|Economic outcomes associated with switching individuals with schizophrenia between risperidone and olanzapine: findings from a large US claims database
|Zhao Z Y, Namjoshi M, Barber B L, Loosbrock D L, Tunis S L, Zhu B J, Breier A
This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.
The use of atypical antipsychotic agents, olanzapine (OLZ) and risperidone (RIS), for the treatment of patients with schizophrenia.
Economic study type
The study population comprised patients diagnosed with schizophrenia according to the International Classification of Diseases (ICD-9: diagnosis code 295.xx), who were privately insured.
The setting was secondary care. The economic study was conducted in the USA.
Dates to which data relate
The effectiveness, cost and resource data used in the analysis were gathered between October 1995 and December 1999. The prices related to 1999.
Source of effectiveness data
The effectiveness data were derived from a single study.
Link between effectiveness and cost data
Both the cost and effectiveness data were collected retrospectively from a claims database on the same patient sample.
The study sample consisted of patients with schizophrenia who had been prescribed RIS and OLZ. The patients were identified from a large claims database in the USA, which contained data on approximately 1.6 million employees, dependants and retirees. The database did not include records for Medicaid patients. A validated algorithm for the identification of patients with schizophrenia from claims databases was used to identify patients who had been given a diagnosis of schizophrenia at least once as inpatients, or twice as outpatients. Only patients who filled at least two prescriptions, and who received at least 60 days' treatment for each of the two medications, were selected for the study. Continuous health insurance coverage for both the pre- and post-switch periods was also required.
Of the 448 patients identified who had a diagnosis of schizophrenia and were prescribed OLZ and RIS, only 265 met the criteria required for inclusion. In addition, only 244 patients fulfilled the requirement for continuous health insurance coverage for both the pre- and post-switch periods. These 244 were ultimately included in the analysis, with 202 having switched from RIS to OLZ (RIS-OLZ group) and 42 having switched from OLZ to RIS (OLZ-RIS group). Reasons for switching were not available from the database and were therefore not taken into consideration. No power calculations were performed either prospectively or retrospectively. There was no evidence that the study sample was appropriate for the clinical study question.
The study was a retrospective within-group comparison study ("before-after") that consisted of two separate comparisons, one within each group studied. The patients were selected from a claims database that contained information on employees, dependants and retirees from 21 US states. The switch date was defined as the date for the first prescription of OLZ in the RIS-OLZ group and of RIS in the OLZ-RIS group. The pre- and post-switch periods were defined, respectively, as one year before and one year after the switch date. However, if the patients did not have one year of observation in either the pre- or post-switch period, then the actual duration of the observation was used and the treatment costs were annualised. Of the 244 patients included in the analysis, 65.2% had a complete one-year observation period. This percentage was similar for the OLZ-RIS (65.3%) and RIS-OLZ (64.3%) groups.
Analysis of effectiveness
All of the patients included in the study were accounted for in the analysis. The primary health outcomes evaluated were the percentages of patients with concomitant use of antiParkinsonian agents, antidiabetic agents, antihyperlipidaemic agents and typical antipsychotics in the pre- and post-switch periods. These drugs were considered because their use might indicate the existence of adverse effects due to RIS and OLZ, or the reduced efficacy of RIS and OLZ. The mean duration of observation was not statistically significantly different between the pre- and post-switch periods for both the OLZ-RIS and RIS-OLZ groups, (p=0.4891 and p=0.1907, respectively, Wilcoxon signed rank tests).
In the RIS-OLZ group:
30.20% of patients were treated with antiParkinsonian agents in the pre-switch period versus 21.29% in the post-switch period, (p=0.0094);
30.69% (pre-switch) and 18.32% (post-switch), respectively, were treated with typical antipsychotics, (p=0.0006);
12.38% (pre-switch) and 10.89% (post-switch), respectively, were treated with antidiabetic agents, (p=0.2568); and
11.39% (pre-switch) and 14.36% (post-switch), respectively, were treated with antihyperlipidaemic agents, (p=0.1336).
