Competing risks

When more than one thing can happen first, and one of them stops the clock.

Taught by

Michael Crowther, PhD

CEO & Director of Statistical Methodology

Formerly Associate Professor of Biostatistics at the University of Leicester, and the author of the merlin framework.

Research survival analysis · joint models · multi-state models · statistical software

PhD in Medical Statistics, University of Leicester, 2014

Register your interest

What you will be able to do

• Distinguish a competing risk from censoring, and say why the distinction changes the estimand
• Estimate and report a cumulative incidence function, and explain why 1 - Kaplan-Meier overstates it
• Choose between a cause-specific and a subdistribution model on the question being asked, and report both
• Fit one parametric cause-specific model per cause and turn the pair into an absolute risk for a stated patient, with an interval and the coherence check that goes with it
• Report hazard ratios and absolute risks together, say what a subdistribution hazard ratio does not license, and recognise when the problem is really a multi-state one

Before you start

• Take first: Survival analysis
• Helps to have done: Flexible parametric survival models

Curriculum

13 steps: 10 lectures, 2 practicals and 1 quiz.

Provisional — this course is still being written, and the outline will change.

Week 1 · what a competing risk is, and what censoring one costs

• 01WatchWhat a competing risk is22 min
• 02WatchWhat goes wrong if you treat it as censoring26 min
• 03PracticalPractical 1 · describing competing-risks data45 min

Week 2 · cumulative incidence, and comparing groups

Opens 7 days after you start.

• 04WatchThe cumulative incidence function and Aalen-Johansen28 min
• 05WatchComparing groups: Gray's test and what it answers24 min
• 06PracticalPractical 2 · estimating the cumulative incidence50 min

Week 3 · regression, and which model answers which question

Opens 14 days after you start.

• 07WatchCause-specific hazard models26 min
• 08WatchThe Fine and Gray subdistribution model26 min
• 09WatchChoosing between them, and reporting both24 min

Week 4 · absolute risk, and what to report

Opens 21 days after you start.

• 10WatchA parametric model for every cause26 min
• 11WatchAbsolute risk for a stated patient28 min
• 12QuizCheck yourself: hazards, risks and the gap between them10 min
• 13ReadWhat to report from a competing-risks analysis22 min