Portal de Periódicos da CAPES

A Comparison of Surgical and Medical Costs for Refractory Epilepsy

2002; Wiley; Volume: 43; Issue: s4 Linguagem: Inglês

10.1046/j.1528-1157.43.s.4.5.x

1528-1167

Michael P. Platt, Michael R. Sperling,

Pharmacological Effects and Toxicity Studies

As health care costs increase, the need to provide cost-effective care has increased. As with other conditions, new drugs, novel diagnostic tests, and new surgical approaches to treat epilepsy have led to escalating costs. These developments require updated economic studies to reassess costs of epilepsy and the cost-effectiveness of therapy for epilepsy (1–3). Analyses should determine not only the cost-effectiveness of epilepsy treatments with regard to medical measures, but also the psychosocial and intangible measures. It is especially important that long-term assessments be performed, because epilepsy is a chronic illness that may last for decades. Because the epilepsies comprise a broad spectrum of syndromes, determining and comparing the cost-effectiveness and cost-utility of various therapies is challenging. Economic considerations are particularly important for patients with medically refractory seizures. Twenty to thirty percent of people with epilepsy do not completely respond to medical treatment (4), and these individuals disproportionately share the detrimental economic and social impact of the illness. Hence, although the most refractory patients compose a minority of those with epilepsy, they account for a large share of the total costs of epilepsy. One study estimated that the 15% of patients who are most refractory account for ≥50% of the total costs of the illness (5). Jacoby (6) found that cost correlated with the severity of the illness and that intractable patients cost 8 times more that those with controlled epilepsy (Table 1). The added expense derives from both medical and nonmedical sources, and indirect expenses are particularly high. Medical expenses include the cost of physician visits, emergency room visits, hospitalizations, medications, diagnostic laboratory testing [blood tests, EEG, magnetic resonance imaging (MRI)], which are considerably higher in patients with recurrent seizures than in those whose epilepsy is in remission. Indirect expenses, which account for as much as 75% of total costs, include lost productivity from unemployment, underemployment, or lost work time, excess mortality, and lost work of relatives or friends who care for the ill person (5,7). Murray et al. (7) estimated the lifetime cost of refractory epilepsy cases in the United States, who compose 29% of all adult cases. Indirect costs included lost wages due to unemployment and underemployment and lost caretaker earnings. Transfer payments and mortality costs were excluded. An incidence analysis for patients with refractory epilepsy showed that direct costs were $101,160,782 and indirect costs were $217,421,887. Total cost was $318,582,669 ($12,962/person), with indirect costs accounting for 68%. A prevalence analysis found direct costs of $912,553,518 and indirect costs of $2,992,629,945, with a total cost of $3,905,183,463 ($11,745/person). Indirect costs accounted for 77% of the total. Begley et al. (5) used an incidence analysis to estimate direct medical and indirect costs for six prognostic groups of patients, ranging from those in permanent remission to institutionalized patients with intractable seizures. Direct nonmedical costs were not included. Direct medical costs of epilepsy averaged $20,352 per patient, ranging between $4,272 for well-controlled patients in remission and $138,602 for the most severely affected individuals. Noninstitutionalized patients with frequent seizures accounted for 59% of the total cost. Indirect costs composed 75% of the expense in these patients. In a subsequent analysis, Begley et al. (2) estimated the 1995 costs of epilepsy in the United States at $12.5 billion. The direct costs included were direct medical costs. Lost wages and mortality were included in indirect costs. For incident cases, those with intractable epilepsy represented 25% of all cases but accounted for 79% of estimated costs; indirect costs accounted for 88% of the total expense. For prevalence cases, patients with intractable epilepsy composed 43% of all cases but accounted for 80% of estimated costs; indirect costs accounted for 84% of the total cost. While all refractory patients with epilepsy might be treated with medication, some individuals might respond to surgical treatment (8). In appropriate candidates, surgical treatment improves seizure control. The success of surgical therapy depends on many factors. These include (a) location of surgery, because temporal lobe surgery offers better results than extratemporal surgery; (b) distribution of pathology, because restricted pathology is associated with a better outcome than widely distributed pathology; (c) clinical variables related to history and physical examination; and (d) diagnostic test results, such as MRI, EEG, single-photon emission computed tomography (SPECT), neuropsychological testing, and positron emission tomography (PET). Anterior temporal lobectomy is the operation most commonly performed and has the highest success rate with 65–80% of patients achieving seizure remission