SEARCH

SEARCH BY CITATION

Keywords:

  • antipsychotic;
  • cost;
  • cost-effectiveness;
  • effectiveness;
  • formulary;
  • olanzapine;
  • practical clinical trial;
  • risperidone;
  • schizophrenia

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Objectives:  This randomized, open-label trial was designed to help inform antipsychotic treatment policies. It compared the 1-year cost-effectiveness of initial treatment with olanzapine (OLZ) (n = 229) versus a “fail-first” algorithm on conventional antipsychotics (then olanzapine if indicated) (CON) (n = 214); and versus initial treatment with risperidone (RIS) (n = 221).

Methods:  Individuals with schizophrenia or schizoaffective disorder were recruited from May 1998 to September 2001. Clinical, functioning, and resource utilization data were collected at baseline and five postbaseline visits. Brief Psychiatric Rating Scale scores defined “clinical effectiveness;” Lehman Quality of Life Scale social relations scores defined “social effectiveness.”

Results:  Requiring failure on less expensive antipsychotics before use of olanzapine did not result in total cost savings, despite significantly higher antipsychotic costs with OLZ. Total 1-year mean costs were $21,283 for CON; $20,891 for OLZ; and $21,347 for RIS (pair-wise comparisons nonsignificant). Intent-to-treat effectiveness comparisons (nonsignificant) were augmented by analyses that adjusted for duration on initial antipsychotic treatment, and by comparisons of patients remaining on initial antipsychotic treatment versus those who required switching. When accounting for differential switching rates (OLZ 0.14 vs. CON 0.53, P < 0.0001; vs. RIS 0.31, P < 0.0001), OLZ was significantly more effective than CON on clinical (P = 0.025) and social (P = 0.043) measures, and significantly more effective than RIS on the social (P = 0.002) measure. Further, patients initiated on an antipsychotic from which they needed to switch required additional resources for hospitalization (P = 0.036) and crisis services (P = 0.029).

Conclusions:  Approaches that integrate costs, effectiveness, and treatment patterns are important for providing optimal information regarding the value of first-line antipsychotic options for schizophrenia.


Introduction

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Published clinical guidelines for a schizophrenia treatment episode include as first-line medication options both conventional and second-generation or atypical antipsychotics [1]. Nevertheless, higher acquisition costs for the atypicals as a group, as well as cost differentials among atypical agents, have led to expenditure control policies within both private and public payer systems. Examples include preferred drug lists, prior authorization requirements, and “fail-first” algorithms (with failure on a less expensive medication required before a more expensive one is covered). Payers and other stakeholders continue to seek information on the overall value of atypical agents, both individually and as a class, as first-line treatment for schizophrenia [2,3].

Most research designed to assess the value of atypical versus conventional antipsychotic treatment has utilized a “cost-minimization” approach. Thus, interest has focused primarily on the extent to which increased expenditures for an atypical agent could be meaningfully offset by reductions in costs for other resources. Indeed, many researchers have shown treatment with atypical agents (clozapine, olanzapine, risperidone) to be associated with significant savings in hospitalization costs [4,5]. A diverse set of studies has demonstrated that through inpatient and/or other service-cost offsets, the total (direct) costs of a treatment episode are essentially equivalent for atypical and conventional antipsychotics [6,7].

Although fewer in number, studies have also compared particular atypical antipsychotic agents with respect to treatment cost outcomes. Economic comparisons of olanzapine and risperidone have been conducted with randomized clinical trial data and with public payer claims data. Both prospective and retrospective analyses have shown the total costs of treatment to be similar for these two atypical antipsychotics, despite higher acquisition costs for olanzapine. As with conventional versus atypical comparisons, higher medication costs for olanzapine were offset by reductions in costs required for other resources [8–12].

The value of conventional versus atypical medications has also been examined through treatment effectiveness studies. Relative to conventional antipsychotics, atypical antipsychotics have been linked to increased medication compliance, decreased need for antipsychotic switching, and improved social functioning and health-related quality of life [13–17]. Less research has directly compared atypical agents (including olanzapine and risperidone) on quality-of-life or other effectiveness outcomes. Additional work is needed to differentiate clearly the atypical agents on effectiveness outcomes [18] and to integrate effectiveness and cost findings.

Although assessments of both costs and effectiveness are important for understanding the overall value of different antipsychotic treatments, studies have rarely been designed to integrate these outcomes. The current clinical effectiveness trial (i.e., practical clinical trial) was conducted to help fill this gap by comparing the cost-effectiveness of specific first-line antipsychotic treatment options [19–21]. The goal was to provide relevant information for clinical practice and payer policy regarding olanzapine as first-line treatment for a schizophrenia treatment episode. The first set of analyses compared the cost-effectiveness of initial (first-line) treatment with olanzapine versus a “fail-first” algorithm requiring failure on conventional antipsychotic treatment before possible treatment with olanzapine. The second set compared the cost-effectiveness of two atypical agents, olanzapine and risperidone, as initial (first-line) treatment.

Methods

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Participants and Treatment Procedures

Participants were recruited within both academic and community treatment settings, primarily in mental-health outpatient clinics. Screening procedures included a physical exam, electrocardiogram, clinical chemistry and hematology labs, and a comprehensive clinical assessment with documentation of psychiatric diagnoses [22] and treatment history. Individuals entered the trial between May 1998 and September 2001. They included men and women at least 18 years old who met DSM-IV criteria for schizophrenia, schizoaffective disorder, or schizophreniform disorder [23], and met a psychotic symptom threshold of 18 or more on the Brief Psychiatric Rating Scale (BPRS) [24] (extracted from the Positive and Negative Syndrome Scale [PANSS]) [25]. Individuals recently experiencing an adverse event attributable to current antipsychotic treatment (unless olanzapine or risperidone) were also eligible, although the vast majority met symptom criteria. Patients with very serious, unstable physical illnesses and other medical conditions or histories contraindicating use of any study medication were excluded. Further details are available in a previous publication [26].

Protocol and consent documents were approved by a central institutional review board (IRB) or by local IRBs for a total of 23 sites (among 15 states). Appropriate informed consent procedures were followed, with signed consent obtained prior to participation. Training on data collection and safety-reporting requirements was provided by a large contract research organization (CRO), as was site monitoring. Of the original 23 investigators, 21 contributed to the data set. One site failed to recruit any subjects; data from another site were excluded because of protocol adherence and data quality issues. Participants were randomly assigned to begin the treatment episode with one of three open-label antipsychotic regimens (in oral solid formulations): 1) olanzapine as first-line treatment (OLZ); versus 2) first-line treatment with a maximum of two (consecutive) conventional agents before a possible switch to olanzapine (CON); and versus 3) risperidone as first-line treatment (RIS). Choice of a particular conventional agent was made by the treating physician and was based on an individual's clinical and treatment history. Barring any clinically significant adverse events (AEs), all patients were to remain on their randomized treatment regimen for at least 8 weeks but could continue on their initial regimen for as long as clinically indicated during the 1-year trial period.

Initial dosing, titration, and dosing adjustments were determined by treating physicians, with instructions to consider clinical indications, as well as most current product labeling and package insert recommendations [26]. The protocol included guidelines for initiating patients on risperidone (beginning with 1 mg/twice daily) and on olanzapine (10 mg/day), with explicit caveats that higher or lower initiating doses could be used if clinically indicated. Switching antipsychotic agents was also at the discretion of treating physicians, who were required to document the reason for the switch (e.g., lack of efficacy, tolerability). The simultaneous use of two antipsychotics was restricted to the interval needed for any switch. Most other psychotropic and nonpsychotropic medications could be used concomitantly. The study sponsor paid for antipsychotic and antiparkinsonian agents. Throughout the trial, detailed information was collected on each patient's medication use.

