Likelihood Ratio Nomogram
- pre-test probability (prevalence) is on the left
- likelihood ratio is in the middle
- you can then draw a line from the pretest probability through the likelihood ratio and it will predict the post-test probability buy intersecting the scale on the right
Altman Nomogram
- has standardised mean on left and power of the study on the right
- the required sample size is indicated by a diagonal line going down between the two axis
- significance level is also indicated
- sample size depends on p-value, power, expected difference and the expected variability of the outcome variable in study population (Std dev)
Why would you use an Altman’s nomogram?
- you may want to calculate the required sample size by keeping constant p-value, power and estimated an expected effect size
- or, in a pubished paper, one may wish to test the power of the sample using the given sample size, found effect size and set p-value
Funnel plot
- used to detect publication bias
- usually used in meta-analysis
- precision (sample size r standard error) is plotted along the y axis
- effect size (cohen’s d, OR or RR) is on the x axis
- large samples are high up
- verticle line is drawn through the point representing the ‘true effect size’
- large studies are closer to the line and small studies more spread out
- if there is an inverted funnel then we can assume there is no publication bias but if there is a gap then there is publication bias
ROC curve
-square of axis
-top= TN (specificity)
-right= FN (1- sensitivity)
-bottom= FP (1-specificity)
-left=TP (sensitivity)
chance line- diagonal line through the middle (curve of a test that performs no better than diagnosing by chance)
-diagonal corners- cut offs for lowest sensitivity and specificity
-elbow off curve is the ideal cut off between false positives and false negatives
RCT CONSORT FLOW
-shows what happens to everyone in the study
Forest plot
-OR with 95% CI is at 1.0 and this is where the line is drawn
Kaplan Meier
-probability on Y axis, time to event is on x axis
Stem and leaf plots
- stem as units of 10
- leaf- each number corresponds to a single observation when the tenth digit value on the stem is added to the leaf value
Path analysis
- curved arrows -slightly associated
- straight arrow- causative
- error measurement is included at each point
- numbers are seen on main arrows which are path coefficients (beta regressor)
- the end point has a number above it which is the R-square: coefficient of determination
Scatter plot
- two continuous variables plotted against eachother
- log values may be used to yield parametric transformation
- line is drawn to show correlation
- distance from that line is the residual (deviation)
Galbraith plot
- standardised effect is on y axis (Z score)
- precision is on x axis
- shows +2 SD and -2SD lines
- middle line depicts pooled effect size
- shows outliers
- precision is 1/SE
L/Abbe Plot
Control group % is on X axis
Intervention % is on Y axis
-equality line is drawn that indicates event rates are equal within the two arms within trials (RR=1)
-trials that are below the equality line show that control groups have better event rates than intervention groups
-each circle represents an individual trial, larger circles represent sample size
-dotted line is the pooled RR line from metanalysis
Box and Whisker Plot
box is 25th-75th quartile
- horizontal line depicts median (50th percentile)
- dashed line is mean
- lower adjacent- the minimum value excluding outliers
- upper adjacent- maximum value excluding outliers
- outliers are values that are 1.5 time greater or less than the upper quartile