after surgery (8). With improved seizure control, long-term utilization of health care resources diminishes, because the need for physician visits, hospitalization, diagnostic tests, and attendant costs diminish. In addition, indirect costs might decrease if seizures were the major cause of reduced productivity. However, surgery is expensive, and this outlay must be weighed against any potential long-term cost savings. Cost-effectiveness and cost-utility studies can address issues related to the cost of surgery and compare the costs of medical and surgical treatment of refractory epilepsy. These analyses compare different types of treatments for a single illness, or compare treatments for different illnesses. Cost-effectiveness studies measure outcomes in natural units, such as seizure frequency or mortality rate, and can provide cost estimates to achieve specific rates of improvements in these measures. However, this type of analysis can yield misleading results within a single domain depending on how the data are presented, and it cannot account for multiple-outcome domains. The following example will illustrate potential problems in interpretation. Therapy A reduces mean seizure frequency by 20% (with 2% seizure free) and costs $100 per patient, whereas therapy B reduces mean seizure frequency by 60% (with 50% seizure free) and costs $1,000 per patient. In one formulation, therapy A costs $50 per 10% seizure reduction, whereas therapy B costs $167 per 10% seizure reduction, so A is more cost-effective at reducing seizure frequency. However, examining the seizure-free rate would yield the result that therapy A costs $50 per 1% rate of achieving seizure freedom, whereas therapy B costs $20 per 1% rate of achieving seizure freedom. In this analysis, B is more cost-effective. Which analysis should we regard as more useful? The second analysis is more meaningful than the first if the benefit gained by seizure reduction is not linear, particularly if seizure freedom (100% reduction of seizures) has a greater inherent worth than any other level of seizure reduction. This requires that there be other tangible and intangible benefits that accrue to seizure-free patients, so that measuring just the reduction in seizure frequency is inadequate to compare costs. This is particularly relevant when comparing medical and surgical therapy of epilepsy, which have seizure outcomes akin to those of treatments A and B (9). Cost-utility studies overcome the limitations of using a single outcome measure, although they have their own problems. They combine different types of outcome into a single measure of well-being. Health-related quality of life has received much attention in recent years, and reliable cross-cultural scales have been developed for use in people with epilepsy (10–13) Quality of life is related to seizure control and has been shown to improve after epilepsy surgery, particularly in seizure-free patients (13–15). In cost-utility analysis, quality of life is incorporated in a measure known as quality-adjusted life years (QALY), which account both for quality-of-life changes and survival. One can then calculate a dollar cost per unit of outcome (QALY). This measure accounts for multiple tangible and intangible benefits, although the outcome measure is an abstract concept that is not readily comprehensible to the general public. This article will serve two purposes. First, we review existing cost-effectiveness and cost-utility studies of surgical therapy for epilepsy and cost comparisons of medical and surgical therapy. Thus far, few investigators have directly addressed this issue. However, many cost assumptions will be made in future studies that will influence the results. The second part of this article addresses these cost factors, and demonstrates how use of different underlying assumptions alters the results and conclusions of any comparison of medical and surgical therapy for refractory epilepsy. Two cost-utility studies have assessed the costs of temporal lobectomy for intractable epilepsy. These studies have calculated cost in dollars per QALY and related those costs to treatment of other illnesses. They concluded that costs fall within a range considered reasonable, yet highlight the inherent uncertainties in their models. King et al. (16) evaluated real direct medical costs in a cohort of patients undergoing evaluation for epilepsy surgery and estimated changes in quality of life based on literature reports. Cost data were taken from 51 patients, of whom 34 (67%) ultimately had anterior temporal lobectomy. Sixty-nine percent were free of seizures after surgery (including 16% with auras alone), and 31% had some postoperative seizures. They calculated that the cumulative discounted benefit from surgery was 1.1 QALYs (equivalent to 1.1 extra years in good health). They found that evaluation for surgery and treatment of a subset of individuals resulted in an average marginal cost of $29,800 per patient, yielding a cost-utility ratio of $27,200 per QALY. This figure was compared with that calculated for therapies for other conditions (Table 2). King et al. (16) further showed how the cost per QALY depended on model assumptions. The amount of change in quality of life affected cost, so that the change in quality of life