HPG 6104 • Epidemiological Methods II

Measures of Disease Frequency & Association

Week 1 Companion Handout (Quick Guide)

The 2×2 Table (Your Universal Language)

Rule of thumb: before calculating anything, label your groups and confirm the denominator.

	Disease +	Disease −	Total
Exposed	a	b	a + b
Unexposed	c	d	c + d
Total	a + c	b + d	N

Measures of Disease Frequency

1) Prevalence (A snapshot: "How much disease is here?")

Point prevalence: (Existing cases at time t) ÷ (Total population at time t)
Period prevalence: (People who had disease during period) ÷ (Average population during period)

Interpretation: Burden of disease → useful for planning services (beds, clinics, meds).
Common trap: Prevalence changes with duration (chronic diseases inflate prevalence even if incidence is modest).

2) Incidence (A video: "How fast are new cases occurring?")

Cumulative incidence (Risk): (New cases during period) ÷ (Population at risk at start)
Assumptions: closed cohort; follow-up time similar; everyone starts disease-free.
Incidence rate (Density): (New cases) ÷ (Total person-time at risk)
Best when: open/dynamic populations, varying follow-up, losses to follow-up.

3) Linking Prevalence and Incidence

Prevalence ≈ Incidence × Average duration

Why it fails: Rapidly changing incidence, short observation windows, changes in survival, migration, case definitions.

Measures of Association

Cohort Studies (Incidence is known)

Relative Risk (RR): I_e / I_u
Ex: RR = 4 → exposed have 4× the risk vs unexposed.
Risk Difference (RD): I_e - I_u
Absolute burden (extra cases due to exposure).
Attributable Fraction (AF_e): (RR - 1) / RR
Proportion of disease among exposed attributable to exposure.

Case-Control (Incidence unknown)

Odds Ratio (OR): (a×d) / (b×c)
Odds of exposure among cases relative to controls.
When OR ≈ RR: Rare outcome + controls represent source population.
Trap: Saying "OR=4 means 4× risk" without checking rarity.

Fast Cheat Sheet & Worked Example

Choosing the Measure

Cross-sectional: prevalence, PR, POR
Cohort: risk, rate, RR/IRR, RD, AF_e, PAF
Case-control: OR

Mini Worked Example (Cohort)

If I_e = 40/1000 and I_u = 10/1000:

RR = 40 / 10 = 4
RD = 40 - 10 = 30 per 1,000
AF_e = (4 - 1) / 4 = 0.75 → 75% of cases among exposed are attributable.

"Today we calculated measures from 2×2 tables. Next week we ask: are these differences likely due to chance? That’s where chi-square, Fisher’s exact, and Mantel–Haenszel come in."