In the OLZ-RIS group:
14.29% of patients were treated with antiParkinsonian agents in the pre-switch period versus 19.05% in the post-switch period, (p=0.4142);
14.29% in both the pre- and post-switch periods were treated with typical antipsychotics, (p=1.0000);
11.90% in both the pre- and post-switch periods were treated with antidiabetic agents, (p=1.000); and
21.43% (pre-switch) and 23.81% (post-switch) were treated with antihyperlipidaemic agents, (p=0.3173).
After switching from RIS to OLZ there was a more favourable medication treatment course. A significantly greater percentage of patients no longer required the concomitant use of antiParkinsonian agents or typical antipsychotics. Switching from OLZ to RIS was not followed by any significant change in the medication treatment course.
Measure of benefits used in the economic analysis
The authors did not derive a summary measure of benefit. In effect, a cost-consequences analysis was performed.
The perspective adopted in the study was unclear, but it appears to have been that of a third-party payer. Mental and total health care costs were estimated and compared separately for the pre- and post-switch periods for each group. The total health care costs comprised pharmaceutical and medical costs. The medical costs covered inpatient costs, emergency room costs, hospital outpatient costs, and the costs of psychiatric day or night treatment, nursing home, physician office visits, psychotherapy, substance abuse, laboratory tests and other miscellaneous services. As a sub-set of total health care costs, the mental health care costs were calculated from service charges associated with a mental illness (ICD-9 diagnosis codes 290.xx to 319.xx) and for medications used in treating mental illness. The resource quantities and the costs were not reported separately. The costs were estimated using actual data derived from the claims database, reflecting service charges but not the amounts actually paid. Discounting was appropriately not carried out since the pre- and post-switch periods, for which the costs were estimated, were each one year in duration. The resources quantities were estimated from those used in the period 1995 to 1999. The prices referred to 1999.
Statistical analysis of costs
The costs were presented as point estimates only. Wilcoxon signed rank tests were used to assess changes in the mental and total health care costs between the pre- and post-switch periods.
The indirect costs were not included in the analysis.
A sensitivity analysis was conducted to examine the impact that the exclusion of RIS-OLZ patients, who initiated treatment with RIS before OLZ became available on the market, would have on the results obtained. This exclusion resulted in 63 patients remaining in the RIS-OLZ sub-sample tested in the sensitivity analysis. This was considered appropriate since OLZ was on the market later than RIS. Patients who initiated RIS before OLZ was available had an opportunity to switch to a new drug once OLZ was on the market, resulting in substantially greater numbers in the RIS-OLZ group (202) than the OLZ-RIS group (42).
Estimated benefits used in the economic analysis
See the 'Effectiveness Results' section.
In the RIS-OLZ group, the mental health care costs per patient were $13,121 in the pre-switch period and $11,371 in the post-switch period, (p=0.3246). The corresponding total health care costs per patient were $17,713 (pre-switch) and $16,545 (post-switch), respectively, (p=0.9137).
In the OLZ-RIS group, the mental health care costs per patient were $11,460 in the pre-switch period and $14,408 in the post-switch period, (p=0.1270). The corresponding total health care costs per patient were $17,077 (pre-switch) and $23,458 (post-switch), respectively, (p=0.6805).
The costs were estimated for one year before and after switching the treatment.
The costs of adverse effects were also included in the analysis, although they were not reported separately.
Synthesis of costs and benefits
The costs and benefits were not combined. The results of the sensitivity analysis showed that there was still no significant change in mental and total health care costs, (p=0.887), when only patients who initiated treatment with RIS after OLZ became available were included in the RIS-OLZ group.
After switching from risperidone (RIS) to olanzapine (OLZ), there was a more favourable medication treatment course but no increase in the total health care costs. Conversely, patients switching from OLZ to RIS might not be expected to be associated with any benefits in medication treatment course and total health care costs.
