Direct and indirect costs associated with epileptic partial onset seizures among the privately insured in the United States

Authors


Address correspondence to Jasmina I. Ivanova, M.A., Analysis Group, Inc., 10 Rockefeller Plaza, 15th Floor, New York, NY 10020, U.S.A. E-mail: jivanova@analysisgroup.com

Summary

Purpose:  Compare annual direct and indirect costs between privately insured U.S. patients with epileptic partial onset seizures (POS) and matched controls.

Methods:  One thousand eight hundred fifty-nine patients (including a subset of 758 employees) with ≥1 (POS) diagnosis (ICD-9-CM: 345.4.x–345.7.x), 1999–2004, ages 16–64 years, were identified from a privately insured claims database. Control group was an age- and gender-matched cohort of randomly chosen beneficiaries without epilepsy (ICD-9-CM: 345.x). All were required to have continuous health coverage during 2004 (baseline) and 2005 (study period). Chi-square tests were used to compare baseline comorbidities. Univariate and multivariate analyses were used for comparisons of annual direct (medical and pharmaceutical) and indirect costs during the study period.

Results:  Patients with POS were on average 42 years of age, and 57% were women. Patients with POS had significantly higher rates of mental health disorders, migraine, and other neurologic disorders, and higher Charlson comorbidity index (CCI) compared with controls. On average, direct annual costs were significantly higher for POS patients ($11,276) compared with controls ($4,087), p < 0.001; difference of $7,190. Epilepsy-related costs (i.e., costs for antiepileptic drugs, claims with epilepsy or convulsions diagnoses) accounted for $3,290 (29% of direct costs). Employees with POS had substantial and significantly higher indirect (disability- and medically related absenteeism) costs compared with controls ($3,431 vs. $1,511, p < 0.001). Multivariate analyses supported the matched-control univariate findings.

Conclusion:  Patients with POS had significantly higher costs compared with matched controls. Epilepsy-related costs underestimate the excess costs of patients with partial onset seizures.

Epilepsy, estimated to affect 1.5–2.3 million people in the United States (Begley et al., 2000; Halpern et al., 2000; World Health Organization 2001), imposes a substantial economic burden. Approximately one-half of newly diagnosed epilepsy patients have generalized onset seizures, and 42% have partial onset seizures (Murray et al., 1996). The annual epilepsy-related direct costs per epilepsy patient in the United States were reported from $1,071 to $1,490 (in 1995 dollars) (Begley et al., 2000; Halpern et al., 2000; Strzelczyk et al., 2008). Indirect costs are reported to range from 12–85% of total annual costs (Strzelczyk et al., 2008). Although epilepsy is treatable with antiepileptic drugs, it is estimated that 20–40% of newly diagnosed epilepsy patients will become refractory to treatment (Kwan & Brodie, 2000; French, 2007). Patients are generally considered treatment resistant or refractory if they have failed three or more antiepileptic therapies. The annual direct cost of refractory epilepsy was estimated as $2,885 per patient valued in 1994 (Murray et al., 1996). Although it is not fully understood how or why patients develop treatment resistance, the majority of patients who are refractory to treatment have partial onset seizures (French, 2007). Among U.S. patients with partial onset seizures, approximately 45% are medically refractory (Casino, 1990).

Although four studies have reported the annual direct cost of epilepsy and one has reported the annual direct cost of refractory epilepsy in the United States, the direct cost burden of partial onset seizures has not been reported (Begley et al., 1994; Murray et al., 1996; Begley et al., 2000; Halpern et al., 2000; Strzelczyk et al., 2008). Furthermore, no study has quantified the indirect (work-loss) cost burden of partial onset seizures associated with short- and long-term disability and medically related absenteeism.

The objectives of this study were to assess the direct and indirect cost burden of partial onset seizures in a U.S. privately insured population from an employer/third party payer perspective. This perspective is important in the United States, where more than 60% of American workers had job-based health insurance in 2001 (Health Affairs 2002). Specifically the study: (1) compared the annual direct costs of patients with partial onset seizures to those of demographically matched controls without epilepsy; and (2) compared the annual indirect costs, and total costs of employees with partial onset seizures to those of employee controls.

