Outbreak Investigation

---
name: outbreak-investigation
description: Field epidemiology outbreak response — case definitions, epidemic curves, analytic studies, and control measures.
category: scientific
---

## Overview

outbreak-investigation covers the classic field-epidemiology sequence for responding to disease
outbreaks: verifying the outbreak, establishing case definitions, describing by time/place/person,
generating hypotheses, testing them analytically, and implementing control measures. It follows the
CDC's established steps while emphasizing the judgment calls — particularly when data are
incomplete and decisions can't wait for perfect evidence.

## When to use

- Responding to a cluster of illness: is this an outbreak?
- Writing case definitions (confirmed/probable/suspected) with the right sensitivity-specificity
  trade-off.
- Building epidemic curves and interpreting their shapes.
- Descriptive epidemiology: time, place, person analyses.
- Analytic studies: retrospective cohort and case-control for identifying sources.
- Implementing and evaluating control measures.
- Writing outbreak reports and after-action reviews.

## Core concepts

- **Verify first.** Confirm the diagnosis (lab confirmation on initial cases), confirm the
  outbreak (cases exceed expected — check baseline rates, changes in surveillance, diagnostic
  practices, and population denominators before declaring one). Many "outbreaks" are surveillance
  artifacts.
- **Case definitions.** Clinical + laboratory + epidemiologic criteria, with tiers:
  suspected (sensitive, for case finding), probable (clinical + epi link), confirmed (lab).
  Definitions evolve as knowledge grows — version and date every revision, and reclassify cases
  when definitions change.
- **Line lists.** One row per case: identifiers, demographics, onset date, symptoms, exposures,
  lab results, outcome. The line list is the outbreak database — build it early, update it
  continuously, protect it.
- **Epidemic curves.** Histogram of onset dates. Point-source (single peak, narrow), continuous
  common-source (plateau), propagated (successive taller peaks one incubation period apart).
  The curve's shape is often the fastest hypothesis generator — learn to read it.
- **Time/place/person.** Person: attack rates by age, sex, occupation, risk factors. Place: spot
  maps, attack rates by location/ward/table. Time: epi curve. Attack rates (ill/exposed) are the
  core descriptive measure — always with denominators.
- **Hypothesis generation.** Case interviews (open-ended initially, then focused questionnaires),
  site visits, and the descriptive patterns. Talk to the first few cases in depth — they often
  reveal the exposure.
- **Analytic testing.** Retrospective cohort (defined exposed population: compute attack rates
  and relative risks per exposure — e.g. food items at a banquet) or case-control (no defined
  population: odds ratios). For foodborne outbreaks, the cohort design with food-specific attack
  rates is standard; stratify to handle confounding between foods eaten together.
- **Control measures.** Source-directed (remove the vehicle, close the venue), transmission-
  directed (isolation, hygiene, vector control), and host-directed (prophylaxis, vaccination).
  Implement on epidemiologic evidence — waiting for definitive proof while people get sick is a
  failure, but document the evidence basis for each measure.
- **Communication.** Regular updates to stakeholders, public messaging that's honest about
  uncertainty, and a final report with lessons learned. Outbreaks are public events; the
  epidemiology is only half the job.

## Practical workflow

1. **Prepare.** Assemble team, review background rates, gather case reports.
2. **Verify.** Confirm diagnoses; compare against expected numbers.
3. **Case definition + case finding.** Draft definition; find cases actively (not just
   passively reported ones — mild cases are systematically missed).
4. **Line list.** Build and maintain; assign onset dates carefully (onset, not report date).
5. **Describe.** Epi curve, spot map, attack rates by person/place/time.
6. **Hypothesize.** From descriptive patterns + case interviews + site knowledge.
7. **Test.** Cohort or case-control study; compute RR/OR with CIs; consider dose-response.
8. **Control.** Implement measures matched to the evidence; monitor the epi curve for effect.
9. **Report.** Methods, findings, control measures, lessons; archive data for future outbreaks.

## Common pitfalls

- Declaring an outbreak from a surveillance artifact (new test, new reporting rule).
- Onset dates replaced by report dates (distorts the epi curve).
- Case definitions that drift without versioning.
- Passive case finding missing mild cases (biases severity and attack rates).
- Confounded food analyses (eaten-together items) without stratification.
- Waiting for perfect evidence before control measures.
- No final report — the next outbreak team relearns everything.