|
Cost-utility analysis of vagus nerve stimulators for adults with medically refractory epilepsy |
Forbes R B, MacDonald S, Eljamel S, Roberts R C |
|
|
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 vagus nerve stimulators (VNS) for the treatment of medically refractory epilepsy (MRE). The comparator was not explicitly stated, but it appears to have been no implantation of the device.
Type of intervention Treatment and palliative care.
Study population The target population comprised patients with MRE. No further details were reported.
Setting The authors did not specifically state the setting, but it is likely to have been tertiary care. The economic evaluation was conducted in Northern Ireland.
Dates to which data relate The effectiveness and resource use data were derived from studies published from 1994 to 1999. The price year was 1996.
Source of effectiveness data The effectiveness data were derived from a review of published studies, augmented by authors' assumptions.
Modelling A model with a 5-year time horizon was used to derive the main outcomes of the study. The type of model used was not reported.
Outcomes assessed in the review Input parameters that fed the model were the number-needed-to-treat (NNT) to have a 50% or greater reduction in seizure frequency, device removal rate, and infection risk.
Study designs and other criteria for inclusion in the review The NNT was derived from a meta-analysis of randomised controlled trials (RCTs) identified by the review. The device removal rate and the infection rate were stated to have been taken from the published literature, although references were not cited.
Sources searched to identify primary studies MEDLINE, EMBASE and the Cochrane Controlled Trials Register were searched for studies of effectiveness data. Abstracts from a symposium were also handsearched.
Criteria used to ensure the validity of primary studies Only RCTs were selected for ascertaining the efficacy of VNS devices. It was not reported how the other studies used were selected.
Methods used to judge relevance and validity, and for extracting data Number of primary studies included Two studies were included in the review for effectiveness. One study was included for the estimation of the quality-adjusted life-year (QALY).
Methods of combining primary studies The NNT was derived from a meta-analysis. The QALY was estimated using a narrative method, with supplemental information obtained from patients.
Investigation of differences between primary studies Results of the review The NNT to have a 50% or greater reduction in seizure frequency was 6 (95% confidence interval, CI: 3 - 14).
The device removal rate was 2.7%.
The risk of infection was 1.1%.
The QALYs gained per implant were 0.285.
Methods used to derive estimates of effectiveness The authors made assumptions to derive estimates of effectiveness.
Estimates of effectiveness and key assumptions Some estimates of effectiveness (side effects as infection or removal rates) were stated to have come from published literature, although no references were cited. It was assumed that the battery life of the device was only 5 years, and that infections would require 7 days' treatment in an inpatient setting with intravenous antibiotics.
Measure of benefits used in the economic analysis The measure of benefits used in the economic analysis was the QALYs. The authors searched the literature (MEDLINE and EMBASE) and reported the results of a published study carried out on people with epilepsy using the time-trade off method (TTO), which they used in the sensitivity analysis. They decide to elicit values from 42 patients in their setting, both through the TTO and the EQ-5D. The TTO experiment was extremely difficult in this population, and only 7 patients behaved as described in the literature. The value that those patients gave to a 50% reduction of seizure frequency was used as the estimate of QALY gained. The UK social tariffs were used to obtain utility values from the EQ-5D. Discounting was appropriate for the time horizon chosen, and the QALYs were discounted at a rate of 2%.
Direct costs The costs included were the cost per implant, cost per day of inpatient stay, length of inpatient stay, cost per unit of theatre time, cost per infected device, life expectancy of implant battery, and health care costs averted by the VNS. Discounting was appropriate for the 5-year time horizon chosen and a rate of 6% was used. The quantity/cost boundary was not stated by the authors, but it appears to have been that of the hospital. The estimations of the quantities and costs were derived using modelling. The quantity and the cost data came from 7 studies identified from the literature, and from the Scottish Healthcare Purchasing Information Centre. The price year was 1996 and reflation or deflation was carried out according to the source year. Purchasing Power Parity (PPP) calculations were used to convert US values to the UK ($1 = 0.641). Data from other countries were converted to 1996 US dollars and then to UK pounds sterling through PPP.
Statistical analysis of costs The costs were treated as deterministic estimates in the model.
Indirect Costs The indirect costs were not considered.
Sensitivity analysis A sensitivity analysis was performed to assess the impact of model assumptions on the final cost per QALY. Eight scenarios were evaluated, which varied one estimate at a time. The variables were the day-case implantation costs, two different costs averted from VNS, EQ-5D valuations, two different estimations of valuations from the literature, complication rates, and targeting treatment to most likely responders. The estimates were taken from the literature or based on assumptions.
Estimated benefits used in the economic analysis The base-case analysis reported that the QALY gain per patient was 0.285 per annum. This was assumed to be constant during the 5-year time horizon of the model. Side effects were only considered for the cost side of the study.
Cost results The authors did not report the total costs of the intervention and the comparator. They were report only in a disaggregated fashion. The cost per implant was 5,500, the cost per day of inpatient stay was 315, the cost per unit of theatre time was 395, and the cost per infected device was 3,100. The health care costs averted by the improved control of VNS were 745.