Methods

Data source

The study sample was selected from a privately insured de-identified administrative database (Ingenix Employer Database) that included approximately 3 million beneficiaries (including employees, spouses, and dependents) from 17 large U.S.-based companies with claims for services provided in the years 1999–2005. The database contains information on patients’ demographics (e.g., age and gender), monthly enrollment history, and medical and pharmacy claims. Utilization of medical services was recorded in the database with date of service, associated diagnoses, performed procedures, billed charges, and actual amounts of payments to providers. The database also includes pharmaceutical drug claims, with prescribed medications identified by National Drug Code (NDC), date of prescription fill, days of supply, quantity, and actual payment amounts. Short- and long-term disability claims are available for employees in 17 of the companies (but not for their beneficiaries), with dates of coverage and amounts paid to employees.

Sample selection

The study sample included beneficiaries who had at least one partial onset seizures diagnosis (ICD-9-CM codes: 345.4.x–345.7.x) from 1999–2004, continuous health coverage eligibility during 2004 (baseline) and 2005 (study period), and ages 16–64 during the study period. Because the partial onset seizure diagnosis was required prior to the beginning of the study period, the study sample represented prevalent cases. Potential controls were selected from among beneficiaries without any epilepsy diagnosis (ICD-9-CM codes: 345.x) at any time in the claims history, with continuous health coverage eligibility during 2004 (baseline) and 2005 (study period), and ages 16–64 during the study period. Beneficiaries age 65 years and older were excluded from the analyses because they could also be eligible for Medicare, and payment information could be incomplete for those with dual eligibility. The final control group was selected as a random sample matched to study group beneficiaries based on age, gender, region, and employment status using a 1:1 matching ratio. From among the patients in the study and control samples, the subset of employees in the 17 companies, ages 18–64, was selected for the analyses of indirect and total costs.

Patient characteristics

Patient characteristics compared between study and control group included demographics (i.e., age, gender), employment status, and baseline comorbidities identified using claims during the 12-month baseline period in 2004.

Baseline comorbidities included any mental disorders; select specific mental health disorders (e.g., depression, anxiety), migraine, other neurologic disorders, as well as the Charlson comorbidity index (CCI) calculated as the weighted sum of 17 physical conditions predictive of 1-year mortality (Charlson et al., 1987; Romano et al., 1993).

Main outcome measures

Main outcome measures included direct costs and indirect costs (for the subset of employees with partial onset seizures) calculated over the 12-month study period in 2005. Direct costs included medical services costs and pharmaceutical costs based on reimbursements from the insurer/managed care plan to providers. Patient copay or deductibles were not included in direct costs. Medical costs were further categorized into inpatient, emergency department (ED), and outpatient/other costs based on place and type of service codes as well as CPT procedure codes for emergency care. Direct costs comprise both epilepsy-related and nonepilepsy-related costs. Epilepsy-related costs included costs for medical services with a primary or secondary diagnosis of epilepsy (ICD-9-CM: 345.x) or convulsions (ICD-9-CM: 780.3) and prescription drug costs for antiepileptic drugs identified using NDC codes from a pharmacy database [Master Drug Data Base (MDDB; Medi-Span, Indianapolis, IN, U.S.A.)]. Medical claims had up to two diagnosis codes. When two diagnosis codes were provided, it was not known which code was the primary. Mental health–related costs were a component of nonepilepsy-related costs and included costs for medical services with a diagnosis of any mental disorders (ICD-9-CM: 290.x–319.x) and prescription drug costs for antidepressants, antipsychotics, and antianxiety agents.

