 adults hospitalized with bipolar I manic or mixed episodes were randomized to divalproex or lithium, in addition to usual psychiatric care, and followed for 1 year. All subsequent treatment of bipolar disorder was managed by the patient’s psychiatrist. Symptoms of mania and depression were evaluated at baseline and at hospital discharge. Assessments at the start of maintenance therapy and after 1, 3, 6, 9 and 12 months included manic and depressive symptoms, disability days and QOL. Medical resource use data were also collected monthly and costs were estimated using national sources. Results: Divalproex-treated patients (12%) were less likely to discontinue study medications for lack of efficacy or adverse effects than lithium-treated patients (23%). No statistically significant differences between the treatment groups were observed over the 1-year maintenance phase for clinical symptoms, QOL outcomes, or disability days. Mean estimated total medical costs were $28,911 for the divalproex group compared with $30,666 for the lithium treatment group. Patients continuing mood stabilizer therapy at 3 months had slightly better health outcomes and substantially lower total medical costs than those who discontinued therapy ($10,091 versus $34,432, respectively).

B Presented at Annual Meeting of the American College of Neuropsychopharmacology, Las Croabas, Puerto Rico, December 1998; Annual Meeting of the American Psychiatric Association, Washington, DC, May 1999; and at Challenges for the 21st Century: Mental Health Services Research, Washington, D.C., July 18–20, 2000. This research was supported by a grant from Abbott Laboratories, Abbott Park, Illinois. T Corresponding author. Tel.: +1 301 654 9729; fax: +1 301 654 9864. E-mail address: [email protected] (D.A. Revicki).

0165-0327/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jad.2005.01.002

184

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193

Conclusions: Divalproex maintenance treatment for bipolar disorder resulted in comparable medical costs, clinical and QOL outcomes compared with lithium. Patients remaining on mood stabilizer therapy had substantially lower total medical costs and better health outcomes compared with those who discontinued therapy. D 2005 Elsevier B.V. All rights reserved. Keywords: Divalproex; Lithium; Cost-effectiveness analysis; Bipolar disorder; Clinical trials

1. Introduction Bipolar disorder is a severe and disabling psychiatric disorder with a 1-year prevalence rate of 0.7% to 1.3% in the United States (Kessler et al., 1994; Narrow et al., 2003). The natural history of bipolar disorder is characterized by frequent relapse and recurrence (Goodwin and Jamison, 1990), with impaired patient functioning and well-being even after symptomatic recovery (Post et al., 2003; Coryell et al., 1993; Cooke et al., 1996; Calabrese et al., 2003; Leidy et al., 1998; Keck et al., 1998). Long-term prospective studies demonstrate that after an initial manic episode less than half of patients demonstrate a good response to treatment (Post et al., 2003; Tohen et al., 1990). In the United States, the economic burden of bipolar disorder has been estimated at $7 billion in direct medical costs and $38 billion in indirect costs (Wyatt and Henter, 1995). Bryan-Comstock et al. (2002) recently found that patients with bipolar disorder had annual medical costs that were four times those of patients without bipolar disorder. Lithium and divalproex have been extensively studied in randomized, clinical trials of acute bipolar mania, with response rates ranging from 50% to 70% (American Psychiatric Association, 1994; Price and Heninger, 1994; Bowden, 1998). The long-term response rates for lithium in patients with bipolar disorder range from 40% to 70%, and relapse rates range from 34% to 61% (Goodwin and Jamison, 1990; American Psychiatric Association, 1994; Price and Heninger, 1994; Bowden, 1998; Harrow et al., 1990; O’Connell et al., 1991). Clinical trials demonstrate that divalproex has comparable efficacy to lithium in treating acute mania and in long-term maintenance studies (Bowden et al., 1994; Freeman et al., 1992; Pope et al., 1991; Keck et al., 1998; Bowden et al., 2000). Few pharmacoeconomic evaluations have compared the cost-effectiveness of divalproex and lithium

for the treatment of bipolar disorder (Keck et al., 1996). Although clinical decision analysis and modeling can provide some evidence of the possible costeffectiveness of alternative medical treatments, the most valid evidence of cost-effectiveness comes from prospective randomized studies that combine measurement of patient outcomes and health service use and costs (Keck et al., 1996). No clinical trials have compared the long-term quality of life (QOL) and medical costs of divalproex and lithium in patients with bipolar disorder. This pragmatic, randomized clinical trial compared the effectiveness and medical costs of lithium and divalproex in patients with an acute manic episode of bipolar disorder requiring hospitalization. The clinical trial was performed under conditions of routine psychiatric practice without blinding, and patients continued to receive their usual medical care while they participated in this yearlong study. Pragmatic clinical trials provide information on health outcomes and medical costs in more generalizable community settings compared to randomized clinical trials designed to evaluate efficacy and safety (Revicki and Luce, 1995; Revicki and Frank, 1999).

