Biostats Flashcards
Mode
the value that occurs most frequently
Another term for normal distribution
Gaussian distribution
“bell-shaped curve”
68% values within 1 SD
95% within 2 SD
Reject or accept null hypothesis?
Rejecting the null hypothesis means there IS a significant difference between the study groups
The researcher tries to disprove the null hypothesis
WHat is the alpha value?
The max permissible error margin
Commonly 5% (0.05)
Results are statistically significant when….
when the p-value is equal to or less than the alpha value
p < 0.05 = 95% probability the conclusion is correct
smaller p value = higher probability
Data is significant when the CI does not contain zero (t/f)
TRUE
Type I error =
False-positives
Null hypothesis is rejected, even though it was true
alpha value = chance of false positves
Type II error =
False-negatives
Null hypothesis is accepted, when it should be rejected
Beta value = chance of false negatives
Study power = the probability that the test will reject the null hypothesis correctly (1 - Beta)
Relative risk (RR)
risk in exposed group / risk in control group
RR = 1 = no difference
RR > 1 greater risk
RR < 1 less risk
Discrete data
data that fits into limited distinct options
Nominal = categories (sex, ethnicity, etc)
Ordinal = categories that have an order (pain scale, HF classification)
Continuous data
data is some type of measurement which has unlimited options
Ratio data: height, weight, BP (0 = none)
Interval data: temperature (0 equals something)
T-test assess significance for what type of data?
continuous data
2 independent groups = unpaired t-test
3 or more samples = ANOVA
What test is used to assess significant of nominal/ordinal data groups?
Chi square test
test used to assess 2 groups with continuous data that is not normally distributed
Mann-WHitney (Wilcoxon Rank-Sum) test
test used to assess 3+ groups with continuous data that is not normally distributed
Kruskal-wallis test (also used for categorical data with 3+ groups)
ANOVA for 3+ groups with continuous normally distributed data
Independent variables can/cannot be changed by the reasearcher
CAN
drug dose, placebos, etc
dependent variables can change from the effects of the independent variables
Types of correlation
Oridinal data: Spearman’s rank-order (rho)
continuous data: Pearson’s correlation coefficient (r; -1 to +1)
Intention to treat vs. per protocol
intention to treat:
includes all pts in the analysis, even the pts that did not complete the study for various reason
more conservative analysis
Per protocol:
Only includes pts that completed the study
Sensitivity of lab result
True positive
likelihood of test being + in a pt that has disease
Specificity of lab result
True negative
likelihood of a test being - in pts that do not have the disease
Rank study types from least reliable to most reliable
expert opinion -> case report/series -> Case-controlled -> cohort -> RCTs -> systematic review/meta-analysis
Case-control study
compares pts with a disease (cases) vs without a disease (controls) and looks back in time (retrospectively)
Cohort study
compares outcomes of pts exposed vs not exposed to a treatment
can be prospective of retrospective