Systematic reviews Flashcards
systematic reviews are also known as
evidence sysntheis
review
what is there
synthesis
what does it mean
Lord Rayleigh
a victorian scientist who discovered Argon, believed it is more important to make sense of what we already know, instead of discovering new things e.g. secondary research
evidence synthesis brings
all the evidence to one place
how many peer reviewed articles published in 2014
- 36 mill
- one every 20 second
systematic review is
more methodical- you can see how they were done- can be replicated
systematic reviews give accounts of
the search, appraisal and synthesis methods
systematic also means
exhaustive- has ALL the info out there
some people say that if you don’t include all the info-
then you are not carrying out a systematic review
which statistical representation are used in systematic reviews
forest plots
forets plots show
- down the far left= the trial
- shows odd rations- depending on how far one rah plot is not he control/ treatment side of the plot
- plots have confidence intervals
- can make cumulative forest pots where fata from diff studies synthesis- should decrease confidence limits
if confidence limits are on1
not signif
if confidence limits do not spread past 1
then the results are significant
-
meta-analyses will have
narrower confidence intervals
forest plots can be used to
look at trial o specific topic one ra lon period of time e.g. thrombosis
e.g. is it ethical to give control, ten years after it has been proven to be the favourable treatment
forest plots example
1987- the clinical value of thrombolysis remains uncertain
p value
the probability that if this was repeated you would get the same result- significance
SRs in terms o hierarchy of evidence
at the top
- quantitative
- used to decide whether treatment/ diagnostic technique should be employed
negative of SR
dint hel popel decide how they feel about the new treatment/ diagnostic technique- not qualitative
Steps in a systematic review
1) ask the Q
2) get the stuff - search- how
3) check the study- appraisal, how
4) synthesise the stuff- what does it mean, meta-analysis
5) apply in the decision
summary of SR
find- appraise- act
when asking a q
- focus on what knowledge is needed
- identify the most important unknowns
- not all qs are equal
important that we use the right evidence
to help patient and service development
(1)Structuring a Q using PICO
P- population
I- intervention
C- control
O- outcome
population
the kind of people you are interested in
Intervention
what may happen to the group
control
what will happen without the intervention
Outcome
consequences of interest to the patients
(2) find the evidence
- electornic bibliographic databases
- not all databases are equal
e. g. Meddling, EmBase, PsycIfo
peer reviewed articles are
vastly over respected
- may things which are peer reviewed are rubbish
where can you find grey ‘non-peer reviewed’ literature
SIGLE
(3) Appraise the evidence
-did they find enough
-what does it mean?
-should you believe it?
Will it help with your uncertainty? What was the Question? what si the context?
(4) Synthesis
- depending on kind of review
- different kind of synthesis for diagnosis and critical reviews
- cause use forest plots
On forest plots systematic reviews will
have the narrowest Ci
Heterogeneity
level is 0.1
heterogeneity is
reers to variation in study outcomes between studies
Heterogeneity is represented by
I^2
I^2
how much of the variation is due to chance vs heterogeneity - between variance/ total vairance
low I^2
may not be a porblem
high I^2
substantial heterogeneity
heterogeneity needs to be
explored- could be due to different populations/ doses
why may heterogeneity occur
- look for wide CIs
- studies data will disagree significantly
Biases
- positive Studie sage mir eliekly to be published and quicker
- positive outcomes
- langauge
funnel plot- publication bias
- larger studies have higher/ lower odds
- negative studies are much less likely to be published
- review will have more positive the negative outcome trials
funnel plots illustrate why it is important to
use unpublished literature too
- showing studies with equal stand error are subjected to publication bias based on whether they yield positive or negative results
types of SR
- effectiveness
- epidemiological
