The Medicare Advisory Payment Commission continues its examination of regional differences in Medicare spending by looking at another measure–variation in service use. (MedPAC Report) But MedPAC actually defines service use in a way that is really just another measure of spending. It took the raw spending files and adjusted them for local wage and disproportionate hospital share payments and for beneficiary health status and demographics. The primary analysis then looked at differences among metropolitan statistical areas over the three years from 2006-2008.
Service use, as computed by the Commission, has less variation than raw spending. Variation in service use in 30% higher for beneficiaries in the areas at the 9oth percentile of service use, compared to 55% higher for raw spending. Service use is about two times higher in the highest area compared to the lowest, whereas it is 2.5 times higher for spending. The variation among service use is similar for decedents and nondecedents in a year, so the differences are not likely due to end-of-life care differences. The greatest variation is in post-acute care–nursing homes and home health care, but variation is correlated across types of care.
The report specifically looks at McAllen, Texas and finds its higher service use is due to post-acute care. Similarly, DME and home health use are high in Miami, where fraud is widely prevalent in those sectors. Looking at drug use, the report finds less variation in service use, although it is still present. The Commission’s explanation for the variation includes differences in practice patterns, “entrepreneurial tendencies” (love that euphemism), beneficiaries’ preferences and some other factors. Interestingly, growth in service use is slightly negatively correlated with the amount of service use spending. This suggest some regression to the mean and that some of the variation is just statistical randomness.