EBM Midterm Flashcards
What is internal validity
Are the results due to the intervention/exposure that was studied or something else
What are the major threats to internal validity
Chance, Bias, confounding
What is chance
random error in measurements
What is confounding
confusion of effects
bias
Problems with the way a study was designed, conducted or analyzed that leads to incorrect results or conclusions
what is external validity
Are the results applicable (generalizable) to other populations (patients), settings and time
what is the P value
The probability the results are due to chance rather than a real treatment effect
Do you want a small p-value or a large p-value
Want it to be less than 0.05
How do you deal with chance
Increase sample size
Does statistical significance always mean clinical significance
NO
BUT can’t have something be clinically significant if it isn’t statistically significant
What does PICO stand for/
The question of the study P: population/patient I: intervention/exposure C: comparison (control/other intervention) O: outcome (result of the intervention)
What are the three things integrated in EBM?
evidence
clinical expertise
patient values
How do you deal with confounding
RANDOMIZATION
ensures group are similar in all aspects (known and unknown)
Where do we look in a study to see if our control and intervention groups are similar in baseline characteristics
TABLE 1
What are other ways confounding can be mitigated other than by randomization?
Stratification: try and make sure there are similar numbers in each group
Matching: matched on known factors
Multivariable models: logistic regression
What is selection bias
Systematic error (or differences) in how the study subjects were selected or who participated
Does selection bias effect internal or external validity
External validity
what are examples of selection bias
Volunteer bias: People who volunteer/participate in studies are different from those who don’t
Adherer bias: employed individuals adhere to medications better
Attrition bias: lost to follow up
What is information bias
Problems with measuring, collecting or classifying information (exposure and/or outcomes)
what are some examples of information bias
Outcome errors: problems with measuring tools and the actual measurements
Recall bias: individuals remember things differently (ie. pain scale)
Interviewer bias: Interviewer asks about exposure/outcome differently
Detection bias: If you look you will find it
How do you minimize bias
BLINDING
WILL NEVER TOTALLY ELIMINATE IT!
What is publication bias
Authors and journals tend to publish positive findings (especially drug trials)
What is an RCT
Tests whether an intervention works by comparing it to a control condition
Subjects have equal chance of being assigned to each group
What is a parallel RCT design
Regular: how you think an RCT works
What is a cluster randomized RCT
Clusters of individuals are randomized instead of individuals
Used in hospitals/pharmacies
what is a cross-over RCT?
The patient becomes their own control because they take turns in the control and intervention group.
Takes more time because need a wash-out period for the drugs.
What does a power calculation help us determine?
The study sample size needed to show if a difference exists
MUST be determined before the study starts
What is unique about the inclusion/exclusion criteria in RCTs
VERY STRICT
limits generalizability
What must be included in intervention criteria?
Drug (dose, regimen, delivery method, follow-ups, length of intervention)
Procedure (What’s involved, timing, follow-up, length of intervention)
What is an objective outcome
Measurable (blood pressure, lipid levels)
what is a subjective outcome
Subjects interpretation (back pain)
what is a hard endpoint
Death, stroke, MI
what is a surrogate endpoint
Blood pressure (which can lead to stroke)
What is a primary endpoint
Main result that is measures at the end of the study to see the effect of the intervention
What is a secondary endpoint
Additional results of interest but not main focus
Caution when interpreting as the study wasn’t designed around them (Power calculation is based on the primary endpoint)
NEVER make a decision based solely on a secondary endpoint
What is a composite endpoint
A primary endpoint that contains several events (Instead of just looking at death as a primary outcome, you include MI, stroke, and death)
What is the difference between an outcome and the results
Outcome is the thing we want to measure, results are the statistical analysis of the measurements
what are the 3 methods of randomization
Complete: regular randomization (ie. flipping a coin)
Block: block of 6 randomized, 3 in intervention, 3 in control
Stratified: trying to achieve similarities in baseline characteristics
What is Intention to Treat
Analyzing the data for ALL patients and according to the group they were originally randomized to
ONCE RADOMIZED ANALYZE
How do you know if ITT to treat was done
look at the results table and the population that was analyzed
what is the importance of ITT?