Assessments and Resource Cost Assignments

Baseline data collection occurred on the day of, and prior to, randomization. Demographic and insurance data were collected, as were data related to clinical symptoms, comorbid psychiatric diagnoses, psychiatric history, prior service utilization, and current community/social functioning. Treatment effectiveness was measured in both clinical and social terms. The primary measure of clinical effectiveness was derived from the clinician-rated BPRS. Each of 18 symptoms is rated from 0 (not present) to 6 (extremely severe). In addition to screening (Visit 1) and baseline (Visit 2), BPRS ratings were obtained at each of five subsequent visits. Visits 3, 4, 5, 6, and 7 were scheduled for 2 weeks, and 2, 5, 8, and 12 months postbaseline, respectively.

Social effectiveness was based on the patient-reported “subjective satisfaction with social relationships” subscale of the Lehman Quality of Life Interview (LQLI) [27,28]. This subscale includes three questions, with each rated from 1 (terrible) to 7 (delighted): 1) How do you feel about things you do with others? 2) How do you feel about the amount of time you spend with others? and 3) How do you feel about the people you see socially? Several antipsychotic treatment studies have included LQLI subscales or selected individual items [14,21,29,30]. For the current trial, the LQLI was administered at baseline and at Visits 4, 5, and 7.

Antipsychotic tolerability comparisons included analyses of treatment-emergent AEs, development of extrapyramidal symptoms (EPS), and changes in weight. AEs were detected through clinical evaluation, spontaneous patient report, and lab test results. All AEs were recorded and classified using the US Food and Drug Administration (FDA) Coding Symbols and Thesaurus for Adverse Reaction Terms [31]. Appropriate standard procedures defined by the FDA were followed for tracking and reporting serious adverse events (SAEs). EPS ratings were obtained at baseline and at each subsequent visit. Development of EPS was defined as meeting a specific threshold on either the Simpson–Angus Scale (SAS) [32] or the Barnes Akathisia Rating Scale (BAS) [33]. A total mean SAS score of more than 0.30 defined treatment-associated Parkinsonian-type symptoms, among patients with a baseline score of less than or equal to 0.30. A global BAS score of at least 2 at any postbaseline visit defined treatment-emergent akathisia, for those with a score of less than 2 at baseline [34]. Weight gain patterns were compared by using two thresholds: 7% and 5% increase from baseline [35,36]. By using observations available at each visit, weight gain was also assessed longitudinally (baseline to visit 3; baseline to visit 4, etc.).

The economic analysis was conducted from the perspective of the public payer health-care system and thus included only direct medical costs [37]. For cost-effectiveness analyses, costs were defined as total medical treatment costs. Included were costs of treatment resources considered “mental health” or “psychiatric” (e.g., antipsychotic and other psychotropic medications, psychiatric hospitalizations, outpatient visits to psychiatrists), as well as those considered “nonpsychiatric” resources (e.g., nonpsychotropic medications, hospitalizations for physical illnesses, primary care physician visits, nonprotocol labs).

Using information from patient report, medical records, and administrative databases, site personnel completed resource utilization forms. Units of specific service use for individual patients were documented and coded (but not assigned costs) at each study site between 1998 and 2002. The International Classification of Diseases (ICD-9-CM) was used to code primary and secondary diagnoses [38], with several codes subsequently updated to reflect revisions since the start of trial-data collection. Medicare public data (based on 2001 national average charge and payment schedules) were used as a costing benchmark [21,39,40]. Diagnostic and/or procedure codes were mapped to one of several data sources by senior staff of the CRO's Health Economics Division. Table 1 provides a summary of unit costing sources for each resource component. Included is information on the use of per diem rates (for psychiatric inpatient stays), as well as applicable professional fees. Estimations of (nonmedication) costs for individual procedures and fees were based on charged amounts [41], with all values in 2001 US dollars.

Table 1.  Summary of costing sources and examples of cost assignments for common procedures and antipsychotics
Service/resource componentCosting sourceCommon procedures/antipsychotics
  • *

    Medicare Provider Analysis and Review (MEDPAR) of Short-Stay Hospitals.

  • For psychiatric inpatient, facility per diem rates calculated from DRG charges/payments specific to psychoses. Per diem rates were multiplied by lengths of stay. Laboratories and medications were inclusive.

  • Hospital and Skilled Nursing Facilities Part B Inpatient Services, and Outpatient Fee Schedule.

  • §

    Skilled Nursing Facility Prospective Payment System.

  • ||

    Hospital Outpatient Prospective Payment System (HOPPS) 2001 Public Use File.

  • HOPPS Clinical Lab Fee Schedule.

  • Drug Topics Red Book, Average Wholesale Price discounted by 15%.

Acute inpatient (short-stay hospitals)Facility: MEDPAR (nonpsychosis DRG)*; 485 per diem (psychosis DRG) Professional fees: Part B• Psychiatric diagnosis interview ($187) • Hospital discharge ($98) • Psychiatric treatment with 45– 50 min of evaluation and management ($134)
Longer-term inpatient  a. Nursing facility  b. State hospitala. Facility: SNFPPS§ (non psychosis DRG); 329 per diem (psychosis DRG); Professional fees: Part B b. Facility: 338 per diem 
Partial hospitalFacility: HOPPS||  Professional fees: Part BGroup psychotherapy ($114)
Acute outpatient (emergency room, other crisis services)Facility: HOPPS  Professional fees: Part BEmergency room visit ($150)
Maintenance outpatient
 a. Psychiatrists  b. Other MDs  c. Substance abuse treatment  d. Professionals (non-MD)  e. Laboratories (nonprotocol mandated)For a, b, c, and d: Facility: HOPPS (if applicable)  Professional fees: Part B For e: HOPPS Clinical Lab  Fee Schedule• Medication management ($79) • Group psychotherapy ($53) • Psychiatric treatment office visit 20–30 min ($89) • Comprehensive metabolic panel ($40) • Hemogram ($27) • Lipid panel ($55)
Medications  a. Antipsychotic  b. OtherRed Book AWP–−15%Mean daily antipsychotic costs:  • Olanzapine ($12.31)  • Haloperidol ($0.13)  • Perphenazine ($0.86)  • Risperidone ($9.57)

Estimations of medication costs began with the average wholesale price (AWP) listed for each medication in the Drug Topics Red Book [42]. For costing antipsychotics, 2001 AWP information was available. At the time of concomitant medication costing in this study, however, only 2003 AWP listings were available with the CRO's automated Price-Check system. To increase comparability, 2003 medication prices were deflated by 8% (overall), using the Prescription Drug Consumer Price Index. Finally, a customary 15% discount rate was applied to all (antipsychotic and concomitant) medications to more accurately reflect “real world” costs [14,43,44]. Table 1 provides examples of common procedures with assigned charges, as well as cost calculations for the four most common antipsychotics.

Statistical Analyses

Analyses of variance or chi-square tests were used to assess the baseline comparability of treatment groups, and they included all randomized patients. The postbaseline analytic sample excluded 13 patients who failed to receive a single day of randomized treatment or to provide any postbaseline data. Using the Statistical Analysis Software (SAS) [45], two-tailed α = 0.05 significance level tests (or 95% confidence intervals [CI]) were conducted for all analyses, without adjustments for multiplicity.

