Evidence and Clinical Questions Flashcards
PICOS
population, intervention, comparison, outcome, study design
Cross sectional study
Descriptive epidemiology
Can help form a hypothesis
2 types of analytical epidemiology
experimental (randomized controlled trials)
observational (case controlled and cohort)
Types of reviews
lit review- subjective and cannot be replicated, systematic and nonsystematic methods
systematic review- clear and specific research question, inclusion/exclusion criteria
meta-analysis- pooled individual studies, increase precision of estimates and synthesize findings across populations
Challenges in lit review
-large number of studies
-Inconsistent terminologies
-different statistical methods
Systematic review steps
- write protocol
- search literature
- review titles, abstracts and full texts
- abstract data
- assess risk of bias
- summarize evidence
extra steps for meta analysis
6. assess heterogeneity
7. obtain pooled measure of association
8. assess publication bias
where do you report systematic reviews?
PRISMA
diamonds
diamonds are associated with 95% confidence intervals,
if diamonds do not cross the null, there is a statistically significant effect
if heterogenetity
use a random effects model
if no significant heterogeneity
use a fixed effects model
or random effects model (assume true population)
subgroup analysis
use with caution, tires to make inferences about a specific group
sensitivity analysis
analysis using alternative decisions (ex different inclusion criteria)
assess publication bias
many journals do not want to publish negative or null results so are naturally biased against them
funnel plots can assess publication bias
no bias= symmetrical mountain
I^2>50%
not due to random chance
due to one or more factors
is it reasonable to pool results if they are very heterogeneous
no