preserves the value of randomization
keeps important data that could otherwise be lost
what is per-protocol
analyzing data only from subjects who completes the study or followed protocol exactly
what is the placebo effect
Perceived or actual effect from an ineffectual or inactive treatment
what are observational studies
Observe the effect of an exposure or intervention) on an outcome without interfering with anything
what are the advantages of observational studies
more practical/feasible
more ethical: not denying treatment
real-world: larger inclusion criteria (pregnant, elderly, and kids); real-life situations (comorbidities, non-adherence)
Disadvantages of observational studies
Confounding
Bias
Chance (not an issue as there is a larger population)
What is a case report
Describes and interprets an individual case
advantages of case reports
Useful in identifying new or unusual trends or diseases
Useful in identifying new drug effects (good or bad)
Describe novel interventions (Treatment, Diagnostic procedure)
Suggest areas for further research or alarm
what are disadvantages of case reports
Difficult to interpret
can’t confirm or prove anything: one time story
what is a case series
Collection of individual reports which typically occur within a fairly short period of time
what is a cross-sectional study
Looks at population or study sample at one point in time
USEFUL FOR DESCRIBING PREVALENCE
advantages of cross-sectional studies
good for determining prevalence
Useful for quick examination of potential associations
inexpensive
disadvantages of cross-sectional studies
CANT determine incidence: how many new cases
what is a case-control study
retrospective
Compare the proportion of cases (have already had the outcome) with the exposure to the proportion of controls (no outcome) with the exposure
advantages of case-control studies
good for studying rare diseases or outcomes
quick to do as the outcome has already occurred
good for establishing association
disadvantages of case-control
not randomized
susceptible to bias
only suggest associations: cannot prove cause and effect
what is a cohort study
Groups (or cohorts) of subjects are created and compared according to exposure
Exposed vs. unexposed group
Groups followed over time to see if outcome occurs
BEST type of observational study for finding a valid association
what is a prospective cohort study
At the start of the study, outcome hasn’t occurred yet
what is a retrospective cohort study
Outcome occurred before the study started
advantages of prosepctive studies
conducted in real time
have some control over what data you want to collect and where it is coming from
can manage some confounders
advantages of retrospective studies
good for long follow-up studies
HOWEVER HAS DATA LIMITATIONS
what is the most important thing to consider when choosing participants for a cohort study
MUST EXCLUDE DATA FROM PEOPLE WHO HAD THE OUTCOME BEFORE THE EXPOSURE (ex. had an MI and then was put on BP medication)
advantages of cohort studies
Provides highest level of evidence for observational studies
Can study multiple outcomes after a single exposure
Can find relative risk
disadvantages of cohort studies
Susceptible to bias
Not randomized so susceptible to confounding
More resource intensive than other observational studies
what kind of results can be given from studies (2)
mean (or mean change)
Proportion (fraction of the total that possesses the outcome)
Can you use odds ratios and risk ratios with mean results
NO only proportions
what is the ARD/ARR
absolute risk reduction
absolute risk difference
The absolute difference between the probability of the event in the control group and probability of the event in the intervention group
what is the relative risk
Shows that the risk of an outcome in the intervention group is compared to the risk in the control group
Probability of event (intervention group)/probability of event (control group)
what is the threshold for seeing a difference when using mean results?
0 (also for CI to be statistically significant cant cross 0)
What does a relative risk or odds ratio=1 mean
NO difference in odds/risk between the groups
What does a relative risk or odds ratio>1 mean
HIGHER risk/odds of outcome in intervention group
What does a relative risk or odds ratio<1 mean
LESS risk/odds of outcome in intervention group
what is an odds ratio (OR)
Shows the odds of the outcome occurring in the intervention group compared to the control group
what is relative risk reduction or increase
RRR=1-RR
what is NNT
Number of subjects who would have to be treated (receive the intervention) in order for one additional subject to “benefit” in comparison to control
what is NNH
Number of patients who would be treated before you see one additional subject with an adverse event compared to control
what does CI show us
can show us the magnitude of effect
can show us if there really is a difference (statistical significance)
If the CI crosses the “no difference” threshold then there is not a difference
For MEANS or PROPORTIONS: no difference=0
For RELATIVE RISK or ODDS RATIO: no difference=1
what is a non-inferiority trial
Clinical trial to establish that an intervention is not clinically worse than a comparison by more than a pre-determined margin
what is the non-inferiority margin
Used to determine the sample size needed for adequate power
what is a propensity score
The probability that a subject would be in a particular treatment (study) group based on their observed baseline characteristics
what do propensity scores help with
Helps reduce selection bias and confounding in observational studies
DOESN’T replace the value of randomization