Statistics, Research, Evidence based Medicine Flashcards
Statistical Tests
✓ ANOVA and t test are examples of parametric tests which should only be used when the dataset follows a normal distribution. The rest of the options are non-parametric tests.
✓ ANOVA stands for Analysis Of Variance and compares the means of two or more samples determining whether they come from the same population.
✓ Wilcoxon signed rank test is used to compare the difference between paired groups.
✓ Chi-squared test looks at the association between two categorical variables.
✓ A Mann-Whitney U test looks to see whether there is a significant difference between two sets of data that are from 2 different sets of subjects.
Cohort Studies
Features:
longitudinal
Prospective
Expensive
Can study multiple exposures and their outcome
Useful to assess the effect of genetic variants, rare exposures (i.e. radiation) or exposures that take a long time to produce disease eg asbestos
Reliant on the right exposures being measured and ability to follow-up a large proportion of the cohort.
Levels of Evidence with hierarchy
The hierarchy of strength of evidence used from the highest quality evidence to the lowest:
Level I Evidence from a systematic review or meta-analysis of all relevant RCTs (randomised controlled trial) or evidence-based clinical practice guidelines based on systematic reviews of RCTs or 3 or more RCTs of good quality that have similar results.
Level II Evidence obtained from at least one well designed RCT (eg large multi-site RCT).
Level III Evidence obtained from well-designed controlled trials without randomisation (ie quasi-experimental).
Level IV Evidence from well-designed case-control or cohort studies.
Level V Expert opinion
Confidence Interval versus p value
The statistical tests underlying a 95% CI and a P value of 0.05 are the same.
If the CI includes a treatment effect that is the same for the treatment and control group (this is a relative risk of 1 or a risk difference of zero), the P value will be greater than 0.05.
CI gives an estimate of how certain you are about the real effect of a treatment given the results of this trial. This range of uncertainty around the estimated real effect of treatment is given by the 95% confidence interval (which in simple terms is where we can be 95% sure that the true effect of treatment lies).
The upper and lower ends of the confidence interval can be used to decide whether an important advantage of treatment has been excluded.
Although the confidence interval is easy to display graphically, this is not the reason why it is used more, nor is it to do with the ease of calculation.
