Epidemiology Flashcards
Bradford Hill Criteria for Casuality
Strength of Association Consistency Temporality (exposure must precede occurrence) Biological Gradient Plausibility Coherence Analogy Specificity Experimental Evidence
Prevalence
Number of existing cases in a population exhibiting a disease or condition at a specified time or period
Point: proportion at a specific point in time
Period: proportion with pre-existing cases and new cases during period
Proportion
Dividing one quantity by another
Numerator included in the denominator
Incidence
Number of new cases of disease or condition which occur in a given time period in a population
Need initial and final examination
Cumulative: proportion of subjects under the study that get disease under observation
Ratio
Dividing one quantity by another
Numerator not in the denominator
Prevalence, ratio or proportion?
Proportion
Incidence, ratio or proportion?
Ratio
Except: cumulative incidence is proportion
Relative risk
Measure of association in cohort studies, incidence can be measured in them
How common disease is among those exposed
How many times more common or less common is measure of disease frequency in population
Interpret Relative Risk
RR = 1. (Null)
RR > 1. (Increases risk)
RR <1. (Decreases risk)
Odds Ratio
Measure of association in case-control and cross-sectional studies, indirect estimate of the relative risk
How common is exposure in those diseased
Odds ratio interpretation
OR = 1 (null)
OR > 1 (exposure common)
OR < 1 (exposure uncommon)
Attributable risk
Absolute risk difference or the percentage of cases due to a given exposure
How much does it actually do
Attributable proportion
Proportion of people who developed disease due to exposure
Bias
Systematic error in conduct of study leading to an incorrect association estimate, not valid
Selection Bias
Study groups are not comparable
Information bias
Difference in quantity and quality of information collected from study groups
Confounding
The observed relationship between the dependent and independent variables is due to a third variable (the confounder)
Counterfactual model
Comparing the experience of a population exposed to a factor with experience of same population at the same time but without the exposure
Ways to control confounding
Randomization/restriction
Matching
Analysis
Descriptive studies
General characteristics of distribution —> used in planning
Human subject study
Types of descriptive studies
Case reports/series
Co-relational or ecologic
Cross-sectional
Case reports/series
No controls so cannot ascertain causation
Look at what happened
Co-relational or ecologic
Individual level exposure not available (study populations)
Cross-sectional
Done at one point in time so temporality is not known, causation cannot be ascertained
Snapshot
Cohort studies (risk rate)
Follow exposed and unexposed people over time
Observational
Subjects identified/classified on independent aka predictor variable
Prospective: follow subjects without disease (see if get)
Retrospective: both exposure/disease have occurred in past, follow lives
Case control studies (odds ratio)
Identify people with disease and suitable non disease individuals as control, must be similar in most respects except that they do not have the disease
Study based on dependent (outcome variable)
Incidence cannot be measured
NEED to have the controls have the same exposure opportunity as cases
Retrospective
Experimental studies/clinical trials
Researcher assigns exposure, highest level of epidemiological evidence
Blinding and randomization
Reliability
Getting the same result
Validity
Freedom form the systemic error such as bias
Internal validity
The study was done without any methodological problems
External validity
Study accurately reflecting events that would occur in real situation
Meta-Analysis and Systematic Reviews
Studies of studies
Compares many studies together to provide highest level of evidence
Resolves uncertainty when reports disagree
Goals: limit bias, improve reliability, improve accuracy
