Lecture 8 Flashcards
Systematic review
Video
What is the aim of a systematic review?
To gather infor in order to answer a research question, is qualitative, gives an effect estimate
Video
What is a meta-analysis?
A statistical procedure, you combine the results of similar studies into 1, is quantitative. Can do it in your systematic review
Video
What are the steps in doing a meta-analysis?
You first have a RQ then you search for studies, then do a quality assessment of the studies
Video
What is the cochraine collaboration?
The main journal for systematic reviews. Preregister the protocol, procedures, H0 etc. to fix the procedure, avoid posthoc and avoid same reasearch as someone else
Video
What does a prisma flow diagram show?
It shows which articles were in-/excluded at each stage
Video
What are important points during quality assessment?
Bias, missing data, changes in data, or a prior deviation (lowers the quality)
Video
What is bias and why is it a problem?
It is a systematic error (due to the methods) that will over-/underestimate the effect
Video
What is the cochraine risk of bias based on?
It is based on a randomized control
Video
What is calculated in a meta-analysis?
The summary statistic of each study (effect estimate + CI)
Video
what is the formula for a meta-analysis with a dichotomous outcome?
odds ratio (OR)/ relative risk (RR)/ risk difference (RD), need N of the intervention and N of the control.
Video
What is the formula for a meta-analysis with a continuous outcome?
The mean difference + standardized mean difference, need a group means and SD, N
Video
How is are the RR, RD and OR calculated?
Look at the table etc in the summary notes
Video
With is the CI for a meta-analysis?
95% CI ln(RR) + 1.96 * SE (ln(RR))
Video
How is the SE calculated for the meta-analysis?
SE = square root of (1/a)-(1/a+b)+(a/c)-(1/c+d)
Video
How is the standardized mean difference or pooled SD (S*) calculated?
square root ((n1-1)s1^2 +(n2-1)s2^2)/(n1+n2-2)
1 = intervention
2= control
Video
What is the weighted pooled effect and how is it displayed?
The pooled SD of all studies (1 overall effect estimate + CI), displayed in a forest plot
Video
What is the pooled effect of studies?
A weighted average that is based on their precision wi = 1/SD^2
Video
What is the formulat to calculate the pooled estimate?
wi * estimate / wi
Video
What is clinical heterogeneity?
When there is a difference in the population in intervention, comparison, outcome, a follow-up studies needed if this is not true
Video
What does methodological heterogeneity mean
There is a difference in study design, outcome, measure and bias if this is not true
Video
What does a variability in intervention between studies lead to?
Leads to a higher variability than what is expected by chance.
Video
What can be done to test the statistical heterogeneity?
An eye ball test, test heterogeneity, quantify the heterogeneity
Video
How do you do an eye ball test?
Is a first impresion when looking at the test
Video
How do you do a heterogeneity test?
By doing a Q-test, H0: the effect is homogenous. Chi distribution. Depends on the amount of studies and the N. Fewer studies there is no power (type 2), many/large studies (type 1), alpha = 0.1
