EBM Flashcards
Systematic revue
overview of particular clinical research question
Meta-analysis (3 points)
Uses statistical method to summarise studies, often included in a systematic review, produces single quantitative result
Single quantitative result
pooled treatment result / overall effect estimate
Forest plot
graphical result of meta-analysis
Point effect estimate (forest plot)
vertical midpoint of each square, mean treatment effect for that study
forest plot vertical line
line of no effect
horizontal line (each) on forest plot
confidence interval
if horizontal line (confidence interval) crosses vertical line (line of no effect) then..
No significant difference between two treatments
width of diamond at bottom of forest plot
certainty of overall affect estimate - 95% confidence interval
confidence interval
range of values within which the true mean of whole population is expected to lie
95% confidence interval
95% confident that true mean lies somewhere between two limits of interval
larger sample sizes lead to CIs which are
narrower
less variability leads to CIs which are
narrower
a larger confidence level results in CIs which are
broader
outcome variable
dependent variable
outcome variable of meta analysis
often odds ratio (NOT ALWAYS)
in a meta analysis where the mean difference is zero
no difference between groups
negative mean difference in meta analysis
favours treatment
positive mean difference in meta analysis
favours control (treatment harmful compared to no treatment)
relative frequency
percentage of total number in sample
categorical variable
non-numerical
numerical variable
numerical value, measurable quantity. Continuous or discrete
continuous numerical variable
measurement on continuous scale e.g. height
discrete numerical variable
numbers of events, limited number of possible values e.g. pregnancies, heart attacks
3 graphical representations of categorical variables
frequency distributions, pie charts, bar charts
4 graphical representations of numerical variables
frequency distributions, histograms, frequency polygons, quantiles
measures of central tendency (3)
mean, median, mode
measures of spread of data (3)
range, interquartile range, standard deviation
purpose of shapiro-wilk test
determining normality (or otherwise) of distribution
purpose of t-test
determine whether 2 means differ reliably
correlation and linear regression purpose
testing for relationship between 2 variables
null hypothesis of hypothesis testing
no effect
p < 0.05 hypothesis testing
reject null hypothesis - statistically significant result
shapiro wilk null hypothesis
sample from normally distributed population
shapiro wilk p <0.05
reject null hypothesis - statistically significant result, sample unlikely to be from normal distribution
3 types of t test
one sample t test, unpaired t test, paired t test
t test null hypothesis
no difference between 2 means
t test p <0.05
reject null hypothesis- statistically significant result, reliable difference between 2 means
pearson’s r between which two values
-1 and +1
r = 0 (approx)
no relationship between variables
pearson’s r p < 0.05
reject null hypothesis - statistically significant result, reliable relationship between two variables
pearson’s r null hypothesis
no relationship between two variables
what is p value
probability of observing the results of a trial if the null hypothesis is true
shapiro wilk test statistic
W
one sample t test used to
compare sample mean with population mean
unpaired t test used to
test whether there is difference between 2 independent groups - comparing same measure in 2 different groups
paired t test used to
compare dependent samples of groups e.g. one subject at two times
t test test statistic
t
pearson’s r is a
correlation coefficient
cochrane review =
systematic reviews of primary research in human health care and health policy. highest standard of evidence-based heath care
