Wrangling Survival Data:
From Time-dependent Covariates to Multistate Endpoints

R/Medicine 2019 conference

---

🗓 September 12, 2019
🕗 01:00pm - 05:00pm
📍 Boston, MA
✅ Register

---

Overview

In this four-hour workshop, I will review basic code for standard survival analysis, then work through examples for more complex scenarios. The majority of the focus will be on creating the correct dataset using tools available in the survival package. The final portion of the class will explore the most complex scenario - multistate data. The format will be a mix of lecture and hands-on exercises.

Learning objectives

• Understand different survival data formats
• Create start/stop data using functions in the survival package
• Check data for errors
• Create multistate data and perform basic analysis

Is this course for me?

This workshop is targeted at people who work in the medical field with survival data (or who anticipate needing to work with it). A very basic understanding of survival analysis will be helpful, though not required. A basic understanding of R is assumed.

• Have you had to manipulate data prior to running time-to-event analysis?
• Do you anticipate needing to run analyses with time-dependent covariates or multiple endpoints?
  • Even better, have you used the survival package?
• Have you used R with the RStudio Integrated Development Environment (IDE)? Are you familiar with the various “panes” and “tabs”? For instance, can you quickly find all objects in your current global environment, and can you send R code from a source file (.R, .Rmd) to the console?

Schedule

Time	Activity
01:00 - 01:50	Session 1 (basic analysis, motivation)
01:50 - 02:00	Break
02:00 - 02:50	Session 2 (intro to tools, examples)
02:50 - 03:00	Break
03:00 - 03:50	Session 3 (check data, common mistakes)
03:50 - 04:00	Break
04:00 - 04:50	Session 4 (multistate data)
04:50 - 05:00	Wrap-up / Overtime

Instructor

Beth Atkinson has been a statistician at Mayo Clinic for 29 years and has worked with Splus then R since starting at Mayo, including development work on the rpart package. She has worked extensively with Terry Therneau, author of the survival package, and has spent many hours wrangling data so that it is set up appropriately for time-to-event analyses. Currently she is working with Terry and Cindy Crowson on a book focused on time-to-event analyses, which will include an on-line compendium of detailed examples.

Pre-work

Welcome to the Wrangling Survival Data workshop! I look forward to meeting you in person. Before attending the workshop, please complete the following prework:

1. Sign up for a free RStudio Cloud account at https://rstudio.cloud/ before the workshop. I recommend logging in with an existing Google or GitHub account, if you have one (rather than creating a new account with another password you have to remember).

2. Please bring a laptop that has the following installed:

   • A recent version of R (>=3.6.0), which is available for free at https://cloud.r-project.org/
   • A recent version of RStudio Desktop (>=1.2), available for free (RStudio Desktop Open Source License)
   • The R packages we will use, which you can install by connecting to the internet, opening RStudio, and running at the command line:

   install.packages(c("survival", "survminer", "tidyverse",
                      "lubridate", "arsenal", "broom"))

3. Don’t forget your power cord!

On the day of the workshop, I’ll provide you with an RStudio Cloud project that contains all of the course materials. We will use the software listed above only as an important backup should there be problems with the on-site internet connection.

---

This work is licensed under a Creative Commons Attribution 4.0 International License.