Biostat Flashcards
number of groups: 1
type of data: parametric tests (data has normal distribution)
One-sample t-test
number of groups: 1
type of data: non-parametric tests (data has skewed distribution)
sign test
number of groups: 1
type of data: discrete/categorical
chi-square test
number of groups: 1 (with before & after measures)
type of data: parametric tests (data has normal distribution)
dependent/paired t-test
number of groups: 1 (with before & after measures)
type of data: non-parametric tests (data has skewed distribution)
Wilcoxon Signed-Rank test
number of groups: 1 (with before & after measures)
type of data: discrete/categorical
Wilcoxon Signed-Rank test
number of groups: 2 (treatment & control)
type of data: parametric tests (data has normal distribution)
Independent/Unpaired student t-test
number of groups: 2 (treatment & control)
type of data: non-parametric tests (data has skewed distribution)
Mann-Whitney (Wilcoxon Rank-Sum) test
number of groups: 2 (treatment & control)
type of data: discrete/categorical
Chi-square test or Fisher’s exact test
Mann-Whitney (Wilcoxon Rank-Sum) test may be preferred for ordinal data
number of groups: >3
type of data: parametric tests (data has normal distribution)
ANOVA (or F-test)
number of groups: >3
type of data: non-parametric tests (data has skewed distribution)
Kruskal-Wallis test
number of groups: >3
type of data: discrete/categorical
Kruskal-Wallis test
correlation
used to determine if one variable changes, or is related to another variable. correlation does not prove causal relationship
Regression
describe a relationship between a dependent variable and one or more independent variables
1) linear regression for continuous data
2) logistic regression for categorical data
3) cox regression for categorical data in a survival analysis
sensitivity
A/(A+C) x 100
look in notes
Specificity
D/(A+D) x 100
look in notes
Intention-to-treat
all patients originally allocated for each treatment group
per protocol
only pts who completed the study according to the protocol
clinical trials can be analyzed two different ways
Intention-to-treat
per protocol
two types of trial
equivalence- new tx is roughly the same as old tx
– these trails test for effect int wo directions for higher or lower effectiveness (two-way margin)
non-inferiority trails- new tx is no worse than the current standard based on the predefined non-inferiority (delta) margin. delta margin is minimal difference in effect between the two groups is considered clinically acceptable based on previous research
forest-plots
generally in meta-analysis
provide CI for difference data or ratio data
– boxes: effect estimate from single study shown
– diamond: at bottom show pooled results form multiple studies
– horizontal lines: confidence intervals, longer lines less reliable
– vertical solid line: is no effect line. significant benefit shown if it falls left of line, right of line is significant harm. verticle line is set for zero for difference data or one for ratio data
cohort study
compares outcomes of a group of pt exposed and not exposed to a tx
follows boths groups prospectively (in the future)
limitations: time-consuming, expensive, can be influenced by confoundets (other factors that affect outcomes…smoking, lipid levels, etc.)
cross-sectional survey
estimates a relationship between variable and outcomes prevalence in one particles time in a defined population
case report/ case series
singe patient case report or few patients case series… describes adverse reaction