2. Methods 2.1. Design and patient sample This multi-center, open-label, parallel-group randomized clinical trial was conducted in 33 community- and university-based psychiatry practices throughout the U.S. between 1995 and 1997. The study was divided into two parts: an acute phase, extending from randomization to hospital discharge, and a maintenance phase that began the day after hospital discharge and continued for a total duration of 12 months. Patients at least 18 years of age with a DSM-IV diagnosis of bipolar disorder and hospital-

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193

ized for treatment of an acute manic or mixed episode were randomized in a 1:1 ratio (within 72 h of hospitalization) to receive divalproex sodium (DepakoteR, Abbott Laboratories) or lithium (EskalithR, GlaxoSmithKline), in addition to usual psychiatric care. Female patients of childbearing age must have been using effective birth control and nonlactating. Patients were excluded from the study if they had clinically significant focal neurological abnormalities, seizure disorders, other diseases of the nervous system, drug-induced or AIDS-induced mania, or an uncontrolled medical disorder (e.g., gastrointestinal, renal, hepatic, endocrine, cardiovascular, pulmonary, immunological or hematologic disease). The use of anticoagulants was prohibited. The research protocol was approved by the institutional review board of each participating research site and all patients provided written informed consent. 2.2. Treatment regimen Divalproex was initiated at 15 to 20 mg/kg/day or based on usual psychiatric practice. Lithium was dosed according to package insert guidelines up to 1800 mg/day during acute mania treatment and between 900 mg and 1200 mg/day for maintenance therapy. During hospitalization, the dosage of study drug was titrated as required to optimize clinical response. At hospital discharge, the patient received the study medication (quantity determined by the physician), instructions on self-administration of study medication, and procedures for obtaining study drug refills. Investigators were allowed to prescribe other medications as clinically necessary to treat bipolar disorder, but were strongly discouraged from treating patients concurrently with the other study medication unless clinically required. Throughout the maintenance phase of the study, dosage of study medication and modifications of treatment were adjusted as clinically necessary by the patient’s psychiatrist. 2.3. Data collection in acute and maintenance phases Symptoms of mania and depression were assessed at baseline and at hospital discharge for each patient using the Mania Rating Scale (Endicott and Spitzer, 1978) and the Depressive Syndrome Scale (Endicott

185

and Spitzer, 1978). During the acute phase of the study, adverse events and concomitant medications were recorded. During the 12-month maintenance phase of the trial, investigators assessed patients (regardless of continuation of study drug) according to their usual clinical practice and collected information on medical resource usage, adverse events, and the use of study drug and concomitant medications. In addition, trained research personnel interviewed patients by telephone within 3 days of hospital discharge and thereafter on a monthly basis. During these interviews, clinical and QOL outcomes (months 1, 3, 6, 9, and 12 only) and medical resource use data (monthly) were collected. The mania and depression components of the World Health Organization Composite International Diagnostic Interview (CIDI) (World Health Organization, 1989) modified for telephone surveys (Kessler et al., 1994) were included in the interviews. The CIDI has demonstrated good reliability and validity when administered in telephone interviews (Wittchen et al., 2001). 2.4. QOL and other patient outcomes The Medical Outcomes Study 36-item short-form Health Survey (SF-36) (Ware et al., 1993, 1994) was used to measure patient functioning and wellbeing. We calculated SF-36 mental component summary (MCS) and physical component summary scores (PCS) (Ware et al., 1994). The 17-item Mental Health Index (MHI-17) was also included in the survey (Stewart et al., 1992). Two standard questions on disability days were also included in the interviews (Broadhead et al., 1990). Patients were asked how many days over the previous month their usual activities had been restricted for more than one-half day, and how many days they had spent more than half the day in bed because of illness. We evaluated the reliability and validity of telephone assessment of the patient reported clinical, QOL and other outcomes in a separate sample before this clinical trial started (Revicki et al., 1997). The results indicated that reliable and valid data could be collected in telephone interviews with bipolar disorder patients in the community and these patient-reported data were comparable to data collected during inperson interviews.

