Statistical Process Improvement
Georgetown University
 

Case Study


Using real data, you are asked to compare the performance of one hospital on several quality measures so that a patient with a particular illness could decide whether they should go to the hospital.  In particular you are asked to do the following:

  1. Select a hospital to analyze and an individual within the hospital to receive your report. Usually this person is a medical director or Chief of Quality. Obtain their email by searching on the web in sites such as Linked In (r). The instructors Linked In site can be useful to you as there are many individuals who are in the instructor's contact list.  If the person is not on the instructor's contact list, ask him to add the person.  Then obtain his email from the instructor.  If you cannot find the exact email, guess it by comparing it to emails that you do find for someone else in the organization.  It is OK if you send the email and there is no response.
    • Vivian-Jo Yu on finding a preceptor Slides►
    • Copy and paste the letterLetter of introduction from the instructor to your preceptor. Do not attach this letter to your email, instead copy the content of the draft letter into your email.  Copy and paste the entire letter into your email. Include the logo and the write-up post the signature line.  Do not change any aspect of the letter, this letter was approved by the University legal office. Include the instructors in your email so they can follow-up with responses to the email. Draft Email►
  2. Choose the type of patients you want to focus on.  Choose at least 5 quality measures that are appropriate for a patient with a particular illness. Choose these indicators so that you can create a cohesive picture of the quality for a patient with a particular illness. The quality indicators are listed in the web site on Hospital Compare. Select the hospital peer organizations (at least 3).  These are sometimes local competitors and other times national competitors.  Make sure your selection makes sense, in terms of hospital size and market reach.   
  3. Download real data for your selected hospital.  Download the data for the hospital's peer group. Data► How to Download►
  4. Examine hospital's performance. Create appropriate control charts for each quality indicators.  In control charts, set the control limits using data from hospital's peer group.
  5. Prepare a narrated report posted to the web.  See instructions below on how to do so. Your grade is not affected if your preceptor does not respond to the emails.
    • Send your report by an email from the instructor to the preceptor. Copy the instructor in your email.  Draft Email 3►
  6. Email your report to the entire class
    • Comment on the web on 3 case studies done by your peer that have no comments posted to them. Comment on both accuracy of analysis and presentation of the findings. In leaving a comment, discuss both what worked well and what needs improvement.  Compare the work you are evaluating to your own work.  Say, for example, "Compared to my work, this case study is better because ...".  Enter your comments in Blackboard under assignments as well.      

Objectives

  • Learn how to analyze performance of an organization on multiple indicators
  • Decide on selection of the right control chart for your data
  • Create control charts for a variety of different types of data
  • Compare performance of an organization to its competitors
  • Use your work to get noticed by organizations that might hire you or provide you with a practicum project.

Check for Errors

Before preparing your case study, check that you have not made any error in your analysis: 

  1. Check that you have included all the data available.  If quarterly data are available, include all quarters and not just yearly data.  Check several years and look for measure ID in different file names.  The method of storing measures has changed over the years. 
  2. Check that you are using the right chart.  Many students make an error and use X-bar charts for measures that are reporting rates.  Go through the questions asked on our guide to chart selection and make sure that you have selected the right guide. Slides►  YouTube► Transcript►
  3. Check that the upper and lower control limits are calculated from the data of the competitors. You should not include your hospital's data in these calculations.
  4. Check that the standard deviation is calculated correctly.  If the sample size of patients examined changes, then you should have different values for standard deviations in each time period.  Standard deviation calculations are done on grand average or average P or similar values and not on averages per time period.
  5. Check that the chart is displayed correctly.  Legend should  include fact that UCL comes from peer organizations.  Y-axis and X-axis should have titles.  Observation line should have markers and be in a non-red color.  Control limits should be red and without markers.  Font size for axis information should be higher than, or be, 12 points.  If control limits are projected, the projected portion should be a dash line. 
  6. Check that the control chart is interpreted correctly.  When control limits are based on competitors, the chart should always be interpreted as saying whether the hospital is different from its competitors.  You should not make a statement that "the hospital is improving."  Instead you should say "the hospital is improving compared to its competitors."  The point is that you should reference the competitors in your statements about the hospital.  Avoid using adjectives such as "great," "small," or "excellent."  Try to be less judgmental and more direct about what the data show.  Describe the data with as little judgment as possible. Examples►

Case Study Report

Prepare a presentation that includes the following slides:

  1. Title slide  (includes hospital name, your name, date of analysis, and reference to the course). 
  2. Source of data
    • Rationale for selecting the measures (this should make logical sense)
    • List of measures and data available on each measure
  3. Limitations of data (use a separate slide for each limitation)
  4. Rationale for selection of hospital competitors
    • List of hospital competitors
  5. Control chart for indicator 1 
  6. Add to the chart interpretation of the data
  7. Control chart for indicator 2 
  8. Add to the chart interpretation of the data
  9. Control chart for indicator 3 
  10. Add to the chart interpretation of the data
  11. Control chart for indicator 4 
  12. Add to the chart interpretation of the data
  13. Control chart for indicator 5 
  14. Add to the chart interpretation of the data
  15. Summary of Findings (do not over interpret)

Narrate the slides in one sitting.  After you narrate the slides, reduce the text so no more than 7 words are on the slide. Narrate the slides using Microsoft PowerPoint narration feature. Make the chart as large as possible, taking the entire slide space.  Circle points that are outside the control limits.  Occasional variations in narrations is OK.  Save your narration as a movie.  Upload your narration.  Share the URL of your uploaded narrated slides with  instructor and other students through Blackboard.  Do not send the slides or the narration.  Send only the URL.  If you do not want to post your work publicly, you can take all the steps and post privately or ask for other options. Here are some videos and sites that can be helpful in carrying out this case study.

Examples

See many examples of case studies done by other students in the past. More►

Case Study Evaluation

Evaluations of your project are based on comments left by students, faculty, and contact person on your work.  Students are expected to leave comments on at least 3 presentations. In leaving a comment, please address both the presentation and the accuracy/quality of the work.  Your grade is not affected by lack of response from your sponsor. 


This page is part of the course on Statistical Process Control. Most recent revision 05/15/22.