flowchart LR A[Verify<br/>diagnosis] --> B[Confirm the<br/>epidemic] B --> C[Case definition<br/>+ line list] C --> D[Describe:<br/>time · place · person] D --> E[Hypothesise] E --> F[Test<br/>attack rates · RR/OR] F --> G[Refine +<br/>lab/environment] G --> H[Control] H --> I[Communicate] I --> J[Maintain<br/>surveillance]
Investigation of an Outbreak
Reading the curve before you name the pathogen
2026-06-02
Which curve is which?
Four outbreaks. Four epidemic curves. No labels yet.
Think–pair–share (2 min): Which one spread person-to-person? Which was one bad meal? Which is contaminated piped water? Decide with your neighbour.
The four shapes, named
Important
The shape narrows the source before you have named the pathogen.
It all hinges on the baseline
Important
You cannot call something an outbreak until you know the expected count. No baseline → no outbreak.
Soho, August 1854
Steps 1–2 · Verify diagnosis · Confirm the epidemic
A violent illness kills within hours around Broad Street. Snow confirms it is cholera and that deaths far exceed baseline — a true epidemic. The neighbourhood and its pumps; no deaths plotted yet.
Ask: The prevailing theory blames miasma (bad air). What one fact must you confirm first?
Case finding
Step 3 · Case finding & line listing
Door to door, Snow records each death by home address — an early line list. Every red mark is a death tied to a place.
Ask: Why fix a precise case definition before you start counting?
PLACE — the spot map
Step 4 · Describe by PLACE
Mapped by location, deaths are not scattered with the wind — they cluster tightly around one pump. The farther the home, the safer.
Ask: If cholera spread by bad air, what different spatial pattern would you expect?
PERSON — the exceptions test the idea
Step 5 · Describe by PERSON · test the hypothesis
The brewery (workers drank beer + own well) and the workhouse (own well) are spared. A widow in distant Hampstead died — she had Broad Street water carted in. The voids fall exactly on brewery and workhouse.
Ask: These act like a control group. What finding would have weakened Snow’s idea?
The hypothesis
Step 6 · Formulate the hypothesis
Snow states it plainly: cholera is carried by water fouled with sewage — decades before the bacterium is accepted. He reaches it by pattern and reason alone. The Broad Street pump is the suspect.
Ask: We have a strong association. What is the danger of acting on association alone — and why act anyway?
The analytic test — the “Grand Experiment”
Step 7 · Test the hypothesis analytically
Same South-London streets, two water firms: Southwark & Vauxhall (sewage-fouled Thames) vs Lambeth (intake moved upstream). The contaminated supply carried ≈ 8× the mortality.
Ask: Exposed vs unexposed households — what modern study design is this?
Control — remove the handle
Step 8 · Control measure
The parish takes the handle off the Broad Street pump. The iconic lesson: you can interrupt transmission before you fully understand the organism.
Ask: The epidemic was already declining (many had fled). How sure can we be the handle caused the fall?
Example A · Hostel — a point source
- 64 students; a sharp burst of vomiting & diarrhoea
- All onsets within ~24–30 h of one dinner
- Single tight peak → point source, one common meal
- Task: which dish was the vehicle?
Tip
Point source → compute a food-specific attack-rate table.
Example B · Nipah / measles — propagated
- Successive peaks, each ~one incubation period apart, building taller
- Signature of person-to-person transmission
- Kerala 2018: index case → family & health-care contacts; rapid RRT; zoonotic (fruit bats) → One Health
- Drill: contact tracing, isolation, incubation-based quarantine
Important
Spacing between peaks ≈ the incubation period — it lets you predict the next wave.
