Knowing the cost of epilepsy increases awareness of its burden on individuals and society, who bear the burden, and the potential benefits of prevention and treatment. Calculation of the cost of epilepsy involves applying economic cost-of-illness methods to estimate the different types of cost experienced by individuals, their families, third-party payers, health providers, and society at large. Direct cost includes the cost of resources consumed when health care, social services, and patient or family member services are used to prevent, diagnose, treat, or rehabilitate persons with epilepsy. Indirect cost is the cost of unemployment, decreased productivity, and household work that is lost when people with epilepsy are less able to do their jobs or to work at home. Intangible cost is the economic value of pain and suffering that individuals experience with the disorder. Reviews of methods of estimating the cost of epilepsy have appeared elsewhere in the literature (1–4). We provide a brief description of general methods, discuss current methodologic issues, review the latest findings, and make recommendations for future research.
Summary: Economic concerns are increasingly important in health system design, provider payment, and research funding decisions. Cost estimates are needed to provide insight into where the potential opportunities for cost-savings lie and to lay the groundwork for research to determine how to treat the disorder more effectively. The methods used by economists to estimate the direct and indirect costs of epilepsy are reviewed and results from studies in different countries are discussed. General patterns across patients with different types of conditions are reported. Gaps in the literature are identified and future research needs are assessed.
Epilepsy-specific direct costs are defined as the monetary value of resources consumed in the prevention, treatment, or rehabilitation of people with the disorder. These costs are assumed to reflect the opportunity cost of the resources (e.g., their value if used in the next most efficient way) that are used or lost as a result of the disorder. The costs of medical services are usually estimated according to the average payment made to providers of health care. Direct costs vary depending on the perspective of the study. In countries where health care and social services are paid by the government, the cost of epilepsy treatment may be very low from the patient's perspective. Alternative perspectives include those of costs incurred by employers or health insurance plans that purchase health care. Most cost-of-illness studies adopt a societal perspective that attempts to capture full costs without regard to which individuals or entities are incurring the costs.
Two broad approaches are used in generating estimates of the direct cost of epilepsy: “bottom-up” and “top-down”(5). In the former, cost estimates of the number and types of health care, social services, and family member resources used by individuals with epilepsy are derived from prospective or retrospective studies of real cases, or, alternatively, based on hypothetical information provided by expert panels and related literature. In the top-down type of study, estimates are made of the cost of services for all illnesses based on nationally representative provider surveys, and a portion is attributed to a specific illness. These studies use standard methods (disease-specific diagnosis coding) to classify health resource use for an illness
The top-down approach is desirable for high-prevalence illnesses that are well represented in national surveys, and when there is a lack of data from representative samples of patients with a specific illness. In using population surveys, the top-down approach may produce estimates that better reflect the magnitude and distribution of cost in the general population. However, this approach yields rough approximations of specific cost and is incompatible with the stratification of cost by patient or disease characteristics. The bottom-up approach is used for more precise estimates and allows examination of the variation in cost across individuals. In relying on patient samples, it may not yield results that can be generalized.
Indirect and intangible cost
The usual method for estimating indirect, or productivity-related, cost, referred to as the “human capital” approach, values people in terms of their economic capacity. Epilepsy-related indirect cost is estimated in terms of lost earnings and/or the imputed value of lost household work associated with morbidity and mortality. The intangible cost of pain, suffering, loss of self-esteem, and from emotional issues for those afflicted and their loved ones is generally acknowledged, but rarely measured, because of the difficulty of estimating its monetary value. An alternative to the human capital approach, called “willingness to pay,” considers both indirect and intangible cost by estimating the amount people would be willing to pay to avoid or to reduce the probability of the illness. Although it is attractive from a theoretical perspective, the method is rarely used because of the practical limitations of generating required data. Cost-of-illness studies typically adopt the human capital approach because it is a well-recognized framework with standardized methods for application.
Time and population perspectives
Cost studies vary in terms of their time perspective and the population considered. Prevalence-based cost studies estimate the cost over a short period (usually a year) for all prevalent cases, regardless of the evolutionary stage of the disorder. The short-term nature of this type of study makes it most appropriate for estimating the cost saving of a new technology or program that has short-term benefits for all patients or when comparing different categories of patients presumed to consume different amounts of the same resources. Incidence-based studies, on the other hand, consider the cost of an illness for new cases from the time of diagnosis to a definite end point, such as cure or death. This perspective is more appropriate when evaluating the cost saving of interventions that prevent new cases, or prevent or ameliorate the future course of a disease. The incidence-based perspective is more difficult because it is based on a longitudinal analysis of individuals, taking into account the temporal aspects of epilepsy.