The a priori primary cost-effectiveness analyses compared OLZ to CON and RIS, respectively, using both clinical and social effectiveness measures. Incremental cost effectiveness ratios (ICERs) were used to summarize results [46,47]. The numerator for each ICER was the difference between the mean 1-year total direct costs for each treatment. The denominator was the treatment difference in the average number of “responder days.” Primary cost-effectiveness analyses were based on an intent-to-treat (ITT) approach, with outcomes attributed to each patient's randomized treatment regardless of switching.

Assessments of clinical and social effectiveness were based on the number of responder days. Clinical response was defined as a BPRS score less than 18, and social “response” was defined by maintaining a high level of satisfaction with social relationships (for patients reporting a baseline score of at least 18), or by improving at least 33% of possible improvement. To calculate the number of clinical and social responder days for each treatment group, linear interpolation was used to determine point estimates for a BPRS (clinical), and an LQLI (social) score for each day (between actual assessment dates) [48]. Using each point estimate as the mean of the probability distribution and a repeated measures estimate of variation, the cumulative normal distribution was used to calculate the probability that a patient “responded” on a given day. The number of responder days for each patient was defined as the sum of the estimated probabilities for each of 365 days [49]. By utilizing a propensity score stratification approach [50], comparisons of OLZ with CON and OLZ with RIS on ICERs were repeated. The stratification was based on covariates that had been selected a priori and confirmed through a stepwise model, to have a statistically significant effect on economic or responder day outcomes. The covariates were: site, age, gender, substance abuse diagnosis, insurance, duration of psychiatric problems, baseline symptom level, baseline inpatient status, and time in hospital (prior year).

To address the skewed data (both costs and responder days) distributions, bootstrap resampling was used. Bootstrapping is a nonparametric procedure that provides reliable measures of statistical significance (CI and P-values), even when data do not follow a standard parametric distribution [51,52]. A total of 10,000 bootstrap samples were generated, providing P-value estimates with a standard deviation not exceeding 0.005. Percentages of bootstrap replications falling within each quadrant of the cost-effectiveness plane were computed to assess ICER variability. To illustrate, for OLZ versus CON, the quadrants were: 1) OLZ more effective/greater costs; 2) OLZ more effective/lower costs; 3) CON more effective/greater costs; and 4) CON more effective/lower costs. Separate analyses were conducted for clinical and social effectiveness, with each OLZ-CON comparison followed by a corresponding OLZ-RIS comparison.

Given the high and differential rates of switching, three additional approaches were taken to augment the primary ITT cost-effectiveness analyses. First, effectiveness was reexamined, considering responder days only during treatment on initial antipsychotic regimen, and prorating responder day data to 1 year of treatment. Second, marginal structural models were generated that incorporated actual treatment as a time-varying covariate in 1-year effectiveness [53,54]. Finally, the consequences of needing to switch from initial antipsychotic regimen were examined through analyses of service utilization.

Consistent with previous research, a conservative definition of antipsychotic switching was employed for patients randomized to CON. Only when use of a conventional agent was followed by a switch to an atypical agent were CON patients considered to have “switched” antipsychotics [55]. Because a change from one conventional agent to another was not considered an antipsychotic switch, calculated rates of switching within the CON group were somewhat lower (i.e., underestimated) than if the switching definition had included a change from one conventional agent to another.

Time-to-event analyses were used to compare treatment groups on the secondary outcomes of antipsychotic switching, development of EPS, and weight gain (of at least 7%, and then 5% from baseline). Kaplan–Meier estimates and log-rank tests were used to quantify treatment differences in time to switching, with trial discontinuation considered a censoring event. Cox proportional hazard models, with trial discontinuation and antipsychotic switching as censoring events, were utilized to assess treatment differences in development of EPS and weight gain. EPS medication use was included as a covariate in the Cox proportional hazards analysis of EPS, while gender and baseline body mass index (BMI) were covariates in the analysis of weight gain.

Results

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Patient Baseline Characteristics

A total of 664 individuals (91% of those screened) were randomized: 229 to OLZ, 214 to CON, and 221 to RIS. Table 2 provides a summary of baseline demographic and insurance information, as well as primary psychiatric diagnosis, and baseline weight. No significant differences were found for any OLZ versus CON, or OLZ versus RIS comparisons. Schizophrenia was the primary diagnosis for 65%, schizoaffective disorder for 34%, and schizophreniform disorder for less than 1% of the patients. The average age of the randomized sample was 43 years. Most reported their race/ethnicity to be Caucasian (54%) or African American (34%). The majority (68%) reported living in a private home or apartment, but only 19% reported current employment. Insurance payers were primarily Medicaid and/or Medicare, with 31% reporting dual eligibility. Almost one-fifth (19%) reported having no health-care coverage of any kind. Calculations of BMI (height-adjusted measure of body weight) revealed that at baseline, 68% of males were at least overweight (BMI ≥ 25.0), with 34% qualifying for the obese (BMI ≥ 30.0) or very obese (BMI ≥ 40.0) category. Among females, 83% were at least overweight, with 57% being obese or very obese [56]. There were no statistically significant group differences in average weight or average BMI (Table 2).

Table 2.  Baseline demographics, diagnosis, insurance, and body weight information (overall and two-way comparisons by initial [randomized] antipsychotic treatment group)
Variable*OverallCONO vs. C§OLZ||O vs. RRIS
  • *

    Variables are presented as n (%) unless otherwise noted.

  • n = 664 for Age, Sex, Race/ethnicity, Primary psychiatric diagnosis, and Currently employed; n = 660 for Marital status; n = 659 for Education; n = 654 for Body weight; n = 637 for BMI (237 women and 400 men); n = 625 for Primary insurance.

  • n = 214 for Age, Sex, Race/ethnicity, Primary psychiatric diagnosis, and Currently employed; n = 213 for Marital status and Education; n = 211 for Body weight; n = 202 for Primary insurance and for BMI (67 women and 135 men).

  • §

    P-value comparing OLZ and CON treatment groups.

  • ||

    n = 229 for Age, Sex, Race/ethnicity, Primary psychiatric diagnosis, and Currently employed; n = 228 for Marital status and Body weight; n = 227 for Education; n = 222 for BMI (83 women and 139 men); n = 215 for Primary insurance.

  • P-value comparing OLZ and RIS treatment groups.

  • n = 221 for Age, Sex, Race/ethnicity, Primary psychiatric diagnosis, and Currently employed; n = 219 for Marital status and Education; n = 215 for Body weight; n = 213 for BMI (87 women and 126 men); n = 208 for Primary insurance.

  • BMI, body mass index.

