|
Cost effectiveness of memantine in Alzheimer's disease: an analysis based on a probabilistic Markov model from a UK perspective |
Jones R W, McCrone P, Guilhaume 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 health intervention studied was memantine (20mg/day), a moderate-affinity uncompetitive NMDA antagonist approved in the EU for the treatment of moderately severe-to-severe Alzheimer's disease.
Economic study type Cost-effectiveness analysis; cost-utility analysis.
Study population The baseline cohort of Alzheimer's patients in the study was divided into 8 subgroups according to severity of disease (Mini-Mental State Examination (MMSE) score: moderate (MMSE greater than 14); moderately severe (MMSE 10-14 inclusive); severe (MMSE less than 10)), physical dependency (Activities of Daily Living (ADL) score) and place of residency. The proportions in each subgroup were drawn from a UK epidemiological study, LASER (Livingston et al 2004 - see (Other Publications of Related Interest( below for bibliographic details). In the largest subgroup (44.2% of patients), patients had severe Alzheimer's disease (MMSE less than 10), were institutionalised and were dependent on a carer. Patients in the next largest subgroup (15.8% of patients) were classified as severe (MMSE less than 10), resident in the community, but dependent on a carer. The third largest subgroup (13.7% of patients) comprised patients classified as moderately severe (MMSE 10-14 inclusive), living in the community and independent. These 3 subgroups contained approximately 73.7% of the patient population at baseline, with the remainder of patients split between the other 5 subgroups. Inclusion/exclusion criteria were not reported.
Setting The setting was the community or institution depending on current health status. The economic study was carried out in the UK.
Dates to which data relate Epidemiological data and resource use information were obtained from the LASER study (Livingston et al 2004 - see (Other Publications of Related Interest( below for bibliographic details) carried out in 2003. Effectiveness data were obtained from a clinical trial published in 2003. Quality of life data came from a study published in 1999.
Source of effectiveness data Effectiveness data were derived from a review of the literature.
Modelling The study took the form of a Markov model used to evaluate the costs and benefits of the treatment strategies. Its purpose was to overcome the short-term horizon of resource utilisation analysis conducted alongside a clinical trial. The model could also take into account that outcomes were dependent on a chronic progressive disease state and low life expectancy among moderate-to-severe Alzheimer's disease patients. The Markov model used 6-month cycles with a time horizon of 2 years.
Outcomes assessed in the review Outcomes assessed from the literature were utilities in dementia and the following probability values: severity transition, dependency transition, institutionalisation transition and mortality. A systematic review was not performed for the model.
Study designs and other criteria for inclusion in the review 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 Three studies provided effectiveness data for the model. The LASER Study was a longitudinal epidemiological study of moderate-to-severe Alzheimer's patients in the UK (see (Other Publications of Related Interest( below for bibliographic details). The Reisberg et al 2003 study was a randomised, double-blind, placebo-controlled clinical trial investigating the efficacy and safety of memantine in 252 moderate-to-severe Alzheimer's patients in the USA (see (Other Publications of Related Interest( below for bibliographic details). The Odense Study (Kronborg Andersen et al 1999 - see (Other Publications of Related Interest( below for bibliographic details) was an epidemiological survey of prevalence and incidence of dementia in Denmark, containing the only published European utility values in dementia.
Methods of combining primary studies The primary studies were not combined.
Investigation of differences between primary studies The authors did not investigate the differences between the primary effectiveness studies nor how these differences affected the estimate of the effectiveness of the technology.
Results of the review The results of the review were as follows:
Severity:
In moderately severe patients receiving memantine, the probabilities of becoming moderate, remaining moderately severe or becoming severe at each cycle were as follows: becoming moderate = 0.20; remaining moderately severe = 0.59; and becoming severe = 0.22. The corresponding values, for patients receiving no pharmacological treatment were: becoming moderate = 0.19; remaining moderately severe = 0.36; and becoming severe = 0.45. In severe patients receiving memantine, the probabilities of becoming moderately severe or remaining severe at each cycle were: becoming moderately severe = 0.11; and remaining severe = 0.89. The corresponding values, for patients receiving no pharmacological treatment were: becoming moderately severe = 0.03; and remaining severe = 0.97.
