Background

This assignment provides practice manipulating hospital inpatient records by creating new variables based on diagnosis codes using SAS do loops and arrays. It further provides practical experience using SAS PROC REPORT to output formatted tables for publication in reports and presentations. We are using Texas Public Use Hospital Inpatient Base Data File 1 for the first quarter of 2015, which includes 763,656 records and 194 variables. This is a real, deidentified dataset – do not share it with others outside of this class. Provided in the Canvas assignment link is the SAS dataset in a zipped file and the data dictionary.

Instructions

1. Review the data dictionary and run a PROC CONTENTS to familiarize yourself with the data

2. Create one overall variable for each of the following conditions using whichever codes you feel are appropriate based on the lessons and data dictionary:

a. Heart attack

b. Stroke

c. Pneumonia

d. Type 2 diabetes

e. COPD

3. Use PROC REPORT to create two tables (see page 2 for a suggested layout):

a. One showing the number and percent of people with these conditions by age group, sex, and race for everyone in the data set (exclude missing values)

b. One showing the number and percent of these conditions by the top 5 largest Texas counties by population (look at the U.S. Census Bureau or a State of Texas website)

4. Create one graph using either SAS or Tableau illustrating the data from either of the tables above

5. Remember to comment your Program with descriptions of your work

6. Save the program, log, and output windows with the naming convention “lastname_qr3_log” (change the last word to reflect which SAS window the file contains)

7. Present your findings in a 1-page report

a. Include text to describe the data used for analysis and what information is presented in this report (briefly, what was the purpose? The source of the data? Your methods?)

b. Choose a layout that flows and format tables and graphs in a consistent manner

i. DON’T COPY/PASTE FROM SAS WITHOUT TWEAKING THE FONT, ETC.

ii. Properly title and label your tables and graphs

c. Write a sentence or two as a conclusion summarizing the major findings

d. Find at least one article in PubMed that discusses the validity (sensitivity/specificity) of using diagnosis codes to define the conditions we examined, and include a couple of sentences summarizing the strengths and limitations of this method on your report

e. On a second page, include:

i. A list of references for the article(s) you used

ii. An appendix showing the ICD-9-CM codes you used to define each of the 5 conditions (you can organize them in a table)

8. Submit the three SAS files, your quarterly report, and the PDFs of the journal article(s) you used