8 – Clinical Trails IV

Question 1

Power

Answer 1

• Probability of being able to detect a treatment difference IF one truly exists
• 89% is usually accepted as minimum

Question 2

If reading a clinical trial with insignificant results we should determine if

Answer 2

• Trial had SUFFICIENT POWER to detect a clinical difference

Question 3

5 questions that need to be answered to determine power

Answer 3

1. What is the primary outcome measure for your trial?
2. What do you want to accept for Type 1 error rate?
3. What results are expected in the control group?
4. How small a clinical difference do you want to detect?
5. What will the beta level or power be?

Question 4

What is the primary outcome measure for your trial?

Answer 4

• Discrete or continuous

Question 5

What do you want to accept for Type I error rate?

Answer 5

• Conventionally set at 0.05
• If trial is expensive or harmfully, may want it lower (ex. 0.01)

Question 6

What results are expected in the control group?

Answer 6

• Need an educated guess as to how the control group will respond
• Discrete outcomes (mortality: need an estimate of mortality rate in control group)
• Continuous outcomes (serum glucose: need BOTH estimate of mean and amount of variation in outcomes)

Question 7

How small a clinical difference do you want to detect?

Answer 7

• Is 25% reduction in mortality rate (4% to 3%) the smallest clinical difference you wish to detect?
• **more animals are need to detect small differences

Question 8

5 elements of power

Answer 8

• Type 1 error
• Type 2 error
• What type of results does one anticipate in control group?
• How small a difference is it important to detect?
• Number of individuals in each treatment group

Question 9

External validity

Answer 9

• Does the population of animals in the study represent the animals that you see?
• Comorbidities?
• Combination of therapies?
  **NOT BIAS

Question 10

4 questions that might arise as you apply the evidence

Answer 10

1. What are the potential effects of treatment? (harmful and beneficial)
2. Are there differences in effects of treatments on different sub-groups of animals?
3. Are there differences in levels of risk between different groups of animals?
4. How do the benefits and harms relate to the individual animal or group of animals you have in front of you?

Question 11

Assess

Answer 11

• Did it make a difference?
• Clinical audit=form of assessment
• Not a research project
• How are you going to track things?