Epidemiology Flashcards
What is the gold standard for drug testing?
Controlled clinical trial
What is equipoise?
A state of genuine uncertainty about the benefits/harms resulting from a regimen.
Equipoise calls for what?
Randomized controlled trial
What is the difference between the experimental population and the study population?
Experimental population is the group you chose before applying excluding criteria
Why is randomization important? What bias does it diminishes?
Prevents selection bias
Minimizes confounding
Why is concealment of allocation (the fact that the physician does not know who is a control or on trial during a study) important?
Prevents missclassification of exposure status
Prevents missclassification of outcome status
What is the Hawthorne effect?
People in the control group (and know they are in the control group) that change the behaviour to help the study
What are the limitations of randomized control trial?
Doesn’t represent routine practice
Not effective to study uncommon but serious adverse effects
Expensive and time consuming
Sometimes unfeasible for ethical concerns
What is the Power of study?
Probability that there’s an association and that the study shows it
What is the formula for Power?
1 - (percentage of type II errors)
At which percentage is the Power usually fixed?
80%
What can increase the Power of a study?
A larger sample population
What is the Number Needed to Treat (NNT)?
The NNT is the average number of patients who need to be treated to prevent one additional bad outcome (e.g. the number of patients that need to be treated for one of them to benefit compared with a control in a clinical trial)
Number Needed to Treat (NNT) is the inverse of what?
Absolute Risk Reduction (ARR)
What is the formula for Absolute Risk Reduction?
Control group event rate - Experimental group event rate
What is the formula for Number Needed to treat?
1
______
(Control group event rate - Experimental group event rate)
What is sampling variation?
Uncertainty caused by the fact that individuals differ in their response to treatment
What is the Null hypothesis
Supposing there is no difference between control and trial groups
Do we reject the null hypothesis if the p value is < 0.05?
We reject the null hypothesis
Do we reject the null hypothesis if the p value is > 0.05?
We keep the null hypothesis; the results are statistically insignificant
3 factors influencing the p-values
Magnitude of the treatment effect
Sample size
Standard deviation of the data
What is Type I error?
Saying there’s a difference (an association between two phenomenas) between two groups, but there’s none
What is Type II error?
Saying there’s no difference (no association between two phenomenas) between two groups, but there is
What is the arbitrary level of confidence?
95%
How do you calculate confidence intervals?
Effect size ± (1.96 x measure of variation)
What distinguishes meta-analyses from systematic reviews?
Meta-analyses are quantitative and systematic reviews are qualitative
Which is higher evidence, meta-analyse or systematic review?
Meta-analyse
Which is higher evidence, traditional meta-analyse or individual patient data meta-analyse?
Individual patient data meta-analyse
Characteristics of traditional narrative reviews?
Written by an expert in the field
Informal/subjective methods to collect and interpret info
(systematic reviews follow a predetermined and explicit method)
What is a “pooled analysis”
An individual patient data meta-analyse
What is quality assessment?
Having the data studied synthesized
What is the defining characteristic of fixed-effects models in statistical analyses?
They assume no between-study heterogeneity
What is the defining characteristic of random-effects models in statistical analyses?
They assume heterogeneity within- and between studies (due to difference in studied populations and study designs)
What is I^2 statistics?
A method to estimate the proportion of the total heterogeneity that is due to between-studies heterogeneity
When can a meta-analysis mislead (4)?
When a meta-analysis is done outside of a systematic review
When poor quality studies are included (and good ones excluded)
When no attention is given to heterogeneity
When there’s presence of biases
What is the publication bias?
Studies finding positive correlation between phenomena tend to get more published