Analysis of Protocol Deviations

Question 1

What is the Intention to Treat principle ITT?

Answer 1

Subjects should be analyzed in the groups that they are randomized to irrespective of the treatment that is received. This maintains ranomidzation

Question 2

What types of missing observations exist?

Answer 2

Missing completely at random MCAR, Missing at random MAR and Not missing at random NMAR

Question 3

What is MCAR?

Answer 3

missing values in Y are MCAR if missingness is in-depth of and x
P(R=1 | X,Y) = P(R=1)

Question 4

What is MAR?

Answer 4

missing at random is when outcome Y is independent of y given X
P(R=1 | X,Y) = P(R=1 | X)

Question 5

What is NMAR?

Answer 5

Data neither MCAR and MAR is NMAR

chance of seeing outcome Y depends on Y even after conditioning on X

Question 6

What are the alternatives to ITT?

Answer 6

As-Treated and Per protocol

Question 7

What is As treated comparison?

Answer 7

Comparison of treatment regardless of random allocation

Question 8

What is per protocol comparison?

Answer 8

Compares those who complied with random allocation with (i) all controls if new treatment was not available to controls
(ii) those who received control in the control group

Question 9

What is CACE?

Answer 9

Complier average treatment outcome

estimates average outcome of the completers in each group. randomization respecting estimate.

Question 10

Compare ITT, as treated and per protocol?

Answer 10

ITT is lower than as treated and per protocol, as it is biased towards the null, it gives a more accurate representation when there is possible non-compliance