Randomised Control Trials, Design And Conduct

Question 1

Definition of randomised control trial

Answer 1

A planned experiment in humans designed to assess efficacy of treatment/intervention

Question 2

Definition of equipoise

Answer 2

No existing evidence that the intervention being tested will be superior to existing treatments/effective in all

Question 3

Definition of randomisation bias

Answer 3

Bias that could arise when the characteristics present in both randomly generated groups are not equal

Question 4

Definition of cluster randomisation

Answer 4

Intervention that groups of individuals receive is randomly allocated

Question 5

Definition of power

Answer 5

1-probability of a false negative/TII error

Question 6

Definition of significance

Answer 6

Probability of a false positive/TI error

Question 7

Definition of allocation bias

Answer 7

Bias that arises when investigators know/predict which intervention that the next participant will get

Question 8

What is an RCT
When is this ethically justified
How are these results used

Answer 8

Planned experiment in humans designed to assess efficacy of treatment/intervention

Ethically justified in equipoise exists

Provides evidence on which clinical treatments and health policy decisions are based

Question 9

How are RCTs used in drug trials

Answer 9

Used in all 4 phases
Phase I, safety
-small group of healthy humans
-check PK

Phase II, efficacy
-medium group of patients

Phase III, confirm results
-large group of patients from a wider population

Phase IV, safety monitoring
-In patients for rare adverse effects

Question 10

What are 5 other ways of using RCTs

Answer 10

Methods f diagnosis
Methods of organising care
Methods of disease screening
Methods fo disease prevention and health promotion
Efficacy of health care policies

Question 11

Why is randomisation important in an RCT

Answer 11

Random allocation to a control and investigational group
Reduces bias as each group is likely to have a similar distribution of different characteristics
In a large sample size, different characteristics will have a small effect
Any differences in outcome must be due to intervention

Question 12

What is cluster randomisation and when is it often used

Answer 12

Groups of individuals are randomly allocated to different interventions
Often used in public health

Question 13

What are the 5 questions you must address when designing a RCT

Answer 13

What is the question
Population of interest
Intervention?
Comparison?
Outcome assessment of intervention?

Question 14

What 5 factors must you consider when defining inclusion and exclusion criteria

Answer 14

Diagnosis
Age
Sex
Comorbidities and severity
Recruitment location

Question 15

What 4 factors dictate the total no of participants

Answer 15

Size of expected effect

• if large => small cohort
• if small => large cohort

Statisticians consulted for

• potential SD
• power and significance of results

SD

• if expected to be large => large cohort
• if expected to be small => small cohort

Power
-if power small => large no

Significance
-to make p value more significant => large no

Question 16

What is the difference between power and significance

What power and significance values are acceptable

Answer 16

Power

• 1-probability of a FN
• 80-90%

Significance

• probability of a FP
• p<0.05

Question 17

Describe the process of randomisation

How are the group numbers made similar

Answer 17

RNG with computer
Results concealed from recruiter, prevent allocation bias
Randomisation in blocks make group no’s similar

Question 18

What are the 2 types of control treatment

Answer 18

Treatment benefit not established
-placebo must be physically similar to active drug but inactive

Treatment benefits established
-best existing treatment used for comparison

Question 19

What are the 5Ds used in assessing the outcome

Answer 19

Death
Disease activity/recurrence
Disability
Distress
Discomfort

Question 20

What are the primary outcome measures

What are the 3 characteristics of a primary outcome measure

Answer 20

Must be clinically relevant
Objective and independent of treatment
easy and accurate to measure

Question 21

What is the 2 types of blinding

Why is this important

Answer 21

Single blinding

• patient/physician unaware of treatment
• avoids participation bias

Double blinding

• patient and doctor are unaware of treatment
• avoids physician associated bias