CRD COMMENTARY - Selection of comparators
The selection of the comparators was implicitly justified. The American Psychiatric Association and other expert consensus had endorsed atypical antipsychotic agents as the first-line treatment for schizophrenia. The two agents compared were the most commonly prescribed atypical antipsychotics. Therefore, switching between these two treatments was a common clinical practice. You should consider whether the strategies examined reflect the health technology widely used in your own setting.
Validity of estimate of measure of effectiveness
The basis of the analysis was a within-group comparison study. This type of study carries the risk of confounding, as the effect of the second intervention on the outcomes may have been affected by prior exposure to the first intervention. Moreover, the observed change may be due to a confounder, such as an independent historical event (e.g. change in the overall clinical management of schizophrenia). As highlighted in the paper, it was also likely that patients who switched because of intolerance or lack of therapeutic response would be subject to a higher level of utilisation and costs, and these costs would decrease over the subsequent year after the switch. However, since the authors seemed to be interested in assessing the effect on the outcomes of switching between the two agents, and not simply comparing the associated costs and effectiveness between the two drugs, the study design seems to have been appropriate for the study question. It was possible that the study sample was representative of the patient population since it was derived from a large patient database representing routine practice. However, little was known on the severity of illness or reasons for switching treatment in the patient sample. Moreover, the database consisted of privately insured patients and did not include Medicaid patients (public insurance for the poor). Continuous health insurance coverage for both the pre- and post-switch periods was required in order for patients to be included in the study. The authors acknowledged that these facts might have limited the generalisability of the results.
The duration of observation was not significantly different between the pre- and post-switch periods for either patient group. Appropriate statistical tests were carried out to compare the effectiveness rates before and after switching the treatment. However, as the authors acknowledged, the sample size of each group was relatively small and this might affect the significance of statistical testing. Another element of the study design that may have introduced bias into the analysis was the fact that, for patients who did not have one year of observation in either the pre- or post-switch period, the actual duration of the observation was used and the treatment costs were annualised. No method of adjusting the clinical outcomes to the annual base was reported.
Validity of estimate of measure of benefit
No summary measure of benefit was derived. Thus, a cost-consequences analysis was conducted.
Validity of estimate of costs
The study perspective, although not explicitly stated, was consistent with that of a third-party payer. All the categories of cost relevant to this perspective were included in the analysis. The costs and the quantities were not reported separately. Statistical tests were used to assess the difference between the total costs compared. The costs were estimated on the basis of charges. According to the authors, charges may have overestimated the costs from a payer's perspective, as insurance companies are characterised by differences in reimbursement schemes. On the other hand, charges do not reflect resource use and, consequently, opportunity costs. Discounting was not necessary since the costs were incurred during one year for each of the two study periods, and was therefore not carried out. The date to which the prices referred was reported, which will aid future reflation exercises.
The authors made appropriate comparisons of their findings with those from other studies. They also discussed the issue of the generalisability of the results to other settings (i.e. the US public payer system). The authors reported limitations of the study, which have already been discussed in this commentary. Generally, the results appear to have been adequately reported and the authors' conclusions reflect the scope of the analysis.
Implications of the study
The authors suggested that their findings have important clinical and economic implications for health care decision-makers and clinicians seeking the most cost-effective way to treat schizophrenia.
Source of funding
Supported by Eli Lilly and Company.
Zhao Z Y, Namjoshi M, Barber B L, Loosbrock D L, Tunis S L, Zhu B J, Breier A. Economic outcomes associated with switching individuals with schizophrenia between risperidone and olanzapine: findings from a large US claims database. CNS Drugs 2004; 18(3): 157-164
Subject indexing assigned by NLM
Adult; Aged; Benzodiazepines /economics /therapeutic use; Cost-Benefit Analysis /statistics & Databases, Factual /statistics & Female; Health Care Costs /statistics & Humans; Male; Middle Aged; Risperidone /economics /therapeutic use; Schizophrenia /drug therapy /economics; Statistics, Nonparametric; United States; numerical data; numerical data; numerical data
Date bibliographic record published
Date abstract record published