|
Should clozapine continue to be restricted to third-line status for schizophrenia: a decision-analytic model |
Wang P S, Ganz D A, Benner J S, Glynn R J, Avorn J |
|
|
Record Status 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. Health technology The use of clozapine as a first-line treatment for patients with schizophrenia experiencing an acute psychotic episode. The comparators were clozapine as a third-line treatment after failure of two conventional antipsychotic agents, and conventional antipsychotics alone (i.e. never using clozapine at any stage of a psychotic episode).
Study population The study population comprised a hypothetical cohort of 30-year-old patients with schizophrenia, both men and women, hospitalised for an acute psychotic episode.
Setting The setting was secondary care. The economic study was conducted in the USA.
Dates to which data relate The effectiveness evidence was derived from literature published between 1986 and 1999. The costs were derived from studies published between 1993 from 1999. The price year was 1999.
Source of effectiveness data The effectiveness data were derived from a review or synthesis of completed studies.
Modelling A Markov model was developed to estimate the costs and benefits arising from the adoption of each of the three strategies evaluated (first-line clozapine, third-line clozapine, or conventional antipsychotics alone) for the treatment of 30-year-old patients with schizophrenia hospitalised for an acute psychotic episode.
For the first-line clozapine strategy, it was assumed that patients received clozapine initially, switching to, and remaining on, conventional antipsychotics if they failed to recover to the point of being dischargeable from the hospital, or relapsed after recovery, or developed agranulocytosis on clozapine.
According to the third-line clozapine strategy, patients initiated treatment with a conventional antipsychotic, switching to a second conventional antipsychotic if they failed to recover to the point of being dischargeable from the hospital, or relapsed after recovery, or developed serious tardive dyskinesia (TD) on the first conventional antipsychotic. Following this, the patients were assumed to switch to clozapine if they failed to recover to the point of being dischargeable from the hospital, or relapsed after recovery, or developed serious TD on the second conventional antipsychotic. Finally, patients switched back to a conventional antipsychotic if they failed to recover to the point of being dischargeable from the hospital, or relapsed after recovery, or developed agranulocytosis on clozapine.
The conventional antipsychotics strategy assumed the use of conventional antipsychotics alone, even in treatment-resistant patients.
In every state of the model described, patients faced the possibility of death. Therefore, the Markov model included 7 types of health states. More specifically, acute psychosis and treatment with clozapine, acute psychosis and treatment with a conventional antipsychotic, recovering with clozapine, recovering with a conventional antipsychotic, agranulocytosis with clozapine, serious TD with a conventional antipsychotic, and death. Other potential side effects resulting from treatment, such as Parkinsonism, dystonias or akathisia, were not considered in the model. The cycle of the model was 3 months. A lifetime time horizon was used.
Outcomes assessed in the review The outcomes assessed were the 3-month transition probabilities used in the Markov model, specific to clozapine and conventional antipsychotics. These included:
the probability of recovery from psychosis (expressed as dischargeability from the hospital);
the probability of relapse;
the probability of developing agranulocytosis and of death due to agranulocytosis (for clozapine);
the probability of developing TD (for conventional antipsychotics);
the risk of death by suicide; and
age-adjusted all-cause mortality rates, applied to each state of the model.
Quality of life weights were also derived from the literature.
Study designs and other criteria for inclusion in the review The majority of the data were derived from a published meta-analysis of data from randomised clinical trials (RCTs) (Wahlbeck et al., see Other Publications of Related Interest). Where evidence was lacking, the effectiveness data were derived from observational studies. The authors used published quality of life weights for the health states in schizophrenia. These were derived from standard gambles, rating scales and paired comparison questions.
Sources searched to identify primary studies Criteria used to ensure the validity of primary studies Methods used to judge relevance and validity, and for extracting data Number of primary studies included Approximately 7 primary studies were included in the review.
Methods of combining primary studies The results of the primary studies were not combined. However, most of the effectiveness data were obtained from a published meta-analysis of data from RCTs.
Investigation of differences between primary studies Results of the review The 3-month transition probabilities while on clozapine were:
recovery from psychosis 42.9%;
relapse 8.0%;
development of agranulocytosis 0.019%;
death due to agranulocytosis 3.141%; and
death by suicide 0.02%.
The 3-month transition probabilities while on conventional antipsychotics were:
recovery from psychosis 40.0%;
relapse 13.0%; development of TD 1.3%; and
death by suicide 0.07%.
Age-adjusted all-cause mortality rates were not reported.
The quality weights assigned to health states were:
recovered from psychosis 0.83;
actively psychotic 0.56; and
TD 0.69.
Methods used to derive estimates of effectiveness The authors made assumptions to supplement the effectiveness evidence used to populate the model.
