Introduction
Methods
Population
Competing treatment strategies
TAU
SMA
Key impacts of the stratifier modelled in the study
Model structure
- AP2 responder, patients who will respond to a second-line conventional antipsychotic.
- Clozapine responder, patients who will fail to respond to a second line antipsychotic but will subsequently respond to clozapine.
- Non-responder, patients who will not respond to any existing antipsychotics including clozapine.
Input data
Parameters | Base-line value | Range tested in one-way or two-way sensitivity analysis | Distribution | Source | ||||
---|---|---|---|---|---|---|---|---|
Diagnostic efficacy of predictive test | ||||||||
Sensitivity (proportion of second-line antipsychotic responders that are correctly identified as such) | 0.60 | 0–1.00 | Assumed fixed | Estimate of what may be achievable | ||||
Specificity (proportion of second-line antipsychotic non-responders that are correctly identified as such) | 0.60 | 0–1.00 | Assumed fixed | Estimate of what may be achievable | ||||
Distribution of patients who failed a first-line antipsychotic by subsequent response/non-response | ||||||||
Clozapine responder | 62.50% | 0%–79.15% | Dirichlet distribution (n = 21.0) | Agid et al. [15] | ||||
AP2 responder | 16.67% | N/A | Dirichlet distribution (n = 5.6) | As above | ||||
Clozapine non-responder | 20.83% | N/A | Dirichlet distribution (n = 7.0) | As above | ||||
Response to different antipsychotics in misclassified individuals | ||||||||
AP2 responder’s response to clozapine | 71.16% | 0–1 | Beta distribution (SD assumed to be 50% of mean value) | |||||
AP2 non-responder response to second-line antipsychotics | 0% | N/A | Assumed fixed | Estimate | ||||
Cost data | ||||||||
Cost of predictive test | £500.00 | £100.00–1,000.00 | Gamma distribution (SD assumed to be 50% of mean value) | Estimate |
Interventions | Base-line value | Range tested in one-way or two-way sensitivity analysis | Distribution | Source |
---|---|---|---|---|
Annual cost of treating patients with active psychosis | £39,141 | N/A | As above | Uplifted from the NICE schizophrenia guideline [2] |
Annual cost of treating remitted patients | £15,086 | £10,000–£39,141 | As above | As above |
Health-related quality of life data | ||||
Relapse | 0.4790 | 0.1900–0.6040 | Beta distribution (SD: 0.0330) | Briggs et al. [24] |
Stable schizophrenia | 0.8650 | 0.8650–0.9190 | Beta distribution (SD: 0.0210) | As above |
Other input data | ||||
Annual discount rate for both costs and outcomes | 0.0350 | 0–0.050 | Assumed fixed | NICE guideline manual [19] |
Cycle length | 1 year | N/A | Assumed fixed | Estimate |
Number of cycles | 80 | 10–100 | Assumed fixed | Estimate |
Sensitivity analysis
Model verification and validation
Verification and validation process | People involved |
---|---|
1. Check appropriateness of the model structure | JM and PM |
2. Check appropriateness of data source | JM and PM |
3. Compare data used in the model against the evidence sources | HJ |
4. Check model logic (white-box testing) | HJ |
5. Check the plausibility of the intermediate and final outputs, including results of sensitivity analyses (black-box testing) | HJ, JM and PM |
6. Compare results with published literature | HJ |
Results
Strategy | Total Costs (£) | Total QALYs | Incremental cost (£) | Incremental QALYs |
---|---|---|---|---|
TAU | 510,239 | 16.15 | – | – |
SMA | 502,876 | 16.25 | − 7363 | 0.10 |