|
A Markov model of treatment of newly diagnosed epilepsy in the UK |
Remak E, Hutton J, Price M, Peeters K, Adriaenssen I |
|
|
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 topiramate (TPM) as monotherapy for treating newly diagnosed adult epilepsy patients.
Study population The study population comprised a hypothetical cohort of adult patients with newly diagnosed epilepsy.
Setting The setting was secondary care. The economic study was conducted in the UK.
Dates to which data relate The effectiveness data were derived from literature published from 1995 to 2003. The costs were derived using data from 1994 to 2001. The prices related to 2001.
Source of effectiveness data The effectiveness data were derived from a review and synthesis of completed studies.
Modelling A Markov model was developed to assess the cost-effectiveness of antiepileptic drug combinations for first- and second-line monotherapy for the treatment of adult patients with newly diagnosed epilepsy. The model examined separately the cases of partial and generalised seizures. A cohort of 1,000 patients was examined for 15 years. The health states of the model were "seizure free (SF) on treatment (Tr)", "not SF on Tr", "new Tr" and "death". The cycle length of each health state was 3 months. The starting point was administration of first-line monotherapy.
An initial period of 3 months for each of the first- and second-line treatments was distinguished from the rest of the model (described as maintenance period), as it was likely to differ in terms of treatment changes, resource use and the quality of life of the patients. Patients not responding to first-line treatment (i.e. not SF) were assumed to switch to second-line monotherapy. A specific proportion of SF patients was also assumed to switch to second-line treatment due to intolerable adverse effects. Patients who needed to switch from the second-line treatment were assumed to receive an unspecified form of treatment, with the related costs and outcomes reflecting average values in an epilepsy population. The patients were followed in the model until their death, or until the end of the simulation period.
Outcomes assessed in the review The outcomes assessed in the review included:
the effectiveness during the initial period of treatment and the frequency of 6 side effects (weight gain, weight loss, alopecia, concentration problems, rash, vomiting/nausea) for every antiepileptic drug examined;
the transition probabilities between the health states in the maintenance period of the model, including seizure frequency-specific mortality rates, utilities at each health state (both base-case and side effect-specific, based on discrete choice experiment (DCE)); and
the proportion of patients not responding to treatment who had less than one seizure per month.
Study designs and other criteria for inclusion in the review The effectiveness rates during the initial period were derived from data reported in clinical trials. The long-term outcomes were based on an observational study. It was stated that great care was taken to choose studies with similar designs and patient populations. No more details on the study designs or other criteria for inclusion were reported.
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 Approximately five primary studies were included in the review.
Methods of combining primary studies The results of the primary studies were generally not combined.
Investigation of differences between primary studies The authors did not discuss any investigation of differences between the primary studies included in the review.
Results of the review The reader is referred to the original paper for the model parameter values since there were too many to report in this abstract.
Methods used to derive estimates of effectiveness A UK neurologist, specialising in the treatment of epilepsy, validated the overall model structure and all the assumptions used.
Estimates of effectiveness and key assumptions The following key assumptions were used when developing the model:
the appearance of adverse effects was independent of seizure frequency;
the transition probabilities of the maintenance period were independent of the specific treatment received;
the treatment response was independent of baseline seizure frequency;
the relative risk ratios for overall mortality in epilepsy were the same as those reported for sudden death in epilepsy;
any patient could experience only one side effect at a time;
the side effect rates were independent of the response category;
patients receiving the unspecified treatment experienced the same percentage reduction in seizure frequency as patients who discontinued treatment in one of the studies of the review; and
the utility data for refractory patients were applied to the population of newly diagnosed patients.
Measure of benefits used in the economic analysis The outcome measure used was the number of quality-adjusted life-years (QALYs). In the base-case analysis, utilities were estimated from a long-term UK study collecting EQ-5D scores from patients not adequately controlled with one or more antiepileptic drugs. Values during the initial and maintenance period of the model were derived from the study utilities measured, respectively, 6 months (initial) and 5 years (maintenance) after the start of treatment. The utilities associated with the unspecified form of treatment were derived using mean utilities measured in a group of patients who switched from their study drug for any reason. In order to include side effects in the estimation of benefits, the analyses were repeated using utilities derived from a DCE, designed to evaluate the impact of both seizure frequency and common side effects of antiepileptic drugs. The utility values were obtained from the DCE using a novel time trade-off approach, in which one of the attributes was life expectancy over a defined period. The amount of time patients were willing to forgo was used to calculate the utility values for different side effects. The QALYs were discounted at an annual rate of 1.5%.