was inversely related to cost per QALY. Changes in the discount rate also significantly affected cost estimates. Only direct medical costs were evaluated, and direct nonmedical costs and indirect costs were not included in their model. He demonstrated how change in income after surgery might alter the cost, and showed that if average annual income of seizure-free patients increases >$5400, then surgery is cost saving. Langfitt (17) also evaluated the cost-utility of anterior temporal lobectomy for medically refractory epilepsy. He also used only provider-based, direct medical costs at their center and used quality-of-life ratings based on literature reports. They calculated the marginal cost-utility between surgical and medical management of intractable epilepsy and estimated different levels of cost depending on model variables. They estimated that the average lifetime cost per patient who had surgical evaluation was $109,362, whereas continued medical management led to a per capita lifetime cost of $84,276. They determined that the surgical group had an improvement of 1.61 QALYs, for a surgical cost of $15,581 per QALY. They did a sensitivity analysis and showed how cost was affected by varying assumptions regarding efficiency of evaluation process, seizure outcome, and quality-of-life outcome scores. As in the King study, indirect costs were not included. These two studies yielded results that were similar enough to suggest that the findings are reasonable. However, they have certain limitations, which were acknowledged by their authors. To compare medical and surgical therapy fairly, all associated costs, particularly indirect costs, must be included. Because the indirect costs compose the bulk of expenditures, their exclusion does not portray the full economic picture. Indeed, inappropriate conclusions might be drawn if indirect costs change in one group as a consequence of therapy. This is most likely in surgically treated patients. Many of these patients become seizure free, and there is some evidence that mortality and employment costs lessen as a consequence of successful surgery (18,19). One other study compared costs of medical and surgical therapy. Wiebe et al. (20) modeled long-term direct medical costs in two cohorts, each containing 100 patients, one treated medically, and the other evaluated for surgery. They estimated that 85% of the surgical group would receive surgery, of whom 55% would be seizure free, that 17% would improve, and that 13% of operated-on patients would not significantly improve. They estimated that 12% of the medical group would become seizure free, that 13% would improve, and that 75% would not significantly improve. They used a 5% discount rate to account for differential timing of some costs. The surgical group had significantly higher costs in the first year, but costs equalized after 8.5 years, and after that, surgical treatment was less expensive. For a 35-year period, they estimated that costs for the medical group were $10,741,425, whereas costs for the surgical group were $8,117,911. Sensitivity analyses did not change the findings. They concluded that surgical treatment of temporal lobe epilepsy is cheaper than medical therapy, although this clearly requires a long-term societal perspective. This study also did not estimate nonmedical direct costs or indirect costs, and therefore the dollar amounts do not reflect all of the difference in cost between medical and surgical therapy. The total costs of epilepsy include medical direct cost, nonmedical direct costs, indirect costs, and intangible costs. Dollar values are readily, although not easily, assigned to the first three categories. Direct medical costs include those expenses imposed by medical evaluation and treatment. Nonmedical direct costs include special education, transportation, and residential care. It is sometimes difficult to apportion nonmedical direct costs with certainty if coexisting neurologic impairments also are present. Nonetheless, nonmedical expenses may be significant, and have been estimated to account for 39–56% of total direct expenses (6). One must carefully note which costs have been included and which have been excluded in any particular study, for these affect final dollar estimates. The costs may have been excluded for valid reasons, but one should avoid the pitfall of comparing apples and oranges if comparative studies are attempted. For example, direct nonmedical costs such as special schooling and social support were not included in studies by Begley et al. (2) and Murray et al. (7). Other studies did not include specific indirect medical costs such as transfer payments (5), mortality (7), or caretaker loses (6). Indirect costs are those losses accrued from decreased productivity, but accounting for them is arbitrary and perhaps inaccurate. These costs include unemployment, underemployment, premature death, transfer payments, and lost caretaker productivity. There may be additional employment losses for parents who must care for their children with epilepsy. It may be impossible to ascertain these costs with certainty, and there is legitimate dispute as to how to account for them. Would that parent or an adult patient have chosen a different occupation with a different income level had epilepsy not