One should recognize that estimating the cost of epilepsy is only a first step toward the economic evaluation of treatment for the disorder (see companion article by Heaney and Begley in this issue). Whereas a cost study can provide a picture of the overall burden of epilepsy and can serve to educate and indirectly to inform public policy, it does not provide information about the effectiveness of interventions. One also must know what can be done to prevent and/or treat persons with the disorder and the amount of resources required. Because the incidence-based approach provides base information in evaluating the potential for saving from prevention of cases or ameliorating morbidity, it is a larger step toward an economic evaluation of an intervention than are prevalence studies.
The economic burden of a disease depends on the definition of the population with disease and its expected frequency. Epilepsy is defined by the occurrence of repeated unprovoked seizures (6). This definition requires exclusion of patients with single unprovoked seizures and those with provoked seizures (i.e., seizures occurring in close proximity to an insult to the central nervous system generally thought to be epileptogenic). In addition, an undefined number of individuals have one or more episodes later interpreted as syncope, psychogenic seizures, or other events entering the differential diagnosis with epilepsy.
Studies on the cost of epilepsy must define the target population precisely. The International League Against Epilepsy (ILAE) has developed a simplified classification system of epilepsy for use in epidemiologic and economic studies (6). Many of the definitions remain controversial, and different definitions lead to different results, with more inclusive definitions leading to higher total costs and lower average costs. For example, prevalence studies are typically based on the recruitment of patients with “active” epilepsy, defined by the ILAE classification as patients with at least one seizure in the last 5 years. Such studies yield biased cost estimates because, even when cases are drawn from general populations, they overrepresent people with problems requiring medical attention. A further overestimate of cost is then expected when referral patients are considered as prevalent cases.
The more inclusive definition of single unprovoked seizure has been used in incidence-based studies to document the high health care cost associated with diagnosing a case and determining initial treatment (7,8). When this is done, it is recommended that the cost of single-seizure cases be reported separately from those who ultimately receive an epilepsy diagnosis.
Clinical trials represent a selected setting for the measurement of the cost of epilepsy, and generally, cost studies based on cases in clinical trials should not be the basis for population projections. The study population is mostly represented by patients with drug-resistant epilepsy or other selected categories (e.g., surgical candidates) and only occasionally by newly diagnosed individuals. Accordingly, the number and types of medical contacts tend to vary significantly, depending on the characteristics and timing during the course of the disease. Based on the requirements from regulatory authorities for the conduct of a trial, the number and types of consultations and services used are typically greater that what is usually done in clinical practice. Visits and laboratory tests tend to be in excess for a better assessment of the efficacy and safety of the investigational treatments and to improve patient compliance. This being said, lacking more representative data, clinical trials provide an opportunity for the measurement of the cost of epilepsy attributable to a given treatment in a selected patient population.
Temporal stage of the disorder
Epilepsy is a heterogeneous clinical condition with variable severity. The severity of the disorder, which may be the major source of its cost, is measured by its response to treatment. About 50% of patients with newly diagnosed epilepsy achieve seizure control after treatment start; 20% enter remission after one or more treatment changes; and about 30% tend to present occasional or frequent seizures despite optimal treatment (8,9). With reference to the long-term prognosis and the response to treatment, patients with epilepsy can be grouped in different prognostic categories, whose distribution may differ in children and adolescents and in adults (10).
The cost of epilepsy may vary depending on the time of assessment of the patient during the course of the disorder. Cost is highest, although not necessarily homogeneous, at the time of the diagnosis, depending on the type and extent of the diagnostic assessment, and tends to decrease thereafter. After diagnosis, cost tends to vary according to the response to drug treatment and, in surgical candidates, to surgical treatment (10). In patients unsatisfactorily controlled with conventional antiepileptic drugs (AEDs), new compounds are used, which may represent a significant increase in direct cost (11,12). For these reasons, it is preferable for cost studies to be conducted in newly diagnosed patients identified from the community and monitored prospectively. Retrospective studies and prevalence-based studies should be carefully designed to reflect general populations of persons with epilepsy and to minimize the selection bias associated with long-term prognosis.