Indirect costs during the 12-month study period included actual employer payments for disability days plus estimated costs for medically related absenteeism. Medically related absenteeism costs were calculated based on individual employee wage information available in the eligibility data and days with medical services use. Each inpatient day was assumed to account for a full day of work loss, whereas an outpatient or other visit was half a day of work loss. Because disability covers missed workdays due to prolonged illness, a plan-specific number of work-loss days before the beginning of disability was also included. Medically related absenteeism excluded days with medical services during a period of disability. Indirect costs in this study do not capture productivity loss or absences related to epilepsy when employees did not seek medical care. Indirect resource use included disability days obtained from actual short- and long-term disability claims and calculated medically related absenteeism.

Total costs were computed for employees as the sum of direct and indirect costs. All costs were in 2005 U.S. dollars.

Statistical analyses

Univariate analyses were used to compare characteristics of beneficiaries treated for partial onset seizures and matched controls without epilepsy. Categorical variables were compared using chi-square tests. Nonparametric Wilcoxon rank-sum tests were used for univariate comparisons of costs and medical services counts, which are usually nonnormally distributed.

Two-part multivariate models were also used to compare costs between study and control groups and estimate risk-adjusted costs controlling for differences in baseline patient characteristics (demographics, mental health disorders, migraine, other neurological disorders, and CCI). In the first part of the model, logistic regressions were used to assess the probability of having positive costs. In the second part of the model, a generalized linear model (GLM) with log link and gamma distribution for the error term was used to estimate costs among patients with positive costs. The parameter estimates from the models were used to calculate the predicted probability of having positive costs and predicted costs among patients with positive costs under two hypothetical scenarios—one assuming that all patients had partial onset seizures and another assuming that none of the patients had partial onset seizures. The expected risk-adjusted costs in each group were calculated as the product of the risk-adjusted probability of having positive costs and the risk-adjusted costs among patients with positive costs.

All analyses were conducted using SAS version 9.1 (SAS Institute Inc., Cary, NC, U.S.A.). p-Values ≤ 0.05 were considered to indicate statistically significant differences.

Results

Baseline characteristics and comorbidities

Between 1999 and 2004, 4,323 patients and 1,886 employees had a diagnosis of epilepsy (ICD-9-CM: 345.x) and continuous eligibility in 2004 and 2005. The estimated prevalence of epilepsy in this study is 0.50% among all patients and 0.68% among employees. Among patients with epilepsy, 1,859 beneficiaries (43%) and 758 employees (40%) had partial onset seizures. (Table 1). Patients with partial onset seizures were on average 42 years old, and 57% were women. Patients with partial onset seizures had significantly higher baseline rates of any mental disorders (29.1% vs. 13.1%) as well as rates of specific mental health conditions such as depression (13.0% vs. 5.5%) and anxiety (5.7% vs. 3.3%); migraine (8.8% vs. 2.4%); other neurologic conditions; and a higher average CCI score (0.6 vs. 0.2) compared with controls without an epilepsy diagnosis (Table 1). Similar to patients with partial onset seizures, employees diagnosed with partial onset seizures also had significantly higher rates of baseline comorbidities compared with controls.

Table 1.   Baseline characteristics
 Patients with partial onset seizures (n = 1,859) Patient controls (n = 1,859)p-ValueEmployees with partial onset seizures (n = 758) Employee controls (n = 758)p-Value
  1. CNS, central nervous system; SD, standard deviation.

Demographics
 Age (mean, SD)42.2 (14.7)42.2 (14.7)1.00048.3 (10.5)48.3 (10.5)1.000
 Male42.7%42.7%1.00054.1%54.1%1.000
Region
 New England and Middle Atlantic12.8%12.8%1.00012.1%12.1%1.000
 South Atlantic24.7%24.7%1.00025.1%25.1%1.000
 Central47.7%47.7%1.00047.5%47.5%1.000
 Mountain, Pacific, Hawaii, and Alaska14.9%14.9%1.00015.3%15.3%1.000
Employed40.8%40.8%1.000100.0%100.0%1.000
Comorbidities
 Any mental disorders 29.1%13.1%<0.00127.0%13.3%<0.001
 Migraine8.8%2.4%<0.0019.2%3.6%<0.001
Other neurologic disorders
 Cerebral degenerations1.8%0.1%<0.0011.2%0.0%0.003
 Multiple sclerosis/other demyelinating diseases of the CNS1.7%0.4%<0.0011.7%0.5%0.028
 Convulsions40.3%0.3%<0.00138.0%0.3%<0.001
Charlson comorbidity index, mean (SD)
 Comorbidities included in the Charlson comorbidity index0.6 (1.2)0.2 (0.7)<0.0010.7 (1.4)0.3 (0.8)<0.001

