EBCP Flashcards
Incidence x duration
Prevalence
Incidence
of new cases in population at risk/unit time
Prevalence = incidence
Acute diseases
Prevalence > incidence
Chronic diseases
Quantitative measurement of disease
Morbidity
Quantitative measure of death
Mortality
Probability of developing a disease/complication within a given time interval
Risk
A variable or exposure that decreases the probability of developing a disease
Protective factor
Ratio of incidence rates in two groups being compared
Relative Risk (RR)
Relative Risk =
incidence of exposed group/incidence of non-exposed group
RR = 1
No difference in risk
RR > 1
factor increases risk
RR<1
Factor decreases risk
Looks for associations within a population at one point in time
Cross-sectional study
Estimates prevalence, but not incidence
Cross-sectional study
Uses odds ratio (OR)
Cross-sectional study, Case control
Starts with groups with a specified outcome and looks at associated exposures
Case control
Always retrospective
Case control
Starts with groups of specified exposure and looks at associated outcome
Cohort
May be retrospective or prospective
Cohort
Uses relative risk ratio
Cohort
Error that randomly varies from one measurement to another
Random errors: not accurate, not reliable
Error that does not randomly vary, but moves measurements systematically away from their true value (bias)
Systematic errors: reliable but not accurate
Precision. Data is reproducible. Related to random error.
Reliability
Accuracy. Related to systemic error.
Validity
Categorical variables
Place individuals into groups. Described using % or proportions
Ordinal
Categorical variable: low, moderate, high
Nominal
Categorical variable: have names, no order
Dichotomous
Categorical variable: exactly 2 categories: yes or no
Continuous variables
Described using measures of central tendency (mean, median, mode) and measures of variability (standard deviation/error)
Have numerical values for which arithmetic operations
Normal distribution
Mean = median = mode: mean is good measure of clinical tendencies
Right skew
mean > median > mode: median is good measure of clinical tendencies
Left skew
mean < median < mode: median is good measure of clinical tendencies
Bimodal distribution
Distribution with two peaks: modes are good measures of central tendency
Standard error
SD/sqrt(n). Accounts for reliability of measurement.
Assess statistical difference across groups
Chi-square
Assess statistical difference across small groups
Fisher’s exact test
Comparing 2 groups, if groups have normal distribution (uses means)
T-test
Comparing 2 groups that do NOT have normal distribution (compares medians)
Wilcoxon-Mann-Whitney
Comparing 3 or more groups, does not require normal distributions
ANOVA
Type 1 Error
Identify difference between the groups when one does not exist: Rejecting a null hypothesis in favor or alternative when NULL is actually TRUE
Type 2 Error
Data fails to identify a difference between groups when one exists: Failing to reject the null hypothesis when the NULL is FALSE
P-value
Probability that difference between groups is due to chance alone. P-value < 0.05 = reject null hypothesis, significant
Alpha
Probability of committing a type 1 error.
Beta
Probability of committing a type 2 error. Beta = 0.10 or 0.20
Power
= 1 - beta. Probability of finding a difference between groups when a difference does exist = probability of rejecting the null hypothesis it is false
If CI excludes 0, we can reject null hypothesis when null hypothesis =
Mean 1 = Mean 2. Alternative hypothesis: Mean 1 =/= Mean 2
If CI excludes 1.00, we can reject the null hypothesis when null hypothesis =
RR=1, Proportion 1 = Proportion 2, Alternative hypothesis: RR =/= 1