Predictors of continuation with olanzapine during the 1-year naturalistic treatment of patients with schizophrenia in Japan

Abstract

Purpose

Treatment continuation is considered an important measure of antipsychotic effectiveness in schizophrenia, reflecting the medication’s efficacy, safety, and tolerability from both patients’ and clinicians’ perspectives. This study identified characteristics of patients with schizophrenia who continue olanzapine therapy for a 1-year period in Japan.

Methods

In a large (N = 1850), prospective, observational study, Japanese patients with schizophrenia who initiated treatment with olanzapine were followed for 1 year. Baseline characteristics were compared using t-tests and chi-square tests. Stepwise logistic regression was used to identify independent baseline predictors of treatment continuation.

Results

Most patients (68.2%) continued with olanzapine therapy for the full 1-year study period, with an average duration of 265.5 ± 119.4 days. At baseline, patients who continued were significantly more likely to be male, older, and inpatients; have longer illness duration, higher negative and cognitive symptoms, better health-related quality of life, and prior anticholinergic use. Continuers were significantly less likely to engage in social activities, live independently, work for pay, or have prior antidepressant use. Continuers showed significantly greater early (3-month) improvement in global symptom severity. Logistic regression found that continuation was significantly predicted by longer illness duration, lower positive symptoms, higher negative symptoms, and better health-related quality of life.

Conclusions

In this large naturalistic study in Japan, most patients with schizophrenia stayed on olanzapine therapy for the full 1-year study period. Treatment completion with olanzapine was independently predicted by longer illness duration, lower positive symptoms, higher negative symptoms, and better health-related quality of life.

Keywords:

• schizophrenia
• atypical
• antipsychotics
• discontinuation

Introduction

Schizophrenia is a chronic and disabling mental illness that is associated with cognitive, behavioral, social, and occupational impairments.¹ The primary recommended treatment for the acute symptoms of schizophrenia is antipsychotic medication.^2,3 When assessing outcomes from treatment with antipsychotic medication, a variety of domains have been studied, including the core symptoms of schizophrenia,^4–6 functional outcomes,^7,8 cognition,^9,10 and health care resource use.^11,12 However, a simple measure of treatment effectiveness that is relevant to different stakeholders and broadly reflects meaningful outcomes in usual clinical care is needed.

One overall measure of effectiveness for antipsychotic treatment outcome in schizophrenia is time to all-cause medication discontinuation, a measure that captures the medication’s efficacy, safety, and tolerability from both patients’ and physicians’ perspectives.^13,14 The Clinical Antipsychotic Trials of Intervention Effectiveness, a large, independent, publicly funded, randomized clinical trial in the USA, used time to all-cause medication discontinuation as its primary outcome variable.^13,15 Similarly, the European First Episode Schizophrenia Trial also used this as the study’s primary outcome variable.¹⁶ Patients who continue with an anti-psychotic longer have been shown to have better clinical and functional outcomes and a reduced risk of relapse and hospitalization.^12,13,17,18

Multiple studies have demonstrated that medication discontinuation is sensitive to differences in outcomes between antipsychotics. Olanzapine therapy in particular, has shown significantly lower rates of discontinuation than ziprasidone,^{13,15,19–22} quetiapine,^{13,15,21,23–29} risperidone,^{11–13,19,24,26,30–37} aripiprazole,^4,27,38 as well as typical antipsychotics.^{7,9–12,13,16,19,24,26,35,39–44}

Few studies have attempted to identify baseline characteristics of patients who stay longer on therapy in Japan. Previous studies outside of Japan have identified predictors of antipsychotic continuation in schizophrenia such as male gender,^28,45,46 older age,^{13,25,28,47,48} fewer psychiatric hospitalizations,²⁸ lack of substance-use disorder,^47,49,50 better therapeutic alliance,⁴⁹ greater reduction in symptoms,^17,18 and greater improvements in health-related quality of life (HRQOL).^17,18 Whether these findings generalize to the Japanese health care system, which utilizes inpatient treatment relatively frequently,^51,52 is unclear. Using data from a 1-year naturalistic observational study of patients with schizophrenia in Japan, this study aimed to identify baseline characteristics that differentiate patients who completed 1 year of olanzapine therapy from patients who discontinued olanzapine therapy.