Age (years), mean (SD) 42.8 (12.0) 43.6 (12.1)0.4639 42.7 (12.2)0.5668 42.1 (11.8)
Sex  0.4261 0.5625 
 Female (%)244 (37) 71 (33)  85 (37)  88 (40)
 Male (%)420 (63)143 (67) 144 (63) 133 (60)
Race/ethnicity  0.5895 0.7095 
 Caucasian (%)361 (54)112 (52) 127 (55) 122 (55)
 African American (%)224 (34) 77 (36)  72 (31)  75 (34)
 Other (%) 79 (12) 25 (12)  30 (13)  24 (11)
Primary psychiatric diagnosis  0.5881 0.2533 
 Schizophrenia (%)431 (65)141 (66) 154 (67) 136 (62)
 Schizoaffective disorder (%)228 (34) 71 (33)  75 (33)  82 (37)
 Schizophreniform disorder (%) 5 (∼1) 2 (<1)  0  3 (1)
Marital status  0.5343 0.3948 
 Never married (%)389 (59)121 (57) 138 (61) 130 (59)
 Currently married (%) 93 (14) 30 (14)  37 (16)  26 (12)
Education  0.7535 0.5838 
 Did not complete high school (%)207 (31) 66 (31)  70 (31)  71 (32)
 High school diploma (%)241 (37) 72 (34)  85 (37)  84 (38)
 Post-high school education (%)211 (32) 75 (35)  72 (32)  64 (29)
Currently employed (%)128 (19) 44 (21)0.5463 41 (18)0.7172 43 (19)
Primary insurance  0.8210 0.8321 
 Medicaid (%)231 (37) 73 (36)  77 (36)  81 (39)
 Medicare (%)172 (28) 56 (28)  63 (29)  53 (25)
 Private (%) 79 (13) 29 (14)  23 (11)  27 (13)
 Other options (%) 22 (4) 6 (3)  8 (4)  8 (4)
 No insurance (%)121 (19) 38 (19)  44 (20)  39 (19)
Body weight (kg), mean (SD) 86.7 (20.9) 87.3 (19.6)0.4398 85.8 (22.1)0.5209 87.1 (20.9)
BMI, mean (SD)
 Women 32.1 (7.6) 33.6 (8.8)0.1622 31.9 (8.0)0.4802 31.0 (7.4)
 Men 28.2 (5.9) 28.0 (5.5)0.7156 28.3 (6.4)0.5505 28.4 (5.6)

Table 3 presents several aspects of psychiatric history, as well as comorbid psychiatric diagnoses. It also includes mean baseline ratings for PANSS, BPRS, and LQLI subjective social relations. Baseline comparability was also demonstrated for this set of variables. Although the vast majority (95%) began the trial as outpatients, almost one-third had been hospitalized in the past year for “mental or emotional problems.” The most prevalent comorbid psychiatric (lifetime) diagnosis was psychoactive substance use disorder. Antipsychotic treatment in the prior year included only conventional agents for 57% and only atypical agents for 14% of the patients. Overall, 19% had been treated with both classes of antipsychotics over the past year.

Table 3.  Baseline information on psychiatric treatment history, current care setting, and scores on clinical and social relations scales: overall and two-way comparisons by initial (randomized) antipsychotic treatment group
Variable*OverallCONO vs. C§OLZ||O vs. RRIS
  • *

    Variables are presented as n (%) unless otherwise noted.

  • n = 664 for Inpatient care setting, Antipsychotic treatment, PANSS, and BPRS; n = 584 for Age at first psychiatric hospitalization; n = 644 for Previous episodes of schizophrenia; n = 650 for Time spent in hospital; n = 663 for Comorbid psychiatric diagnoses; n = 592 for LQLI satisfaction.

  • n = 214 for Inpatient care setting, Antipsychotic treatment, PANSS, and BPRS; n = 186 for Age at first psychiatric hospitalization; n = 211 for Previous episodes of schizophrenia; n = 210 for Time spent in hospital; n = 213 for Comorbid psychiatric diagnoses; n = 191 for LQLI satisfaction.

  • §

    P-value comparing OLZ and CON treatment groups.

  • ||

    n = 229 for Inpatient care setting, Comorbid psychiatric diagnoses, Antipsychotic treatment, PANSS, and BPRS; n = 200 for Age at first psychiatric hospitalization; n = 222 for Previous episodes of schizophrenia; n = 223 for Time spent in hospital; n = 209 for LQLI satisfaction.

  • P-value comparing OLZ and RIS treatment groups.

  • n = 221 for Inpatient care setting, Comorbid psychiatric diagnoses, Antipsychotic treatment, PANSS, and BPRS; n = 198 for Age at first psychiatric hospitalization; n = 211 for Previous episodes of schizophrenia; n = 217 for Time spent in hospital; n = 206 for LQLI satisfaction.

Age at first psychiatric hospitalization (year) mean (SD) 26.2 (9.5) 26.0 (9.2)0.3888 26.9 (10.0)0.1924 25.6 (9.3)
Number of previous episodes of schizophrenia, mean (SD) Time in hospital (past year) for “mental/emotional problems” 6.8 (9.6) 6.9 (10.3)0.7022 7.4 (9.9)0.3298 6.1 (8.5)
 Days, mean (SD) 9.1 (34.1) 8.0 (36.6)0.9127 7.7 (21.3)0.2305 11.6 (41.6)
 None (%)447 (69)145 (69)0.6276150 (67)0.5434152 (70)
 Less than 1 month (%)125 (19) 41 (19)  46 (21)  38 (18)
 1 month or more (%) 78 (12) 24 (11)  27 (12)  27 (12)
Inpatient setting at trial entry (%) 31 (5) 7 (3)0.3541 12 (5)1.0000 12 (5)
Antipsychotic treatment (past year)
 Conventional(s) only (%)377 (57)124 (58)0.5031125 (55)0.5064128 (58)
 Atypical(s) only (%) 95 (14) 25 (12)0.1732 38 (17)0.6033 32 (14)
 Both (%)128 (19) 36 (17)0.2299 49 (21)0.6412 43 (19)
Threshold for comorbid psychiatric diagnoses (lifetime)
 Mood disorder (%)134 (20) 40 (19)0.4104 51 (22)0.4880 43 (19)
 Anxiety disorder (%) 35 (5) 9 (4)0.5191 13 (6)1.0000 13 (6)
 Psychoactive substance use disorder (%)296 (45) 93 (43)0.3903109 (48)0.2969 94 (43)
PANSS total score, mean (SD) 86.9 (19.8) 86.0 (19.5)0.5132 87.2 (19.4)0.9052 87.5 (20.6)
BPRS score, mean (SD) 31.8 (11.5) 31.2 (11.1)0.6002 31.8 (11.3)0.5784 32.4 (12.1)
LQLI satisfaction with social relations subscale, mean (SD) 13.9 (3.7) 13.9 (3.5)0.5849 14.1 (3.7)0.4761 13.8 (3.8)

The groups were also comparable at baseline on prior resource use. Overall, 69% reported at least one outpatient psychiatrist visit in the past 3 months and 28% reported at least one social worker or case manager visit. Twenty-nine percent reported at least one emergency room (ER) visit within the past 3 months. Of more than 250 “preexisting” conditions reported at study entry, four were reported by at least 10% of the baseline sample: depression (15%); hallucinations (10%); hypertension (21%); and insomnia (12%). Of eight two-way comparisons for these conditions, one difference was statistically significant. Insomnia was reported by 16% of those subsequently randomized to OLZ, compared with 8% of those randomized to RIS (P = 0.0059).

Treatment Patterns

For those randomized to CON, the most frequently prescribed agents were perphenazine (23%), loxapine (17%), haloperidol (16%), thiothixene (16%), and fluphenazine (10%). Mean-modal doses of the primary antipsychotics were: olanzapine (13.49 mg/day; SD 8.03); risperidone (4.95 mg/day; SD 2.67); haloperidol (10.98 mg/day; SD 9.45); and perphenazine (14.21 mg/day; SD 10.60). Percentages prescribed anticholinergics (predominantly benztropine) were: olanzapine (41%); conventional antipsychotics (53%); and risperidone (48%). Other common medications were valproic acid, lorazepam, fluoxetine, and trazodone.

