Evidence Based Medicine and Reviews of Evidence Flashcards
What is biological plausibility?
A causal link is more likely if a biologically plausible mechanism is demonstrated
A biologically plausible mechanism streghtens the case for a causal link
What is evidence based medicine?
The concientious (searched), explicit (appraised) and judicious (appropriate) use of current best evidence in making devisions about the care of individual patients
What are the 3 main values in evidence based medicine?
- Relevant Scientific Evidence
- Patient’s Values and Preferences
- Clinical Judgement
What does ‘scientifically proven’ actually mean?
The treatment is better than the other treatment available
NOT that you are more likely to benefit from the treatment than not
What is absolute risk?
Chance of benefitting from treatment vs not
What is relative risk?
How much better the treatment is vs alternative treatment
What are the 2 types of literature review of studies, which is best?
Narrative reviews- impliciit assumptions, opaque methodology, not reproducible → biased, subjective
Systemic reviews- explicit assumptions, transparant methodology, reproducible → unbiased objective
What is a systematic review?
An overview of primary stduies that used explicit and reproducible methods
What is a meta-analysis?
A quantitative synthesis of the results of two or more primary studies that addressed the same hypothesis the same way
What are the purposes of meta-analysis studies?
- Facilitate the synthesis of a larger number of stduy results
- Systemically collate study results
- Reduce problems of interpretation due to variations in sampling
- Quantify effect sizes and their uncertaintly as a pooled estimate
How do you interperet a Forest plot?
- Size of box correlates to size of study
- Individual odds ratios with their 95% confidence intervals (lines) given for each study
- The diamond is the pooled estimate
- Centre/ dotted line is the pooled odds ratio
- Points of the diamond are the pooled 95% confidence intervals
- Solide line= null hypothesis
What are the problems that face meta-analysis?
-
Heterogeneity between studies
- Model for variation (fixed effect vs random effect)
- Variation in analysis
- Variable quality of the studies
- Publication bias in study selection
What are the 2 models that exist of modelling for variation between studies when doing a meta anlysis?
Fixed effect - assumes that the studies are estimating the exact same true effect size
Random effects - assumes studies are estimating similar, but not the same true effect size
What is the fixed effect model for meta-analysis?
Fixed effect believes there is only one true effect
Deviations from the true effect are due to random error
Explain the random effects model for meta-analysis
There are many true effects but only one mean of the average effects
Individual studies vary from their own true effect
Weight of squares on forest plots are more equally distributed as it believes one large trial should not sqew the results
How do funnel plots differ if there is evidence of bias vs no evidence of bias?
No bias = more equal distribution of results
Bias = lean towards one, more favourable effect
