|Economic evaluation of rivastigmine in patients with Parkinson's disease dementia
|Willan A R, Goeree R, Pullenayegum E M, McBurney C, Blackhouse G
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 rivastigmine, 3 to 12 mg/day over a 24-week period, for the treatment of dementia associated with Parkinson's disease (PD).
The study population comprised patients with PD in whom mild-to-moderate dementia developed at least 2 years after their initial clinical diagnosis. Patients were included if they were over the age of 50 years, had a Mini-Mental State Examination (MMSE) score between 20 and 24, and had contact with a responsible caregiver at least 3 days a week.
The setting was primary care. The economic study was carried out in Canada.
Dates to which data relate
The effectiveness data and resource use data related to a single study (Emre et al. 2004, see 'Other Publications of Related Interest' below for bibliographic details). The costs were measured in 2004 prices.
Source of effectiveness data
The effectiveness data were derived from a single study.
Link between effectiveness and cost data
The costing was carried out retrospectively for the period before the treatment phase and prospectively for the period during the treatment phase. The costing was undertaken on the same patient sample as that used in the effectiveness study.
The authors used a parent study for their clinical data, limited details of which are reported in the current study. Further details are given in the parent study (Emre et al. 2004).
The authors reported that a total of 541 patients were included in the study. The average age was 73 years (range: 50 to 91), 65% were male and all but two were Caucasian. Of the 541 patients, 362 received rivastigmine (of whom 4 died) and 179 received placebo (of whom 7 died). There was no report that power calculations were carried out at the outset, but later calculations indicated that the study had 80% power to detect differences in cost greater than Canadian dollars 5,942. Details of how the initial sample was selected were not included in the current report.
The authors designed a prospective, randomised, double-blind, parallel-group, placebo-controlled study. Randomisation was carried out in a 2:1 ratio. The study was set in 68 centres across 12 countries. The clinical trial had three phases. First there was a 3-week pre-randomisation phase. Then there was a 24-week, double-blind post-randomisation treatment phase comprising 16 weeks of titration and 8 weeks of drug maintenance. Finally, there was a 24-week open label extension phase. Patients who were intolerant to the higher doses used in the trial continued to use a lower dosage. Loss to follow-up occurred in 62 patients (40 in the intervention group and 22 in the control group) including 11 patients who died; data for these patients were extrapolated from that available.
Analysis of effectiveness
The authors did not state whether the analysis was conducted on an intention to treat basis or for treatment completers only. The basis appears to have been treatment completers only as results were reported for 172 patients in the control group and 358 patients in the intervention group. The primary health outcomes were the cognitive sub-scale of the Alzheimer's Disease Assessment Scale (ADAS-cog) and the Alzheimer's Disease Cooperative Study Clinician's Global Impression of Change (ADCS-CGIC). The probability of adverse events was also reported. The authors did not report the comparability of the patient groups at baseline.
In the 70-point ADAS-cog scale, there was an average improvement of 2.1 in rivastigmine patients and an average deterioration of 0.7 in placebo patients, (p<0.001).
The mean score in the 7-point ADCS-CGIC scale at 24 weeks was 3.8 in rivastigmine patients and 4.3 in placebo patients, (p=0.007).
The adverse events experienced were nausea (affecting 29% of rivastigmine patients versus 11.2% of placebo patients; p<0.001), vomiting (16.6% versus 1.7%; p<0.001), and tremor (10.2% versus 3.9%; p=0.01).
The "statistically significant and clinically important" differences led the authors to conclude "rivastigmine can improve care in this study population", compared with placebo
Measure of benefits used in the economic analysis
The summary measures of health benefit were derived from MMSE scores measured at randomisation and scheduled at week 24 of follow-up. The scores were converted into the quality-adjusted survival time (QAST) using utility estimates and a conversion algorithm taken from Jonsson 2003 (see 'Other Publications of Related Interest' below for bibliographic details). Jonsson used data from 272 patients with Alzheimer's disease and measured utility with the EuroQol instrument. QAST was expressed in quality-adjusted life-days (QALDs) and quality-adjusted life-years (QALYs).
The costing was carried out from a societal perspective for the UK and Canada. The costs and resource use were reported separately. Health care resource use was measured retrospectively for the 24 weeks prior to randomisation, but prospectively for the 24 weeks of the treatment phase during the clinical trial. This information, which was provided by the caregiver, covered study drug usage, caregiver time, concomitant drugs relating to PD and dementia, stays in hospital, stays in long-term care facilities, and ambulatory services. The unit costs were taken from national sources within the UK and Canada, and were updated to 2004 prices using the health care component of the Consumer Price Index. Discounting was not required because of the short time horizon of the trial.
Statistical analysis of costs
The authors used a hierarchal model approach with a random effect for country to estimate the between-treatment difference in mean effectiveness (QAST) and mean total cost, along with the corresponding variances and covariance. Cost-effectiveness was measured using a cost-effectiveness ratio which, in turn, was used to generate cost-effectiveness acceptability curves and incremental net benefit.
Caregiver time was estimated as a direct cost. The patient (due to the average age of 73) was unlikely to be involved in economically productive work and, therefore, indirect costs associated with the patient were not relevant to this study.
UK pounds sterling () and Canadian dollars (CAD).
There was no report of sensitivity analyses being carried out.
Estimated benefits used in the economic analysis
The overall treatment difference was 2.81 QALDs greater for rivastigmine patients (standard error 1.85, 90% confidence interval, CI: -0.243 to 5.86; p=0.13).
The authors reported no significant difference in resource use between the treatment arms.