Estimates of effectiveness and key assumptions Owing to the lack of any direct evidence, the quality of life weight for agranulocytosis was assumed to be equal to that assigned to severe infection while immunosuppressed from cancer chemotherapy. The quality weight assigned to agranulocytosis was 0.46.
The recovery from psychosis was assumed to lead to the same quality of life with either clozapine or conventional antipsychotics.
Measure of benefits used in the economic analysis The main measure of health benefit used was the number of quality-adjusted life-years (QALYs) gained with each strategy assessed. Quality of life weights for the health states in schizophrenia were derived from standard gambles, rating scales and paired comparison questions. The benefits were estimated over the patients' lifetime and were discounted at an annual rate of 3%.
Direct costs The direct costs consisted of health service costs. These included hospitalisation costs for psychotic episodes, costs of outpatient care, residential treatment costs, antipsychotic medication costs, and costs of treating side effects (TD due to conventional antipsychotics and agranulocytosis leading to hospitalisation due to clozapine). In the case of treatment with clozapine, a weekly cost for white blood cell (WBC) monitoring was added to the total costs (authors' assumption).
The costs and the quantities were not analysed separately. Hospitalisation costs were derived from the Inventory of Mental Health Organisations and General Hospital Mental Health Services, 1995. Outpatient and residential treatment costs were derived from two studies published in 1990 and 1993, one of which utilised Medicaid data. The other costs were estimated from published literature (1994 to 1996). The total costs were derived by modelling. All of the costs were inflated to 1999 prices using the medical care component of the Consumer Price Index. Discounting was appropriately undertaken since the costs were incurred during more than one year. The total costs were estimated over the lifetime of the hypothetical population of 30-year-old patients. The annual discount rate was 3%.
Statistical analysis of costs The costs were treated deterministically. No statistical analysis of the costs was undertaken.
Indirect Costs The indirect costs were not included in the analysis.
Sensitivity analysis A sensitivity analysis was carried out to investigate the robustness of the results to variability in the input parameters used in the model. All model input parameters (i.e. transition probabilities, utility weights and costs) were tested in one-way sensitivity analyses. A probabilistic sensitivity analysis was also undertaken, in which a Monte Carlo simulation was used to vary all input parameters simultaneously. Regarding transition probabilities, the upper and the lower bounds of the 95% confidence intervals around each estimate, as reported in the literature, were used. For costs and utility weights, the range of values used was determined by decreasing or increasing the base-case estimates by 25%.
Estimated benefits used in the economic analysis The undiscounted life expectancy was 31.03 years for the first-line clozapine strategy, 31.01 years for the third-line clozapine strategy and 30.92 years for the conventional antipsychotics alone strategy.
After discounting at 3% annually and quality-adjusting for time spent in each health state, the three strategies yielded 14.59 (first-line), 14.58 (third-line) and 14.51 (conventional) QALYs per patient, respectively.
Cost results The undiscounted annual costs per patient were $26,650 under the first-line clozapine strategy, $26,640 under the third-line clozapine strategy and $26,530 under the conventional antipsychotics alone strategy.
After applying a 3% annual discount rate, the total discounted costs per patient for the three strategies were $514,100 (first-line), $513,800 (third-line) and $509,200 (conventional), respectively.
The costs of side effects of medication were included in the estimation of the total costs.
Synthesis of costs and benefits The costs and benefits were combined in the form of incremental cost-effectiveness ratios (ICERs).
The ICER of first-line clozapine versus third-line clozapine was $24,100/QALY gained.
The ICER of first-line clozapine versus conventional antipsychotics alone was $58,000/QALY gained.
The ICER of third-line clozapine versus conventional antipsychotics alone was $64,400/QALY gained.
The ICER of first-line clozapine versus third-line clozapine was most sensitive to estimates of the rate of recovery from acute psychotic episodes on both conventional antipsychotics and clozapine. Assigning the value most unfavourable to the first-line clozapine strategy for either of these gave ratios over $30,000/QALY.
The analysis was also relatively sensitive to the quality of life weight assigned to (in descending order) being recovered from psychosis, rates of TD on conventional antipsychotics, the cost of clozapine, the cost of inpatient hospitalisation, the discount rate, relapse rates on clozapine and conventional antipsychotics, and the cost of residential treatment. The results were relatively insensitive to the rest of the parameters.
According to the probabilistic sensitivity analysis, the 25th, 50th, and 75th percentiles of the ICER of first-line clozapine versus third-line clozapine were $16,700, $23,500 and $31,100/QALY gained, respectively.
Authors' conclusions Employing clozapine as first-line (instead of third-line) treatment for acute psychotic episodes of patients with schizophrenia led to small gains in life expectancy and quality-adjusted life expectancy, at moderate but acceptable costs. In health care systems where clozapine had never been used even for treatment-resistant patients, employing it as a first-line treatment also appeared to lead to modest gains in quality-adjusted life-years (QALYs) at a fairly reasonable cost.
