Introduction to Survival/Time-to-Event Data Analysis

24 February 2021–25 February 2021, 9:30 am–1:00 pm

This course introduces the concept of modelling time-to event data, commonly known as survival analysis. Log-rank tests and Cox proportional hazard regression models are used to determine associations between different factors and the event occurring.

NOTE: Due to the coronavirus outbreak, all courses will now be delivered online through a live video feed. You can expect the same level of group and individual support as you would have received in our face-to-face courses.

Survival/time-to-event analysis is appropriate when the outcome of interest is an event and that event has not occurred for everyone in the dataset. The outcome can be something negative (for example death, recurrence of tumour) or something positive (for example, recovery, task completion).

The simplest analysis is the log-rank test that assesses differences according to a single factor. Cox proportional hazards regression is appropriate to investigate the rate at which the event occurs according to several potential predictors.

This workshop gives an introduction to time-to-event (survival) data for non-statisticians, and covers the following topics:

• Use of Kaplan-Meier
• Life tables
• Cox regression analyses
• Hazard ratios
• How to set the data up for analysis
• Including interaction terms in the models
• Assessing modelInterpretation of SPSS output

SPSS outputs are given and we consider how to interpret these to determine the best model and to assess goodness-of-fit.The course will be of use to users of alternative statistical packages too as the concepts discussed throughout the course are generally applicable. The dataset may be taken away and analysed within other packages.

Note that this course does not involve hands on use of a stats package, we will consider only pre-prepared printouts.