186

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193

2.5. Medical resource use and costs Data were collected on the number of outpatient psychiatric physician or clinic visits, outpatient medical visits, home health service visits, emergency room visits, hospitalizations for psychiatric and other reasons, psychologist visits, and other mental health provider visits. For emergency room visits and hospitalizations, data were collected on reason for visit/admission, length of stay (for hospitalizations), and medical charges. The investigators also provided data on the number of patient visits to their clinical center, study medication doses and any change in doses, and concomitant medications. Only direct medical costs were evaluated. Because economic data were collected in different time periods, all prices (costs) were adjusted to 1997 dollar values (using the appropriate medical component of the Consumer Price Index) to reflect purchasing power in common dollar terms (Luce et al., 1996). The Universal Bill (UB-92) was used to determine the cost of hospitalizations. Inpatient physician charges were based on the assumption of one initial visit on the day of admission and one subsequent visit per additional inpatient day, using mean national CPTcoded charges for moderately complex visits from the Physicians Fee and Coding Guide (Health Care Consultants of America, Inc., 1997). Mean national charges for visits to psychiatrists or other physicians were applied to the number of reported outpatient physician visits, using charges reported for moderately complex visits in the Physicians Fee and Coding Guide (Health Care Consultants of America, Inc., 1997). Outpatient visits to the clinical investigators were also recorded and assigned mean national charges. We assumed a mean charge of $49 per visit to a non-physician health care provider, based on the 1997 Health Care Financing Administration’s average reimbursed charge for a visit with a clinical social worker (Health Care Financing Administration, 1997). Medication costs for the study drugs and concomitant medications were obtained from the Red Book (Allen, 1997), which includes the average wholesale price for both prescription and over-thecounter drugs. A patient-year approach was used to estimate missing resource utilization values (mean for all available measurements was carried forward for

dropouts) (Rutten-van Molken et al., 1994). The mean of the observed counts of resources per month before withdrawal were linearly extrapolated to the remainder of the study period. 2.6. Statistical analysis The main statistical analyses were conducted using an intent to treat approach, with patients categorized according to their initial randomization assignment. Patients who received at least one dose of the study medication (lithium or divalproex) and who completed at least one follow-up assessment during the maintenance phase were considered evaluable. For the comparison of clinical symptom and QOL outcome data, patients needed to have a baseline and at least one follow-up assessment. The last observation carried forward technique was used to impute missing follow-up assessments. Baseline demographic, clinical, and QOL measures were compared between treatment groups using chi-square tests for categorical variables and t-tests for continuous variables. The primary study endpoints were mean months without DSM-IV level manic and depressive symptoms, MCS scores, and total medical costs over the course of the study. The sample sizes in this clinical trial have 87% statistical power to detect a clinically meaningful between-treatment difference in MCS scores, assuming a standard deviation of 12 and a difference of 3 points (Revicki et al., 1997). A twotailed p-value of 0.05 was used for all statistical tests, and no adjustments were made for multiple comparisons. Student’s t-tests were used to compare between treatment differences in mean months without manic and depressive symptoms over the 12-month study. Analysis of covariance (ANCOVA) models were used to compare mean MCS scores between treatment groups at 1, 3, 6, 9, and 12 months after hospital discharge for acute mania after controlling for baseline MCS scores. Because the cost data were highly skewed, the Wilcoxon rank sum test was used to compare medical costs between treatment groups. Planned secondary analyses compared study treatment discontinuation rates using chi-square tests and PCS scores and disability days between treatment groups using similar ANCOVA models. Cox proportional hazard models compared time to first hospital

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193

admission or emergency department visit. Student’s ttests compared mean number of hospital days over the maintenance phase. An exploratory analysis was completed comparing clinical outcomes, QOL (i.e., MCS and PCS scores), restricted activity days, and total medical costs for patients continuing or discontinuing any mood stabilizing therapy after 3 months. Ordinary least squares regression models were used to compare these two groups over the remaining 9 months of follow-up. The regression models included age, gender, alcohol/ substance abuse, rapid cycling/mixed mania, and discharge Mania Rating Scale and Depression Rating Scale scores. For the planned secondary and exploratory analyses, we report actual p-values for these statistical comparisons. These p-values are reported for descriptive purposes only.

3. Results Two hundred and twenty-one bipolar disorder patients were randomized to either divalproex (n=112) or lithium (n=109) and received at least one dose of study medication during the acute hospital

187

phase (Fig. 1). The majority of patients (93% of those randomized to divalproex and 89% of those randomized to lithium) entered the maintenance phase of the clinical trial. The two groups were well matched at baseline for demographic variables, clinical characteristics, and QOL measures (Table 1). Sixty-five percent of divalproex group and 63% of the lithium group were previously treated with lithium; 43% of the divalproex group and 32% of the lithium group were previously treated with divalproex. No statistically significant differences were observed between the lithium and divalproex groups based on clinical endpoints at hospital discharge or initial hospital length of stay (divalproex: 11.0F9.4 days; lithium: 11.7F9.6 days). Complete follow-up data (i.e., N10 months of follow-up) were available for 79 divalproex patients (76%) and 85 lithium patients (88%) ( p=0.051). The average duration of study participation was 306 days (S.D.=119) in the divalproex group and 339 days (S.D.=85) in the lithium group (t=2.25, df=187, p=0.025). There were no differences between early study treatment dropouts (within 3 months; n=17) and those who continued in the study (n=184) based on gender, age, race/ethnic group, initial hospital length of stay,

221 subjects randomly allocated

112 allocated to divalproex

109 allocated to lithium

112 received allocated treatment

109 received allocated treatment

104 entered maintenance treatment

97 entered maintenance treatment

Followed up at Month 3: n = 86 Month 6: n = 80 Month 9: n = 75 Month 12: n = 69

Followed up at Month 3: n = 86 Month 6: n = 82 Month 9: n = 77 Month 12: n = 65

86 in analysis 18 excluded

86 in analysis 11 excluded

Fig. 1. Flow diagram for divalproex–lithium pragmatic clinical trial.

