Medicare spending is concentrated among a few high-cost beneficiaries who are often targeted by cost-saving interventions. The Continuous Medicare History Sample file was used to analyze trends in the spending concentration over thirty years. Annual expenditures became less concentrated over time, although the year-to-year persistence of person-level high costs remained strong. There was an increase in the prevalence of chronic conditions among high-cost beneficiaries, which supports the rationale for focusing cost-saving interventions on chronic disease management. However, the decrease in concentration may reduce the potential savings from interventions focused on such beneficiaries. Annual spending has become less concentrated over time, with the top 5 percent of spenders accounting for less of the Medicare spending total than in the 1980s. The concentration of Medicare spending among a small number of beneficiaries is well documented. In any given year, the most expensive 5 percent of beneficiaries account for roughly half of fee-for-service (FFS) Medicare costs.1 Aggregated over multiple years, spending becomes less concentrated: During the five-year period 1997–2001, the top 5 percent of beneficiaries accounted for 27.3 percent of Medicare costs.
Nonetheless, many high-cost beneficiaries have persistently high costs from year to year. Similar spending patterns have been found in non-Medicare healthcare markets. The CMHS contains enrollment and claims data for a 5 percent sample of the Medicare population, selected on the basis of the last two digits of the Health Insurance Claim (HIC) number. The file covers the most recent thirty-year period for which data are available. Data on cost and utilization are aggregated by type of service by calendar year. Claim-specific data such as diagnoses and procedures are not contained in the file. Data on managed care enrollment and state buy-in status are included for each year of Medicare entitlement. The CMHS is longitudinal; that is, beneficiaries remain in the file unless a change in their HIC numbers causes them to drop out of the digital sample.