Direct costs The study perspective was stated to be that of the UK NHS. The costs consisted of drug costs and annual direct medical costs for every health state (SF on Tr, not SF on Tr, other treatment). They also included the costs of treating adverse effects (rash, and also renal stones found in 1.5% of the patients receiving TPM). During the initial period, the costs of an additional general practitioner (GP) visit and a biochemistry test (in the case of older antiepileptic drugs) were included. Patients receiving the unspecified form of treatment were assumed to acquire, on average, the same costs as those experiencing less than one seizure per month.
The costs and the quantities were reported separately for the drug costs and the costs of treating adverse effects. The costs of GP visits and management of health states were based on studies published in 1994 and 1998, respectively. The drug costs were obtained from the British National Formulary, while the NHS reference costs of 2001 provided the costs of nephrolithotomy. Other published sources were used for the remaining cost estimates. The total costs were derived using modelling. The costs were inflated to 2001 prices using Hospital and Community Health Services inflation indices. Future costs were discounted at 6% annually, as appropriate, since the model simulated long-term costs incurred during 15 years.
Statistical analysis of costs The costs were treated deterministically. No statistical analysis of the costs was performed.
Indirect Costs The indirect costs were not included in the analysis.
Sensitivity analysis A sensitivity analysis was carried out to test the uncertainty surrounding the base-case estimates. A one-way sensitivity analysis was performed to test assumptions on discount rates. A stochastic multivariate sensitivity analysis was also undertaken. This examined variables such as mortality rates, costs, effectiveness data and side effect frequency, with the exception of transition probabilities. The distribution and range of mortality rates was derived from the literature. The distributions and ranges of the other key variables were based on assumptions. The results of the multivariate analysis were used to calculate cost-effectiveness acceptability curves (CEACs) and the cost-effectiveness frontier.
Estimated benefits used in the economic analysis The abbreviations for the treatments are as follows: carbamazepine (CMZ), lamotrigine (LTG), valproate (VPA) and topiramate (TPM).
Partial seizures.
First- and second-line monotherapy with TPM and CMZ was more effective, irrespective of the sequence of treatments used. The base-case effectiveness results ranged from 9.636 to 9.849 QALYs per patient. The pairs of first- and second-line treatment, from the least to the most effective, were LTG-CMZ, CMZ-LTG, LTG-TPM, TPM-LTG, CMZ-TPM, TPM-CMZ. The same ranking was reported for the DCE analysis, with effectiveness ranging from 9.752 to 10.082 QALYs. TPM-CMZ resulted in 0.004 and 0.018 incremental QALYs versus CMZ-TPM in the base-case and DCE analysis, respectively.
Generalised seizures.
The base-case results ranged from 9.862 to 9.988 QALYs per patient. The pairs of treatment, from the least to the most effective, were VPA-TPM, TPM-VPA, VPA-LTG, LTG-VPA, LTG-TPM, TPM-LTG. TPM-LTG resulted in 0.118 additional QALYs versus VPA-LTG. The latter showed 0.004 additional QALYs versus TPM-VPA, which showed 0.004 additional QALYs versus VPA-TPM. The DCE analysis showed changes in ranking, with values ranging from 9.909 to 10.118 QALYs. The order of ranking in this case was LTG-VPA, VPA-LTG, LTG-TPM, TPM-LTG, VPA-TPM, TPM-VPA. TPM-VPA resulted in 0.012 additional QALYs versus VPA-TPM.
The incremental benefits were not presented with respect to dominated (least effective, more costly) options. The benefits were estimated for 15 years, discounted at 1.5%. Adverse effects were account for in the DCE analysis.
Cost results Partial seizures.
Combinations of TPM with CMZ were the least costly. The costs ranged from 11,095 to 12,552 per patient. The ranking of treatments from the least to most costly was CMZ-TPM, TPM-CMZ, CMZ-LTG, LTG-CMZ, TPM-LTG, LTG-TPM. TPM-CMZ incurred an extra cost of 153 compared with CMZ-TPM.
Generalised seizures.
Combinations of VPA with TPM were the least costly. The costs ranged from 11,179 to 12,214 per patient. The ranking of treatments from the least to most costly was VPA-TPM, TPM-VPA, VPA-LTG, TPM-LTG, LTG-VPA, LTG-TPM. TPM-LTG incurred an extra cost of 208 versus VPA-LTG, which resulted in an extra cost of 505 compared with TPM-VPA. The latter (TPM-VPA) incurred an incremental cost of 123 versus VPA-TPM.
The incremental costs were not reported for dominated options. The costs were estimated for 15 years, discounted at 6%. The total costs included the costs of treating adverse effects.
