Given that it is a fifth of the economy, there are a variety of sources attempting to track, categorize and analyze health spending. The Centers for Medicare and Medicaid Services produces the National Health Expenditure Accounts, which capture aggregate spending, and divide it by payer and by category of spending. There are surveys which attempt to capture individual health spending, like the Medical Expenditure Panel Survey and the Medicare Current Beneficiary Survey. And there are large payer databases, like the Medicare claims data and private payer data warehouses like that assembled by the Health Care Cost Institute. Each of these has deficiencies, depending on the purpose for which it is used. Of particular interest to policymakers is understanding trends among high-cost patients and among specific illnesses and service or product categories. In a new paper, researchers lay out an approach of harmonizing survey results of individual spending with the larger national health spending analyses. (NBER Paper)
Personal health care, or goods and services used by individuals in relation to their health, account for around 84% of all health spending. The researchers used 2002, which seems very data, but apparently was a year in which they could obtain good data across all sources. They identified spending both for institutionalized and community-dwelling populations, which is important because some surveys miss institutionalized patients and they tend to be high-cost. The authors also made certain exclusions and adjustments to the various data sources to align them. The goal was to be able to extrapolate data from the individual surveys to the full national spending analyses. At a high level, surveys appear to under-estimate national health spending by about 13%, and surveys appear to undercount high-spending individuals. Out of about $1.04 trillion in national health care spending in 2002 covered by this analysis, $205 billion was on institutionalized persons. Next steps, now that the basic analysis has been completed, might be to conduct more accurate tracking of spending and trends for various subgroups in the population.