Direct and indirect resources use

Patients with partial onset seizures had significantly higher direct and indirect resources use during the study period compared with matched controls without epilepsy (Table 2). On average, patients with partial onset seizures had 0.30 inpatient visits per patient per year compared with 0.10 visits for controls; 0.69 versus 0.30 ED visits; and 14.5 versus 8.1 outpatient/other visits. Similarly, employees with partial onset seizures had significantly higher resource use compared to controls (Table 2).

Table 2.   Resource use during the study period
Resource utilizationPatients with partial onset seizures (n = 1,859)Patient controls (n = 1,859)p-ValueEmployees with partial onset seizures (n = 758)Employee controls (n = 758)p-Value
  1. SD, standard deviation.

Proportion of patients with at least one claim for:
Direct resource use
 Inpatient visits17.4%7.4%<0.00116.8%7.7%<0.001
 Emergency department visits32.7%19.6%<0.00133.5%20.5%<0.001
 Outpatient/other visits94.7%85.3%<0.00195.4%87.7%<0.001
  Diagnostic procedure35.7%14.3%<0.00138.0%16.6%<0.001
  Neurology specialist visits53.4%4.3%<0.00154.4%4.2%<0.001
  Psychiatry/psychology specialist visits10.8%5.4%<0.00110.7%5.5%<0.001
  Primary care visits 92.0%79.7%<0.00192.4%82.9%<0.001
  Clinic visits52.6%35.2%<0.00153.6%37.3%<0.001
  Visits to chiropractor, physical therapy, physical medicine, or rehab.15.8%12.2%0.00218.1%13.1%0.007
  All other outpatient visits59.6%45.5%<0.00159.8%46.6%<0.001
Indirect resource use
 Disability    17.0%5.8%<0.001
 Medically related absenteeism    86.8%89.2%0.155
 Work loss   96.4%90.6%<0.001
Number (mean, SD) of:
 Direct resource use
  Inpatient visits0.30 (0.86)0.10 (0.43)<0.0010.28 (0.78)0.11 (0.50)<0.001
  Emergency department visits0.69 (1.89)0.30 (0.93)<0.0010.65 (1.57)0.30 (0.94)<0.001
  Outpatient/other visits14.5 (15.9)8.1 (10.9)<0.00114.3 (13.8)8.9 (11.3)<0.001
 Indirect resource use
  Disability days   39.2 (108.8)7.5 (47.1)<0.001
  Medically related absenteeism days   10.2 (15.7)7.0 (16.3)<0.001
  Work-loss days   49.5 (107.0)14.5 (49.3)<0.001

Among patients with partial onset seizures, during the study period 73.8% had at least one antiepileptic drug claim; 37.1% filled prescriptions for at least two antiepileptic drugs and had at least 90 days overlap in medication supply. Rates of epilepsy-related medical services during the study period were 11.1% for inpatient visits, 8.8% for ED visits, and 72.6% for outpatient/other visits.

As for work-loss burden, employees with partial onset seizures had a significantly higher rate of disability (17.0% for employees with epilepsy and 5.8% for controls) leading to significantly higher average disability days compared with employed controls (39.2 days for employees with partial onset seizures and 7.5 days for controls). The rate of medically related absenteeism was not significantly different between employees with epilepsy and controls, but employees with epilepsy had a significantly higher number of medically related absenteeism days. The total annual number of work-loss days was on average 49.5 days among employees, with partial onset seizures versus 14.5 days among employed controls (Table 2).