Synthesis of costs and benefits For a programme of six VNS implants, the cost per QALY estimate was 28,950. Using the 95% CI for the NNT, the range of possible values lay between 13,000 and 71,000 per QALY.
In the scenarios presented in the sensitivity analysis, the values ranged from 4,785 per QALY for a model that assumed the device could be targeted to those with 33% chance of response, to 63,245 per QALY when unfavourable values for health gains were used. Life expectancy of the device, as well as the device costs, could have significant impact on the results. The cost per QALY was not sensitive to the length of stay or complication rates.
Authors' conclusions There is no strong economic argument against a programme of vagus nerve stimulator (VNS) implantation, although care should be taken to try to identify and treat those most likely to benefit. Therefore, the decision to implant should be made primarily on clinical and efficacy grounds.
CRD COMMENTARY - Selection of comparators The comparator strategy was not clearly stated or justified. It appears to have been no device implantation and, as such, represented current practice in the authors' setting. The authors reported at least one strategy with comparable efficacy (add-on anticonvulsants) in the discussion, but they did not include this in the comparative analysis. You should decide if the comparator represents current practice in your own setting.
Validity of estimate of measure of effectiveness The estimates of effectiveness were taken from a systematic review of the literature. The methods and conduct of the review were partially reported. However, to fully assess the validity more detail would be required. The authors depicted the meta-analysis graphically and stated that the results of the meta-analysis were used in the model, but no details were reported. It was unclear if the authors had considered the impact of differences between the studies when estimating the effectiveness. Although the implicit comparator was no device implantation, the two RCTs included in the meta-analysis compared high-frequency versus low-frequency VNS. The authors did not state if low-frequency VNS has a similar effect to no VNS implant. You should decide if this assumption is reasonable in your own setting.
Other methods to derive estimates of effectiveness were based on authors' assumptions, made on the basis of the literature. A sensitivity analysis was conducted to evaluate the robustness of the results to estimates of effectiveness. However, due to the limited details reported, it was difficult to determine the internal validity of the effectiveness evidence.
Validity of estimate of measure of benefit The estimation of benefit was modelled using a 5-year time horizon. The authors stated that the estimate of the QALY gain was based on seven usable responses to a TTO exercise from 42 patients who agreed to undertake it. To test the robustness of the results, the authors evaluated other scenarios in the sensitivity analysis with utilities taken from the literature, or from the patient EQ-5D values. Using different values had a big impact on the estimated cost per QALY. Given the low numbers that produced usable responses, and the fact that the authors identified that different values had such a large impact, the robustness of the results obtained is questionable.
Validity of estimate of costs The authors did not report the cost perspective. Some categories of costs that were not stated to have been included were drug treatment, physician and hospital visits, laboratory tests and other resource use, and indirect costs. It is possible that some of these were included in the estimation of the annual costs averted per person by improved seizure control, taken from seven published studies, but it is difficult to be certain of this. In most of these studies drug therapy was evaluated, and the benefit of the VNS was assumed to be similar in cost terms for persons with more active versus less active epilepsy. As the comparator was not described in detail, it is not clear which of the cost categories were common to both and would not, therefore, alter the incremental results.
The costs and the quantities were, in part, reported separately. The resources used were taken from published literature and based on assumptions. They were tested in the sensitivity analysis under different scenarios, which makes it easier to generalise to other settings. The prices were taken from published sources and from the authors' setting, and a sensitivity analysis was conducted on the unit costs. The authors applied differential discounting to the costs (6%) and health outcomes (2%), but did not present undiscounted results. Prices from non-UK countries were converted to US dollars and then converted to UK pounds sterling using PPP calculations. The price year was reported.
Other issues The authors compared their cost per QALY results with other studies in the epilepsy field. However, the issue of generalisability was not specifically addressed.
Implications of the study The authors stated that VNS have become an acceptable treatment for patients with refractory epilepsy, and that their economic model demonstrates that the economic argument against a programme of implantation is weak. The decision to implant should be based primarily on clinical and efficacy grounds. The authors also stated that there remains concern that the estimates of benefit conducted during the study using the TTO could not be completed by the majority of the patients, and when the estimates derived from the EQ-5D were used the cost per QALY almost doubled.
Bibliographic details Forbes R B, MacDonald S, Eljamel S, Roberts R C. Cost-utility analysis of vagus nerve stimulators for adults with medically refractory epilepsy. Seizure 2003; 12(5): 249-256 Indexing Status Subject indexing assigned by NLM MeSH Cost-Benefit Analysis; Electric Stimulation Therapy /economics /instrumentation; Epilepsy /economics /physiopathology /therapy; Humans; Prostheses and Implants /economics; Vagus Nerve /physiopathology AccessionNumber 22003001011 Date bibliographic record published 30/04/2004 Date abstract record published 30/04/2004 |
|
|
|