Methods

Sample selection

The data for this study came from the Olanzapine Post Marketing Surveillance (OPMS) study. OPMS was a large multicenter naturalistic 1-year study that took place in Japan and consisted of 1850 patients who met the study entry criteria. To be eligible to take part in the study, participants had to have been diagnosed with schizophrenia and to have initiated treatment with olanzapine. The diagnosis for schizophrenia was based on criteria in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders published by the American Psychiatric Association.¹ During this naturalistic study, all treatment decisions, including the decision to initiate olanzapine, were left to the discretion of the treating physician. Study enrollment started in November 2003 and was completed in July 2004. The follow-up period for the study continued for 1 year after enrollment or until the patient discontinued the treatment with olanzapine. Data were collected at the baseline, 3-month, 6-month, and 12-month visits. The procedures were approved by the internal review boards at each of the participating institutions. Informed consent for this observational study was acquired based on the requirements at each institution.

Measures and definitions

The primary dependent variable in this study was a dichotomous variable indicating if a patient continued with or discontinued olanzapine before the end of the 1-year study. Patients who left the study were classified as discontinuing olanzapine.

In this observational study, simple noninvasive measures were chosen to limit interference with usual clinical practice. Patient symptomatology was measured using the Clinical Global Impression-Schizophrenia (CGI-SCH) scale. The CGI-SCH scale consists of ratings for positive, negative, cognitive, and depressive symptoms as well as a global severity rating.⁵³ The ratings for the CGI-SCH scale are anchored and range from no symptoms (0) to severe symptoms (6).⁵³ Assessment of the concurrent validity between the CGI-SCH and the Positive and Negative Syndrome Scale found that the validity coefficients ranged from 0.61 for depressive symptoms to 0.86 for positive symptoms, with the remaining coefficients ranging from 0.75–0.80.⁵⁴

Consistent with prior research, treatment response was operationally defined based on the CGI-SCH Global Severity rating.⁵⁵ For patients with a baseline score between 4 and 6 points, treatment response was defined as a 2-point improvement. For patients with a baseline score between 1 and 3, treatment response was defined as a 1-point improvement.

HRQOL was assessed using the European Quality of Life 5-Dimensions Visual Analog Scale (EQ-5D VAS).⁵⁶ The EQ-5D VAS ranges from 0 to 100 and has demonstrated moderate concurrent validity (ranging from 32 to 62) with the various scales of the World Health Organization Quality of Life – Brief instrument.⁵⁷

Prior medication use was coded with indicator variables (Yes/No) for each of the following medication classes: antipsychotics, anticholinergics, antidepressants, anxiolytics/ hypnotics, mood stabilizers, or other. The presence of a baseline medical comorbidity was coded based on the presence of any of the following conditions: hypertension, hyperlipidemia, hepatic dysfunction, renal dysfunction, or other. Participation in social activities was coded if the patient indicated they had participated in any social activities in the past 4 weeks. Working for pay was coded if the patient indicated working with income at baseline. Finally, living independently was coded if the patient indicated living independently at baseline rather than being hospitalized, homeless, living communally with persons caring for the patient, or living as a dependent family member.

Statistical methods

Differences in baseline patient characteristics were compared using t-tests for continuous variables and chi-square tests for categorical variables. Stepwise logistic regression was used to identify independent baseline predictors of continuing olanzapine therapy. Table 1 lists all of the baseline predictor variables available to the stepwise logistic regression model. However, the CGI-SCH Global Severity scale was excluded from the stepwise logistic regression due to multicolinearity with the subscales. The significance level was set at α = 0.05 and all analyses were completed using SAS 9.1.3 (SAS Institute Inc, Cary, NC).