Unadjusted direct and indirect costs

Annual direct costs were $11,276 per patient with partial onset seizures and $4,087 per control, with an excess annual per epilepsy patient cost of $7,190 (p < 0.001). Outpatient services accounted for 34%, inpatient services for 28%, and drug costs for 29% of partial onset seizure patients’ annual direct costs (Table 3).

Table 3.   Univariate direct and indirect cost comparisons
 Patients with partial onset seizures (n = 1,859) Patient controls (n = 1,859)p-ValueEmployees with partial onset seizures (n = 758) Employee controls (n = 758)p-Value
MeanSDMeanSDMeanSDMeanSD
  1. SD, standard deviation.

Total direct cost$11,276$26,265$4,087$13,866<0.001$10,652$28,091$4,393$12,729<0.001
Drug cost$3,234$5,139$828$2,066<0.001$2,987$3,959$1,069$2,651<0.001
Medical cost$8,042$25,003$3,258$13,451<0.001$7,665$27,091$3,324$12,176<0.001
 Inpatient $3,156$16,999$878$6,616<0.001$3,247$19,773$1,010$6,978<0.001
 Emergency department $315$1,188$131$697<0.001$301$1,239$125$559<0.001
 Outpatient$3,880$10,460$2,054$9,185<0.001$3,486$8,013$2,069$6,855<0.001
 Other medical costs$692$5,868$197$1,849<0.001$630$7,500$120$397<0.001
Total indirect costs     $3,431$8,668$1,511$4,543<0.001
Disability     $2,077$8,623$508$3,317<0.001
Medically related absenteeism     $1,354$2,178$1,003$3,094<0.001
Total costs     $14,083$31,990$5,904$14,663<0.001

Among patients with partial onset seizures, epilepsy-related costs per patient were on average $3,290 (29% of direct costs). Drug costs ($1,350) were approximately 41% of epilepsy-related costs, but accounted for only 12% of the annual direct costs of patient with partial onset seizures. Epilepsy-related inpatient costs were $1,125, outpatient/other were $753, and ED costs were $62.

Mental health–related costs in patients with partial onset seizures accounted for 11% of nonepilepsy–related direct costs. Patients with partial onset seizures had on average significantly higher annual mental health–related costs ($900 total direct mental health–related costs; $361 for prescription drugs, $316 for inpatient, $9 for ED, and $213 for outpatient/other costs) compared with patients without epilepsy ($257), p < 0.001.

Annual indirect costs were $3,431 for employees with partial onset seizures and $1,511 for controls, p < 0.001; $1,919 annual per patient difference. Employees with partial onset seizures had significantly higher disability- and medically related absenteeism costs. Total costs including direct and indirect costs were also significantly higher in employees ($14,083) with partial onset seizures compared with controls ($5,904), p < 0.001; annual per employee difference of $8,178 (Table 3).

Risk-adjusted direct, indirect, and total costs

Direct cost differences between patients with partial onset seizures and controls remain statistically significant and substantial even after adjusting for differences in baseline comorbidities. Annual risk-adjusted direct costs per patient were $10,312 ± $761 [mean ± standard deviation (SD)] among patients with partial onset seizures and $5,272 ± $741 among controls, p < 0.001; annual per patient difference of $5,040. Among employees, risk-adjusted direct costs were $9,151 ± $920 among employees with partial onset seizures and $5,847 ± $802 among controls, p < 0.001; risk-adjusted indirect costs were $2,897 ± $337 among employees with partial onset seizures and $1,960 ± $236 among controls, p < 0.001; and total risk-adjusted costs were $12,262 ± $994 among employees with partial onset seizures and $7,740 ± $911 among controls, p < 0.001.