188

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193

Table 1 Baseline demographic, clinical characteristics, and quality of life outcomes by treatment group

Age in years, mean (S.D.) Male, % Caucasian, % Education, % bHigh school High school grad NHigh school Rapid cycling, % Mixed mania, % History of alcohol abuse, % History of drug abuse, % MRS score, mean (S.D.) DSS score, mean (S.D.) Days of hospitalization during acute phase, mean (S.D.) SF-36 summary scores Physical component Mental component

Divalproex (N=104)

Lithium (N=97)

38.8 (13.8) 48.1 59.6

37.4 (11.7) 49.5 57.7

21.2 22.1 56.7 11.1 48.1 44.2 43.3 22.0 (8.9) 24.0 (7.3) 11.0 (9.4)

18.6 20.6 60.8 13.4 55.2 48.5 49.5 23.2 (8.5) 25.2 (8.0) 11.7 (9.6)

48.3 (9.6) 40.4 (14.2)

49.4 (9.8) 39.3 (14.0)

after 3 months. Three percent of divalproex patients also received lithium, while 6% of lithium patients received divalproex. After 3 months, 29% of divalproex patients continued treatment with antipsychotics or antidepressants, compared with 30% of lithium patients. 3.2. Clinical and patient reported outcomes

No statistically significant differences between treatment groups.

treatment group or on clinical history characteristics. As compared to those who continued in the study, a higher proportion of early dropouts had less than a high school education (18% versus 41%, p=0.049), had a history of suicidality (29% versus 65%, p=0.005), were not compliant with the study protocol (27% versus 82%, pb0.0001), and were discharged to unsupervised settings (24% versus 47%, p=0.046).

Less than half of study patients (39% of divalproex-treated and 40% of lithium-treated) remained on study medication over the entire 12-month study. The primary reasons that patients discontinued their study medication were: lack of efficacy or adverse effects (12% versus 23% of divalproex- and lithiumtreated patients, respectively; p=0.035), and noncompliance with the clinical protocol (25% versus 16%, respectively; p=0.094), loss to follow-up or other administrative reasons (24% versus 23%, respectively; p=0.820). Five percent of divalproex-treated patients compared with 8% of lithium-treated patients discontinued due to insufficient efficacy, and 7% of divalproex group patients compared with 14% of lithium-treated patients discontinued due to adverse effects. There were no statistically significant differences between the two treatment groups over the 12-month Table 2 Qualify of life outcomes by treatment group Month

3.1. Psychopharmacologic treatment The average daily dosage of divalproex was 1504 mg (S.D.=583, range 466 to 4000) and the average daily dosage of lithium was 1213 mg (S.D.=398, range 471 to 2400) over the 12-month study. During the first 3 months (including the initial hospitalization), 26% and 22% of the patients received divalproex and lithium monotherapy, respectively. In the initial 3 months of the study, 14% of divalproextreated patients also received lithium, while 18% of lithium-treated patients also received divalproex. During early treatment, including the initial hospital stay, most study patients received an antipsychotic combined with either lithium (50%) or divalproex (47%). The rate of monotherapy increased to 69% in the divalproex group and 63% in the lithium group

1

3

9

12

(S.D.) 49.8 (11.2) 49.9 (10.1)

48.8 (11.4) 49.1 (10.1)

49.1 (10.9) 50.4 (9.8)

Mental component summary, mean (S.D.) Divalproex 42.0 41.8 43.5 (14.4) (13.4) (13.8) Lithium 42.2 40.4 42.6 (12.9) (14.1) (12.5)

44.7 (12.9) 43.2 (13.7)

44.1 (12.6) 43.1 (12.2)

Mental health index-17 mean (S.D.) Divalproex 64.5 66.0 (19.0) (18.1) Lithium 65.8 63.1 (17.1) (19.5)

67.6 (17.3) 65.1 (19.0)

66.8 (18.8) 65.7 (17.1)

Physical component summary, mean Divalproex 48.2 49.0 (11.5) (10.7) Lithium 50.4 50.7 (9.8) (11.1)

6

66.3 (18.2) 64.5 (17.1)

No statistically significant differences between treatment groups.