Figure 1 provides enrollment, treatment allocation, analytic sample, and disposition information. The trial completion rate for CON was 73.8%, but more than half (57.4%) of these 155 completers had switched at some point to an atypical antipsychotic. Only 11% of patients in the CON group switched from their initial conventional agent to another conventional agent. The OLZ trial completion rate was 66.8%, with 16.1% of completers having switched. The RIS completion rate was 69.3% with one-third (32.5%) of the 151 having switched before completing the trial. As shown in Figure 2, rates of initial antipsychotic switching were significantly lower for OLZ than for CON (P < 0.001) and RIS (P < 0.001). Compared to the CON group, rates of continuing treatment without a switch for OLZ were about 8% greater by 8 weeks and 34% greater by 6 months. Compared with RIS, OLZ rates were 9% greater at 8 weeks and 17% greater at 6 months.

image

Figure 1. Patient disposition by initial antipsychotic treatment group.

Download figure to PowerPoint

image

Figure 2. Time-to-event group comparison. Event = switching from initial antipsychotic treatment.

Download figure to PowerPoint

Tolerability Outcomes

Percentages of patients experiencing an SAE of any kind were not statistically significantly different (CON 17.3%, OLZ 19.7%, RIS 19.0%). Among the randomized patients, there were nine deaths at some point during the trial and one death within a few weeks of trial completion (10 of 664 [1.5%]). Three of these 10 individuals had been randomized to OLZ, three to RIS, and four to CON, although one of the four never began antipsychotic treatment and was not included in the analytic sample. None of the deaths were, in the opinion of investigators, related to study medications or to protocol procedures.

Among those not meeting baseline criteria for existing EPS (CON: n = 142, OLZ: n = 151, RIS: n = 149), the probability (over 1 year) of not developing EPS was significantly greater for OLZ compared with CON (P = 0.0060). The advantage of OLZ over RIS approached but did not reach statistical significance (P = 0.0708). Time to 7% weight gain was significantly shorter for OLZ compared with CON (P < 0.0001) and for OLZ compared with RIS (P = 0.0226). With 5% weight increase as the “event” threshold, results were similar, but the OLZ-RIS difference did not reach statistical significance (P = 0.0777). Among patients still on their initial antipsychotic regimen, increase from baseline weight was significantly greater for OLZ compared with CON at each visit. Weight gain was significantly greater for OLZ compared with RIS at 8 weeks and 12 months, and the difference approached significance at 8 months (P = 0.0543). On average, CON patients (n = 67) gained 2.43 kg, compared with 6.00 kg for OLZ (n = 125) and 3.19 kg for RIS (n = 99).

Patterns of Service Use

The average 1-year total cost per patient during the trial was $21,170. One of the largest contributors to total treatment costs (27.8%) was short-stay inpatient hospitalization. Overall, 21% of patients were hospitalized at least once during the postbaseline study period (ITT group percentages were CON: 21%; OLZ: 20%; RIS: 22%). The large majority of hospitalizations (83%) were linked to psychiatric diagnosis-related groups (DRGs). ITT analyses showed that 14% in the CON group, 15% in the OLZ group, and 17% in the RIS group had at least one acute psychiatric admission. Among “nonpsychiatric” admissions, the most common DRGs corresponded to 1) diseases of the circulatory system (in 11 out of 53); 2) injury or poisoning (in 8 out of 53); and 3) the respiratory system (in 7 out of 53). Partial hospitalization also accounted for a relatively large proportion of treatment costs. Of the sample, 24% utilized this resource (ITT percentages were CON: 29%; OLZ: 22%; RIS: 22%). Medications were responsible for 22.9% of total costs, with antipsychotics accounting for 13.2% (ITT proportions: CON, 0.08; OLZ, 0.18; RIS, 0.14).

Only 1% of the patients had a postbaseline admission to a long-term care facility, contributing to 9.2% of total costs. Fourteen percent used acute outpatient services at least once, but this component represented less than 1% of total costs. Outpatient maintenance (nonacute) services represented 12% of treatment costs. Half of these costs were linked to physician visits, with 74% of trial participants having at least one outpatient visit to a psychiatrist or other physician. Outpatient (nonprotocol) labs accounted for less than 1% of costs. Based on estimates that 74% of acute hospitalizations, 92% of longer-term inpatient stays, 36% of ER visits, and 73% of outpatient physician costs (charges) could be classified as “psychiatric” or “mental health,” the proportion of total costs was calculated to be within the range of 79.9% to 89.5%. This range is dependent upon the precise proportion of concomitant medications used for psychotropic purposes, which could not be accurately estimated. Average antipsychotic costs for OLZ ($3756) were significantly greater compared with CON ($1652) (P < 0.001) and RIS ($2907) (P < 0.001). Mean nonmedication costs (charges) were $15,165 for OLZ; $17,565 for CON; and $16,343 for RIS (differences nonsignificant).

ITT Cost-Effectiveness Analyses

Total 1-year mean costs were $21,283 for CON, $20,891 for OLZ, and $21,347 for RIS (with pair-wise comparisons statistically nonsignificant). Table 4 compares OLZ and CON on average 1-year per patient costs, mean clinical (BPRS) responder days, and mean social relations (LQLI) responder days. OLZ was associated with 9.7 more clinical responder days and 6.7 more social relations responder days, but 95% CI were wide and mean differences were not statistically significant. Differences in average ITT total costs were minimal and also statistically nonsignificant. For each ITT comparison shown in Table 4, P-values are provided for both “adjusted” and “unadjusted” analyses.

Table 4.  Cost and effectiveness comparisons: estimated total mean 1-year costs (US dollars) and responder days for clinical and social effectiveness
   DifferenceP-value (adj.)P-value (unadj.)§
  • *

    n = 223 for mean total costs.

  • n = 210 for mean total costs.

  • Bootstrap P-value adjusted for site, age, gender, substance use diagnosis, insurance, duration of psychiatric problems, baseline symptom level, baseline inpatient status, hospital time (prior year), by use of propensity score stratification.

  • §

    Bootstrap P-value without covariate adjustment.

  • ||

    n = 218 for mean total costs.

  • Note: All means estimated by propensity score adjusted (see covariate list in footnote) bootstrap resampling.

  • CI, confidence interval; ITT, intent-to-treat; NS, nonsignificant.

OLZ vs. CONOLZ*CON   
 ITT mean total costs20, 97221, 049−77NSNS
 (95% CI)(17, 431; 24, 793)(17, 121; 25, 496)(−5, 786; 5, 365)0.9570.890
 ITT mean clinical responder days131.8122.09.7NSNS
 (95% CI)(117; 147)(107; 137)(−11; 30)0.3920.638
 ITT mean social responder days106.199.46.7NSNS
 (95% CI)(93; 120)(86; 113)(−12; 25)0.4710.511
 Mean clinical responder days—prorated125.6101.823.80.025
 (95% CI)(111; 140)(87; 116)(3; 45)  
 Mean social responder days—prorated100.482.517.90.043
 (95% CI)(88, 113)(71; 95)(1, 36)  
OLZ vs. RISOLZ*RIS||   
 ITT mean total costs21, 15321, 644−491NSNS
 (95% CI)(17, 419; 25, 306)(17, 731; 25, 788)(−5, 976; 5, 077)0.8620.860
 ITT mean clinical responder days129.0127.71.3NSNS
 (95% CI)(114; 144)(113; 143)(−20; 23)0.8680.993
 ITT mean social responder days105.596.59.0NSNS
 (95% CI)(91; 119)(84; 110)(−10, 28)0.3050.498
 Mean clinical responder days—prorated122.5108.713.9NS
 (95% CI)(108; 137)(94; 124)(−7; 35)0.197 
 Mean social responder days—prorated99.773.026.70.002
 (95% CI)(87; 112)(62; 84)(10; 43)  

Because the estimate showed OLZ to be associated with a mean cost saving and with greater effectiveness (although not statistically significant), the specific ICER point estimate value was not considered meaningful. The quadrant analysis showed that for 82% of bootstrap samples, OLZ had greater clinical effectiveness than CON. This greater effectiveness was associated with higher costs for 40% and lower costs for 42%. For social effectiveness, OLZ was superior to CON for 76% of samples. For this analysis, greater effectiveness was associated with higher costs for 36% and lower costs for 40% of bootstrap samples. Overall, OLZ and CON were each associated with greater costs about half the time (48% for OLZ and 52% for CON).