What cost items should be counted
To estimate the full costs of the disorder, include a comprehensive set of cost items. There is general agreement that direct-cost studies should include, at a minimum, all well-known epilepsy-related items of health care consumption, such as physicians' visits, hospitalizations, diagnostic investigations, laboratory tests, and treatments (AEDs, surgical procedures, etc.). Non–health care items such as transportation cost, informal nursing and social services provided in institutions or at home, and unpaid care or services provided by patients' relatives also should be documented whenever possible. In practice, however, studies are not able to capture all the relevant components of cost. It is necessary to set priorities, with a focus on those items that are most likely to have a significant effect on costs. Researchers should justify the items selected for study and indicate possibly important items that are not addressed.
There is no consensus regarding which indirect costs should be included in cost-of-illness studies and wide variation in existing estimates. Most studies have focused on employment-related losses. Both the occurrence and severity of epilepsy appear to be negatively related both to the probability of employment and to the productivity of those employed (13). Epidemiologic studies indicate excess mortality risk for persons with epilepsy, with the greatest increase in the younger age groups and for more severe epilepsy (14). It has been argued that the measurement of indirect cost with the human capital method overstates the true cost to society (15). Where productivity losses can be identified, the “friction cost” method, a technique used to adjust the value of marginal productivity, provides a more accurate estimate of indirect cost. The inclusion of indirect cost also raises serious equity considerations, because people less likely to be employed (e.g., children and older people) will appear to have a lower cost. Standard measures of indirect cost should to be defined and surveys conducted of general populations of persons with epilepsy to estimate each measure.
Attributing direct or indirect cost to epilepsy compared with any associated (whether or not specifically correlated) clinical condition is a significant difficulty. In general, the cost experienced by patients with epilepsy is an overestimate of the cost attributable to epilepsy, unless the costs of comorbid conditions are identified and excluded. The identification and exclusion of these conditions is difficult when they cannot be differentiated from epilepsy or its complications as potential sources of cost to be valued. For example, in a patient with posttraumatic hemiparesis and epilepsy, the direct cost relative to the management of the motor disturbance may easily be separated from the cost for the management of epilepsy. By contrast, driving or occupational restrictions tend to be attributed to both conditions. As well, the degree to which the cost is attributable to the cognitive and/or behavioral problems associated with epilepsy, an epileptogenic condition, or its treatment is hardly defined. By contrast, the incremental cost attributable to epilepsy as a comorbid condition in institutionalized persons has been calculated and found to be significant (16).
Attribution of cost to adverse effects or to the use of AED treatment is another difficult task. Where measured, serious adverse treatment events may be a relevant source of cost in patients with epilepsy (17) and may represent the major cost difference between new AEDs (18). Conversely, the introduction of a new AED may be associated with a reduction of cost (19).
Any attempt to separate the cost attributable to epilepsy from those of the other conditions may be biased by the methods of investigation. For example, insurance claims at the encounter level were classified separately by Griffiths et al. (20) depending on the presence of epilepsy in the diagnostic code. The accuracy of this approach heavily depends on the correctness of the attribution in the billing codes. Similarly, Begley et al. (7) applied criteria for each medical encounter to determine which direct costs are attributed to epilepsy and which are not. This approach theoretically captures the attributable cost of the disorder but is difficult to conduct and subject to measurement error. Ideally, the use of nonepileptic controls may help overcome the problem of attributional adjustment, because they are the source of expenditures that a patient with epilepsy may incur as a normal individual. To our knowledge, the only study using a case–control approach for attributional adjustment was that of Begley et al. (7), who referred to the 1996 U.S. Current Population Survey as a comparator in estimating indirect cost in the United States. In this approach, the total direct or indirect costs of cases with epilepsy are compared with the overall costs of similar persons without epilepsy. An alternative strategy, when attribution cannot be directly addressed, is to perform a sensitivity analysis, which involves examining the effects of changes in the key assumptions of the study. For example, what effect does it have on the results if a percentage of the doctor visits are attributed to epilepsy and a percentage to depressive illness that may or may not be associated with the epilepsy?
