Propensity Score Analysis in Healthcare Data

Ehsan Karim
ehsank.com

Link to webinar RMD files on GitHub.

Summary

This webinar will focus on learning causal inference approaches in a healthcare data analysis context with a particular focus on explaining the application of propensity score analysis in a real-world data analysis context. The session will outline how these analyses are different than conventional regression methods and will address key assumptions/diagnostics of these models.

Objectives

  • Describe the basic concepts associated with causal inference and propensity score approaches
  • Review guidelines for applying propensity score methods (e.g., matching and inverse probability weighting)
  • Demonstrate a propensity score analysis using R software and a sample training dataset
  • Explain assumptions and diagnostics of propensity score analyses
  • Discuss the best practices associated with propensity score analyses
  • Outline some extensions of this approach in solving complex real-world problems in the healthcare data analysis context (e.g., in longitudinal and big-data context).
  • To learn more about the research case example and related training dataset that will be used for this webinar session, please see the following web links:

Requirement

Background in causal inference is not required. Attendees should have prerequisite knowledge of multiple regression analysis and working knowledge in R (e.g., basic data manipulation and regression fitting). In the webinar, R will be the primary software package used to demonstrate the implementations.

Sample Data Source

  • Research Article: The effectiveness of Right Heart Catheterization (RHC) in the initial care of critically ill patients.
  • Details about the training dataset:
  • Link to download the training dataset.

Lab Materials

# Topic Lab Links
1 RHC Data manipulation Part 1
2 Propensity score matching Part 2
3 Propensity score matching, more options Part 3
4 Propensity score weighting - ATT Part 4
5 Propensity score weighting - ATE Part 5
6 Multivariate Regression Part 6

Software Requirements

It is assumed that you have the following software packages installed. Webinar does not provide any installation support. Note that, working on software during the webinar is not mandatory. But if the participant like, they are welcome to browse through the webinar slides (as well as check out other materials) in their own laptop.

  • R from the following sources (installing either one is fine)
  • IDE
    • RStudio desktop (installation necessary)
    • Online accounts (no installation necessary, a supported browser is fine)

References