Questions

We Are Asking

We shed lights on the relationship between money and healthcare, specifically:

Medicare spending efficiency:Does higher hospital healthcare spending translate to better healthcare quality?

Regional differences:Do poorer regions' spending efficiency lower?

Datasets

We Are Using

We measure hospital medicare quality by ten objective measures used by the Delmarva Foundation’s Award for Excellence and many other prestigious hospitals, as well as two subjective measures from a patients survey conducted by the Centers for Medicare and Medicaid.

Medicare spending data are also available from the same data source.

We measure regional proverty by Individual Income Tax data by zip code from IRS Tax Center.

Data Processing

Distributions are not normal. This analitical limit pursuaded us to 1)focus more on data distribution rather than correlation ;2)emphasize visual analytics by an interactive map rather than statistical approaches.

Missing data.Around 700 hospitals did not provide information on all of the measures required for the Delmarva Foundation's criteria. We left these hospitals out of our analysis. These hospitals included more hospitals that had lower Medicare expenditure, so this may have affected our data. Since the values omitted were evenly distributed on the whole, though, we believe our findings are still valid.

Geo-coding.We transferred the hospital addresses and the zipcode regions to lan/long geographical data, yet there are about 10 data points outside of US territory. We mannualy removed these data spots.

Exploration

of the question

Statistics

We compared the Medicare spending per claim and the median income of the hospital area of hospitals that met the criteria for the Delmarva Award with those that did not by looking at the difference between their distributions.
We found that hospitals that spent more on Medicare were more likely to meet the criteria for the award (p < .00001). Besides, counterintuitively, hospitals with higher median incomes are no more likely to meet the criteria for the Delmarva award than those with lower median incomes.

Visualization

Although the geographical distribution of variables is not statistically significant to our question, visual analysis shows some interesting patterns. For example, We compared the poverty level among area without a hospital, with one hospital and with multiple hospitals. We have also included these observations in our web page, hoping to inspire new interesting questions and follow-up discussions.