Assessing cost variation
Separate cost assessments are needed for persons with epilepsy of differing types and severities. From a direct-cost perspective, seizure control appears to be the principal determinant of cost variation (see next section). However, the differential cost associated with seizure frequency and seizure type has not been assessed and is difficult to estimate because a given seizure type and frequency have different combinations across patients. In addition, direct, indirect, and intangible cost tend to reflect the individual reactions to the occurrence of a seizure, which may be different across subjects and, in the same subject, depending on site and circumstances of the seizure. By contrast, the classification of epilepsy into syndromic patterns may represent a practical way to explore cost variation in patients with epilepsy. However, the management of the different epileptic syndromes tends to reflect mostly the degree of seizure control, which may be different even within each syndromic category.
To understand cost variation fully, we must develop a general model of the determinants of variation in utilization and cost that identifies the relative importance of clinical factors, such as etiology, prognosis, seizure frequency, and type, as compared with social and economic factors, such as age, gender, ethnicity, income, and insurance. From a clinical viewpoint, it may be expected that clinical factors should explain most of the variation. However, in recent years, wide variation in health care utilization and cost, unrelated to clinical factors, has been demonstrated in a number of disease areas (21,22). The importance of these factors in epilepsy has not been fully explored in the literature.
EVIDENCE ON DIRECT COST
A number of bottom-up direct-cost studies recently in the literature have adequately addressed most of the issues discussed and have followed roughly comparable methods in terms of case definition, cost items measured, and estimation procedures (7,23–28). Major methodologic differences remain with respect to the methods for sample selection and whether prospective or retrospective data collection was followed. A review of these studies follows, indicating substantial variation with respect to the direct cost of epilepsy for both incident and prevalant cases, but a fairly consistent pattern with respect to the determinants of variation in direct cost.
In an Italian referral population including 525 children and adults followed up prospectively for 1 year during the period 1996 through 1997, the cost of epilepsy varied significantly with the severity of the disease, its temporal stage, and its response to treatment (23)(Table 1). The average annual cost was $2,240 in active drug-resistant cases, $1,657 in active non–drug-resistant epilepsy, and $420 in inactive epilepsy. Hospital admissions and drugs accounted for the largest percentage of expenditures. Significant cost differences were confirmed in an Italian referral pediatric population (189 cases) of patients in remission for 12 months before admission, patients with active non–drug-resistant epilepsy, and patients with drug-resistant seizures who were followed up for 1 year during 1996 through 1998 (24)(Table 1). In this study, for patients in the same epilepsy category, cost tended to vary significantly according to the health care setting: highest in teaching hospitals with most advanced equipment and resources, intermediate in general hospitals where patients were managed by an epileptologist without access to latest technology and staff, and lowest in outpatient settings where patients were treated by a child neurologist lacking specialization in epilepsy.
|Study||Beghi et al., 2000||Guerrini et al., 2000||Jacoby et al., 1998||Cockerell et al., |
|van Hout et al., 1997|
|Country||Italy||Italy||U.K.||U.K.||France, Germany, U.K.|
|Population||525, all ages||189 youth (age 0–18)||785, all ages||1,628, all ages||300 adults (age 18–65)|
|Direct cost per person per year||$420 inactive |
$1,657 active nondrug-resistant
$2,240 active drug-resistant cases
|$851 inactive |
$1,104 active nondrug-resistant
$3,256 active drug-resistant
|$243 inactive |
$1,238 active <1 seizure/mo
$2,288 active >1 seizure/mo
|$126 inactive |
$803 active (at least one seizure in previous 2 yrs)
|$1,748 inactive |
$3,208 active (ranging from <1/mo to >1/day)
Jacoby et al. (25) determined the average annual cost in 1992 through 1993 for a general sample of 1,000 cases from a U.K. Health Region (Table 1). They found that persons with epilepsy having one or more seizures in the last year had annual epilepsy-related health care cost 7 times higher than that of persons with epilepsy who were seizure free ($1,763 vs. $243). Average cost ranged from $243 for those with no seizures to $2,288 for those having one or more seizures per month. Cockerell et al. (26) derived cost of care for a general sample of 1,096 cases in the U.K. National Epilepsy Survey (NES) during 1990 through 1992 (Table 1). Average cost for persons with inactive epilepsy (defined as no seizures in the last 2 years) was $126 compared to $803 for persons with active epilepsy (defined as having at least one seizure in the previous 2 years), a sixfold difference.