Table 4 also provides an ITT comparison of OLZ and RIS on average total costs, mean clinical responder days, and mean social relations responder days. The small advantage for OLZ in total costs was again statistically nonsignificant. BPRS responder day means were almost identical. On average, OLZ was associated with nine additional social relations responder days, but this advantage over RIS was not statistically significant. For each ITT comparison, P-values are again provided for both “adjusted” and “unadjusted” analyses. For the reason described earlier, the specific ICER value is not presented for the OLZ-RIS comparison. With quadrant analyses, greater clinical effectiveness was found with OLZ in 55% and with RIS in 45% of samples. OLZ was more effective in social relations for 83% of bootstrap samples and RIS more effective for 17%. ITT cost differences for OLZ versus RIS were relatively minimal, with 43% of samples showing greater costs for OLZ and 57% greater costs for RIS.

Effectiveness Prorated—Based on Initial Treatment Duration: Accounting for Treatment Failure

More than half of the clinical (52%) and the social responder days (53%) attributed to CON were days patients were actually on an atypical antipsychotic (olanzapine or risperidone). For OLZ, 22% of the clinical and 24% of the social responder days were days patients were on an antipsychotic other than olanzapine. Finally, 39% of the clinical responder days and 48% of the social responder days attributed to RIS were days patients were on an antipsychotic agent other than risperidone. Also reported in Table 4 are results of analyses using responder days while on initial antipsychotic, prorated for 1 year of treatment. OLZ was associated with significantly more responder days compared with CON for both the clinical (P = 0.025) and social (P = 0.043) effectiveness measures. The OLZ advantage over CON more than doubled for clinical effectiveness (increasing from 9.7 to 23.8 more responder days), and social effectiveness (increasing from 6.7 to 17.9 more days). The clinical effectiveness advantage of OLZ over RIS increased from 1.3 to 13.9 responder days, but the difference was not statistically significant. In the social relations domain, OLZ nearly tripled its advantage over RIS (from 9 to 26.7 more responder days), with the difference showing statistical significance (P = 0.002). The results of these prorated analyses were corroborated by results of a separate marginal structural model analysis, incorporating antipsychotic treatment as a time-varying covariate [54].

Impact of Treatment Failure (Needing to Switch from Initial Antipsychotic)

Compared with patients able to remain on their initial antipsychotic regimen (n = 430), those who required a switch (n = 182) had treatment episodes with $3546 greater total (ITT) costs (nonsignificant). Costs (charges) were numerically higher for each nonmedication resource component (inpatient acute, inpatient long-term care, partial hospitalization, nonacute outpatient, and acute outpatient services). Differences were statistically significant with two components: acute inpatient care (P = 0.036) and acute outpatient services such as ER visits (P = 0.029). For patients starting a treatment episode on an antipsychotic from which they had to switch, an average of $195 was saved in yearly medication costs. These modest savings were accompanied, however, by an additional $3741 needed for other, nonmedication resources (with more than 70% accounted for by acute/intensive services, i.e., hospitalization, partial hospitalization, crisis intervention, ER).

Conclusions

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

This randomized, open-label trial was designed to help inform practices and policies regarding first-line antipsychotic options for a schizophrenia treatment episode. One set of analyses compared the cost-effectiveness of initial (first-line) treatment with olanzapine versus a “fail-first” algorithm requiring failure on conventional antipsychotic treatment before the availability of olanzapine. The second set compared the cost-effectiveness of two atypical agents, olanzapine and risperidone, as initial (first-line) treatment. Participants were representative of individuals with schizophrenia or schizoaffective disorder currently seen in many treatment settings, presenting with a variety of physical and psychiatric comorbidities [57, 58].

The trial demonstrated that a treatment algorithm requiring patients to first fail on less expensive, conventional antipsychotic therapy before access to the more expensive one (olanzapine) did not result in total treatment-cost savings. Despite significantly higher antipsychotic costs associated with OLZ, the OLZ-CON difference in total 1-year direct medical costs was very small and statistically nonsignificant. These findings are consistent with those of other recent studies comparing economic (and other) outcomes with first- versus second-generation antipsychotics [7,11]. Similarly, current comparisons of olanzapine and risperidone as first-line treatment for schizophrenia corroborated previous findings regarding their total cost equivalence [59]. Compared with either CON or RIS, significantly higher antipsychotic costs for OLZ were offset by the combined (nonsignificant) impact of lower costs (charges) for several other services, including acute-care hospitalization and partial hospitalization.

Effectiveness differences in this study were not significant with the use of an ITT approach that did not account for the significantly lower risk of antipsychotic treatment switching with OLZ [7,10,60]. Nevertheless, analyses that incorporated patterns of initial treatment failure showed OLZ to have significantly greater effectiveness than CON for both clinical symptoms and social relations and significantly greater effectiveness than RIS for social relations. These differences in effectiveness are consistent with the findings of other researchers [34,61] and support the potential for social and other functional outcomes to be instrumental in differentiating the value of antipsychotic treatment options [62,63].

Interpretations of current cost and effectiveness results were enhanced by additional findings regarding the negative clinical and economic consequences associated with having initiated patients on an antipsychotic regimen from which they needed to switch [55,64,65]. Resources most associated with initial antipsychotic switching were acute and/or intensive treatment services—services that routinely account for the largest proportion of nonmedication costs in schizophrenia treatment [66]. Antipsychotic treatment failure (defined by needing to switch from initial treatment regimen) was associated with very modest savings in medication costs. Increased expenditures for other resources, however, overshadowed the medication reductions by a factor of 19. Economic consequences of a similar magnitude were found in a recent study that examined the impacts of a policy restricting the availability of psychotropic medications [67].

The ideal strategy, of course, would be to begin a schizophrenia treatment episode with the antipsychotic that is most likely to lead to effective and persistent treatment. Because the best initial treatment choice for any single patient will depend upon his or her particular treatment needs, illness profile, prior medication use, preferences, and so on, the rationale for open formulary access (i.e., not restricting options for initial antipsychotic treatment) appears to be clear. Many stakeholders would agree that if a particular “fail-first” policy results in significant decreases in antipsychotic expenditures and in total treatment cost neutrality, it would also be important to consider the potential economic, clinical, and humanistic impacts of initial treatment failure. The examination of medication patterns, as well as analyses of specific types of service utilization can substantially enhance interpretations of total costs associated with available antipsychotic treatments [68].

Pharmacoeconomic studies vary greatly in terms of design, methodological assumptions, and data analytic approaches. This variation, as well as the lack of a widely accepted proxy for defining “actual costs,” present many challenges in interpreting and comparing findings across studies [69,70]. In addition, economic analyses based on prospective study data are almost always faced with suboptimal statistical power. A sample consisting of several hundred patients is often required to show statistical significance, even when the magnitude of group differences in costs is large (e.g., $5000–$10,000). Nevertheless, the current comparability of total costs has been demonstrated in previous work including retrospective studies with very large samples. Thus, current results regarding total cost comparability cannot be regarded as simply an artifact of low power.

