Ideally, in cost-effectiveness analysis, the benefits of treatment should be defined as broadly as possible so that comparisons can be made among the variety of treatments and services in epilepsy, and among treatments in epilepsy and other diseases. The value of conducting cost-effectiveness analysis systematically is that it permits the analyst to assess efficiency of different treatments within and across diseases. In addition, precise information on medical resource utilization associated with each treatment also is needed. Unfortunately, because epilepsy is a diverse and complex condition, the relevant outcomes used to assess the benefits of treatment continue to be debated (see the Langfitt and Wiebe article in this issue). Furthermore, precise data on resource utilization and treatment effects have generally been lacking, so most studies are conducted for hypothetical cohorts of patients based on data derived from clinical trials and expert panels.
Epilepsy treatment outcomes are typically described in terms of improvements in specific clinical indices, such as seizure frequency and/or severity. There is some dispute about which measure to use (7). The main argument for these clinical measures is their relative ease of measurement and interpretation. However, they fail to reflect “quality-of-life” outcomes that are associated with the treatment of epilepsy. Trials of monotherapy show that ≤90% of patients treated with AEDs will experience adverse symptoms related to their medication (8). These side effects can vary from mild feelings of fatigue to life-threatening hypersensitivity reactions. AEDs may also have adverse psychological and/or cognitive effects. Treatments sometimes fail to achieve adequate seizure control to offset these adverse effects, leading patients to discontinue the treatment.
The quality-of-life (QOL) outcomes of treatment are significant to patients and physicians, but are difficult to measure. A variety of generic and epilepsy-specific scales have been developed to describe different health states in epilepsy (9). However, there is no consensus on which is most appropriate for general use. Furthermore, to reflect the patient's preferences, the QOL scales must be weighted by the value people place on different health states. This requires deriving utilities for different health states, using a rating scale, standard gamble, or time trade-off method. For example, the time trade-off method involves patients stating how much of their life expectancy with their current state of health they would be willing to give up to live the remaining years in excellent health. The resulting responses are converted into a health utility index ranging from 0 (worst response) to 1 (best response, unwilling to give up any time). The index can then be used to adjust expected years of survival in a given health state by the utility of that health state to derive QALYs or HYEs.
Another requirement of an economic evaluation is an accurate and complete assessment of both the short-term and long-term net cost of a considered treatment. The net cost of a new therapy, for example, should include the acquisition cost of the treatment, the cost of resources used in the management of adverse side effects of treatment, the cost of therapeutic switching if the treatment fails, and the cost increases or saving from current and future outcome differences, if such differences occur (see above equation). Direct cost and saving arise in relation to the treatment of an illness. Health care resources generate direct cost, such as payments for outpatient visits to physicians, pharmaceuticals, hospital admissions, and laboratory investigations. Social service cost can also arise from treatment including such items as payments for transportation, childcare, and respite care for a family caregiver.
Cost analyses in cost-effectiveness studies should attempt to measure the long-term incremental differences in health care, social service, and patient and caregiver cost attributed to a given treatment versus its comparator (10). Relevant health care cost items for epilepsy include AEDs (drug cost and pharmacy fee), hospitalization, emergency room services, physician visits (inpatient and outpatient), diagnostic procedures, surgery, diagnostic tests, and laboratory tests. Relevant social service items may include social worker services, psychologist–psychiatrist visits, day care, remedial schooling, and residential care. Patient and caregiver cost measures may include time spent seeking care, time spent undergoing care, child care cost for parents during care, transportation expense to access care, time family members spend in home care, cost for respite care, cost of safety equipment, special furniture, etc. Some of the direct cost measures may be omitted if they are known to be relatively small or do not differ significantly across the treatment regimens being compared.
Direct cost and saving are measured in terms of the monetary value of resources used. Cost refers to the economic value of resources consumed as a consequence of treatment. Economists measure such resources in terms of “opportunity costs:” the value those resources would have generated in their next best alternative use. In most countries, health care and social services are exchanged for money; for example when pharmaceuticals are bought from pharmaceutical companies or when hospitals employ staff and purchase equipment to provide health care. Economic evaluations usually rely on the payments made for these goods and services when calculating the cost of treatment, assuming that, in the absence of information on the cost of production, payments reflect the social opportunity cost of the resources.
