1.6: Cross-Over trials

Question 1

Parallel and paired study design:

Answer 1

• Parallel: Different groups of patients studied concurrently -> two independent samples t-test for inference
• Paired: Patients receive both treatments (e.g. matching parts of anatomy) -> estimate treatment effect based upon ‘within-subject’ comparison -> paired t-test

Question 2

Cross over design: Principle

Answer 2

• Patients receive a sequence of treatments; the order determined by randomisation -> estimate of treatment effect based upon ‘within-subject’ comparison
• Times of administration: treatment periods

Question 3

Advantages of cross over trials (x3):

Answer 3

• Within subject comparisons (eliminating between-patient variation)
• Sample size is smaller
• Precision increased

Question 4

Disadvantages of cross over trials (x5):

Answer 4

• Inconvenience to patients
• Drop outs
• Period by treatment interaction; treatment effect may not be constant over time
• Carry-over effect (‘persistence of treatment applied in one period in a subsequent treatment’)
• Analysis is more complex; pairs of measurements, so may be a systematic differences between periods

Question 5

Wash out periods; types of wash-out:

Answer 5

• A period in a trial during which the effect of a treatment given previously is believed to disappear.
• If no treatment is given during the wash-out period then it is passive
• If a treatment is given during the wash-out period then the wash-out is active

Question 6

3 circumstances when cross-over design is particularly useful:

Answer 6

• Chronic disease (relatively stable over time)
• Single dose trials of bioequivalence (PK/PD) rather than long term trials
• Drugs with rapid, reversible effects rather than ones with persistent effects

Question 7

Basic cross-over trial design:

Answer 7

• 2x2 (two treatment, two cross over)
• Either AB or BA
• In this module, normally distributed endpoints are considered

Question 8

Simple analysis for cross-over trials:

Answer 8

• If no period effect, then one can proceed as per the paired design previously using a paired t-test
• Calculate the treatment differences for each subject, calculate the mean of differences and SE
• Perform a one-sample t-test for the differences (i.e. a paired t-test)
• Construct a confidence interval for the true difference
• Assuming normally distributed differences and lack of bias

Question 9

Factors that might cause the differences to not be distributed at random about the true treatment effect:

Answer 9

• Period effect
• Period by treatment interaction
• Carry-over
• Patient by treatment interaction (cannot be investigated by AB/BA design)
• Patient by period interaction

Question 10

Notation in AB/BA trials:

Answer 10

• Mu: Expectation of treatment B
• Tao: Treatment effect
• Pi: Period effect

Question 11

Estimating treatment effect in the presence of a period effect:

Answer 11

Tao-hat = (mean of period differences for period 1 -2)/2

Question 12

Estimating period effect:

Answer 12

Pi-hat = (sum of mean of period differences) / -2

Question 13

Adjusting the estimated treatment effect for the effect of period:

Answer 13

• FC 13a

Question 14

Demonstrate the cell means model

Answer 14

• mu every tab
• tao for A t-mt by sequence
• pi for second period
• lamba 1 and 2 for fixed carry-over effects