Cost-effectiveness research is likely to contribute greatly to policy decisions around the implementation of the Medicare Modernization Act and its drug benefit programs [71]. This study has illustrated that comprehensive assessments will necessarily incorporate the “real world” complexities of medication switching. It provides clinicians, policymakers, and other stakeholders with an integrated set of findings that can assist in efforts to understand the value of different antipsychotics as first-line options for the treatment of schizophrenia.

Principal investigators contributing data in this multicenter trial were: Denis Mee-Lee, MD, Honolulu, HI; Michael Brody, MD, Washington DC; Christopher Kelsey, MD and Gregory Bishop, MD, San Diego, CA; Lauren Marangell, MD, Houston, TX; Frances Frankenburg, MD, Belmont, MA; Roger Sommi, Pharm.D., Kansas City, MO; Ralph Aquila, MD and Peter Weiden, MD, New York, NY; Dennis Dyck, PhD, Spokane, WA; Rohan Ganguli, MD, Pittsburgh, PA; Rakesh Ranjan, MD; Nagui Achamallah, MD and Bruce Anderson, MD, Vallejo, CA; Terry Bellnier, R.Ph., Rochester, NY; John S. Carman, MD, Smyrna, GA; Andrew J. Cutler, MD, Winter Park, FL; Hisham Hafez, MD, Nashua, NH; Raymond Johnson, MD, Ft. Myers, FL; Ronald Landbloom, MD, St. Paul, MN; Theo Manschreck, MD, Fall River, MA; Edmond Pi, MD, Los Angeles, CA; Michael Stevens, MD, Salt Lake City, UT; Richard Josiassen, PhD, Norristown, PA. Portions of this research have been presented at the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) 8th Annual International Meetings, Arlington, VA, May, 2004; and the Institute on Psychiatric Services (IPS), Atlanta, GA, October, 2004.

Financial support: Eli Lilly and Company.

