Mini Comparative Effectiveness Project
The objective of this assignment is to teach you how to conduct a
comparative effectiveness study using electronic health records. You are asked to
analyze existing observational data. These can be obtained from
electronic health records or if you do not have access to the necessary
data you can rely on web data (see the section on sentiment analysis).
To make it easier, you do not need to rely on massive data and can
conduct a pilot using 5 cases and potentially 50 controls. You
analyze the data, prepare a narrated presentation, and provide details
of how the analysis was completed.
You need to rely on observational data available to you on the
web or through an electronic health record. There should be an
intervention that you are evaluating. Patients who receive the
intervention are called cases and patients matched to cases are called control.
There should be at least 5 cases and two matched controls for each case.
You may need to look through 50 potential controls to find the matched controls.
Here is a list of data you need:
- Definition of cases and matched controls. The definition must be done in terms of variables found in an electronic health record or on the web. Others should be able to duplicate the definition.
- Definition of observation and follow-up periods. In the
observation period, cases and controls are matched so that they have the
same outcomes. In the follow-up period, we test if the cases and
controls have different outcomes.
- For each case and control the following data are needed:
- A de-identified unique patient ID
- Dates of outcomes and nature of outcomes during observation and
- To learn how analyze data see the reading on comparative
- Analyze the data using Excel. Except for original data,
all variables should be calculated using functions available within
Excel. For example, use "if" function to match cases to
controls. Use Excel comparison of means to verify that cases
and controls match each other during the observation period.
All analysis must be done so that if the data changes the result
will change without any additional data entry.
- Calculate the difference in cases and controls through survival
Present your work as a narrated slides posted to
the Author Stream or other public sites.
Make sure the following are part of your presentations:
- Title slide with your name or alias
- The need for comparative effectiveness study
- Review of similar studies published in PubMed
- Definition of terms
- Verification of matching
- Difference of cases and controls
- Introspection on value of matched case controls
Make the presentation look good:
- Consistent font and size for the text
- Take into account that web displays are 1/4 size of web
displays. Make sure that all elements are readable in the
- No more than one point per slide. No multi-point slides
- Narrate at CD quality
- No spelling or grammatical errors
Submit the URL for your presentation to your instructor
This page is part of the course on
Information Systems. This page was edited
by Farrokh Alemi, Ph.D. ęCopyright protected.