lurked in the background? How much does epilepsy and how much does the coexisting memory impairment contribute to lowered job performance? These questions cannot be easily answered, yet could significantly affect cost calculations. Transfer costs are another source of dispute, and only some economists include them in indirect cost calculations. Intangible costs are the social and psychological costs such as decreased quality of life, social isolation, and impaired satisfaction. Although it was once thought that these costs were immeasurable, recent investigators have used quality-of-life measures to estimate the burden of disease on the patient's emotional well-being. Although the measures can have cost assigned to them (e.g., dollars per QALY) as discussed earlier, specific intrinsic dollar values cannot be given for them. Because existing studies have examined only direct medical costs when comparing medical and surgical therapy for refractory epilepsy, there is a need to compare these treatments including nonmedical direct costs and indirect costs. A more thorough comparison of medical and surgical therapies is needed to determine their clinical and economic benefits and shortcomings. We constructed a cost analysis that included direct and indirect costs to compare surgical management with medical management of patients with refractory temporal lobe epilepsy. We examined how including and excluding different variables altered the cost estimates and used literature estimates of costs in the model. The goal was to assess how considering the indirect costs influenced a comparison of medical and surgical therapy. The cost estimates were drawn from published prevalence studies and converted to 1995 U.S. dollars. One inherent problem in comparing data from multiple studies is that different investigators have used varying definitions of intractable epilepsy. We used the patient groups defined by Jacoby et al. (6). They separated patients into three groups: Group 1, comprising patients with no seizures in the past year, Group 2, comprising patients having less than one seizure per month, and Group 3, comprising patients experiencing at least one seizure per month. All direct and indirect costs were assigned based on this grouping. Before assignment to either the medical or surgical groups, refractory patients were assumed to have at least one seizure per month, equivalent to membership in Group 3. In calculating costs after assignment to a treatment group, patients who became seizure free moved from Group 3 to Group 1, those with improved seizure status moved from Group 3 to Group 2, and those without change in seizure frequency remained in Group 3. For costs associated with caretaker expense and transfer payments, only patients who became seizure free were assigned Group 1 costs, and all other patients were assigned costs at the level of Group 3. In this analysis, medically treated patients were all assumed to remain in Group 3, based on preliminary results of an analysis of 44 medical refractory patients followed up for ∼4.7 years in our center. Both direct medical and nonmedical costs were considered, and were derived from Jacoby et al. (6)(Table 3). Direct surgical costs were averaged from King et al. (16) and Langfitt (17)(Table 4). Indirect costs included losses from unemployment, underemployment, transfer payments, caretaker loses, and mortality. Employment levels and income for the different seizure groups were drawn from previously published data (19) and new unpublished data derived from 250 surgical patients at the Jefferson Comprehensive Epilepsy Center. We have observed significant improvement in employment after surgery, but only for seizure-free patients. The preassignment mean income was $17,456 per year, and after assignment, changed to $24,806 for seizure-free patients, $23,156 for improved patients, and $15,094 for unimproved patients. The difference between seizure-free and refractory patients is similar in magnitude to the figure of $7,622 (1995 US$) published by Murray et al. (7). Costs of transfer payments were derived from Jacoby et al. (6)(Table 3). Caretaker losses taken from Murray et al. (7) were $1,579 for those with intractable epilepsy. As noted earlier, patients who did not become seizure free were deemed to gain no benefit in caretaker losses or transfer payments. The full effects of decreased transfer payments and increased employment were not realized until 2 and 5 years, respectively, after assignment. Some variables were changed to see how they affected costs. The effect of altering the times until change in caretaker losses, employment, and transfer payments was determined. The effect of altering the screening yield, the percentage receiving intracranial EEG monitoring, and the surgical success rate on cost was determined. To model costs, a number of assumptions were made based on experience in our center. One hundred patients were assigned to a medical-management group, and 100 patients were assigned to a surgical group. In the surgical group, all individuals were screened as outpatients and as inpatients. Twenty-six percent of those screened were evaluated with deep-electrode EEG monitoring, and 76% of the surgical group ultimately had anterior temporal lobectomy. The remaining 24% of patients