Baseline measurements

Question 1

What are the three approaches to using baseline data?

Answer 1

Post, Change and ANCOVA

Question 2

What is the post approach?

Answer 2

Ignore baseline, use two-sample t-test on the outcome

Question 3

What is the change approach?

Answer 3

Use the difference between outcome and baseline, two sample t-test on outcome-baseline

Question 4

What is the ANCOVA approach?

Answer 4

use prerandomisation values for eavh subject as a covaraite in a linear model

Question 5

Details on post approach?

Answer 5

depends on expected mean difference between groups being zero if there is no treatment effect

Question 6

Details on change approach?

Answer 6

assumes that the between-group difference at the outcome (if no treatment effect) is the difference in groups at baseline

Question 7

Details on ANCOVA approach?

Answer 7

predicts what group different outcome would be in the absence of any treatment effect
narrowest CI and highest power

Question 8

Why is it pointless to perform test on baseline data?

Answer 8

Under proper randomization these results should be due to chance and as statistical tests see whether data are consistent with chance hence pointless