CRD COMMENTARY - Selection of comparators A justification was given for the comparators used. Clozapine as third-line treatment represented current practice at the time of the analysis, as approved by the US FDA, whereas no use of clozapine was a common tactic observed in some health care systems. You should consider whether any of the comparators represent widely adopted practices in your own setting.
Validity of estimate of measure of effectiveness The authors did not state that a systematic review of the literature had been undertaken. Since the methods used to find and select the primary studies and to extract the data were unclear, it is difficult to assess the validity of the estimates. However, most of the effectiveness data used were obtained from a published meta-analysis of data from RCTs. Where evidence of effectiveness was lacking, data were derived from observational studies, hence introducing the possibility of confounding bias. The authors suggested that RCTs may have limited internal and external validity. For example, medication non-compliance in clinical trials can attenuate the apparent efficacy of antipsychotic medications and affect the cost-effectiveness analyses in unpredictable ways. The authors acknowledged that the use of dischargeability from hospital as a proxy for clinically meaningful improvement might have underestimated the advantage of clozapine over conventional antipsychotics, because it did not account for the greater frequency of partial improvements observed for clozapine versus conventional antipsychotics.
Validity of estimate of measure of benefit The estimation of benefits was modelled. The Markov model used for this purpose was appropriate, as it included all basic states resulting from the use of the health technologies examined and it estimated associated benefits, probably over the lifetime of the patients. The authors acknowledged that the model did not fully describe the range of health states that could be experienced by patients with schizophrenia, such as all side effects from medications. However, the model captured side effects that have been clearly established and were burdensome.
Validity of estimate of costs Although the authors stated that they adopted a societal perspective, the indirect costs and other direct costs, such as patient costs and informal care, were not included in the analysis. No justification was given for their exclusion. The costs and the quantities were not analysed separately, which hinders the reproducibility of the results. A sensitivity analysis of the costs was undertaken, using ranges that appear to have been appropriate. The authors acknowledged that the assumptions of weekly WBC monitoring, rather than the current practice of monitoring every 2 weeks after 6 months, might have biased the results against the clozapine-first strategy. Discounting was applied, as necessary, since the costs were incurred during more than one year. The date to which the prices referred was reported, and this improves the generalisability of the results.
Other issues The authors did not compare their findings with those of other studies, but they stated that their study was the first to estimate the cost-effectiveness of first-line clozapine versus a third-line clozapine strategy. The issue of generalisability to other settings was partially addressed using sensitivity analyses. The results of the analyses were adequately reported. The authors reported a number of limitations of their study. For example, issues on the validity of the effectiveness and cost data used in the base-case and sensitivity analyses. Also, some relevant health states experienced by patients with schizophrenia (e.g. other side effects, potential benefits of clozapine including improvements in psychotic symptom severity) were omitted from the model. Moreover, they acknowledged the fact that their study could not answer the question of what the preferred first-line treatment should be in acute psychotic episodes in schizophrenia, as the number of trials comparing clozapine with newer atypical antipsychotics in treatment-sensitive patients was insufficient. However, the authors' conclusions reflected the scope of the analysis.
Implications of the study The authors suggested that clozapine should not necessarily be confined to its role as third-line treatment for treatment-resistant patients with schizophrenia experiencing an acute psychotic episode. They stated that clinical and economic reasons appeared to justify its addition among possible treatments considered for treatment-sensitive, as well as treatment-resistant, schizophrenia. With regards to the issue of the preferred first-line strategy for acute psychotic episodes, it was suggested that data from large trials comparing clozapine with newer atypical agents for treatment-sensitive patients are needed. This would allow the structure of the analysis to be updated and cover this additional, critically important issue.
Source of funding Supported by a Research Career Award from the National Institute of Mental Health.
Bibliographic details Wang P S, Ganz D A, Benner J S, Glynn R J, Avorn J. Should clozapine continue to be restricted to third-line status for schizophrenia: a decision-analytic model. Journal of Mental Health Policy and Economics 2004; 7: 77-85 Other publications of related interest Wahlbeck K, Cheine M, Essali A, Adams C. Evidence of clozapine's effectiveness in schizophrenia: a systematic review and meta-analysis of randomised trails. American Journal of Psychiatry 1999;156:990-9.
Indexing Status Subject indexing assigned by NLM MeSH Adult; Antipsychotic Agents /economics /therapeutic use; Clozapine /economics /therapeutic use; Decision Support Systems, Clinical; Humans; Markov Chains; Outcome Assessment (Health Care); Probability; Quality-Adjusted Life Years; Schizophrenia /drug therapy; United States AccessionNumber 22004008235 Date bibliographic record published 30/06/2005 Date abstract record published 30/06/2005 |
|
|
|