Midterm 1 Flashcards
What is EBM?
integration of best research evidence and clinic expertise and patient values
What are the five A’s of EBM?
Ask- make question
Acquire- find info
Appraise- evaluate info
Apply- use results
Assess- evaluate performance
What is PICO?
P-population
I-intervention
C- versus
O-outcome
What are the top notch filtered information that we can access?
Systematic reviews
Critically apprasied
What are unfiltered information and rank them of descending quality?
RCT
Cohort
Case-Control
Cross sectional
Case report
What is external validity?
how generalizable it is
What is internal validity and what 3 threats are there?
- due to something weird during the study
Chance confounding and bias
What is chance?
random error in measurements or observations
What is bias?
Any systematic error in results
What is confounding?
Confusion of effects- some other factor is causing issues
What does P values mean?
> 0.05= support null= not statistically different
<0.05= reject null= statistically different
How can we lower chance?
increase sample size
use p values
REMEMBER STAT SIG MAY NOT BE CLINICALLY SIG
How can we lower confounding?
randomize or statistical models to adjust for it.
What is stratification vs matching randomization?
Strat- radnomize 2 groups separately so they get the same representation
Matching- often by sex and age so that they put 2 similar people in different groups
What is selection bias?
how they were selected
often effects external validity
What is volunteer bias?
people who volunteer are different than those who don’t
What is healthy worker bias?
employed= healthier
What is attrition bias?
people who leave study or die and not following up
What is recall bias?
individuals may remember differently. maybe remember negative better
What is interviewer bias?
ask/ probe different people if they know what group they are in
What is surveillance bias?
one group is studied/ followed more closely
How can we lower bias?
minimize through blinding
create everyone the SAME
What is publication bias?
tend to only publish positive findings
What is the difference between parallel and cluster RCT?
parallel= split into two groups
cluster= cluster of individuals are randomized
What is study power?
ability to show difference between the groups= need a good in-between number
What is a hard vs soft endpoint?
hard= death, heart attack
Soft= blood pressure
What is a primary vs a secondary endpoint?
primary= what the study was designed for
secondary= additional findings
What is the issue with secondary endpoints?
study was not designed for it so confounding may be an issue. harder to trust results
What is a composite endpoint?
Primary endpoint with several events
ie) MACE
Issue with composite endpoints?
hard to determine true effect of intervention on each event
What are the tree methods of randomization?
Complete= no limits= not same numbers in each group
Block= force equal numbers by 6 subjects three must go in each group
Stratified= by age or sex usually
What is ITT?
Analyze data of all patients even those that died or dropped out
What is per protocol?
Analyze data from subjects who followed it exactly
What can’t observational studies do?
PROVE CAUSATION
What’s a benefit of observational studies?
feasible, real world, ethical always
What are the 3 main types of Observational studies?
Case reports/series
cross-sectional
case-control
cohort
What is a case report study? also what is a case series :)
interpret one case
pros= useful for new disease or side effects, further research, identify new drug effects
Cons= hard to interpret, CONFOUNDING
series= individual reports in short time
What is cross-sectional studies? Pro, Cons?
Look at sample at one point in time
Pros= prevalence, look for association, cheap, quick
Cons= don’t know incidence, can’t prove, confounding
What is case control?
Compare 2 groups retrospective
look for people who had outcome and this ewho did not
Try to make two groups as similar as possible
Cons= not randomized, bias, can’t prove
Pros= good for rare, quick
What is a cohort study?
Follow the two cohorts
similar to RCT but NO randomize
Can be prospective and retro
Cons= bias, no randomize, costly
Pros= highest level of evidence
How to calculate ARR/ARD and what does it mean?
ARR= %placebo- %outcome
tells us that drug X is % outcome less than placebo
How to calculate RR
Prob of event in intervention/ Prob of net in placebo
What does an RR of <1 mean?
less risk of outcome in intervention
How to calculate RRR?