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193

189

Table 3 Medical costs by treatment groupa Mean (SE) Total medical costs Inpatient medical costs Outpatient medical costs Medication costs Study medications Other medications a b

Divalproex

Lithium

$28,911 (3599) $23,790 (3537) $3043 (646)

$30,666 (7364) $27,655 (7376) $2415 (202)

$1288 (269) $790 (247)

$40 (5) $555 (150)

Differenceb

p-value

$1755 $3865 $628

0.693 0.695 0.470

$1248 $235

b0.001 0.316

Total and disaggregated medical costs with imputation using the person-mean method. Difference is divalproex lithium.

study for the primary clinical symptom or MCS outcomes. The mean (S.D.) number of months without DSM-IV mania and depression was 5.3 (4.6) for divalproex and 5.4 (4.4) for lithium (t=0.24, df=199, p=0.814). Likewise, there were no statistically significant differences between the treatment groups based on the MCS (Table 2), the PCS, or measures of disability days (not shown).

S.D.=27.7) (t=0.81, df=180, p=0.417). While study medication costs were higher, total medical costs and inpatient costs were lower in the divalproex group compared with the lithium group (Table 3). Mean total medical costs were $28,911 for the divalproex group and $30,666 for the lithium group ( p=0.693).

3.3. Medical resource use and costs

A total of 129 (64%) patients continued any mood stabilizing therapy (i.e., lithium, divalproex, carbamazepine, etc.) beyond 3 months and 72 patients (36%) discontinued mood stabilizer therapy at or before 3 months. Mean PCS scores were no different (data not shown), but there was a trend for improved MCS scores at 6 months (mood stabilizer mean= 43.7; no mood stabilizer mean=40.7, p=0.194), 9

Time to first hospitalization or emergency department visit was not significantly different between the treatment groups ( p=0.616). The divalproex-treated group had fewer days in the hospital over the 12month follow-up period (mean=11.3, S.D.=21.4) compared with the lithium-treated group (mean=14.1,

$40,000

3.4. Outcomes and costs of treatment continuation

Continued mood stabilizer at 3 months Discontinued mood stabilizer by 3 months

$30,000

$20,000

$10,000

$0 Fig. 2. Annual medical costs by mood stabilizer treatment status. Continued treatment status defined as continuing any mood stabilizer medication at 3-month assessment. Discontinued mood stabilizer treatment defined as no mood stabilizer medication at 3-month assessment.

190

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193

months (mood stabilizer mean=44.9; no mood stabilizer mean=40.7, p=0.057), and 12 months (mood stabilizer mean=44.0; no mood stabilizer mean=41.9, p=0.280) in the group continuing treatment. Patients who remained on mood stabilizer therapy reported significantly fewer restricted activity days compared with patients discontinuing therapy (mean=12.8F2.4 and 23.6F4.8, respectively; t=2.01, df=177, p=0.048). Mean hospital days over the 12 month follow-up were significantly lower among patients continuing therapy compared with patients discontinuing therapy (mean=14.7F4.1 and 33.4F7.3, respectively; t=2.43, df=199, p=0.016). The mean total medical costs for the group discontinuing therapy were three-fold higher than the medical costs of those remaining on therapy ( p=0.023) (Fig. 2). The differences in total costs were primarily due to greater hospital costs ($29,770 versus $6300) ( p=0.022) and slightly greater outpatient treatment costs ($3410 versus $2366) ( p=0.273) in the group that discontinued therapy.

4. Discussion Divalproex and lithium were found to be comparable based on alleviation of clinical symptoms, QOL, and disability days. This is not unexpected, since psychiatrists were managing their patients with the study medications as well as other psychopharmacologic medications to optimize clinical response. While improvement in psychological and social functioning was observed, the bipolar disorder patients in this study still exhibited impaired functioning and psychological well-being after 12 months of treatment, reflecting previously demonstrated poor patient outcomes despite symptomatic improvement (Post et al., 2003; Coryell et al., 1993; Cooke et al., 1996; Calabrese et al., 2003; Leidy et al., 1998; Keck et al., 1998; Revicki et al., 1997, in press). In this study, divalproex was associated with numerically lower total medical costs ($28,911 versus $30,666; difference of $1755 over 1 year) and an average of almost three fewer hospital days during the maintenance phase compared with lithium, but these differences were non-significant. Early dropout from study treatment (i.e., within first 3 months) was not associated with a history of