Synthesis of costs and benefits The estimated costs and benefits were combined in the form of incremental cost-effectiveness ratios (ICERs). The ICERs were not calculated in the case of dominated options (less effective, more costly), as this was not necessary.
Partial seizures.
The scenarios LTG-CMZ, CMZ-LTG, LTG-TPM and TPM-LTG were dominated by CMZ-TPM. TPM-CMZ had an ICER of 38,241/QALY (base-case analysis) or 8,498/QALY (DCE analysis) versus CMZ-TPM. A multivariate sensitivity analysis confirmed that combinations of TPM with CMZ were the preferred options.
Generalised seizures.
In the base-case analysis TPM-LTG had an ICER of $1,764/QALY versus VPA-LTG, 6,635/QALY versus VPA-TPM, and dominated LTG-VPA and LTG-TPM. TPM-VPA had an ICER of 30,700/QALY versus VPA-TPM, whereas VPA-LTG had an ICER of 126,276/QALY versus the same comparator. Both strategies were dominated by TPM-LTG by the rule of extended dominance. In the DCE analysis, LTG-VPA, VPA-LTG, LTG-TPM and TPM-LTG were dominated by VPA-TPM. TPM-VPA had an ICER of 10,233/QALY compared with VPA-TPM. A multivariate analysis showed that if society valued additional QALY gains at less than 6,635, then VPA as first-line treatment would have the highest probability of being cost-effective. If the QALY gains were valued higher, then first-line TPM (possibly followed by second-line LTG) would be the strategy most likely to be cost-effective.
Changing the discount rates did not have a substantial affect on the results of any of the analyses.
Authors' conclusions Topiramate (TPM) was a cost-effective treatment for newly diagnosed adult patients with epilepsy. In partial seizures, combinations of TPM with carbamazepine (CMZ) were most cost-effective. In generalised seizures, TPM followed by lamotrigine (LTG) had the most favourable incremental cost-effectiveness ratio (ICER) given currently accepted UK thresholds.
CRD COMMENTARY - Selection of comparators A justification was given for the selection of the comparators used. All the comparators were treatment options recommended by the UK North Clinical Neuroscience Partnership guidelines. You should decide whether these treatment strategies are widely used in your own setting.
Validity of estimate of measure of effectiveness The authors did not state that a systematic review of the literature had been undertaken. However, it was reported that the efficacy data were derived from well-designed randomised controlled trials. Also, that studies with similar designs and patient populations were included in the review. The effectiveness estimates were generally not combined. The impact of differences between the primary studies was not discussed.
Validity of estimate of measure of benefit The estimation of benefits was modelled. The Markov model used was appropriate for this purpose since it incorporated, by using transition probabilities, all possible health states resulting from antiepileptic drug monotherapy, including the consequences of likely treatment switches.
Validity of estimate of costs The study perspective was stated to have been that of the UK NHS. All the categories of cost relevant to the perspective adopted were included in the analysis. The costs and the quantities were reported separately for the drug costs and the costs of treating adverse effects. A multivariate sensitivity analysis, assuming log-normal distribution of the costs, was conducted. Discounting was carried out, as appropriate, since the costs were estimated during 15 years. The date to which the prices referred was reported.
Other issues The authors did not compare their results with those from other studies. However, they stated that this was the first economic assessment of TPM for the treatment of newly diagnosed adult epilepsy patients. The issue of generalisability to other settings was not addressed. The authors reported a number of limitations of their study. These mainly related to the assumptions used in the model structure and with respect to some parameters, especially utility values. Nevertheless, all the assumptions had been validated by expert opinion and were tested in sensitivity analyses. The perspective adopted was also considered to be a limitation of the study. It was believed that a wider, societal perspective would have strengthened the results. The authors seem to have presented their results adequately and their conclusions reflected the scope of the analysis.
Implications of the study Based on their results, the authors emphasised the promising role of TPM as a therapeutic option for patients with newly diagnosed epilepsy. They proposed that their model could provide a framework within which the cost and health gain implications of treatment with newer antiepileptic drugs could be studied. Finally, they suggested that their results from the DCE approach should be confirmed by long-term clinical trials collecting both clinical and cost data.
Bibliographic details Remak E, Hutton J, Price M, Peeters K, Adriaenssen I. A Markov model of treatment of newly diagnosed epilepsy in the UK. European Journal of Health Economics 2003; 4: 271-278 Indexing Status Subject indexing assigned by CRD MeSH Cost-Benefit Analysis; Epilepsy; Great Britain; Humans; Markov Chains AccessionNumber 22004008006 Date bibliographic record published 30/09/2004 Date abstract record published 30/09/2004 |
|
|
|