Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

Outwitting Procrustes

No description
by

Aaron M Tejani

on 18 July 2012

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Outwitting Procrustes

Conflict Declaration
Deliberate
Aaron M Tejani
Lower mainland pharmacy services
Therapeutics Initiative (UBC)
Cochrane hypertension review group
Who was Procrustes?
Quick read of trials
Come up with your own question!
skepticism
Don't even bother
Hierarchy of Outcomes
Serious adverse events
Clinical significance
vs statistical significance
Surrogates vs tangible outcomes
Issues to consider
Example: LABA+ICS for COPD
Death
Life-threatening
Hospitalization (initial or prolonged)
Disability
Congenital anomaly
Intervention to prevent permanent damage
Questions
Randomization is magic
Sequence generation
The drug "did" it
You need to check
p-values
small studies
Allocation concealment
Blinding
All outcomes can be influenced
(especially subjective outcomes)
Clinical effects may compromise blinding
Lack of blinding
could bias assessment
could bias actual outcome
Critical appraisal leads to different conclusions
So what?
Lack of:
blinding
randomization
allocation concealent
Overestimation of effect size
Surrogate Outcomes
Outwitting Procrustes:
critical thinking and clinical trials

Blood pressure
Hemoglobin A1c
LDL cholesterol
Bone mineral density
FEV 1
Scales and Scores
MCID
Linear??
Intention-to-treat
Beyond numerators and denominators
20% overestimation of effect
19% of findings become non-significant
Loss to follow up
What you need to know
Report numbers lost in each group
Problem if loss >20% of randomized population
Consider worst case scenario
The Worst
Last observation carried forward
What you need to know
How many people per group
When did they leave?
Significance and power
Power
Significance
Absence of evidence
is not evidence of absence
of an effect
Logical conclusions
Examples
p-value adjustments
Poor man's Bonferroni
0.05
_____
questions/outcomes/subgroups
=
Adjusted p-value
is it lower?
set the
bar
higher
subgroup analyses
effect within a subgroup
Issue #1:
underpowered
Issue #2:
false positives
Issue #3:test for interaction
Interaction test significant if
p <0.1, why?
Issue #4:
not random
Problems unless
pre-specified
stress overall effect
needs to be based on
pre-randomization characteristics
plausible reason for difference
and predict direction of effect
composite outcomes
consider relative importance
consider relative frequency
counting rules
no negative effects
non-inferiority
sample size issues
margin is critical
What to look at
where does the margin come from?
is it justified?
is it less than a clinically important difference?
look at "worst" end of confidence interval
look at per protocol
i.e enhance differences
levity
declaration
no mugs pens, coffee, honorariums, awards, etc...
Need to do the accounting
Maintains balance
Keep the list secret
false negatives
false positives
p values, NNT, ARR/ARI
p-value = probability of chance
relative risk= risk with Tx / risk with control
(1 = same risk or no difference)
absolute risk increase/reduction = risk with Tx -risk with control
(0 = same risk or no difference)
relative risk reduction
5-10/10 = 0.50
(says nothing about magnitude)
Confidence Interval
(if repeated, the true parameter would be within this range 95% of the time)
aaron m tejani, vijaya musini, ken bassett, tom perry
no conflicts to declare
objectives
list the sources of bias within a randomized controlled trial
identify the importance of methodological characteristics to help reduce bias (e.g. blinding)
outline steps required to interpret different analyses within and RCT (e.g. subgroups, intention to treat, composite outcomes)
describe the implications of different types of bias on the interpretation of data from RCTs
describe what p-values, confidence intervals, and effect estimates mean (odds ratio, absolure risk reduction)
Full transcript