Exam #1: Appraising Therapy Studies Flashcards
What are the steps to effective EBM practice?
1) Ask answerable question
2) Search evidence
3) Evaluate evidence
4) Integrate evidence w/ practice and apply
5) Evaluate results
What are the preliminary questions for ALL papers?
- Why was it done?
- Is it primary of secondary?
- Does the study have an appropriate design to address the question?
What are the questions to discern if a therapy study is valid?
- Was there a clear question?
- Random?
- Did randomization produce comparable groups?
- Were all subjects accounted for?
- Blinded?
- Was the treatment equal between groups?
What is the purpose of randomization?
To reduce or eliminate bias while also creating COMPARABLE groups
*****Note that non-random studies generally have a higher rate of FALSE POSITIVES
What is concealed allocation?
Third-party assigns groups
This is THE MOST IMPORTANT thing to look for in RTCs
What is the purpose of concealed allocation?
Avoid SELECTION BIAS
What is the “intention to treat” strategy?
All patients are analysed in the group assigned, regardless of whether or not they completed or received that treatment
Failure to do this will skew the results toward the intervention
What is blinding of researchers and subjects?
Researchers and subjects don’t know what groups the subject is assigned to
What is the rule of thumb regarding “lost to follow-up?”
- If more than 20% of the subjects dropout, then the study is invalid
- If less than 80% follow-up, likely invalid
What is the opposite of an “intention to treat” analysis?
Per protocol analysis
What is accuracy in the context of therapy studies?
Does the treatment effect represent the true direction and magnitude of the treatment effect
What does a p-value less than 0.05 mean?
Less than 1/20 probability that the effect is a result of CHANCE ALONE
Generally, what does a p-value mean the smaller it is?
Smaller the p-value, the less likely chance alone is responsible for the effect seen
What is the general premise of a confidence interval?
CI is a range of numbers within which a specific result will be, at a given confidence level e.g. 90-95%
What should you think if a CI crosses 1.0?
The result of interest is NOT significant