Dependency:
In moderate patients, the probability of an independent patient remaining so after a cycle was 0.824, and of a dependent patient remaining so was 0.75, for both treatment groups. In moderately severe patients receiving no pharmacological treatment, the probability of an independent patient remaining so after a cycle was 0.688, and of a dependent patient remaining so was 0.909. In severe patients receiving no pharmacological treatment, the probability of an independent patient remaining so after a cycle was 0.594, and of a dependent patient remaining so was 1. The corresponding values for patients treated with memantine were obtained by applying the odds ratio of 3.03 (95% CI: 1.38 - 6.66) to these values.
Institutionalisation:
In patients receiving no pharmacological treatment, the probability of being institutionalised was 0.074 for moderate patients and 0.125 for both moderately severe and severe patients. The corresponding values for patients treated with memantine were obtained by applying the odds ratio of 0.147 (95% CI: 0.017 - 1.290) to these values.
The mortality rate in both arms was 0.032 in moderate patients, 0.071 in moderately severe patients and 0.188 in severe patients.
Mean utility values were estimated per dependency level: independent 0.6511 +/- 0.1981; dependent 0.3207 +/- 0.3084.
Measure of benefits used in the economic analysis The outcomes used in the analysis were years of independence, years resident in the community, and quality-adjusted life-years (QALYs). In order to value health states, 244 Danish patients with dementia were interviewed. Questions were mapped onto the EQ-5D questionnaire.
Direct costs Discounting was applied to costs at the rate of 3.5% per annum during the second year of the model. Costs and quantities were presented separately. Costs were estimated per setting and dependency level, assuming that any influence of severity on costs would be taken into account via these variables. Cost estimates for patients in the community were based on direct medical and non-medical resources and included hospitalisations, consultations, paramedical services, social services, community care centres and provision of home equipment. For institutionalised patients, only hospitalisation and institutionalisation costs were included. Estimation of quantities was based on resource use data collected in the 2004 LASER study (see (Other Publications of Related Interest( below for bibliographic details). Unit costs were derived from 2000 standard tariffs and inflated to 2003.
Statistical analysis of costs Costs entered the economic model as empirical distributions, from which a value was sampled for each model simulation.
Indirect Costs Indirect costs were not included in the analysis.
Sensitivity analysis The stochastic model was constructed using a priori distributions for the main variables, and the base case analysis gave descriptive statistics for the costs and outcomes following 10,000 Monte Carlo simulations. Thus, a probabilistic sensitivity analysis was constructed. Authors investigated a worst-case scenario whereby patients received treatment for 2 years but memantine was effective only for 6 months. One-way sensitivity analyses of the key parameters (independence and institutionalisation) were conducted using the confidence intervals of the odds ratios for the effect of memantine versus no pharmacological treatment.
Estimated benefits used in the economic analysis Clinical effectiveness was greater for memantine than no pharmacological treatment.
Memantine patients gained a mean 0.10 years (SD: 0.04) of independence: mean 0.37 years (SD: 0.06; 95% CI: 0.27-0.49) versus 0.26 years (SD: 0.03; 95% CI: 0.20-0.33) for patients given memantine and no pharmacological treatment, respectively.
Memantine patients gained a mean 0.06 years (SD: 0.04) in the community: mean 0.54 years (SD: 0.05; 95% CI: 0.42-0.63) versus 0.47 years (SD: 0.07; 95% CI: 0.33-0.59) for patients given memantine and no pharmacological treatment, respectively.
Memantine patients gained a mean 0.04 incremental QALYs: 0.61 QALYs (SD: 0.30; 95% CI: 0.16-1.24) versus 0.57 QALYs (SD: 0.31; 95% CI: 0.11-1.24) for patients given memantine and no pharmacological treatment, respectively.
A discount rate of 3.5% for health benefits was applied.
The efficacy of memantine was applied to only the first 12 months of the model, being mainly drawn from a 6-month clinical trial, which was succeeded by a 6-month open-label study.
Cost results The mean total cost per patient treated with memantine was 60,516 (SD: 45,908; 95% CI: 22,972-149,403) and the mean total cost per patient treated with no pharmacological intervention was 62,479 (SD: 48,989; 95% CI: 21,981-155,368), resulting in a cost reduction of 1960 (SD: 4,504) per patient over 2 years.
Synthesis of costs and benefits Estimated costs and benefits were evaluated to find the incremental cost per year of independence gained, the incremental cost per year in the community gained, and the incremental cost per QALY gained of memantine compared to no pharmacological treatment. The results showed that memantine was the dominant intervention in all 3 cases as it was associated with lower costs and improved outcomes over the 2-year model period. In the worst-case sensitivity scenario, memantine was still the dominant alternative. The effects of memantine on QALYs and costs were sensitive to changes in effect on dependency but memantine remained dominant at the lowest limit of the confidence interval. Costs were sensitive to a change in effect of memantine on institutionalisation: an assumption that memantine had no effect increased the incremental cost per QALY gained to 20,000.
