Duke Health - Module 7 Flashcards
Some systematic reviews will also include a meta-analysis
* A meta-analysis mathematically pools data from …
* This allowed you to see an —
similar individual studies
aggregated effect
- Did the review address a focused clinical question?
(3)
a. Patient problem
b. Intervention or exposure
c. Outcomes
- Was the search for relevant studies detailed and exhaustive?
a. Search multiple appropriate article databases such as Medline, Embase, and Web of Science
b. A minimum of — article databases is recommended
three
c. Search strategies should be
reproducible: they should be described in the methods section and the full strategies should be included as an appendix
d. — should be included in the search
Unpublished evidence (trial data, conference abstracts, or unpublished results)
e. — at the end of included studies should also be checked
Cited references
- Was the selection of studies reproducible?
(4)
a. Description of study screening and selection process
b. Explicit inclusion and exclusion criteria
c. Blinded, dual screening
d. Neutral conflict resolution
- Was the risk of bias of the primary studies assessed?
a. Each included study should be appraised for —
b. (2) tools are available for different kinds of studies
validity
Quality and risk of bias
Magnitude/Precision
1. What is the magnitude of treatment effect?
2. How precise are the results?
a. The forest plot provides
confidence intervals for all studies, as well as for the pooled summary estimate
To address the issue of consistency, we assess —
heterogeneity
To address the issue of consistency, we assess heterogeneity
- Variability in characteristics between the studies in a systematic review
- Variability may come from:
(3)
o Interventions of interest
o How data was captured
o Outcomes were measured
You can test for heterogeneity
- The eyeball test
- The eyeball test
(2)
o You should be able to draw a line intersecting all of the CI
* A low value indicates little difference between the studies
Was there evidence of publication bias?
* Several tests have been developed to detect publication bias
o Requires a
large number of studies (>30)