Discussion

Partial onset seizures are associated with a substantial direct and indirect economic burden among privately insured in the United States. The overall prevalence of epilepsy in this study (0.50% among patients and 0.68% among employees) was similar to the reported mean prevalence of 0.68% in the United States (Strzelczyk et al., 2008). Patients with partial onset seizure represented 43% of patients with any epilepsy diagnosis identified during sample selection, which is consistent with the literature (Casino, 1990). Patients diagnosed with partial onset seizures had on average $7,190 higher annual direct costs compared with controls without an epilepsy diagnosis. Employees with partial onset seizures had on average $1,919 higher annual indirect costs compared with employed controls. After adjusting for baseline characteristics, patients with partial onset seizures had on average $5,040 higher risk-adjusted annual direct costs and employees with partial onset seizures had on average $937 higher risk-adjusted indirect annual costs compared with controls.

Epilepsy-related costs ($3,290) accounted for 29% of annual direct costs. The excess cost burden associated with partial onset seizures is likely underestimated when only epilepsy-related costs were considered. This is in part due to the association of partial onset seizures with a substantial comorbidity burden (e.g., higher rates of mental health disorders, migraine, and other neurologic disorders, and higher CCI). Comorbidities, for example depression, may be related to and even caused by the underlying epilepsy condition, and the excess costs of treatment for such comorbidities could be in part attributed to epilepsy. During the study period, patients with partial onset seizures had on average statistically significantly higher mental health–related costs ($900) compared with patients without epilepsy ($257). Moreover, medical services with epilepsy as the underlying cause may not be coded with epilepsy diagnosis in the claims data. For example, an epilepsy diagnosis may not be recorded if an epilepsy patient uses ED services due to a serious trauma (e.g., a fracture from a fall or automobile accident).

Comorbidities and medically related services that are associated with epilepsy, but may not be coded with an epilepsy diagnosis, are important in understanding the potential cost savings of new epilepsy treatments. The excess all-cause costs compared with average patients without epilepsy may be more representative of the excess costs associated with epilepsy. Partial onset seizure treatment and control create opportunity for leverage from not only reduced epilepsy-specific medical costs (e.g., fewer inpatient or ED visits with an epilepsy diagnosis), but also reduced costs for accident care, migraine, mental health services, and other conditions, as well as costs for disability- and medically related absenteeism that may be associated with the underlying epilepsy status.

Four studies (all published between 1994 and 2000) have focused on the cost of epilepsy but not specifically on partial onset seizures (Begley et al., 1994; Murray et al., 1996; Begley et al., 2000; Halpern et al., 2000; Strzelczyk et al., 2008). These studies reported direct costs associated with epilepsy-related health care use only from a societal perspective and relied on data sources such as an expert panel of physicians (Murray et al., 1996; Halpern et al., 2000), health care resource use from a single health care system (Begley et al., 1994, 2000), or respondent-provided answers to the National Medical Expenditure Survey (Halpern et al., 2000). The epilepsy-related direct costs reported in this study for partial onset seizures patients ($3,290) are higher that the previously reported direct epilepsy-related costs in epilepsy patients ($1,071–$1,490 in 1995 dollars) (Begley et al., 2000; Halpern et al., 2000) and higher than those in refractory epilepsy patients ($2,885 in 1994 dollars) (Murray et al., 1996). Because approximately 45% of partial onset seizures patients are estimated to be refractory to treatment (Casino, 1990), it is expected that costs of partial onset seizures patients will be intermediate, between those of epilepsy patients in general and refractory epilepsy patients. The higher costs estimated in this study can be explained by increased costs associated with medical services and diagnostic procedures from 1994/1995 to 2005 as well as higher costs for antiepileptic drug treatment. The reported costs of antiepileptic drug treatment in earlier studies reflected primarily the cost of phenytoin, divalproex, and carbamazepine. The cost of newer drug treatments as well as their impact on symptom control was not considered in earlier studies. This study sample included patients with existing partial onset seizures diagnosis rather than new onset. Costs in patients with existing diagnosis are likely lower than in patients with epilepsy onset (Begley et al., 2001). Moreover, costs may vary in patients with partial onset seizures only versus multiple seizure types and in active and inactive patients.