RRR= 1- RR
HOW to calculate OR
Prob risk of event/prob risk of not = for one drug
do it again for the other drug then do
intervention Odds/placebo Odds
What does a OR>1 mean?
higher odds of outcome in intervention
What is NNT? Same with NNH
NNT= 100/ARD% Orrrr if it is in decimal then 100/the decimal always round up
How do you use survival curves?
look at how many years to get 50% survival
WHat does Hazard ration of <1 mean. What about>1?
<1= less hazard in intervention
>1= higher hazard
WHat does confidence interval mean?
means that in 95% of the time the drug reduced __________ and true value is between __________
What is threshold of no difference for mean or proportions?
0
What is a washout period?
period of time to let the drug be fully eliminated from the body
What is a cross over RCT
intervention then gets placebo and vice versa
Why do you need to make sure that a study is designed before it begins (a priori)?
limit bias
Which has better generalizability? RCT or observe?
Observe
If I increase my sample size in my study, what will this help to reduce?
Chance
Randomization helps to minimize:
confounding
How can we find out if ITT or per protocol
If they randomized certain number and if the results have the same table
Who is blinded in single blind trials
the evaluators
What does NNT mean?
it takes _____ treated for ______days for one ADDITIONAL person to see benefit
What is typically a good NNT?
depends on treatment and length of time
But for CVD anything less than 50
What can CI help us see?
best case and worst case scenario
What is threshold of no difference for RR and OR?
1
Is this CI significant mean weight -3-19
NO
IS this CI significant mean weight 30-40
YES
Is this CI significant? RR 1.9-5
Yes
Is this CI significant OR 0.4-5
NO
Why is CI more informative than p values?
shows us range and direction of the effect.
Also help us see clinical and statistical significance
What does ITT help prevent?
confounding
Why is ITT not great for non-inferiority RCT?
may accidentally show non-inferiority
If a lack of superiority in a superiority trial does this mean it is non-inferior?
NOOOOOOO
What do we ABSOLUTELY need to have established before a non inferior trial?
Need the margin of inferiority
What is a propensity score
probability of a subject would be in a certain treatment group based on observed characteristics
When do we use propensity scores?
OBSERVATIONAL studies
What does propensity scores limit?
selection bias and confounding
Does propensity scores replace randomization?
FUCK NO
What is a adaptive clinical design?
planned review of the data, make modifications and redo conducting
What examples for Adaptive clinical designs?
Adjust sample size, stop treatment arms (if really bad), change allocation, stopping early(positive or negative)
What is a con of adaptive clinical trials?
interim data analysis= operational bias as you are aware of the effects.
True or false all systematic reviews are very good
False
What do systematic reviews reduce?
confounding
What is a meta-analysis?
use stats to integrate results, extract data from each study and weight the numbers based on if study provides more info
When can’t we do Meta-analysis
if different population, comparisons
check if similar enough by use Q
Why is systematic reviews important?
hard to stay up to date
What is Cochrane collaboration?
highest standard. international volunteers to review
What are the components of a systematic review?
Make question- establish protocol, eligibility, RCT and/or Observe
Comprehensive literature search
Identify inclusion studies (look at abstract)
Assess Quality
Extract data
Analyze data (only for meta)
Interpret
Issues with systemic reviews
no transparency
bad criteria
poor quality papers
What are clinical practice guidelines and why are they important?
Systematically developed statements to assist practitioners and patients make decisions. Important for good informed choices
What are the components of a CPG
Systematic reviews (+/- MA)
Set of recommendations
target population
target audience
expert panel-range of HCP+ info on them (qualifications)
How they developed recommendation (voting/consensus)
external review
updates
funding statements
What is the grading system for CPG?
Numbers=1 (we recommend) 2 (we suggest)
Letters= A-D A is best
What are some potential problems of CPG?
Poor quality studies
Publication bias
not user-friendly
liability- sued if not followed
How can we appraise CPG?
AGREE II- has 23 items and 6 domains to check CPG
check scope, stakeholders, rigour, clarity, applicability
editorial independence
