Iterative metrics and Medicare Spending per Beneficiary

Today's New York Times carries a front-page article describing Medicare's proposal to include a period 3 days before a patient hospitalization and 90 days after to calculate a "Medicare Spending per Beneficiary Episode" to be included in payment calculations. You can read the original Federal Register notice here (76 FR 25788; it was posted May 5). The relevant section begins about 109 pages in. In effect, Medicare will now hold hospitals accountable for coordinating related care and services for 90 days after a hospitalization.

Representatives of some hospital associations are complaining that hospitals will lose track of patients two to three months after an episode, and others say that the proposed rules will be hard on hospitals that serve many low-income patients. These are typical comments providers make when a new metric is announced. (The Times also reports that some hospitals are saying, in effect, "Bring it on!")

If I were working in a hospital, or thinking about applying a similar methodology to my operations, I would recommend remembering two key points about methodology.

First, the baseline may not be correct. Funders or others looking at results like this need to be flexible enough to change the baseline, if the baseline appears to be set at the wrong level.

Second, once the baseline is established correctly and the metric has worked for a few years, hospitals will respond to it and, in all likelihood, a new baseline will have to be calculated.

It will be interesting to see how the comments period plays out and how this rule is eventually established. For now it's a good illustration of how important each iteration of a metric can be.


A step in the right direction

The EPA has announced a new fuel efficiency label for cars, displaying vastly more information for consumers, including fuel economy for city driving, highway driving, and a new combined city/highway fuel efficiency data point.

The label also tells you how many gallons driving 100 miles will take, the annual fuel cost (assuming 15,000 miles driving in a year) and two ratings, one for fuel economy and one for tailpipe smog emissions.

Depending on the car, the label also tells you how much you save (or spend) in fuel costs over 5 years compared to the average new vehicle. There's also a scannable QR code for smartphones that links to additional information and tools. And you can change the default driving assumptions -- ratio of highway to city, miles traveled per year -- to reflect yours.

Equally important, the National Highway Traffic Safety Administration (NHTSA) has been raising average fuel efficiency requirements for fleets of motor vehicles that include light, and now medium and heavy-weight trucks. . . in the past, manufacturers had hidden the poor gas mileage of heavier cars like SUVs by calling them light-weight trucks. Now these cars, and trucks, will have to meet higher standards. You can look at NHTSA's Fuel Economy page for more information.

I've been away for two weeks but am now back and ready to keep posting.


Two useful NY Times articles

Today's NY Times contains not one but two interesting statistics-based articles. The first, by my favorite statistician Nate Silver, predicts what Derek Jeter's season will look like. I'm linking to it because Silver explains the basis for his predictions very clearly.

The second, also sports-related, is a useful discussion of the importance of context in understanding and using statistics, this time in the tragic (no matter what size the denominator turns out to be) efforts to understand chronic traumatic encephalopathy in former (so far) football players.

I'm looking forward to any comments on the biases I've revealed here.


Drug Testing in Baseball . . . by the numbers

The NY Times reporter Mike Schmidt has a story in today's paper about the very small number of off-season drug tests conducted by Major League Baseball. The closest I can come to finding the original report is here, a blog post which provides a little additional context: namely, that MLB has consistently conducted very few drug tests during the off-season (though the numbers have increased over three years). As Schmidt points out, 138 off-season tests, or slight more than 10% of players, seems very low. It's also a small number as a percent of all tests: 3.6%.

In explaining the small number of tests, Major League Baseball notes that the off-season is short, and that athletes may be anywhere in the world. The World Anti-Doping Agency (WADA) points out that because out-of-competition testing can be done without notice, it's an important deterrent. WADA uses a program it calls Whereabouts, in which elite athletes let testing or sports authorities know their locations during their sport's off-season, so that they can be found for surprise testing. Each country's sports federation actually picks who will be tested and conducts the testing, and testing isn't limited to athletes in the Whereabouts program. But if other international sports can conduct off-season testing, why can't MLB?

Mike is based in Baghdad at the moment -- you can keep up with his daily life through his interesting and quirky blog.

Popular Posts