mixed episodes, comorbid alcohol abuse, baseline clinical symptom severity or demographic characteristics. There were differences between patients who remained on study treatments compared to those who discontinued early on several clinical-related variables. Early treatment dropouts were more likely to have a history of suicide related behavior and were more likely discharged into an unsupervised residential setting. These findings suggest that post discharge planning is essential for enhancing delivery of outpatient mental health services, mood stabilizer therapy adherence, and functioning in the community. There are medical and economic benefits for patients to continue mood stabilizer therapy. Clinical outcomes and psychological well-being were slightly better in patients remaining on treatment. Patients on mood stabilizer therapy reported half the number of disability days compared with those who discontinued treatment (i.e., 1 day per month versus 2 days per month). Patients discontinuing mood stabilizer therapy by 3 months had total costs that were three-fold higher than those of patients continuing therapy ($34,432 versus $10,091, respectively). Li et al. (2002), based on California Medicaid program data, also found that mood stabilizer use was associated with reduced inpatient and outpatient medical costs. Medication adherence is associated with hospitalization and medical costs (Svarstad et al., 2001). The lithium-treated patients remained in the study longer than the divalproex-treated patients. Differences between mood stabilizing agents that lead to differences in compliance may have important implications on treatment outcomes. For instance, differences between agents in their potential to cause troublesome side effects may affect patients’ ability to continue treatment. In this study, fewer divalproex-treated patients discontinued medication due to lack of efficacy or side effects than did lithium-treated patients. In another clinical trial, Bowden et al. (2000) also observed fewer treatment discontinuations due to intolerance or non-compliance in the divalproex-treated group compared with the lithium-treated group. Using a clinical decision analysis model, Keck et al. (1996) found that divalproex was cost-saving for all bipolar disorder patients, but this cost difference was primarily attributable to patients with rapid cycling or mixed episodes. Lithium was cost-saving

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193

for patients with classic mania. In comparison to the 9% savings in total medical costs for divalproextreated patients estimated by Keck et al. (1996), we found that the savings were approximately 6% (about $1700 per patient with bipolar disorder per year). This magnitude of cost savings are likely significant for the health care system. For example, in a health care system with 5000 patients with bipolar disorder, treatment with divalproex may be associated with about $8.5 million in savings. A recent clinical trial comparing divalproex and olanzapine treated acute mania patients found that the 12-week outpatient medical costs were significantly lower in the divalproex-treated group ($541 versus $1080) (Revicki et al., 2003). Several characteristics of this study need to be considered when interpreting the cost and effectiveness outcomes. First, the patients in this study initially had acute mania requiring hospitalization, and their clinical course was likely more severe than the general population of patients with bipolar disorder treated in the United States. Acute manic episodes are often managed in intensive outpatient settings, and hospitalization is reserved for only more seriously ill patients. Second, the less adherence, less educated, and suicidal patients were more likely to discontinue study treatment by 3 months, although most of these patients were followed for the duration of the study. Therefore, the results of this study may be generalizable to only a restricted sub-group of more severely ill bipolar disorder patients. Third, lithium therapy is not recommended for rapid cyclers and while there were few of these patients (b12%) in the study, it may have affected the results of the study. No serum drug level data were systematically collected during this naturalistic clinical trial. Treating psychiatrists evaluated drug levels according to their individual clinical practice and the average daily doses for lithium and divalproex were comparable to other clinical trials (Bowden et al., 2000). The study sample size had limited statistical power to detect moderate differences in total medical costs, given the large variances often observed on cost variables. Larger sample sizes (i.e., 300–500 subjects per group) are typically required to evaluate differences in medical costs (Health Care Consultants of America, Inc., 1997). Finally, patients and physicians were not blinded to treatment, introducing a potential bias. All patient outcome assess-

191

ments during the maintenance phase were performed by research interviewers blinded to treatment condition, which should obviate this source of bias in the outcomes assessments. The results of this pragmatic, randomized study suggests that divalproex treatment for bipolar disorder resulted in comparable medical costs, clinical and QOL outcomes, and fewer discontinuations due to lack of efficacy or adverse events as compared to lithium. More important, patients remaining on any mood stabilizer therapy had substantially lower total medical costs and better functional outcomes compared to those who discontinued therapy. Clearly, treatment interventions that are effective at maintaining bipolar disorder patients on mood stabilizer therapy have significant financial implications for the mental health care system and potential benefits in patient functioning and well-being.

