An outbreak is identified as an excess over what would ordinarily be expected, which means detection depends on knowing what ordinary looks like. That requirement shapes how surveillance systems are built.

Baselines are built from historical data

Health agencies model the expected number of cases for each week of the year, drawing on several years of past records.

Seasonality is substantial for respiratory and gastrointestinal illness, so the same case count can be unremarkable in one month and alarming in another.

An alert is triggered when observed counts exceed the modelled expectation by a margin chosen to balance false alarms against missed events. Setting that margin is a policy decision as much as a statistical one.

Detection uses several independent streams

Clinician notification of specified diseases remains the backbone, since it comes with clinical detail and a confirmed diagnosis.

Faster but cruder streams include emergency department presentations grouped by symptom, laboratory test volumes and school or workplace absence.

These indicators move earlier than confirmed diagnoses because they do not wait for testing, at the cost of being far less specific. Agencies run them in parallel so that an early crude signal can be checked against a slower reliable one.

Wastewater samples the whole population at once

Testing sewage for pathogen genetic material captures shedding from everyone connected to a catchment, including people with no symptoms.

It is unaffected by whether people seek care or whether testing is available, which removes two of the largest biases in case-based data.

It cannot identify individuals or give a precise case count, so it works as an early warning of direction rather than a measure of size.

Sequencing establishes whether cases are linked

Cases occurring close together may share a source or may be coincidental, and clinical information alone often cannot distinguish them.

Comparing pathogen genomes shows how closely related the isolates are, and near-identical sequences indicate recent common transmission.

This is how a scattered set of food-borne illnesses across different regions is recognised as a single outbreak from one distributed product.

Detection speed depends on people reporting

Every stream except wastewater ultimately begins with someone seeking care and a clinician recording what they found.

Delays in that first step propagate through the whole system, which is why statutory reporting requirements exist for particular infections and carry defined time limits.

Public messaging about when to seek care for specific symptoms exists partly for this reason, since detection cannot outrun the presentations it depends on.