The Centers for Medicare and Medicaid Services appear to be on to something with their promised new datasets. It’s a leap for an agency whose previous offerings were a confusing mishmash of poorly-labeled files. If they continue to add granularity as they roll out more features, journalists could have a useful and innovative set of tools on their hands.
The “Dashboard” CMS intends to create, a demo of which is currently online, looks at Medicare spending on services. It’s a good tool for quickly grasping the big picture; you can view data as a bubble chart, for example, which makes it easy to see changes over time. The information is available by state, diagnosis type and hospital (at least up to the top 10 hospitals in each state), and can be downloaded in a spreadsheet. But more granularity will be needed to make this truly useful for journalists. Adding county-level data, for instance, would allow comparisons by community. Also, looking at the top 10 hospitals by state is of limited use when there are hundreds of hospitals in California alone.
There’s also a new and improved interface up at data.medicare.gov. That site allows you to compare the quality of care at various facilities. More on that in a later post.
The details about the rest of CMS’ new datasets are less clear, but a couple of them look very promising. For example, a new claims database is due out in September. The plan is to take a sample of five percent of the traditional Medicare population, strip all personal identification data out, and present it by service type (inpatient, outpatient, home health, prescription drug, etc). The claims will be accompanied by demographic data, although its not clear exactly what this will mean. It could allow comparisons by ethnicity, income level, and age. Hopefully, we’ll actually be able to see how much we’re spending on various segments of the population.
Another dataset that looks exciting and should be available by the end of the year is a large set of community health indicators. CMS says the survey will include “hundreds (ultimately thousands)” of measures of health care quality, costs, diseases and types of services received. This is the kind of data that could potentially answer questions like: as a young woman living in Brownsville, Texas with suspected breast cancer, how likely are you to be given an MRI rather than a mammogram? How about if you’re living in Marin County, California? These figures could be invaluable for public health analysis, as they’ll include obesity rates, smoking rates, etc., paired with costs and utilization of services.
As with the other data Medicare’s got in the hopper, though, the key will be the level of granularity it contains. Currently, they say county-level data is “possible,” and we’ll dig in as soon as it’s made available.