Acknowledgments We would like to acknowledge the following investigators and institutions for their participation in the Depakote Outcomes Study: Lori Altschuler, MD, Los Angeles, CA; Charles L. Bowden, MD, San Antonio, TX; Steven Buchanan, MD, Little Rock, AR; James Chou, MD, Orangeburg, NY; Cynthia Conrad, MD, PhD, Hartford, CT; Bradley Diner, MD, North Little Rock, AR; Renee DuPont, MD, La Jolla, CA; Cathy Frank, MD, Detroit, MI; Ira Glick, MD, Stanford, CA; Laszlo Gyulai, MD, Philadelphia, PA; Donald Hilty, MD, Sacramento, CA; Robert Hirschfeld, MD, Galveston, TX; Naveed Iqbal, MD, Bronx, NY; Donna Jermain, PharmD, Temple, TX; Richard Johnson, MD, Memphis, TN; Gary Kaplan, MD, Providence, RI; Paul Keck, Jr., MD, Cincinnati, OH; Ronald Landbloom, MD, St. Paul, MN; Patricia Marken, PharmD, University of Missouri Kansas City, Kansas City, MO; Charles Merideth, MD, San Diego, CA; Marvin Miller, MD, Indianapolis, IN; Jeff Mitchell, MD, Albuquerque, NM; James Moore, MD, Daytona Beach, FL; S. Craig Risch, MD, Charleston, SC; Alan Rosenbaum, MD, Farmington Hills, MI; Gary Sachs, MD, Boston, MA; David Sack, MD, Costa Mesa, CA; G. Michael Shehi, MD, Birmingham, AL; Alan Swann, MD, Houston, TX; Mauricio Tohen, MD, Carlos Zarate, Jr., MD, Belmont, MA;

192

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193

Julia Vertrees, PharmD, San Antonio, TX; Richard Weisler, MD, Raleigh, NC; and Jay Yeomans, MD, Charlotte, NC. References Allen, C.B. (Ed.), 1997. Drug Topics Red Book. Medical Economics Company, Inc., Montvale, NJ. American Psychiatric Association, 1994. Practice guidelines for the treatment of patients with bipolar disorder. American Journal of Psychiatry 151, S1 – S36 (12 Suppl.). Bowden, C.L., 1998. Key treatment studies of lithium in manicdeppressive illness: efficacy and side effects. Journal of Clinical Psychiatry 59 (Suppl. 6), 13 – 19. Bowden, C.L., Brugger, A.M., Swann, A.C., 1994. Efficacy of divalproex vs. lithium and placebo in the treatment of mania. Journal of the American Medical Association 271, 918 – 924. Bowden, C.L., Calabrese, J.R., McElroy, S.L., et al., 2000. A randomized, placebo-controlled 12-month trial of divalproex and lithium in treatment of outpatients with bipolar disorder. Archives of General Psychiatry 57, 481 – 489. Broadhead, W.E., Blazer, D.G., George, L.K., et al., 1990. Depression, disability days, and days lost from work in a prospective epidemiologic survey. Journal of the American Medical Association 264, 2524 – 2528. Bryan-Comstock, L., Stender, M., Devercelli, G., 2002. Health care utilization and costs among privately insured patients with bipolar I disorder. Bipolar Disorder 4, 398 – 405. Calabrese, J.R., Hirschfeld, R.M.A., Reed, M., et al., 2003. Impact of bipolar disorder on a U.S. community sample. Journal of Clinical Psychiatry 64, 425 – 432. Cooke, R.G., Robb, J.C., Young, L.T., et al., 1996. Well-being and functioning in patients with bipolar disorder assessed using the MOS 20-item short form (SF-20). Journal of Affective Diseases 39, 93 – 97. Coryell, W., Scheftner, W., Keller, M., et al., 1993. The enduring psychosocial consequences of mania and depression. American Journal of Psychiatry 150, 720 – 727. Endicott, J., Spitzer, R., 1978. A diagnostic interview: the Schedule for Affective Disorders and Schizophrenia. Archives of General Psychiatry 35, 837 – 844. Freeman, T.W., Clothier, J.L., Pazzaglia, P., et al., 1992. A doubleblind comparison of valproate and lithium in the treatment of acute mania. American Journal of Psychiatry 149, 108 – 111. Goodwin, F.K., Jamison, K.R. (Eds.), 1990. Manic-Depressive Illness. Oxford University Press, New York. Harrow, M., Goldberg, J.F., Grossman, L.S., et al., 1990. Outcome in manic disorders: a naturalistic follow-up study. Archives of General Psychiatry 47, 665 – 671. Health Care Consultants of America, Inc., 1997. Physicians Fee and Coding Guide: A Comprehensive Fee and Coding Reference. Augusta, GA: MAG Mutual Group, 1997. Health Care Financing Administration, 1997. Part B Procedure Data by Specialty. Health Care Financing Administration, Washington, D.C.

