1 / 21

Every achievement originates from the seed of determination.

Every achievement originates from the seed of determination. Survival Analysis. Nonparametric Methods for Comparing Survival Distributions. Abbreviated Outline. How to formally compare 2 or more survival distributions using hypothesis tests

muncel
Download Presentation

Every achievement originates from the seed of determination.

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Every achievement originates from the seed of determination.

  2. Survival Analysis Nonparametric Methods for Comparing Survival Distributions

  3. Abbreviated Outline • How to formally compare 2 or more survival distributions using hypothesis tests • These tests look at weighted differences between the observed and expected hazard rates, allowing us to put more emphasis on certain parts of the curves

  4. Hypotheses

  5. Notation

  6. Test Statistics

  7. Test Statistics

  8. Test Statistics

  9. Test Statistics

  10. Test Statistics

  11. Test Statistics Reject Ho if U is too large.

  12. Log-rank Test • Constant weight function: Treat all observed failure times equally. • It has optimum power to detect alternatives where the hazard rates in the M populations are proportional to each other

  13. Proportional Hazard Assumption • An underline assumption of many methods • Suppose there are 2 groups of survival data. Then h1(u)=c*h2(u) where hi(u) is the hazard function of group i and c is a constant

  14. Wilcoxon Test • Survival time t(j) is weighted by nj, the number of individuals at risk at time t(j). • This test is less sensitive than the log-rank test to deviation of the observed to the expected in the tail of the distribution of survival times.

  15. Example: 6-MP • To compare the survival distributions of the placebo group and the 6-MP group using the log-rank test Test of Equality over Strata Pr > Test Chi-Square DF Chi-Square Log-Rank 16.7929 1 <.0001 Wilcoxon 13.4579 1 0.0002 -2Log(LR) 16.4852 1 <.0001

  16. Stratified Tests • Previously, we assumed that the various groups of individuals under comparison are homogeneous with respect to other factors which may affect survival time • One way of detecting differences in survival between groups, while accounting for the effects of other factors is to stratify.

  17. Stratified Tests • When the number of strata is large, a test typically has low power to detect treatment differences.

  18. Stratified Tests • Hypothesis:

  19. Stratified Tests

  20. Example: 6-MP • The patients are stratified according to remission status (partial or complete). • Consider a test of Ho of no treatment effect, adjusting for the patient’s remission status. • The stratified log-rank test (chisq=17.9 and p-value = 2.28x10^-5) indicates that the distribution of survival times is significantly different between 6-MP and placebo groups.

More Related