Epidemiology clinical Qs Flashcards
define efficacy
the ability to produce results under ideal clinical trial conditions
define effectiveness
the ability to produce results in real clinical use
what model of effectiveness is this?
intervention has CONSTANT effect throughout course of disease
what model of effectiveness is this?
intervention has constant effect, but LAGGED start and benefits are not immediately apparent
what model of effectiveness is this?
intervention has COMPOUNDED RETURNS with effect at time of enactment and improvement that magnifies throughout course of disease
how do you measure causality
cannot be measured directly- must be inferred (called “causal inference”)
what is the “causal effect” of the intervention?
effectiveness
define cause
if the event would NOT have occurred without it
what is the “counterfactual paradigm”
what you predict would’ve happened if the treatment hadn’t been done
can we observe associations or causations?
we can ONLY observe associations
does association = causation
NO
define cost-effectiveness
value of treatment given money spent
NOT whether more money is saved than spent (bc that doesn’t happen with health treatments)
define systematic review
clear research question + protocols for selecting articles
can be replicated
time consuming
less subjective than literature/narrative reviews
define meta-analysis
forms a “pooled effect estimate” by combining the effect estimates from multiple studies
increases precision + statistical power of estimates
systematic review vs meta-analysis (quantitative or qualitative)
systematic review = qualitative
meta-analysis = quantitative
define literature/narrative review
conducted by an expert
combines subjective + non-subjective methods when selecting articles
subjective overall
can NOT be replicated
systematic review steps
1- write systematic review protocol
2- search literature
3- review texts
4- abstract data from texts
5- assess risk of bias
6- summarize evidence
meta-analysis steps
1- write systematic review protocol
2- search literature
3- review texts
4- abstract data from texts
5- assess risk of bias
6- assess heterogeneity
7- obtain pooled measure of association
8- assess publication bias
meta-analysis 4 key points
-pooled effect estimate
-assess for heterogeneity
-analyze data in forest plot
-funnel plots to assess publication bias
what is the clear diamond on a forest plot for heterogeneity?
overall pooled result of the meta-analysis
what does it mean if the clear diamond is on the left side of the vertical line of a forest plot?
decreased heterogeneity, YES overall statistically significant effect
what is the vertical dashed line on a forest plot for heterogeneity?
the pooled estimate
what is the vertical solid line on a forest plot for heterogeneity?
the null
what is the vertical dashed line on a funnel plot?
pooled odds ratio
what does a funnel plot look like with NO publication bias?
symmetrical
what does a funnel plot look like with publication bias?
asymmetrical
what does a funnel plot look like with publication bias due to poor quality of smaller studies?
asymmetrical
what do you use to see if heterogeneity is caused by random chance OR by a different variable?
Q-statistic or I^2 statistic
what are you looking for with a Q-statistic result?
p-value < 0.2
= reject the null hypothesis (heterogeneity IS statistically significant)
what are you looking for with a I^2 statistic result?
I^2 = (% variability due to heterogeneity rather than chance)
I^2 > 50%
= heterogeneity is substantial
