Multilevel and frailty survival models

Patients arrive in clusters and the clusters differ. A survival model that says nothing about that is answering a different question.

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

• Say what clustering does to a survival analysis, and why a cluster-robust standard error is a different answer from a frailty model rather than a cheaper one
• Fit shared frailty, random-slope and nested multilevel survival models with merlin's random-effect element syntax, and check each nested fit against the model inside it
• Choose a frailty distribution knowingly, and report the variance on a scale a reader can use
• Predict cluster-specific and population-averaged survival, and say which of the two the question asked for

Before you start

• Take first: Survival analysis

Curriculum

13 steps: 11 lectures, 1 practical and 1 quiz.

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

Week 1 · what clustering does to a survival analysis

• 01WatchClustered survival data, and the cohort this course uses22 min
• 02WatchCluster-robust standard errors, and what they do not do24 min
• 03WatchTwo estimands, not one answer26 min
• 04PracticalPractical · what clustering does to a standard error45 min

Week 2 · the shared frailty model

Opens 7 days after you start.

• 05WatchThe shared frailty model26 min
• 06WatchChoosing a frailty distribution22 min
• 07WatchWhat the variance means, and how to report it22 min
• 08QuizCheck yourself: clustering, estimands and variances10 min

Week 3 · beyond a shared intercept

Opens 14 days after you start.

• 09WatchWhen the treatment effect varies across clusters26 min
• 10WatchThree levels, and where merlin stops24 min

Week 4 · prediction after a multilevel survival model

Opens 21 days after you start.

• 11WatchCluster-specific and population-averaged survival26 min
• 12WatchThe BLUPs, and checking them before you trust them26 min
• 13ReadReporting a multilevel survival analysis20 min