who were unsuccessfully screened for surgery had medical management but remained in the surgical group (intent to treat). Anterior temporal lobectomy was assumed to result in the following outcome: 70% seizure free, 20% improved seizures, and 10% no improvement (21). Figure 1 shows the comparison of direct and indirect costs for the medical and surgical groups over a 40-year period. It shows that overall costs are higher in the medical group, and that the initial higher cost of surgery is overcome in the first decade. Medical group costs progressively increase with the passage of time, mainly because of higher indirect costs in those patients. One year after assignment, direct medical costs accounted for 97% of the total direct costs in the surgical group and 42% in the medical group. Twenty years after assignment to medical or surgical therapy, direct medical costs accounted for 68% of all direct costs in the surgical group, whereas the percentage remained constant at 42% in the medical group. This difference was due to the high initial direct costs of surgery, and the reduced long-term direct costs for those who became seizure free after surgery. When all direct costs are considered, the total direct costs for the surgery cohort became less than those of the medical cohort at 14.4 years (Fig. 2). If only direct medical costs are considered, surgical group costs equal medical group costs 35 years after assignment in this model. Comparison of costs for medical and surgical patients over 40 years. Both direct and indirect costs are included. Costs are higher in the surgical group 1 year after assignment, but higher for the medical group by year 10. Medmgmt, patients assigned to medical therapy; Surg Eval, patients assigned to surgical evaluation and treatment arm. Graph comparing direct costs in the surgical and medical groups. The lines cross at 14 years, after which surgical therapy is cheaper. Figure 3 shows indirect costs over time. One year after evaluation, indirect costs accounted for 17% of the total costs in the surgical group and 40% in the medical group. No indirect costs were assumed to change in the first year after assignment. Indirect costs became less for the surgery cohort than the medical cohort beginning 1 year after assignment, because patients who became seizure free after surgery had decreased caretaker costs. Transfer payments were decreased 2 years after surgery for seizure-free patients, and employment gains reached a maximum at 5 years after surgery. Unemployment and underemployment initially accounted for 58% of the indirect costs and 39% of the total costs. Full benefit with regard to employment was assumed to occur 5 years after surgery in seizure-free patients. Ten years after assignment, there was a 31% reduction in indirect cost from unemployment in the surgical group compared with the medical group. When all indirect costs are combined with direct costs (Fig. 4), the total costs for the surgical group became less than those for the medical group 7.3 years after assignment, ∼50% less time than if only direct costs are used in the calculations. Graph comparing indirect costs in the surgical and medical groups. The indirect costs are substantially higher for the medically treated patients, because of the substantial proportion of seizure-free patients in the surgical group. Graph comparing total costs for surgical and medical groups. This analysis demonstrates how estimates of cost can be influenced by the inclusion and exclusion of different variables from the analyses. We confirm the results of Wiebe et al. (20) that surgical therapy appears to cost less over the long term than does medical therapy. Depending on which costs are included in the analysis, a surgical-management approach can be shown to become more cost-effective in as little as 7.3 years, or as much as 35 years. The data illustrated represent only the first step in a series of analyses comparing medical and surgical therapy. They illustrate the need for caution when interpreting economic reports of cost of therapy. They also highlight the need to take a long-term perspective. We have not addressed the effects of changes in mortality, varying the discount rate, varying the yield of the surgical evaluation process, and differing rates of seizure relief after surgery, all of which would affect costs. Surgical therapy is a cost-effective treatment. Long-term expenses associated with epilepsy surgery compare favorably with those incurred by medical treatment. It is in the best interests of every society to take a long-term rather than a short-term view. Sadly, private businesses more often focus on short-term rather than long-term gains. The division of expense in the United States and many other countries encourages a short-sighted approach. Current medical expenses including surgical costs are incurred by the rate payers and insurance companies this year and are not amortized. The decrease in direct medical expense in surgical patients takes place over a decade or more, which is a long time frame for many executives whose salaries are tied to quarterly profits. Furthermore, the gains in income benefit society far more than private insurers. Hence, it is crucial that we adopt a perspective that allows both individual and societal benefit.