By Chelsea Reynolds
AHCJ Graduate Research Assistant
Sifting through Medicare data is no easy task for even the most statistics-savvy health reporters. But new tools from Dartmouth Atlas, a Medicare data analysis organization run by the Dartmouth Institute for Health Policy and Clinical Practice help streamline the process.
Kristin Bronner, Dartmouth Atlas' managing editor, presented the Web site's updated capabilities at the Health Journalism 2010 conference in Chicago on April 23, 2010. The Atlas' new site design launched the day before.
She said the old Web site was equally informative as the updated version, but it was difficult to navigate and utilize Medicare data. "We tried to make it easier to explore the database," she said.
Bronner, who serves as webmaster and content developer for the site, outlined its increased capabilities. She focused on the new "Customize this Map" menu. After you determine which Medicare-specific data you wish to analyze (such as Medicare reimbursements per enrollee or hospital discharges per 1,000 Medicare enrollees), you can choose "Customize this Map" to view the data reports in five formats.
The "Table" tool displays the Medicare data in a standard chart: by specific location (state, city, or hospital) and by the data set you wish to view.
The "Trend" button allows you to select data from up to 10 different geographic regions and display chronological numerical trends for that data set for all the years Dartmouth Atlas has compiled data for. The trends are displayed in an easy-to-understand line graph.
The "Map" option presents the specified data in a color-coded map of the United States. Depending on your data set, you may select "Hospital Service Area" to visualize Medicare city-specific data for each represented major hospital in the country. This is particularly useful for reporters working on local health beats.
The "Bar Graph" application is similar to the "Trend" tool. It displays the data in a traditional two-axis bar graph so you can easily compare information from one data set to another. Bronner detailed how journalists can use this to show disparities in Medicare costs for many different hospitals within Chicago. Certainly the same steps could be taken to analyze Medicare quality for any major metro area.
And last, the "Distribution" option provides the frequency of specific Medicare data responses in a scatterplot. This makes it simple to visualize the "bell curve" for the data you specify. It also provides a rough context by which to view standard deviations and outliers for the data set you're viewing.
After discussing the five primary Dartmouth Atlas data selection tools, Bronner also highlighted the "Hospital Care Intensity" data set, which indicates the level of care that Medicare patients receive in their last two years of life. It also details each hospital's percentile rating for quality of care within its state.
"If you don't really know anything about our data but you want to get a quick snapshot, use this tool," Bronner said.
The lecture closed with a question and answer session that encompassed specific aspects of the Web site's usability. Bronner emphasized one take-home point: Dartmouth Atlas' Medicare data is reliable and statistically sound.
