Statistics Flashcards
Likelihood ratio (negative)
False negative (1 - sensitivity) Divided by True negative (specificity)
Likelihood ratio (positive)
True positive (sensitivity) Divided by False positive (1 - specificity)
Positive predictive value (PPV)
Likelihood that those that tested positive actually have the disease
Also called post-test probability
Disease positive tested positive
Divided by
Total tested positive (true + false positive)
Pre-test probability
Prevalence of the disease
All disease positive
Divided by
Total
Sensitivity
‘True positive rate’
True positives
Divided by
Total with the disease (true positive + false negative)
Specificity
‘True negative rate’
True negatives
Divided by
Total without the disease (true negative + false positive)
Probability (risk)
Probability of event
Divided by
Total at risk
Publication bias
Unpublished negative trials exaggerate true effect size
Greatest threat to validity
Odds ratio (OR)
Probability of event
Divided by
Probability of no event
Case control/cross sectional studies
Hazard ratio (HR)
RR averaged over time (duration of study)
Survival analysis
Number needed to treat
1
Divided by
Absolute risk reduction
Absolute risk reduction
Risk of outcome not on therapy (placebo/alternative therapy)
Minus
Risk of outcome on therapy
Relative risk reduction
1 - RR
Expressed as a percentage
Relative risk
Probability of outcome on therapy
Divided by
Probability of outcome not on therapy (placebo/alternative therapy)
Expressed as decimal
RCT or cohort study