How investigators turn a pile of case reports into a picture they can act on
Step 5 (descriptive epidemiology) is where these tools get used. Each one displays the outbreak's data a different way, and a Disease Detective needs to be able to read all of them.
A histogram of case counts by date of onset: its shape hints at how the outbreak is spreading (see below).
A spreadsheet with one row per case and columns for key variables (symptoms, exposures, dates): the investigator's master worksheet.
A map with one mark per case, used to spot geographic clustering: the modern descendant of John Snow's cholera map.
A table breaking case counts down by category (age group, exposure, location) to compare rates between groups.
Lab techniques (Pulsed-Field Gel Electrophoresis and Whole Genome Sequencing) that create a genetic "fingerprint" of a pathogen sample.
PulseNet is a CDC network that compares the DNA of germs from sick people. It used to compare PFGE fingerprints; today it mainly uses whole-genome sequencing (WGS), which reads nearly all of a germ's DNA. Similar DNA can help investigators find cases that may be part of the same outbreak. SNP (Single Nucleotide Polymorphism) mapping compares tiny, single-letter DNA differences to show how closely related two samples are.
An epi curve's shape is a clue in itself. Before any lab result comes back, the silhouette alone tells you how the outbreak is probably spreading. Here are the four classic shapes to recognize:
One sharp peak, all within a single incubation period. Everyone exposed at once, from one shared source.
A broad, sustained plateau: the source keeps exposing new people until it's found and fixed or removed.
Several irregular, unevenly spaced peaks: exposure keeps turning on and off, like a well only used sometimes.
Multiple rolling waves, each about one incubation period apart: the disease is spreading person to person, wave feeding wave.
Quick check: click a shape, then click the description that matches it:
If you know the agent's incubation period, the curve tells you when things happened. Tests ask this a lot:
Point-source: when was everyone exposed? Count back one median incubation period from the peak. Norovirus (median about 33 hours) peaking Saturday at 9 PM means the exposure was around noon Friday. You can check it from the edges too: earliest case minus the shortest incubation, and latest case minus the longest, should both land near the same time.
Continuous common-source: when was the source fixed? New cases keep appearing for about one incubation period after the source is shut off, so the drop-off comes one incubation period after the fix. Count back from where the cases stop.
Propagated: when will the next wave peak? Waves are about one incubation period apart, so add one incubation period to the latest peak.
Two more patterns to recognize: a single sharp peak followed by a smaller bump about one incubation period later means a point-source outbreak with secondary spread (the first sick people passed it on at home). And an ill person whose symptoms started before the event, like a volunteer cook, is a clue that they were the source, not a victim.
Tests call a common-source outbreak followed by person-to-person waves a mixed outbreak. (Common source is the umbrella for point, continuous and intermittent outbreaks: everyone was exposed to the same source.) Don't assume the earliest case on a line list is the primary case: one case far ahead of the rest may be an unrelated background case, a mistyped date, or an imported case. And people infected before a control measure can still get sick for up to about one incubation period afterward, so the measure's effect shows up about one incubation period later. A drop that starts much sooner than that was probably already happening (the outbreak was already fading).
The simulator below lets you build the two most commonly tested shapes yourself, case by case: