SEARCH

SEARCH BY CITATION

Keywords:

  • Epilepsy;
  • Medical expenditure;
  • Economic impact

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Purpose:  To assess differences in medical care expenditures and informal care received for adults and children by individuals’ self-reported epilepsy status and to estimate the total economic impact of epilepsy in the United States.

Methods:  Pooled medical expenditure panel survey data from 1996–2004 were used. Children’s regression analyses were adjusted for race, sex, general self-reported health status, family size, and age. Adults’ analyses were also adjusted for income and education. The national annual economic impact was estimated by multiplying the average individual differences by previously published national prevalence data.

Results:  The results of regressions appropriately weighted to account for study design indicate excess medical expenditures for those with epilepsy of $4,523 [95% confidence interval: $3,184–$5,862]. Excess expenditures were similar for adults and children. Adults with epilepsy received 1.2 extra days of informal care [95% confidence interval: 0.2–2.3]. The national impact included $9.6 billion of medical expenditures and informal care.

Discussion:  Epilepsy has significant impact on individual medical expenditure and generates a national impact in the billions of dollar.

Epilepsy imposes a burden on society by imposing a significant burden on both the individuals who have the condition and on those around them. Numerous attempts have been made to assess the burden of epilepsy on society in countries including India, Burundi, Australia, European countries (Beran, 1999a; Heaney et al., 2001; Thomas et al., 2001; Nsengiyumva et al., 2004), and the United States (Begley et al., 1994, 2000; Griffiths et al., 1999).

In the United States, Begley et al. (1994) estimated the lifetime cost for the incidence cases based on the prognostic group. They divide their study group into six prognostic groups, ranging from the permanent remission group after initial diagnosis and treatment to the institutionalized intractable seizures group, according to severity. This level of detail allowed them to present a range of results. By applying the different health care utilization probabilities determined by the expert panel and the different productivity losses derived by national survey and other studies according to those severity and prognostic groups, they concluded that the total lifetime cost per patient ranged from $4,272 for persons with remission after initial diagnosis to $138,602 for persons with severe seizures, in 1990 dollars. Griffiths et al. (1999) reported the payer cost of epilepsy by using administrative claims data for a 9,090 epilepsy patient cohort. They reported that the mean total annual cost of all medical services for those who have epilepsy was $9,617 during 1992 and 1996. Begley et al. (1994) later estimated that the total cost of epilepsy for the United States in 1995 was $12.5 billion (Begley et al., 2000). Their cost estimate consisted of $1.7 billion direct medical cost projected from 608 cases in two metropolitan areas and $10.8 billion indirect cost from productivity loss projected from 1,168 surveyed adult patients at the 18 epilepsy treatment centers.

Although various approaches are possible to illustrate the health care burden for epilepsy patients and families, no report to date has been able to take advantage of a large nationally representative survey with high quality data on self-reported utilization and medical conditions. Nine years of data from the medical expenditure panel survey (MEPS) were used for this study. This survey has been administered annually since 1996 by the Agency for Healthcare Research and Quality (AHRQ) and includes 10,000 to more than 20,000 individuals each year (Agency for Healthcare Research and Quality, 2007). This quantity of data allows researchers to estimate the excess resource utilization associated with relatively rare conditions. An analysis of these data will allow us to infer the association of epilepsy with medical care and informal care in a way that was not previously possible in the United States.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Subjects and data

The MEPS consists of household surveys of the civilian noninstitutionalized population, with supplemental information from a survey of medical providers and a survey of insurance providers (Cohen, 2002; Agency for Healthcare Research and Quality, 2007). The MEPS uses an overlapping panel design in which a new panel is initiated each year and each cohort is followed for 2 years. Using a computer-assisted personal interview approach, the MEPS collected detailed data on demographic characteristics, self-reported health status, health care utilization, charges and payments by payer sources, employment status, and health insurance. Interviews occur six times over 30 months, and the AHRQ constructs measures of expenditures for two calendar years. The AHRQ provides deidentified data for public use that includes information to allow analyses accounting for the survey sample design and weights to make the results nationally representative. Data from 1996–2004 were used in this analysis; the 9 years of data provide a sufficient number of observations of individuals with epilepsy (which has a relatively low prevalence) to estimate the association of epilepsy with medical care expenditures and quantities of informal care received.

Measures

The dependent variables were total health care expenditures and the number of days of informal care provided by individuals living outside the study subject’s household. The total health care expenditures were calculated as the sum of all medical expenditures from office-based, inpatient, outpatient, emergency room, home health care–related, medication related, and medical equipment/supply–related components. Survey respondents consented so that the providers identified in response to the MEPS questionnaires, such as hospitals, physicians, home health care providers, and pharmacies, could be contacted by telephone and asked to provide the information on dates of visits, diagnosis and procedure codes, charges, and payments, to supplement and replace the information from the respondents to ensure accuracy. All dollar amounts were adjusted to 2004 dollars using the consumer price index (Bureau of Labor Statistics, 2007). We included informal care days from MEPS to illustrate the burden of unpaid care services provided by family members. This informal care has been used in various economic burden studies (Harrow et al., 2004; Frick et al., 2007).

Each study subject’s epilepsy status was inferred based on self-reported conditions and reasons for utilizing medical care. Individuals were asked about their medical care utilization and asked to indicate the reasons for utilization. Professional data coders assigned 3-digit International Classification of Disease (ICD-9) codes. An ICD-9 code of 345 from any of a study subject’s records in the condition file was used to infer that an individual had epilepsy.

In multivariate regression analyses, we controlled for various demographic and socioeconomic variables that could confound the relationship between epilepsy and the dependent variables. The variables used in analyses including children were a subset of those in the analyses including only adults. The variables in the analyses of children included age, sex, race (white compared with all others), self-reported health status, insurance type, and family size. For regressions including only adults we also included education level, marital status, and individual income level. Education was divided into four categories: less than high school, high school graduate, some college education, and a 4-year college degree or more. Income was divided into quartiles. Insurance was divided into three categories: at least some private insurance, those with only public insurance, and those with no insurance. Self-reported health status (divided between those reporting excellent, very good, or good health and those reporting fair or poor health) was also used to adjust for the possibility that individuals with epilepsy are less healthy in general, but not necessarily because of epilepsy. Therefore, in the multivariate regression results, any relationship between the dependent variables and epilepsy remains after controlling for other potential differences between persons with and without epilepsy.

Analyses

We used Stata version 9.2 (StataCorp, College Station, TX, U.S.A.) to analyze the data. Analyses were performed for the entire sample and with the separate samples of children (younger than 18 years of age) and adults. We began with descriptive analyses of unadjusted expenditures and informal care days experienced. Next we ran multivariable linear regressions accounting for the complex study design to yield the adjusted differences between medical expenditures and informal care days. In addition, to separately illustrate the probability of any positive medical expenditure and the excessive expenditure compared to other conditions with positive expenditure, we supplemented the analyses with the model called the “2-part model” that has been used previously (Mullahy, 1998; Frick et al., 2007). This approach first fits a weighted logistic model to estimate the odds ratio, comparing the odds of having nonzero expenditures among individuals with epilepsy with the odds of having nonzero expenditures among individuals without epilepsy. The second part of the model is a linear regression to infer the difference in the magnitude of positive expenditures by epilepsy status. This approach allows us to infer whether those with epilepsy are more likely to utilize any care and to infer whether those who use care have higher expenditures than those without epilepsy.

After making inferences about excess medical care expenditures and informal care days received, we projected the annual national impact of epilepsy by multiplying the average per person effects estimated from MEPS data by prior estimates of the prevalence of epilepsy (Hirtz et al., 2007). Hirtz et al. (2007) provide the best available estimate of the prevalence of epilepsy; they combined other prevalence studies using the study quality as a guide for the relative importance given to each study.

In addition to projecting national excess medical care expenditures, we calculated a monetary value of excess informal care received. Informal care was valued at the 2004 minimum hourly wage of $5.15 US dollars (Employment Standards Administration Wage and Hour Division, 2007). Using minimum wage is a very conservative estimate of the value of informal care days. For the purpose of making comparisons among conditions, this approach guarantees that are we are not overstating the case for the economic burden from epilepsy.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Table 1 summarizes the characteristics of the population. A total of 264,513 had no missing data and were included in the analysis. Among these, 78,475 (29.7%) were children. The prevalence of epilepsy for the 1996–2004 MEPS was 0.9% (95% CI: 0.8–1.1%) for the children and 1.0% (95% CI: 0.9–1.1%) for the adults. Among both children and adults, in comparison to those without epilepsy, those with epilepsy were more likely to have only public insurance, less likely to be uninsured, and less likely to express good health status or better. Among adults, those with epilepsy were less likely to be married, more likely to have no more than a high school education, and more likely to have an income below the median level.

Table 1.   Characteristics of the respondents of the 1996–2004 medical expenditure panel survey by epilepsy status
CharacteristicsWithout epilepsyWith epilepsyTotal
N%aCIaN%CIN%CI
  1. aWeighted percentage and confidence interval accounting survey design.

  2. CI, confidence interval.

Total study population261,67699.098.9–99.12,8371.00.9–1.1264,513100.0 
Children
 Total77,70599.198.9–99.27700.90.8–1.178,475100.0 
 Male39,65651.250.6–51.846661.455.9–66.940,12251.350.7–51.9
 White58,56477.976.5–79.359782.077.1–86.959,16177.976.5–79.4
 Reported good, very good, or excellent health status75,54797.897.6–98.056581.477.6–85.276,11297.697.5–97.8
 Insurance
  Any private insurance42,94968.066.7–69.435855.950.6–61.343,30767.966.6–69.2
  Public insurance only26,34623.422.2–24.538640.535.2–45.726,73223.522.3–24.7
  Uninsured8,4108.68.0–9.1263.61.4–5.88,4368.58.0–9.1
 Age, years (mean)8.7 8.6–8.87.7 7.0–8.38.7 8.6–8.8
 Family size (mean)4.4 4.4–4.54.3 4.1–4.64.4 4.4–4.5
Adults
 Total183,97199.098.9–99.12,0671.00.9–1.1186,038100.0 
 Male84,99148.047.7–48.487544.540.7–48.485,86648.047.7–48.3
 White148,32483.182.0–84.11,57479.476.0–82.9149,89883.082.0–84.1
 Education level
  Less than high school49,00820.119.5–20.878732.128.4–35.949,79520.319.6–20.9
  High school graduate60,00032.932.2–33.573936.332.5–40.160,73932.932.3–33.5
  Some college38,27822.622.1–23.130717.013.9–20.138,58522.522.1–23.0
  College and up36,68524.423.5–25.323414.611.6–17.636,91924.323.4–25.2
 Married106,70255.752.1–59.389244.340.7–47.9107,59457.356.7–57.9
 Reported good, very good, or excellent health status157,22687.687.2–88.01,09856.853.3–60.4158,32487.386.9–87.7
 Insurance
  Any private insurance123,19274.373.5–75.189249.345.4–53.1124,08474.073.2–74.9
  Public insurance only29,81712.812.2–13.495340.736.9–44.530,77013.112.5–13.7
  Uninsured30,96212.912.4–13.422210.08.0–12.031,18412.812.3–13.3
 Income quartile
   –$970652,47323.523.0–24.01,04545.842.5–49.253,51823.723.2–24.3
  $9707 –$20,83943,82122.422.0–22.953225.623.1–28.144,35322.522.0–23.0
  $20,840 –$3810543,97525.224.8–25.629917.014.5–19.444,27425.124.7–25.5
  $38,106 –43,70228.828.0–29.619111.69.6–13.643,89328.727.9–29.4
 Age (mean)45.1 44.8–45.448.1 46.6–49.645.2 44.9–45.5
 Family size (mean)2.8 2.8–2.92.6 2.5–2.72.8 2.8–2.9

Table 2 shows unadjusted means and confidence intervals for expenditures and informal care days accounting for weights and study design as well as adjusted differences. Total annual medical expenditures for children with epilepsy were $6,379 (95% CI: $4,203–$8,555) and for children without epilepsy were $1,032 (95% CI: $989–$1,976). The adjusted difference was less than the unadjusted difference but remained statistically significant at $4,703 (95% CI: $2,609–$6,797). For children, the difference in days of informal care was not statistically significant. For adults, both adjusted differences were statistically significant. For expenditures, the difference was $4,465 (95% CI: $2,925–$6,005); for informal care days the difference was 1.2 (95% CI: 0.2–2.3). The average difference in expenditures when combining adults and children was $4,523 (95% CI: $3,184–$5,682).

