Note from the National Guideline Clearinghouse (NGC): The National Institute for Health and Clinical Excellence (NICE) commissioned an independent academic centre to perform a systematic literature review on the technology considered in this appraisal and prepare an assessment report. The assessment report for this technology appraisal was prepared by the Southampton Health Technology Assessment Centre (SHTAC), Wessex Institute for Health Research and Development (WIHRD), University of Southampton (see the "Availability of Companion Documents" field).
Clinical Effectiveness
The manufacturer applied the quality assessment criteria to the INNOVATE trial and an open-label "naturalistic" randomly controlled trial (RCT). Since the latter did not meet the inclusion criteria for the manufacturer's submission (MS), the Evidence Review Group (ERG) have not subjected it to critical appraisal. The manufacturer's quality assessment of the INNOVATE trial was not adequate for some parameters (refer to Table 2 of the Evidence Review Group Report [see the "Availability of Companion Documents" field]), and details of whether the process was performed by two independent reviewers were lacking. There is also uncertainty about the validity of including unpublished post hoc analysis for two subgroups.
The submitted evidence generally reflects the decision problem defined in the submission.
There is uncertainty about some of the statistical methods used in the analysis because of post hoc adjustments to the data, and reporting of results for the primary intention to treat (PITT) (and not true intention to treat [ITT]) population. Confidence intervals are not presented consistently for all outcomes. There is also uncertainty about the quality of the INNOVATE trial as no information was provided in the MS on how treatment allocation was concealed, and some details were also lacking with regards the power calculation for the sample size. The INNOVATE trial paper did not report on either of these aspects. The ERG also noted some concern that personnel preparing and administering the injections were aware of the identity of the drug/placebo treatment and whether this may have potentially impacted on the blinding of the study.
No evidence synthesis in the form of a meta-analysis was possible as there was only one RCT.
Refer to Sections 3.1 to 3.4 of the Evidence Review Group Report (see the "Availability of Companion Documents" field) for more information.
Economic Evaluation
Critical Appraisal of the Manufacturer's Submitted Economic Evaluation
Modelling Methods
An outline critical review of modelling methods has been undertaken. The review has used the framework for good practice in modelling presented by Philips and colleagues as a guide, addressing issues of model structure, structural assumptions, data inputs, consistency, and assessment of uncertainty.
Modelling Approach/Model Structure
The MS presents a Markov state transition model comprising five health states: day-to-
day symptoms, clinically significant (CS) non severe exacerbation, CS severe exacerbation, asthma death and non asthma death. The modelling approach and health states used in the model seem reasonable to the ERG and the clinical experts consulted. The structural assumptions have been justified and seem reasonable.
Structural Assumptions
The MS provides little detail on the development of the model structure and makes no explicit reference to its clinical validation. The MS suggests that the model structure has been derived from observation of clinical trial findings and not from an underlying clinical model of the disease. The structure reflects the conception of asthma as a disease of fluctuating day-to-day symptoms and intermittent exacerbations. Thus patients' usual condition is one of impaired quality of life, due to the variable presence of day-to-day symptoms (such as wheezing, coughing or shortness of breath causing the patient to wake at night). Patients' movement between health states does not reflect progression of disease – as would be more typical of a Markov model of chronic disease – but temporary and reversible deterioration.
While noting the lack of detail on the development and validation of the model, it appears to be appropriate given the decision problem, the data available and the specified causal relationships.
One-way Sensitivity Analyses
The MS presents univariate sensitivity analyses for a limited range of methodological (discount rates), structural (time horizon) and parameter (treatment duration, asthma related fatality, health state utility, exacerbation cost and basis for estimating omalizumab drug cost) uncertainties. No rationale has been given for the choice of variables included in (or excluded from) this sensitivity analysis. In addition the analysis has been conducted by replacing base case values with alternative assumptions – no consideration has been given to variation around base case values using credible ranges or confidence intervals. Some key input parameters (such as proportion of responders, exacerbation rates or relative risk of exacerbation with omalizumab add-on therapy) which might be expected to be highly influential on the cost-effectiveness estimates have been omitted from the sensitivity analysis.
ERG Sensitivity Analysis
The ERG presents sensitivity analyses for these parameters in Table 13 of the Evidence Review Group Report (see the "Availability of Companion Documents" field). The ERG used the confidence intervals for the parameters as ranges in the sensitivity analyses. These were taken from the INNOVATE trial data, the manufacturer's calculations on Excel sheet 'AP2-Model Parameters', or calculated using standard confidence interval calculations. The ranges for other parameters were chosen arbitrarily based on reasonable likely ranges.
Comment on Validity of Results Presented with Reference to Methodology Used
In general, the approach taken to modelling cost effectiveness in this patient group seems reasonable. A number of concerns have been raised by the ERG, with respect to the approach to costing omalizumab treatment, the appropriateness of the exacerbation fatality rate used and inadequate consideration of potential biases introduced by missing data. However the overall structure of the model seems reasonable and the significance of the concerns raised by the ERG can partly be determined through careful sensitivity analysis.
Refer to Sections 4.2 to 4.4 for more information on methods used to analyze cost-effectiveness.