Authors' conclusions The authors concluded that, based on their model, memantine provides increased duration of independence, delayed time to institutionalisation and increased QALYs. These benefits translate in to reduced utilisation of healthcare resources and hence, cost savings.
CRD COMMENTARY - Selection of comparators The reason for the choice of comparator (placebo) was not clear, and nor did the authors provide a justification for the choice. Implicitly, current practice does not include pharmacological treatment in moderately severe-to-severe Alzheimer's patients. You should decide if this reflects practice in your own setting.
Validity of estimate of measure of effectiveness Effectiveness data came mainly from a single randomised, placebo-controlled clinical trial based in the USA (the economic model was set in the UK). Details of the study were not reported in the present paper; therefore it is difficult to discuss the validity of the sample selection or the internal validity of the study. Data from an epidemiological study conducted in southeast England were also included. It was not stated why a systematic review for sources of clinical effectiveness data was not conducted. The authors used data from the available studies selectively and did not combine them. The methods used to derive estimates of effectiveness were reported and had face-validity. Treatment benefits were incorporated as stochastic variables with a priori distributions, thus tending to reinforce the robustness of the results.
Validity of estimate of measure of benefit QALYs were estimated via utility values obtained from Danish Alzheimer's patients. The authors compared the values used with those obtained in a UK study, but it is not clear why the UK values were not employed in the model.
Validity of estimate of costs All categories of cost relevant to the perspective (NHS and PSS) were included in the analysis. It is not clear why older reflated prices were used rather than up-to-date costs. Costs and quantities were reported separately, enhancing the generalisability of the study. Quantities were estimated alongside a UK epidemiological study, which also fed into the population distribution of the model. This increases the internal validity of the model. A conservative assumption was made to the effect that, the efficacy of memantine endured for 12 months, while the cost of memantine treatment was applied to the full 2 years of the model. This suggests that the results may have been biased against memantine treatment. Costs were incorporated as stochastic variables, thus tending to reinforce the validity of the results.
Other issues The authors made appropriate comparisons of their findings with other studies, emphasising the recently increasing importance of decline in functional abilities, which is included in the model as the dependency variable. They noted that the model was based on recent UK cost and epidemiological data, increasing external validity and relevance to the setting of interest. It was implied that the study could be most easily generalisable to the USA. The stochastic approach took account of parameter variability while sensitivity analyses confirmed the internal validity of the model. The authors did not present their results selectively. The authors acknowledged limitations such as imprecision, combining data from various sources, model assumptions, and lack of internal validity and generalisability of results.
Implications of the study The authors suggest that, from the NHS perspective, treatment with memantine should free up resources available for the care of Alzheimer's patients. They highlight the need for further studies into the utility values of Alzheimer's patients, particularly those with severe disease.
Source of funding The funding for this project was provided by Lundbeck A/S and Merz Pharmaceuticals GmbH.
Bibliographic details Jones R W, McCrone P, Guilhaume C. Cost effectiveness of memantine in Alzheimer's disease: an analysis based on a probabilistic Markov model from a UK perspective. Drugs and Aging 2004; 21(9): 607-620 Other publications of related interest Kronborg Andersen C, Sogaard J, Hansen E, et al. The cost of dementia in Denmark: the Odense Study. Dementia and Geriatric Cognitive Disorders 1999;10(4):295-304.
Livingston G, Katona C, Roch B. A dependency model for patients with Alzheimer's disease: its validation and relationship to the costs of care: the LASER-AD Study. Current Medical Research and Opinion 2004;57:1007-16.
Reisberg B, Doody R, Stoffler A. et al. Memantine in moderate-to-severe Alzheimer's disease. New England Journal of Medicine 2003;348(14):1333-41.
Indexing Status Subject indexing assigned by NLM MeSH Aged; Aged, 80 and over; Alzheimer Disease /drug therapy /economics; Cost-Benefit Analysis; Excitatory Amino Acid Antagonists /economics /therapeutic use; Humans; Markov Chains; Memantine /economics /therapeutic use; Monte Carlo Method; Psychiatric Status Rating Scales; Quality-Adjusted Life Years; Randomized Controlled Trials as Topic AccessionNumber 22004001122 Date bibliographic record published 31/05/2005 Date abstract record published 31/05/2005 |
|
|
|