Table 2.   Mean total health care expenditure amount and informal care days and unadjusted and adjusted differences for individuals without or with epilepsy using 1996–2004 medical expenditure panel survey data
 Outcome measuresaWithout epilepsyWith epilepsyUnadjusted differenceAdjusted differenceb
MeanCIMeanCIMeanCIMeanCI
  1. aCalculated by accounting for survey sample design, difference may not be exact due to rounding.

  2. bChildren: adjusted for age, gender, race, insurance type, subjective health status, family size.

  3. bAdult: adjusted for age, gender, race, education, insurance type, subjective health status, marital status, income level, family size.

  4. bTotal population: adjusted for age, gender, race, education, insurance type, subjective health status, marital status, income level, age, family size, adult indicator.

  5. CI, confidence interval.

ChildrenTotal expenditure1,032989–1,0766,3794,203–8,5555,3463,169–7,5234,7032,609–6,797
Informal care days0.00.0–0.00.5−0.4–1.30.4−0.4–1.30.4−0.4–1.2
AdultTotal expenditure3,2683,189–3,34710,0828,531–11,6336,8145,274–8,3554,4652,925–6,005
Informal care days0.40.4–0.52.51.5–3.62.11.0–3.11.20.2–2.3
CombinedTotal expenditure2,6892,629–2,7499,1817,889–10,4736,4925,207–7,7774,5233,184–5,862
Informal care days0.30.3–0.42.01.2–2.91.70.9–2.51.20.3–2.1

After adjusting for potential confounders, the odds ratio of any positive expenditure for children with epilepsy compared to children without out epilepsy was 7.9 (95% CI: 4.3–14.5). The same odds ratio of any positive expenditure for adults was 11.4 (95% CI: 7.1–18.5). The excessive medical expenditure is significantly higher for those with epilepsy when we analyzed the individuals only with positive expenditure. Among those who have positive expenditure, the excessive expenditure of the children with epilepsy compared to the children without epilepsy was $4,642 (95% CI: 2,516–6,767) and the excessive expenditure for adults with epilepsy was $4,193 (95% CI: 2,638–5,749). The two-part model suggests that among those who used any health care, the individuals with epilepsy had considerably higher expenditures than those without epilepsy. Therefore, those with epilepsy are both more likely to use any care, and among those who have used any care, those with epilepsy have a higher average expenditure.

Table 3 shows the total aggregated annual economic impact of epilepsy and its calculation process. With the assumption of 2.1 million epilepsy prevalence, the estimated medical expenditure was $9.5 billion dollars. With the additional assumption of $41.2 a day minimum wages, the estimated value of informal care was $99.6 million dollars. The total aggregated annual economic impact of epilepsy on the U.S. economy includes $9.6 billion dollars from medical expenditures and informal care.

Table 3.   Total economic impact of epilepsy in the United States
EstimatesEstimated value95% CI
  1. aFrom the reference: (Hirtz et al., 2007).

  2. bAssuming that an informal care day involves 8 hours of work, 2004 minimum hourly wage = $5.15.

  3. CI, confidence interval.

Individual annual excess medical expenditure, mean $4,5233,184–5,862
Estimated number of people with epilepsya2,098,000Not available
Excess medical expenditure, mean $9,488,530,190(6,679,136,154–12,297,924,226)
Individual excess informal care days, mean1.150.25–2.05
Minimum wage per day b, $41.2Not available
Value of informal care days, $99,576,288(21,618,165–177,534,446)
Total annual monetary impact, $9,588,106,478(6,700,754,319–12,475,458,672)