The most comparable study of the cost of epilepsy is one that the authors recently completed and found that the excess annual per patient cost among 4,323 patients with epilepsy (43% of them had partial onset seizures) was $6,396 in 2005 dollars compared with demographically matched controls without epilepsy. (Birnbaum et al., 2008a). That is, the excess direct cost per epilepsy patients was 11% less than the excess direct cost per patient with partial onset seizures reported here ($7,190), and epilepsy-related costs were $2,057, or 37% less than the epilepsy-related costs reported in this study ($3,290).

Indirect costs in published U.S. studies were based on estimates from physician panel (Murray et al., 1996) and survey data (Halpern et al., 2000) and included loss of earnings associated with unemployment and underemployment, productivity losses, as well as loss of caregiver earnings not considered in this study. The study estimates reported here of indirect costs, which include actual disability payments and medically related absenteeism costs from an employer perspective, are not comparable to indirect costs from a societal perspective reported previously. Indirect costs reported here account for 24.3% of total costs of employees with partial onset seizures, and incremental indirect costs are 23.5% of the excess total costs of employees with epilepsy compared with controls. In contrast to the finding that epilepsy patients had lower excess direct costs than patients with partial onset seizures, excess indirect costs for employees with partial onset seizures are similar to those of employees with epilepsy reported elsewhere (Birnbaum et al., 2008b) ($1,919 vs. $1,950, respectively).

Compared to patients with chronic conditions, employees with partial onset seizures have higher costs than employees with diabetes without diabetic retinopathy ($11,898 in total, $9,525 direct and $2,374 indirect costs in 2005 dollars) and lower costs than employees with multiple sclerosis ($21,046 in total, $15,277 direct and $5,769 indirect costs in 2006 dollars) from published studies using the same retrospective data source (Lee et al., 2008; Ivanova et al., 2009).

The advantage of this study is drawing a research sample of beneficiaries and employees with partial onset seizure from a geographically diverse database of many large employers with medical, pharmaceutical, and short- and long-term disability claims. Moreover, resource use and costs are based on actual real-world practice and are reflective of a relatively recent period (2005).

The study results, however, may not be generalizable to the population of patients with partial onset seizures in the United States. The study included only privately insured beneficiaries, ages 16–64, and so may not apply to those patients covered by government health insurance benefits. The indirect cost analyses were further limited to employees, ages 18–64, who were required to be actively employed for the duration of the study period; patients who are permanent disabled and are supported by government programs (such as through the U.S. Social Security’s Supplemental Security Income program) are not included in this analysis. Note also that the usual limitations that are associated with use of claims data and the absence of clinical measures that could provide severity of illness information also apply here. For example, the identification of epilepsy and type of seizures relies on diagnosis codes in medical codes. Information about remission status is also not available.

A topic for future research is the role of antiepileptic treatment on costs over time. To the extent that epilepsy treatment leads to stabilization in the patient’s condition and improved control over partial onset seizure events, costs will stabilize, too. It would also be useful to investigate treatment patterns and their implications for utilization, such as the role of switching, as well as discontinuation, of medication and association with adverse events.

In summary, partial onset seizures pose a substantial and statistically significant direct and indirect economic burden. The excess direct costs of patients with partial onset seizures are underestimated when only epilepsy-related costs are considered.

Acknowledgments

Research support was provided to Analysis Group, Inc. by Janssen Pharmaceutica N.V., Beerse, Belgium. The authors also thank Sanjay Merchant, PhD, Johnson & Johnson Pharmaceutical Services, L.L.C, for his review and editorial feedback on the manuscript.

The authors confirm that they have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Disclosure: Dr. Birnbaum and Ms. Ivanova are employees of Analysis Group, Inc. Mr. Kidolezi is a former employee of Analysis Group, Inc. Dr. Qiu was an employee of Analysis Group, Inc. at the time of the study and is currently with Merck & Co., Inc., Whitehouse Station, NJ. Ms. Caleo is a former employee of Janssen Pharmaceutica N.V., Beerse, Belgium. Ms. Caleo contributed to the study conception, manuscript revision, and provided final approval for the manuscript.

The remaining authors have no conflicts of interest.

Ancillary