Table 1 Baseline univariate characteristics

Characteristic	Total N	Discontinuer (N = 588)	Continuer (N = 1262)	P
Demographics
Age (y), mean ± SD	1850	42.4 ± 15.5	45.9 ± 15.4	<0.001
Female (%)	1850	51.0	44.8	0.013
Clinical status
Outpatient status (%)	1850	49.3	40.4	<0.001
Duration of illness (y), mean ± SD	1451	15.1 ± 14.0	19.7 ± 14.9	<0.001
Tardive dyskinesia (%)	1822	5.2	7.7	0.052
BMI (kg/m^2), mean ± SD	1638	22.6 ± 4.3	22.6 ± 4.0	0.84
Any medical comorbidities (%)	1849	31.5	38.4	0.004
Clinical measures
CGI-SCH global severity, mean ± SD	1822	3.3 ± 1.0	3.4 ± 1.1	0.45
CGI-SCH positive, mean ± SD	1822	3.1 ± 1.5	2.9 ± 1.5	0.052
CGI-SCH negative, mean ± SD	1822	2.9 ± 1.4	3.2 ± 1.3	<0.001
CGI-SCH cognitive, mean ± SD	1822	2.8 ± 1.4	3.0 ± 1.3	0.004
CGI-SCH depressive, mean ± SD	1822	1.7 ± 1.4	1.6 ± 1.3	0.09
Responded at 3-month visit (%)	1822	28.9	36.0	0.003
Functional measures
EQ-5D VAS, mean ± SD	1815	43.6 ± 21.3	49.6 ± 22.8	<0.001
Working for pay (%)	1820	12.0	7.5	0.007
Social activities (%)	1820	32.7	26.8	0.027
Living independently (%)	1822	20.4	15.7	0.014
Baseline and prior medication use
Starting dose of OLZ (mg/day), mean ± SD	1847	8.4 ± 5.3	8.6 ± 5.0	0.56
Switch from atypical antipsychotic (%)	1850	21.6	20.5	0.60
Switch from typical antipsychotic (%)	1850	13.3	15.2	0.27
Prior anticholinergic use (%)	1823	35.7	41.8	0.012
Prior antidepressant use (%)	1823	7.7	4.4	0.005
Prior anxiolytic/hypnotic use (%)	1823	57.6	60.5	0.24
Prior mood stabilizer use (%)	1823	13.0	12.9	0.93
Prior other medication use (%)	1823	32.2	44.9	<0.001

Abbreviations: BMI, body mass index; CGI-SCH, Clinical Global Impression-Schizophrenia; EQ-5D VAS, European Quality of Life 5-Dimensions Visual Analog Scale; OLZ, olanzapine; SD, standard deviation.

Results

Sample description

The OPMS study registered and enrolled 1949 patients. This analysis utilized the 1850 (94.9%) patients who met all study entry criteria: 27 were excluded for contract or registration violations, 20 had no case report forms, 49 did not return after the initial visit, and three did not initiate treatment with olanzapine. The baseline characteristics of this sample can be seen in Table 1. For the entire sample, the average age was 44.8 ± 15.5 years, 53.2% were male, 43.2% initiated olanzapine in an outpatient setting, and the average duration of illness was 18.3 ± 14.7 years. Most participants (68.2%; 1262/1850) continued the olanzapine treatment for the full 1-year study, while 31.8%; (588/1850) discontinued the study at some point (see Figure 1). The mean duration of olanzapine therapy during the 1-year study treatment was 265.5 ± 119.4 days.

Figure 1 Proportion of patients continuing olanzapine for 1 year. Most patients (68.2%, 1262/1850) continued with olanzapine treatment for the full 1-year study.

Note: The mean duration of olanzapine treatment was 265.5 ± 119.4 days.

Univariate comparisons between discontinuers and continuers

Table 1 contrasts the baseline characteristics of patients who completed the 1-year study with olanzapine (continuers) with those who discontinued olanzapine treatment prior to the end of the study (discontinuers). Significant differences were observed in multiple demographics, clinical status variables, symptom measures, functional measures, and prior medication use variables (see Table 1).

Multivariate comparisons between discontinuers and continuers

Stepwise logistic regression was used to identify independent predictors of continuation. The odds ratios and confidence intervals for the final version of the stepwise logistic regression model are presented in Figure 2. Overall, the model was modestly accurate in differentiating patients who would continue or discontinue treatment with olanzapine (c-statistic = 0.635). The c-statistic indicated that the model could accurately classify a randomly selected continuer from a randomly selected discontinuer 63.5% of the time. Continuation was independently predicted by lower positive symptoms, higher negative symptoms, longer illness duration, and better HRQOL. All of the significant predictors were continuous variables; therefore, the odds ratios represent the change in odds of continuation for every one-unit increase in the predictor. For example, for every year that a patient had been diagnosed with schizophrenia, the odds of continuing with olanzapine increased by a factor of 1.01.

Figure 2 Significant predictors of treatment continuation in the stepwise logistic regression analysis. This figure presents the odds ratios and 95% confidence intervals for each of the predictors of continuation in the final stepwise logistic regression model. Each predictor in the final model was a continuous variable; therefore, the odds ratios represent the change in odds of continuation for every one-unit increase in the predictor variable.

Abbreviations: CGI-SCH, Clinical Global Impression-Schizophrenia; EQ-5D VAS, European Quality of Life 5-Dimensions Visual Analog Scale.

