Epidemiology Essentials
The study of the distribution and determinants of health-related states.
The Epidemiological Triad
Disease isn’t random. It requires the interaction of three key elements. Understanding this triad is the first step in breaking the chain of transmission.
Agent
Bacteria, Virus, Chemical, Trauma
Host
Human, Animal
(Genetics, Immunity)
Environment
Climate, Crowding, Sanitation
Levels of Prevention
Public health intervenes at different stages. From preventing risk factors to managing established disease, knowing the level of prevention guides policy.
Stopping the risk from ever appearing.
Protecting at-risk individuals.
Halting progression via early detection.
Managing established disease.
Incidence vs. Prevalence
Incidence is the rate of new cases (risk), while Prevalence is the total burden (snapshot). Prevalence depends on incidence and duration. If a disease lasts a long time (e.g., Diabetes), prevalence accumulates.
- Incidence = New Cases / Population at Risk
- Prevalence = Total Cases / Total Population
Figure: Hypothetical comparison of yearly Incidence rate vs Total Prevalence load.
Hierarchy of Evidence
Randomized Controlled Trial (RCT)
Gold standard for causality. Experimental.
Cohort Study
Exposure → Outcome. Measures Relative Risk (RR). Best for rare exposures.
Case-Control
Outcome → Exposure. Measures Odds Ratio (OR). Best for rare diseases.
Cross-Sectional
Snapshot in time. Measures Prevalence.
Calculations & The 2×2 Table
Almost all epidemiological calculations stem from this simple matrix.
| Disease (+) | Disease (-) | |
|---|---|---|
| Exposed | a | b |
| Unexposed | c | d |
Relative Risk (RR)
RR = [a/(a+b)] / [c/(c+d)]
If RR > 1: Exposure is harmful.
Odds Ratio (OR)
OR = (a × d) / (b × c)
Interpreting Risk
Visualizing how RR values indicate protection vs harm.
Screening Metrics
Sensitivity (SNOUT): Rules OUT disease.
Specificity (SPIN): Rules IN disease.
Graph shows hypothetical Dose-Response (Biological Gradient) proving causation.
Bradford Hill Criteria
Not all associations are causal. These criteria help prove causality.
-
1
Strength
Large RR or OR implies stronger association.
-
2
Consistency
Results replicated in different studies/settings.
-
3
Temporality
Essential: Cause MUST precede effect.
-
4
Biological Gradient
Dose-response relationship (see chart).
Beware of Bias
Selection Bias
Error in choosing participants (e.g., Healthy Worker Effect).
Recall Bias
Cases remember exposure better than controls. Common in Case-Control.
Confounding
A third variable distorts the relationship (e.g., Smoking in Coffee vs Cancer).