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

We demonstrated that the average individual health care cost of epilepsy is substantial, even after adjusting for various individual characteristics. The average excessive health care expenditure of $4,523 due to epilepsy is substantially larger than previously estimated. Begley et al. (2000) estimated the average direct cost for epilepsy prevalent cases at year 1995 was $733 or $909 when adjusted for inflation to a 2004 value (Begley et al., 2000). Although Begley et al. (2000) focused on incident cases that are likely to cost more (Begley et al., 2000; White, 2007), our results were of greater magnitude. This is because the prior article looked at costs only specifically related to epilepsy as judged by patients and physicians rather than all medical care expenditure differences, as well as the fact that the overall cost of care has increased (Heaney et al., 2002; Centers for Medicare and Medicaid Services, 2008). On the other hand, our estimation for total health care expenditure for people with epilepsy of $9,181 is less than that calculated by Griffiths et al. (1999) using insurance claim data. Based on private insurance claims data, they estimated that the mean annual cost of all medical services for patients with epilepsy was $9,617 between 1992 and 1996, which is $11,578 in 2004 dollar value. The difference in insurance types, that is, private insurance population versus general population, and study data, that is, claims data versus survey data, would be the reason for this difference.

The economic burden of the disorder (or the cost of all prevalent cases of epilepsy) includes nearly $9.5 billion of direct medical care costs and additional informal care costs that lead to an overall monetary burden of almost $9.6 billion. There is an ongoing discussion in the literature about the scope of the cost of illness (Annegers et al., 1999; Beran, 1999b; Pachlatko, 1999; Kotsopoulos et al., 2001). This analysis is somewhat limited because it does not factor quality of life into the cost of illness, does not consider the cost of an incident case over time, and does not include the productivity loss cost. However, this is the first study that has taken a largely bottom-up approach to calculating the costs of epilepsy using such a large dataset. The bottom-up approach used an estimate of excess expenditures at the individual level and multiplies it by the number of cases to estimate the impact at the national level. It would be difficult to ever use a top-down approach for this calculation, as it would require national data on expenditures and some indication of the prevalence of epilepsy (Begley & Beghi, 2002).

In our study, adults with epilepsy received significantly more informal care than adults/people without epilepsy. The lack of a statistically significant finding for children is probably not due to the fact that children do not need assistance but that children are likely to receive assistance from individuals within their households. The dollar value of the burden associated with informal care may have been higher than we estimated, as we assumed a minimum wage for all informal care providers.

Our study results showed that having epilepsy is associated with higher expenditures, both because of being more likely to make any expenditures and because those who make expenditures make higher expenditures on average. One thing that could help to account for the higher probability of any utilization is the fact that a previous U.S. national survey indicated that 90% of epilepsy patients were taking one or more antiepileptic medications (Fisher et al., 2000). The higher expenditures among those with epilepsy occur despite steps that insurers take to control expenditures such as cost-containing strategies including “prior authorization” for high cost drugs, a step approach, and generic drug promotion (Theodore et al., 2006; Bialer, 2007).

There are several limitations in our study. First, the estimate of the economic impact is limited because we were not able to include productivity loss and intangible costs. Productivity loss and intangible costs such as cost associated with stigma were known to constitute a significant portion of the cost of epilepsy (Beran, 1999b). Second, we inferred the epilepsy status by ICD-9 based on the medical care utilization. The epilepsy status might be not accurate because the diagnosis of epilepsy was not confirmed. Holden et al. (2005) reported the discrepancy between ICD-9 and a true epilepsy diagnosis. This likely biases the results toward the cost of more active cases, which are of greater interest at the individual level, but might lead to overestimates of the burden at the population level. Third, inferring an average cost-effect from 9 years of data is a limitation in the study, given the changes in treatment during that time (Perucca et al., 2007; Baaj et al., 2008). However, given the relatively small number of cases, making an inference from fewer years would lead to other analytic difficulties. Fourth, epilepsy is known to be associated with learning disability, emotional problems, mental retardation, and brain injury (Begley et al., 1999). The additional costs of these conditions in the educational and other nonmedical care systems are not included in our estimates.

Despite these limitations, we have demonstrated that epilepsy has a significant impact on the national economy as well as individuals. Our estimation can be used to determine the individual and societal burden of epilepsy.

Acknowledgments

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

This study was supported by the grant from educational grant by Abbot Laboratories. We are grateful to the Epilepsy Foundation of America for the unrestricted support. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Disclosure of conflicts of interest: This study was supported by an educational grant by Abbot Laboratories. Author Dokyoung Yoon received financial support for writing and travel expenses. Author Kevin D. Frick received financial support for writing and travel expenses.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References