What can hospital data tell you?

Summary:

We were given data from Massachusetts General Hospital. This data followed 974 patients over the years of 2011 - 2022. We were asked to answer the following questions:

1. How many patients have been admitted or readmitted over time?
2. How long are patients staying in the hospital, on average?
3. How much is the average cost per visit?
4. How many procedures are covered by insurance?

Discussion:

Question 1 never defined what is meant by "admitted". Did they just want to know about the class of inpatients? It was decided to interpret "admitted" as an encounter with the hospital. Using the number of all encounters and not just the inpatient counts showed a similar pattern for H1N1 influenza and Covid years.

Question 2 appears to be only interested once again in the inpatients. However, viewing all encounter Classes shows the majority of encounters are resolved within 15 - 30 minutes. It was decided to add a KPI to watch the time spans on a monthly level.

Question 3 wants the average, which is above the 3rd quartile, (75% of data). It was decided to use the median and a chart was included to show why this is the correct choice.

Question 4 never defined the term "many". It could be viewed as how many "total" encounters did the insurance cover? Or it could mean the "distinct count" of procedures each insurance provider covers? It was decided to use the "total" instead of the "distinct count" because "distinct count" would not necessarily display the most used insurance provider.