8/10/16 EBM: Impact on Clinical Practice - Willett

Question 1

levels of EBM

Answer 1

• studies : original published articles in journals
  
  • syntheses : Cochrane reviews
    
    • synopses : evidence-based journal abstracts
      
      • systems : computerized decision support system

Question 2

types of error

how do you minimize them?

Answer 2

type 1 error (alpha) : seeing something that’s NOT there

• rejecting null hypothesis when it should not be rejected
• REDUCE: lower p-value!

type 2 error (beta) : failing to see something that IS there

• failing to reject null hypothesis when it should be rejected
• REDUCE: increase sample size!

Question 3

types of validity

Answer 3

1. internal validity : is the study flawed? can we trust it?

2. external validity : are results generalizable? can the study results be applied to your patient pop?

Question 4

common limitations of study types

observational studies (incl big data)

Answer 4

observational

• confounding by indication: more likely to prescribe a drug to sick person, more likely to recommend surgery to a generally well person, etc
• selection bias: selection of a group/pop in such a way that randomization is not achieved and group is not a representative sample of the larger pop
• recall bias: hard to remember things at time of testing
• ascertainment bias: if someone is sick, tend to get more tests to try to figure out what’s wrong (sick → more likely to get gold standard tests)

Question 5

common limitations of study types

randomized controlled trials

Answer 5

RCTs

• expensive, time-consuming
• inadequate power
• randomization/concealment issues (can lead to selection bias)
• inadequate blinding/placebo
• early stopping for benefit or harm
• adherence, cross-over
  
  • intention to treat vs. per-protocol

Question 6

common limitations of study types

systematic reviews

Answer 6

systematic reviews

• depends on quality of studies you incorporate into your review
• smaller published trials tend to have much larger effect sizes than larger trials

Question 7

how might PICO be “massaged”?

Answer 7

P - unrepresentative pop

I - unreproducible intervention

C - comparator destined to fail

O - unimportant but important sounding outcome