(Absolute) Risk
Proportion of people who experienced a given outcome
Relative Risk
Treatment Risk/Control Risk
Relative Risk Reduction
(Control Risk - Treatment Risk)/(Control Risk)
Intermediate estimate of risk reduction
Absolute Risk Reduction
Control Risk - Treatment Risk
Most conservative estimate of risk reduction (SMALL)
Number Needed to Treat (to get positive outcome)
1/Absolute Risk Reduction
Odds (in favor of patient living)
(# pts living within treatment)/(# pts NOT living within treatment)
Note: this is within a single treatment, looking at whether patients live or die
Odds Ratio of dying
(Treatment Odds of dying)/(Control Odds of dying)
Least conservative estimate of risk reduction (BIGGEST: "this treatment really lowers your risk!"...but not really)
Type I error
Alpha
Risk of randomly finding effect that is not actually there
False positive
"Statistical significance"
Type II error
Beta
Risk of randomly NOT finding an effect that is actually there
False negative
"Power" (1 - Type II error)
Systematic error
Due to bias, confounding, study design issues
Random error
Due to chance alone (Type I error and Type II error)