Spending trends vary greatly across medical conditions. Nervous system conditions comprising epilepsy has shown an increase in spending growth contrary to a decrease observed in aggregate spending growth of 15 condition categories from 2000 to 2010. Increases in total spending of a medical condition can be explained by an increase in either costs per case or the number of cases or in both elements. Determining the number of epilepsy cases and the cost to treat an individual with epilepsy helps to explain spending trends of the disease. Significant variation in overall Medicare spending across geographic regions unrelated to health outcomes has been well-documented. It is uncertain whether reducing payment rates to high-cost areas would curb spending growth without adversely affecting health care quality for Medicare beneficiaries. Reducing geographic variation is therefore, desirable only if the measured variation represents inefficiencies in the health system. In terms of health care, efficiency is a function of cost of care and quality of care. The identification of factors contributing to inefficiency may guide policy change for its improvement. OBJECTIVES The overall objective of this research was to evaluate the potential for change in prevalence of epilepsy cases, the magnitude of maximum inefficiency and factors contributing to inefficiency for the treatment of epilepsy among Medicare beneficiaries. The first specific aim was to determine whether there has been change in the prevalence of epilepsy among Medicare beneficiaries since 2005. The second aim examined the effects of two value-based programs on the geographic variation of Medicare spending per beneficiary. The third aim sought to identify influential factors driving inefficiency in inpatient care among the beneficiaries with epilepsy by examining cost and quality, accounting for spatial dependence. METHODS Analyses for all specific aims included individual-, county-, and state-level data. Individual-level medical data including beneficiaries’ age, race, sex, zip code, and utilization information five percent random sample were obtained from US Medicare administrative data (2011 to 2013). Epilepsy prevalence information for Medicare beneficiaries (2001 to 2005) was estimated by a previous study. County-level data were obtained from Area Health Resources Files (AHRF) and the American Community Survey. State-level data were obtained from State Physician Workforce Data Book; Dartmouth Atlas of Health Care; Centers for Medicare and Medicaid Services; Tracking Accountability in Government Grants System; US Department of Commerce; National Association of Epilepsy Centers; and US Census Bureau. Epilepsy cases were defined using Medicare claims data with any of the following International Classification of Disease-Version 9-Clinical Modification (ICD 9-CM) diagnostic codes: At least one ICD 9-CM 345.xx (epilepsy), or at least two ICD 9-CM 780.3x (seizure) claims occurring at least 30 days apart. Inpatient inefficiency was defined as a function of cost over quality. Inpatient cost was defined by state-level average adjusted inpatient services spending per hospital stay (AIH). The proxy measure for the quality of inpatient care for beneficiaries with epilepsy was the proportion of hospital stays with an epilepsy or seizure admission diagnosis (PHE). Association analysis was performed using the Spearman correlation coefficient. Generalized linear models with log link and gamma distribution were used for the adjusting and modeling of cost dependent variables. Spatial regression models were used when appropriate to account for spatial dependence. RESULTS The prevalence of epilepsy among older Medicare beneficiaries was estimated to be 22.2 cases per 1,000 persons (2011 to 2013). An increase was observed for all racial groups. However, the subgroup with highest prevalence estimate shifted from the younger age group of 65 to 69 years to the female, 85 years and older. Black beneficiaries persistently had the highest prevalence compared to other racial groups. Analysis for the second specific aim showed that state-level total medical expenditures per beneficiary with epilepsy varied from 11,690 to 29,048 (average 19,890, SD 3,774, US$ 2013), 5.3 times the spending variation for those without epilepsy which ranged from 6,466 to 9,458 (average 7,631, SD 710, US$ 2013). Post-implementation of two value-based programs (hospital readmissions reduction program (HRRP) and the hospital value-based purchasing program (HVBP)), spending variation decreased for both the epilepsy and non-epilepsy cohorts (-14.6% and -9.0% respectively). The primary factor contributing to spending variation was health status for beneficiaries with epilepsy (51.9% of variation) and location of the beneficiary for those without epilepsy (26.1% of variation). Analysis conducted to address the third specific aim showed that different factors influenced inefficiency in inpatient care of beneficiaries with epilepsy among US census regions. For the Northeast region, the number of primary physicians was an inefficiency factor. For the South region, inpatient inefficiency factors included the number of medical residents and fellows, proportion of physicians who were primary care physicians, and retention of physicians who graduated from an institution in the state of practice. Some evidence of defensive medicine was detected in the West region while no specific factors were influential to inpatient inefficiency in the Midwest region. The highest and lowest state-level average adjusted inpatient services spending per hospital stay (AIH) were observed in the District of Columbia (13,376 US$ 2013) and South Dakota (7,901 US$ 2013). Rhode Island (1.06%) had the lowest while Idaho (11.29%) had the highest proportion of hospital stays with an epilepsy admission diagnosis (PHE). Rhode Island also had the lowest inpatient inefficiency index or least inefficient (86) compared to the highest inpatient inefficiency index or most inefficient observed in Idaho (1,417). CONCLUSION The prevalence of epilepsy among Medicare beneficiaries appeared to have increased from previous estimates. Heterogeneity among the 48 contiguous states and District of Columbia with respect to inefficiency in inpatient care was detected. Across-the-board cost reduction policy based on cost alone may not be appropriate for all geographic areas across the US and may even be detrimental to health outcomes in some areas. On both national and regional level, inpatient inefficiency was significantly associated with PHE but not with AIH, indicating that the focus to decreasing inpatient inefficiency for beneficiaries with epilepsy should be based on increasing quality or decreasing PHE. Changes made to decrease PHE (increase in quality) may also increase AIH (increase in cost); therefore, it would be wise to monitor both cost and quality when considering policy change while focusing on quality improvement. Programs such as the HVBP and HRRP that link cost to outcomes appeared to be successful in reducing geographic variation of medical expenditures. Instead of total spending per individual, updated knowledge of the prevalence and cost per case treated for specific chronic medical conditions may better assist resource allocation, budget planning, and health program development.