The effect of epilepsy and its treatment on patients' time and productivity are described as indirect cost, and this cost is usually valued in terms of lost productivity in the workplace. Epilepsy can reduce productivity if patients perform less well at their jobs, have to take time off because of their seizures, or become unemployed. Premature mortality can also be considered in terms of the time that would have been devoted to work or leisure that is lost. Epilepsy treatments may reduce the indirect cost that accrues both to the individual and society at large.
Productivity cost for individuals is estimated in terms of their age and sex-group average value of lost earnings. This process has its basis in the human capital method (11), which values a person's time in terms of expected earnings. This method has also been applied to value the effects of temporary illness on an individual's ability to do the job. For example, the cost of missing 5 days work through epilepsy would be valued as the equivalent of the loss of 5 days' average earnings for that individual worker.
Indirect cost is more difficult to value when the time lost is not paid by wages. For example, time devoted to housework and childcare are important to the individuals concerned and also to the general economy, but such work is rarely paid for in a financial transaction. In the developing world, large numbers of people in rural areas perform work that is unwaged. In these cases, “shadow” pay rates can be used, based on the prices charged for hiring workers to perform similar jobs. Although the human capital method has been shown to have theoretical flaws and raise ethical issues (5), it is relatively easy to apply and has been used to estimate the economic impact of epilepsy in the workplace in cost-of-illness studies in epilepsy (12) and in some economic evaluations of AEDs (13).
Intangible savings and costs arise because of the sense of well-being associated with improved health or vice versa. They are described as intangible because they are difficult to measure. In a chronic condition such as epilepsy, which is often socially stigmatized, this cost is likely to be significant. Unfortunately, no method of assessing the financial value of improved health and well-being in epilepsy has been developed. Contingent valuation methods such as willingness-to-pay for a treatment or outcome can be used to estimate such cost from patients' perspectives, but these techniques have not been used widely in health care.
In economic evaluations of epilepsy treatments using the cost–utility model (discussed in the prior section), indirect and intangible cost is theoretically reflected in the QALY outcome measure in the denominator. Consequently, an advantage of using this model is that net cost calculations in the numerator may be limited to the direct cost of an intervention, avoiding the problems with having to estimate productivity effects and intangibles (14).
The cost of resources used in treatment depend on the perspective that is considered. In most developed countries, patients rarely pay directly for treatments. Individuals most often receive health care that is paid for by national insurance schemes or private insurance companies. Consequently, from the perspective of the patient, health care cost, at the time of consumption, is minimal, other than any user charges or copayments that are required. In this case, it is most useful to perform economic analysis from a societal perspective considering all costs and benefits of a treatment regardless of the source of payment or who benefits. This may not be the case in poorer countries. People living in less wealthy countries are often expected to pay for most or all of any antiepileptic treatment they receive. Those performing economic evaluations must select a perspective that is most appropriate for the decision-makers targeted by the analysis. However, whenever possible, the societal perspective should also be adopted in the analysis so that results will be comparable across studies.
A preferred method for economic assessment in epilepsy would be to collect cost data alongside prospective clinical trials. The benefits of this approach are that results would reflect actual data from real patients, and the internal validity of the data could be assessed through controlled clinical conditions. The disadvantages of this approach are that the treatment pathways and patient samples may not be generalizable to the larger population. Naturalistic studies (retrospective or prospective) provide a more realistic alternative by using patient survey to obtain outcome data and medical claims to assess cost. The strength of this approach lies in the representativeness of patients, treatments, and treatment settings. However, there may be internal validity problems due to selection bias and other confounders. The most common approach used in epilepsy research to date is to model the costs and outcomes of different treatments. This involves reviewing clinical data from drug trials to assess outcome differences and developing cost estimates using clinical data from drug trials, expert opinion, and/or data from existing literature (15).
Acceptable approaches to cost measurement vary by the degree of precision required. The most accurate is microcosting, and involves specific enumeration and summation of payment for inputs (i.e., personnel, facility, supplies) used in providing health care and social services that are consumed in treatment. Because input data are generally not available, gross costing methods are typically used in empiric cost studies in which payment for more global units of service (i.e., the average payment for a physician visit or a day in the hospital) are documented.