Keck, P.E., Nabulsi, A.A., Taylor, J., et al., 1996. A pharmacoeconomic model of divalproex vs. lithium in the acute and prophylactic treatment of bipolar I disorder. Journal of Clinical Psychiatry 57, 213 – 222. Keck, P.E., McElroy, S.L., Strakowski, S.M., et al., 1998. Twelvemonth outcome of bipolar patients following hospitalization for a manic or mixed episode. American Journal of Psychiatry 155, 646 – 652. Kessler, R.C., McGonagle, K.A., Zhao, S., et al., 1994. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States. Archives of General Psychiatry 51, 8 – 19. Leidy, N.K., Palmer, C., Murray, M., et al., 1998. Health-related QOL assessment in euthymic and depressed patients with bipolar disorder: psychometric performance of four self-report measures. Journal of Affective Diseases 48, 201 – 214. Li, J., McCombs, J.S., Stimmel, G.L., 2002. Cost of treating bipolar disorder in the California Medicaid (Med-Cal) program. Journal of Affective Disorders 71, 131 – 139. Luce, B.R., Manning, W.G., Siegel, J.E., et al., 1996. Estimating costs in cost-effectiveness analysis. In: Gold, M.R., Siegel, J.E., Russell, L.B., Weinstein, M.C. (Eds.), Cost-Effectiveness in Health and Medicine. Oxford University Press, New York. Narrow, W.E., Rae, D.S., Robins, L.N., Regier, D.A., 2003. Revised prevalence estimates of mental disorders in the United States: using a clinical significance criterion to reconcile 2 surveys’ estimates. Archives of General Psychiatry 59, 115 – 123. O’Connell, R.A., Mayo, J.A., Flatow, L., et al., 1991. Outcome of bipolar disorder on long-term treatment with lithium. British Journal of Psychiatry 159, 123 – 129. Pope Jr., H.G., McElroy, S.L., Keck Jr., P.E., et al., 1991. Valproate in the treatment of acute mania: a placebo-controlled study. Archives of General Psychiatry 48, 62 – 68. Post, R.M., Denicoff, K.D., Leverich, G.S., et al., 2003. Morbidity in 258 bipolar outpatients followed for 1 year with daily prospective ratings on the NIMH life chart method. Journal of Clinical Psychiatry 64, 680 – 690. Price, L.H., Heninger, G.R., 1994. Lithium in the treatment of mood disorders. New England Journal of Medicine 331, 591 – 598. Revicki, D.A., Frank, L., 1999. Pharmacoeconomic evaluation in the real world: effectiveness versus efficacy studies. PharmacoEconomics 15, 423 – 434. Revicki, D.A., Luce, B.R., 1995. Methods of pharmacoeconomic evaluation of new medical treatments in psychiatry. Psychopharmacology Bulletin 31, 57 – 65. Revicki, D.A., Tohen, M., Gyulai, L., et al., 1997. Telephone versus in-person clinical and health status assessment interviews in patients with bipolar disorder. Harvard Review of Psychiatry 5, 75 – 81. Revicki, D.A., Paramore, L.C., Sommerville, K.W., Swann, A.C., Zajecka, J.M., 2003. Divalproex sodium versus olanzapine in the treatment of acute mania in bipolar disorder: health-related quality of life and medical cost outcomes. Journal of Clinical Psychiatry 64, 288 – 294. Revicki, D.A., Matza, L., Flood, E., Lloyd, A., in press. Bipolar disorder and health-related quality of life: review of burden of disease and clinical trials. PharmacoEconomics.

D.A. Revicki et al. / Journal of Affective Disorders 86 (2005) 183–193 Rutten-van Molken, M., Doorslaer, E., van Vlier, R., 1994. Statistical analysis of cost outcomes in a randomized controlled clinical trial. Health Economics 3, 333 – 345. Stewart, A.L., Ware, J.E., Sherbourne, C.D., et al., 1992. Psychological distress/well-being and cognitive functioning measures. In: Stewart, A.L., Ware, J.E. (Eds.), Measuring Functioning and Well-being: The Medical Outcomes Study Approach. Duke University Press, Durham, NC. Svarstad, B.L., Shireman, T.I., Sweeny, J.K., 2001. Using drug claims data to assess the relationship of medication adherence with hospitalization and costs. Psychiatric Services 52, 805 – 811. Tohen, M., Waternaux, C.S., Tsuang, M.T., 1990. Outcome in mania: a 4-year prospective follow-up of 75 patients utilizing survival analysis. Archives of General Psychiatry 47, 1106 – 1111.

193

Ware, J.E., Snow, K.K., Kosinski, M., Gandek, B., 1993. SF-36 Health Survey Manual and Interpretation Guide. The Health Institute. New England Medical Center, Boston, MA. Ware, J.E., Kosinski, M., Keller, S.D., 1994. SF-36 Physical and Mental Health Summary Scales: A User’s Manual. The Health Institute. New England Medical Center, Boston, MA. Wittchen, H.U., Kessler, R.C., Ustin, R., 2001. Properties of the composite international (CIDI) for measuring mental health outcomes. In: Thornicroft, G., Tansella, M. (Eds.), Mental Health Outcome Measures, Second edition. Gaskell, London. World Health Organization, 1989. Composite international diagnostic interview. Version 1.0. World Health Organization, Geneva. Wyatt, R.J., Henter, I., 1995. An economic evaluation of manicdepressive illness. Society of Psychiatry and Psychiatric Epidemiology 30, 213 – 219.