References

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References
  • 1
    American Psychiatric Association. Practice Guideline for the Treatment of Patients with Schizophrenia. Arlington, VA: American Psychiatric Association, 2004.
  • 2
    Thieda P, Beard S, Richter A, Kane J. An economic review of compliance with medication therapy in the treatment of schizophrenia. Psychiatr Serv 2003;54: 50816.
  • 3
    Soumerai SB. Benefits and risks of increasing restrictions on access to costly drugs in Medicaid. Health Aff 2004;23: 13546.
  • 4
    Essock SM, Frisman LK, Covell NH, Hargreaves WA. Cost-effectiveness of clozapine compared with conventional antipsychotic medication for patients in state hospitals. Arch Gen Psychiatry 2000;57: 98794.
  • 5
    Rosenheck R, Cramer J, Xu W, et al. A comparison of clozapine and haloperidol in hospitalized patients with refractory schizophrenia. N Engl J Med 1997; 337: 80915.
  • 6
    Hudson TJ, Sullivan G, Feng W, et al. Economic evaluations of novel antipsychotic medications: a literature review. Schizophr Res 2003;60: 199218.
  • 7
    Rosenheck R, Perlick D, Bingham S, et al. Effectiveness and cost of olanzapine and haloperidol in the treatment of schizophrenia: a randomized controlled trial. JAMA 2003;290: 2693702.
  • 8
    Liu GG, Sun SX, Christensen DB, Luo X. Cost comparisons of olanzapine and risperidone in treating schizophrenia. Ann Pharmacother 2004;38: 13441.
  • 9
    Edgell ET, Andersen SW, Johnstone BM, et al. Olanzapine versus risperidone. A prospective comparison of clinical and economic outcomes in schizophrenia. Pharmacoeconomics 2000;18: 56779.
  • 10
    Rascati KL, Johnsrud MT, Crismon ML, et al. Olanzapine versus risperidone in the treatment of schizophrenia: a comparison of costs among Texas Medicaid recipients. Pharmacoeconomics 2003;21: 68397.
  • 11
    Gibson PJ, Damler R, Jackson EA, et al. The impact of olanzapine, risperidone, or haloperidol on the cost of schizophrenia care in a Medicaid population. Value Health 2004;7: 2235.
  • 12
    Mauskopf JA, David K, Grainger DL, Gibson PJ. Annual health outcomes and treatment costs for schizophrenia populations. J Clin Psychiatry 1999;60: 1422.
  • 13
    Tunis SL, Johnstone BM, Loosbrock DL, et al. Changes in perceived health and functioning as a cost-effectiveness measure for olanzapine versus haloperidol treatment of schizophrenia. J Clin Psychiatry 1999;60: 3845.
  • 14
    Rosenheck R, Doyle J, Leslie D, Fontana A. Changing environments and alternative perspectives in evaluating the cost-effectiveness of new antipsychotic drugs. Schizophr Bull 2003;29: 8193.
  • 15
    Glick ID, Berg PH. Time to study discontinuation, relapse, and compliance with atypical or conventional antipsychotics in schizophrenia and related disorders. Int Clin Psychopharmacol 2002;17: 658.
  • 16
    Menzin J, Boulanger L, Friedman M, et al. Treatment adherence associated with conventional and atypical antipsychotics in a large state Medicaid program. Psychiatr Serv 2003;54: 71923.
  • 17
    Hugenholtz GW, Heerdink ER, Nolen WA, Egberts AC. Less medication switching after initial start with atypical antipsychotics. Eur Neuropsychopharmacol 2004;14: 15.
  • 18
    Freedman R. Schizophrenia. N Engl J Med 2003;349: 173849.
  • 19
    Schwartz D, Lellouch J. Explanatory and pragmatic attitudes in therapeutic trials. J Chronic Dis 1967; 20: 63748.
  • 20
    Tunis SR, Stryer DB, Clancy CM. Practical Clinical Trials. Increasing the value of clinical research for decision making in clinical and health policy. Special communication. JAMA 2003;290: 162432.
  • 21
    Stroup TS, McEvoy JP, Swartz MS, et al. The National Institute of Mental Health Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) project: schizophrenia trial design and protocol development. Schizophr Bull 2003;29: 1531.
  • 22
    First M, Spitzer R, Gibbon M, Williams J. Structured Clinical Interview for DSM-IV Axis I Disorders-Administration Booklet. Washington: American Psychiatric Press, 1994.
  • 23
    American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders (4th ed.). Washington: American Psychiatric Association, 1994.
  • 24
    Overall JE, Gorham DR. The brief psychiatric rating scale. Psychol Rep 1962;10: 799812.
  • 25
    Kay SR, Fiszbein A, Opler LA. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr Bull 1987;13: 26176.
  • 26
    Tunis SL, Johnstone BM, Kinon BJ, et al. Designing naturalistic prospective studies of economic and effectiveness outcomes associated with novel antipsychotic therapies. Value Health 2000;3: 23242.
  • 27
    Lehman AF. A quality of life interview for the chronically mentally ill. Eval Program Plann 1988; 11: 62.
  • 28
    Lehman AF. Evaluating outcomes of treatments for persons with psychotic disorders. J Clin Psychiatry 1996;57: 617.
  • 29
    Cook PE, Goldberg JO, Van Lieshout RJ. Benefits of switching from typical to atypical antipsychotic medications: a longitudinal study in a community-based setting. Can J Psychiatry 2002;47: 8704.
  • 30
    Cramer JA, Rosenheck R, Xu W. Quality of life in schizophrenia: a comparison of instruments. Schizophr Bull 2000;26: 65966.
  • 31
    US Food and Drug Administration. COSTART. Rockville, MD: US FDA, 1990.
  • 32
    Simpson GM, Angus JWS. A rating scale for extrapyramidal side effects. Acta Psychiatr Scand 1970; 212(Suppl.):S1119.
    Direct Link:
  • 33
    Barnes TR. A rating scale for drug-induced akathisia. Br J Psychiatry 1989;154: 6726.
  • 34
    Tran PV, Hamilton SH, Kuntz AJ, et al. Double-blind comparison of olanzapine vs. risperidone in the treatment of schizophrenia and other psychotic disorders. J Clin Psychopharmacol 1997;17: 40718.
  • 35
    Conley RR, Mahmoud R. A randomized double-blind study of risperidone and olanzapine in the treatment of schizophrenia or schizoaffective disorder. Am J Psychiatry 2001;158: 76574.
  • 36
    American Diabetes Association, American Psychiatric Association, American Association of Clinical Endocrinologists, et al. Consensus development conference on antipsychotic drugs and obesity and diabetes. Diabetes Care 2004;27: 596601.
  • 37
    Revicki DA, Paramore LC, Sommerville KW, et al. Divalproex sodium versus olanzapine in the treatment of acute mania in bipolar disorder: health-related quality of life and medical cost outcomes. J Clin Psychiatry 2003;64: 28894.
  • 38
    Hart A, Hopkins C, Ford B, eds. International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). Salt Lake City, UT: Medicode, 2001.
  • 39
    Centers for Medicare and Medicaid Services (CMS). Available from: http://www.cms.hhs.gov/[Accessed December 17, 2004.].
  • 40
    Dixon L, Lyles A, Smith C, et al. Use and costs of ambulatory care services among Medicare enrollees with schizophrenia. Psychiatr Serv 2001;52: 78692.
  • 41
    Gianfrancesco F, Durkin MB, Mahmoud R, Wang RH. Use of healthcare services by patients treated with risperidone versus conventional antipsychotic agents. Pharmacoeconomics 2002;20: 41327.
  • 42
    Medical Economics Staff. Drug Topics Red Book: Pharmacy's fundamental reference. Montvale, NJ: Medical Economic Data Production Company, Inc., 2001.
  • 43
    Academy of Managed Care Pharmacy. A Guide to Understanding Common Prescription Drug Pricing Terms. Alexandria, VA: Academy of Managed Care Pharmacy, 2002.
  • 44
    Medicaid Drug Rebate Program. Available from: http://www.cms.hhs.gov/medicaid/drugs/drughmpg.asp[Accessed December 13, 2004.].
  • 45
    SAS Institute. SAS/STAT User's Guide. Cary, NC: SAS Institute, 1999.
  • 46
    Polsky D, Glick HA, Wilke R, Schulman K. Confidence intervals for cost-effectiveness ratios: a comparison of four methods. Health Econ 1997;6: 24352.
  • 47
    Obenchain RL. Resampling and multiplicity in cost-effectiveness inference. J Biopharm Stat 1999;9: 56382.
  • 48
    Obenchain RL, Johnstone BM. Mixed model imputation of cost data for early discontinuers from a randomized clinical trial. Drug Inf J 1999;33: 191209.
  • 49
    Nyhuis A, Stensland M, Faries D. Calculating Responder Days for Cost-Effectiveness Studies. Colorado Springs, CO: International Congress on Schizophrenia Research, 2003.
  • 50
    Newgard CD, Hedges JR, Mullins AM. Advanced statistics: the propensity score – a method for estimating treatment effect in observational research. Acad Emerg Med 2004;11: 95361.
  • 51
    Diehr P, Yanez D, Ash A, et al. Methods for analyzing health care utilization and costs. Ann Rev Public Health 1999;20: 12544.
  • 52
    Good P. Permutation Tests: A Practical Guide to Resampling Methods for Testing Hypotheses. New York: Springer-Verlag, 2000.
  • 53
    Hernan MA, Brumback B, Robins JM. Marginal structural models to estimate the causal effect of zidovudine on the survival of HIV-positive men. Epidemiology 2000;11: 56170.
  • 54
    Faries D, Nyhuis A, Tunis SL, Hernan M. Assessing effectiveness in the presence of treatment switching: data from an effectiveness study of antipsychotics. Alexandria, VA: ISPOR Annual Meeting, 2004.
  • 55
    Mahmoud R, Engelhart L, Ollendorf D. The risperidone outcomes study of effectiveness (ROSE): a model for evaluating treatment strategies in typical psychiatric practice. J Clin Psychiatry 1999;60: 428.
  • 56
    Hedley AA, Ogden CL, Johnson CL, et al. Prevalence of overweight and obesity among US children, adolescents, and adults, 1999–2002. JAMA 2004; 291: 284750.
  • 57
    Regier DA, Farmer ME, Rae DS, et al. Comorbidity of mental disorders with alcohol and other drug abuse. Results from the Epidemiologic Catchment Area (ECA) Study. JAMA 1990;264: 251118.
  • 58
    Margolese HC, Malchy L, Negrete JC, et al. Drug and alcohol use among patients with schizophrenia and related psychosis: levels and consequences. Schizophr Res 2004;67: 15766.
  • 59
    Hargreaves WA, Gibson PJ. Effectiveness and cost of risperidone and olanzapine for schizophrenia: a systematic review. CNS Drugs 2005;19: 393410.
  • 60
    Dossenbach M, Erol A, El Mahfoud Kessaci M, et al. Effectiveness of antipsychotic treatments for schizophrenia: interim 6-month analysis from a prospective observational study (IC-SOHO) comparing olanzapine, quetiapine, risperidone, and haloperidol. J Clin Psychiatry 2004;65: 31221.
  • 61
    Davis JM, Chen N, Glick ID. A meta-analysis of the efficacy of second-generation antipsychotics. Arch Gen Psychiatry 2003;60: 55364.
  • 62
    Davis JM, Chen N. The effects of olanzapine on the 5 dimensions of schizophrenia derived by factor analysis: combined results of the North American and International Trials. J Clin Psychiatry 2001;62: 75771.
  • 63
    Ritsner M, Modai I, Endicott J. Differences in quality of life domains and psychopathologic and psychosocial factors in psychiatric patients. J Clin Psychiatry 2000;61: 8809.
  • 64
    McCombs JS, Nichol MB, Stimmel GL. Use patterns for antipsychotic medications in Medicaid patients with schizophrenia. J Clin Psychiatry 1999;60: 511.
  • 65
    Weiden PJ, Kozma C, Grogg A, Locklear J. Partial compliance and risk of rehospitalization among California Medicaid patients with schizophrenia. Psychiatr Serv 2004;55: 88691.
  • 66
    Weiden PJ, Olfson M. Cost of relapse in schizophrenia. Schizoph Bull 1995;21: 41929.
  • 67
    Soumerai S, McLaughlin T, Ross-Denan D, et al. Effects of a limit on Medicaid drug-reimbursement benefits on the use of psychotropic agents and acute mental health services by patients with schizophrenia. N Engl J Med 1994;331: 6505.
  • 68
    Tunis SL, Ascher-Svanum H, Stensland M, Kinon BJ. Assessing the value of antipsychotics for treating schizophrenia: the importance of evaluating and interpreting the clinical significance of individual service costs. Pharmacoeconomics 2004;22: 18.
  • 69
    Wolff N, Helminiak TW, Tebes JK. Getting the cost right in cost-effectiveness analyses. Am J Psychiatry 1997;154: 73643.
  • 70
    Russell LB, Gold MR, Siegel JE, et al. The role of cost-effectiveness analysis in health and medicine. Panel on cost-effectiveness in health and medicine. JAMA 1996;276: 11727.
  • 71
    Neumann PJ. Why don’t Americans use cost-effectiveness analysis? Am J Manag Care 2004;10: 30812.