Canadian cost results were as follows:
the total costs with placebo were CAD 31,185 before randomisation, and CAD 30,088 after randomisation (difference -CAD 1,097);
the total costs with rivastigmine were CAD 31,644 before randomisation and CAD 30,608 after randomisation (difference -CAD 1,036);
the difference in costs (rivastigmine minus placebo) was CAD 459 before randomisation and CAD 520 after randomisation (difference CAD 61).
UK cost results were as follows:
the total costs with placebo were 21,797 before randomisation, and 21,138 after randomisation (difference -659);
the total costs with rivastigmine were 21,830 before randomisation and 21,184 after randomisation (difference -646);
the difference in costs (rivastigmine minus placebo) was 32 before randomisation and 45 after randomisation (difference 13).
Synthesis of costs and benefits
The total cost per patient was CAD 55.76 (90% CI: -3,431 to 3,543) in Canada and -26.18 (90% CI: -2,355 to 2,407) in the UK.
The incremental cost per QALY gained with rivastigmine compared with placebo was CAD 7,249 in Canada. Although rivastigmine was more effective and less costly than placebo in the UK (i.e. rivastigmine was the dominant strategy), the authors calculated the incremental cost-effectiveness ratio of rivastigmine as being -3,403.
In Canada, the probabilities of being cost-effective for willingness-to-pay values of CAD 50,000 and CAD 100,000 for health outcomes were, respectively, 0.56 and 0.63. In the UK, the probabilities of being cost-effective for willingness-to-pay values of 20,000 and 40,000 were, respectively, 0.66 and 0.59.
The incremental net benefit for a willingness-to-pay of CAD 50,000 or 20,000 was CAD 329 in Canada and 180 in the UK.
The incremental net benefit for a willingness-to-pay of CAD 100,000 or 40,000 was CAD 714 in Canada and 334 in the UK.
"Consistent with the improvement in clinical outcomes, there was an observed increase in QAST (quality-adjusted survival time) with rivastigmine. However, although no between-treatment differences in costs were seen, the small sample size and highly variable cost distributions prevent us from making strong conclusions with regards to the effect of rivastigmine on total costs and, by inference, on cost effectiveness." The authors also highlighted the lack of variation in the results due to the inclusion of multiple nations in the study.
CRD COMMENTARY - Selection of comparators
The authors compared rivastigmine with placebo, thereby enabling the active value of the treatment of interest to be evaluated.
Validity of estimate of measure of effectiveness
The authors designed a trial with very high internal validity as the study was randomised, prospective and double-blind. These factors help to reduce systematic differences between patient groups and remove other sources of potential bias. Full details of the study sample were not reported but may be available in the parent study. It is therefore not possible to assess whether the sample was truly representative of the study population or whether the groups were comparable at analysis. Appropriate statistical analyses were carried out to explore variability in the effectiveness data.
Validity of estimate of measure of benefit
QALYs were used as a summary measure of health benefit. MMSE scores from the effectiveness trial were combined with published utility data. QALYs provide a measure that is comparable among a broad range of health technologies. The authors acknowledged the limitation of using the caregiver proxy rating for the utilities in the Alzheimer's disease study.
Validity of estimate of costs
The authors estimated costs from the societal perspective, and cost elements relevant to this perspective were covered in the analyses. Data in the 24-week period prior to randomisation were collected retrospectively, so there may be some recall bias in these estimates. Data for the post-randomisation period were collected prospectively. The overall difference in costs was very small between the two treatment alternatives, thus differences in perspective or due to the setting may have an important influence on the results observed and conclusions drawn. The authors provided a detailed breakdown of the unit costs and resource use, enabling a transparent view of the key cost-drivers; this will enhance the generalisability of the authors' results. Statistical analyses including cost-effectiveness acceptability curves give the reader an even greater understanding of the results and the impact of uncertainty. These factors improve the ability to generalise the results to other settings and populations. However, the authors acknowledged that the study was underpowered to detect difference in the costs. The costs were left undiscounted as they were incurred during less than one year. The date to which the prices related was appropriately reported, which will aid any possible inflation exercises.
The authors drew some comparisons with the parent study but did not compare their final cost-effectiveness results more broadly. The issue of generalisability was addressed by the multinational design of the study. The use of price weights from two countries (as mentioned above), using techniques to explore the impact of uncertainty, further improved the generalisability of the study. The results appear to have been presented in full with a detailed breakdown of the cost components. The results and conclusions accurately reflected the scope of the study. Several limitations were discussed. First, the relatively short time horizon, which may have significantly affected the cost-difference. Second, utilities were not measured in the parent effectiveness analysis. Finally, the use of price weights applied to average resource usage in the trial rather than resource use specific to each country.
Implications of the study
The authors did not make any recommendation for policy or practice following on from their study. They suggested the need for a study with a longer time horizon.
Source of funding
Supported by the Natural Sciences and Engineering Research Programme of Canada, the Hospital for Sick Children Foundation Student Scholarship Programme, and Novartis.
Willan A R, Goeree R, Pullenayegum E M, McBurney C, Blackhouse G. Economic evaluation of rivastigmine in patients with Parkinson's disease dementia. PharmacoEconomics 2006; 24(1): 93-106
Other publications of related interest
Emre M, Aasland D, Albanese, A et al. Rivastigmine for dementia associated with Parkinson's disease. N Engl J Med 2004;351:2509-18.
Jonsson L. Economic evaluation of treatments for Alzheimer's disease. Stockholm: Karolinska Institutet; 2003.
Subject indexing assigned by NLM
Aged; Canada; Cost-Benefit Analysis; Dementia /drug therapy /economics; Female; Humans; Male; Middle Aged; Multicenter Studies as Topic; Neuroprotective Agents /economics /therapeutic use; Parkinson Disease; Phenylcarbamates /economics /therapeutic use; Randomized Controlled Trials as Topic; Rivastigmine
Date bibliographic record published
Date abstract record published