Finally, early change in CGI-SCH Global Severity, defined as the difference between baseline and the first visit (3 months), was compared between the continuers and discontinuers. Patients who continued with olanzapine treatment experienced greater improvements (decreases) in CGI-SCH Global Severity compared to those who discontinued olanzapine therapy (−0.69 vs −0.53; P = 001).

Discussion

In this large naturalistic observational study of patients with schizophrenia in Japan, most patients (68.2%) continued olanzapine therapy for the full 1-year study, with an average time to discontinuation of 265.5 days. A large number of baseline characteristics predicted the continuation in the univariate analyses, but only a few were found to be independent predictors in the multivariate analysis. In the stepwise logistic regression model, higher EQ-5D VAS scores, longer duration of illness, higher CGI-SCH negative scores, and lower CGI-SCH positive scores were significant predictors of treatment continuation. The olanzapine continuers had significantly greater early improvements in CGI-SCH Global Severity scores than the discontinuers.

The findings in the current study are consistent with some past schizophrenia research with outpatients outside of Japan. In the univariate analysis, older age^{13,25,27,47,48} and male gender^28,44,46 predicted antipsychotic continuation. The primary findings in the multivariate analysis were also consistent with previous research on predictors of medication continuation: longer duration of illness,⁴⁵ higher baseline HRQOL,⁵⁸ higher baseline negative symptoms,⁵⁸ and lower baseline positive symptoms⁵⁹ have all been identified as predictors of antipsychotic continuation. Interestingly, in one study, lower baseline positive symptoms were reported to predict discontinuation rather than continuation.⁴⁸ Inconsistencies in predictors of continuation may be a result of differences in study methodologies, outcome variables, the inclusion of inpatients, study population, or health care systems.

In addition to the identified patients’ baseline characteristics, poorer efficacy of antipsychotic medications has also been linked to the discontinuation of treatment. In fact, past research has consistently identified lack of efficacy as being one of the primary reasons for antipsychotic discontinuation.^{13,17,21,26,45,60–63} Consistent with prior research, secondary analysis in the present study found that patients who continued antipsychotic treatment had greater reductions in symptoms.

Limitations

Although a large number of predictors were used in this analysis, some potentially important predictors of antipsychotic persistence were not included in this study. A lower level of substance abuse has been frequently demonstrated as being an important predictor of antipsychotic persistence.^{47,49,50,63–67} Additionally, insight into illness⁶⁷ as well as positive attitudes toward medication^49,68 have been identified as being potentially important in the predictors of antipsychotic persistence. The primary focus of this analysis was on baseline predictors of later antipsychotic persistence. Changes in metabolic and other tolerability parameters were not included in the analyses and may be important predictors of antipsychotic persistence. Furthermore, the OPMS study focused only on olanzapine-treated patients with schizophrenia in Japan, so the findings may not generalize to other antipsychotics or geographic locations.

Conclusion

Consistent with prior research, patients who continued olanzapine therapy were more likely to experience significant improvements in clinical outcomes compared to those who discontinued treatment.^18,69 In addition, patients with certain baseline characteristics appear to be more likely to continue with olanzapine treatment for a longer period of time. Stepwise logistic regression revealed that, compared to discontinuers, those who continued with olanzapine therapy had significantly longer illness duration, lower baseline positive symptoms, higher baseline negative symptoms, and higher baseline HRQOL. Identifying these characteristics has clinical implications in usual care in Japan, as they may be useful in early detection of schizophrenia patients who could benefit from targeted interventions aimed at improving their persistence with medication, thus increasing the chance for better long-term treatment outcomes.

Acknowledgments

Funding for this study was provided by Eli Lilly and Company. Technical writing support was provided by Michael Stensland of Agile Outcomes Research Inc, Rochester, MN, and Susan Dennett of Strategic Health Outcomes Inc, Carmel, IN.

Disclosure

Wenyu Ye is a full-time employee of Lilly Suzhou Pharmaceutical Co, Shanghai, People’s Republic of China. Haya Ascher-Svanum is a full-time employee of Eli Lilly and Company, Indianapolis, IN, USA. Jennifer A Flynn and Yuka Tanji are full-time employees of Eli Lilly Japan, KK, Kobe, Japan. Michihiro Takahashi is a consultant for Eli Lilly Japan, KK. All authors are minor stockholders in Eli Lilly and Company.