The cost for the preceding 3 months was estimated for 300 patients identified by neurologists in France, Germany, and the U.K. by van Hout et al. (27)(Table 1). Average cost per person ranged from $437 for persons with no seizures to $802 for those with one or more seizures, indicating a twofold difference in the cost of active versus inactive patients. If annualized, the cost estimates range from $1,748 to $3,208, considerably higher than that found in the two U.K. studies.
The higher cost reported by the van Hout et al. (27) study and the Italian studies, as compared with the U.K. studies, appears to be primarily due to how the populations were selected in the studies. The European and Italian study populations were drawn from the practices of neurologists who tend to treat more severe cases, whereas those in the U.K. were drawn from general practitioners who tend to have more easily controlled cases.
Three incidence-based studies also have variable findings with respect to cost levels but indicate similar patterns of high initial costs at onset for most patients, followed by much lower costs in subsequent years. De Zelicourt et al. (28) determined the direct health care cost of 1,942 patients with newly diagnosed seizures for 2 years in France (Table 2). The mean annual direct epilepsy-related cost per patient was estimated to be $2,432 and $640 for the first and the second years of follow-up, respectively. Controlling for different factors like age, etiology, and AED use, De Zelicourt et al. (28) found that having active epilepsy (with a number of seizures at 1 year of fewer than 30 per month) resulted in first-year cost of $1,929 compared with $1,302 for a single seizure at diagnosis, and having more than 30 seizures per month at 1 year resulted in a cost of $6,221.
|Study||Cockerell et al., |
|De Zelicourt |
et al., 1996
|Begley et al., |
|Population||602, all ages||1,942, all ages||500, all ages|
|Direct cost per person per year||$917 in year 1 |
$282 in year 4
$254 in year 8
|$2,432 in year 1 |
$640 in year 2
|$3,157 in year 1 |
$702 in year 2
$471 in year 3
$411 in year 4
Begley et al. (7) estimated a time series for 500 incident cases identified retrospectively over the period 1987–1994 among enrollees in health maintenance organizations in Houston, Texas, and residents of Olmsted County (Rochester), Minnesota (Table 2). They found a similar pattern of relatively high cost in the first year of treatment ($3,157) in year 1, decreasing rapidly to $411 in year 4. In the National General Practice Study of Epilepsy (NGPSE), Cockerell et al. (23) examined cost prospectively in an incidence cohort of 602 paients throughout the U.K., determining first-year health care cost of $917 per patient, decreasing after 8 years of follow-up to $254 per patient per annum (Table 2).
As more studies have been done, a range of estimates of the direct cost of epilepsy in different countries and the variation across patients is becoming clear. Population-based prevalence studies from the Italy, the U.K., France, and Germany indicate average annual health care cost ranging from $100 to $2,000 for inactive cases, $900 to $3,000 per year for active cases, and two- to sevenfold higher costs in active cases with frequent seizures as compared with active cases with infrequent seizures. Incidence-based studies from the U.K., France, and the U.S. indicate first-year cost of all new cases ranging from $900 to $3,000 and subsequent-year cost ∼60–70% below first-year cost, reflecting the large number of cases that achieve early remission. Furthermore, the Italian studies indicate that the variation in cost across patients with epilepsy may be explained not only by the temporal stage and severity of the disease, but also by the health care setting where each individual case is being followed up.
It would be interesting to know if the differences in average cost among countries are due to discrepancies in terms of unit cost or of health care resource use. Heaney et al. (29) determined that prices for health care services including AEDs vary greatly among European countries. Another possibility may be related to the distribution by severity of seizures, which has been shown to have a significant effect on cost.
Very few studies have attempted to account for the cost of non–health care services in addition to health care and comparable estimates of the indirect cost of epilepsy, which accounts for the majority (perhaps 70–85%) of the total cost of epilepsy, do not exist. Individual studies have generated estimates for different countries, but the methodologic differences in the indirect cost items included and estimation procedures make comparisons impossible. Moreover, as indicated earlier, only one study dealt with the attribution problem with respect to indirect cost estimation. Although these gaps in the literature remain, considerable progress has been made in understanding the social burden of epilepsy in terms of the magnitude and variation in its economic cost.